UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
/*
|
|
|
|
|
* Copyright (c) International Business Machines Corp., 2006
|
|
|
|
|
* Copyright (c) Nokia Corporation, 2007
|
|
|
|
|
*
|
|
|
|
|
* This program is free software; you can redistribute it and/or modify
|
|
|
|
|
* it under the terms of the GNU General Public License as published by
|
|
|
|
|
* the Free Software Foundation; either version 2 of the License, or
|
|
|
|
|
* (at your option) any later version.
|
|
|
|
|
*
|
|
|
|
|
* This program is distributed in the hope that it will be useful,
|
|
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See
|
|
|
|
|
* the GNU General Public License for more details.
|
|
|
|
|
*
|
|
|
|
|
* You should have received a copy of the GNU General Public License
|
|
|
|
|
* along with this program; if not, write to the Free Software
|
|
|
|
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
|
|
|
|
*
|
|
|
|
|
* Author: Artem Bityutskiy (Битюцкий Артём),
|
|
|
|
|
* Frank Haverkamp
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
/*
|
2007-12-16 07:59:31 -07:00
|
|
|
|
* This file includes UBI initialization and building of UBI devices.
|
|
|
|
|
*
|
|
|
|
|
* When UBI is initialized, it attaches all the MTD devices specified as the
|
|
|
|
|
* module load parameters or the kernel boot parameters. If MTD devices were
|
|
|
|
|
* specified, UBI does not attach any MTD device, but it is possible to do
|
|
|
|
|
* later using the "UBI control device".
|
|
|
|
|
*
|
|
|
|
|
* At the moment we only attach UBI devices by scanning, which will become a
|
|
|
|
|
* bottleneck when flashes reach certain large size. Then one may improve UBI
|
|
|
|
|
* and add other methods, although it does not seem to be easy to do.
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
#include <linux/err.h>
|
|
|
|
|
#include <linux/module.h>
|
|
|
|
|
#include <linux/moduleparam.h>
|
|
|
|
|
#include <linux/stringify.h>
|
|
|
|
|
#include <linux/stat.h>
|
2007-12-16 07:59:31 -07:00
|
|
|
|
#include <linux/miscdevice.h>
|
2007-06-11 23:01:05 -06:00
|
|
|
|
#include <linux/log2.h>
|
2007-12-17 11:33:20 -07:00
|
|
|
|
#include <linux/kthread.h>
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
#include "ubi.h"
|
|
|
|
|
|
|
|
|
|
/* Maximum length of the 'mtd=' parameter */
|
|
|
|
|
#define MTD_PARAM_LEN_MAX 64
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* struct mtd_dev_param - MTD device parameter description data structure.
|
|
|
|
|
* @name: MTD device name or number string
|
|
|
|
|
* @vid_hdr_offs: VID header offset
|
|
|
|
|
* @data_offs: data offset
|
|
|
|
|
*/
|
|
|
|
|
struct mtd_dev_param
|
|
|
|
|
{
|
|
|
|
|
char name[MTD_PARAM_LEN_MAX];
|
|
|
|
|
int vid_hdr_offs;
|
|
|
|
|
int data_offs;
|
|
|
|
|
};
|
|
|
|
|
|
|
|
|
|
/* Numbers of elements set in the @mtd_dev_param array */
|
|
|
|
|
static int mtd_devs = 0;
|
|
|
|
|
|
|
|
|
|
/* MTD devices specification parameters */
|
|
|
|
|
static struct mtd_dev_param mtd_dev_param[UBI_MAX_DEVICES];
|
|
|
|
|
|
|
|
|
|
/* Root UBI "class" object (corresponds to '/<sysfs>/class/ubi/') */
|
|
|
|
|
struct class *ubi_class;
|
|
|
|
|
|
2007-12-16 03:32:51 -07:00
|
|
|
|
/* Slab cache for lock-tree entries */
|
|
|
|
|
struct kmem_cache *ubi_ltree_slab;
|
|
|
|
|
|
2007-12-16 03:49:01 -07:00
|
|
|
|
/* Slab cache for wear-leveling entries */
|
|
|
|
|
struct kmem_cache *ubi_wl_entry_slab;
|
|
|
|
|
|
2007-12-16 07:59:31 -07:00
|
|
|
|
/* UBI control character device */
|
|
|
|
|
static struct miscdevice ubi_ctrl_cdev = {
|
|
|
|
|
.minor = MISC_DYNAMIC_MINOR,
|
|
|
|
|
.name = "ubi_ctrl",
|
|
|
|
|
.fops = &ubi_ctrl_cdev_operations,
|
|
|
|
|
};
|
2007-12-16 03:49:01 -07:00
|
|
|
|
|
2007-12-17 08:37:26 -07:00
|
|
|
|
/* All UBI devices in system */
|
|
|
|
|
static struct ubi_device *ubi_devices[UBI_MAX_DEVICES];
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
/* Serializes UBI devices creations and removals */
|
|
|
|
|
DEFINE_MUTEX(ubi_devices_mutex);
|
|
|
|
|
|
2007-12-17 08:37:26 -07:00
|
|
|
|
/* Protects @ubi_devices and @ubi->ref_count */
|
|
|
|
|
static DEFINE_SPINLOCK(ubi_devices_lock);
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
/* "Show" method for files in '/<sysfs>/class/ubi/' */
|
|
|
|
|
static ssize_t ubi_version_show(struct class *class, char *buf)
|
|
|
|
|
{
|
|
|
|
|
return sprintf(buf, "%d\n", UBI_VERSION);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* UBI version attribute ('/<sysfs>/class/ubi/version') */
|
|
|
|
|
static struct class_attribute ubi_version =
|
|
|
|
|
__ATTR(version, S_IRUGO, ubi_version_show, NULL);
|
|
|
|
|
|
|
|
|
|
static ssize_t dev_attribute_show(struct device *dev,
|
|
|
|
|
struct device_attribute *attr, char *buf);
|
|
|
|
|
|
|
|
|
|
/* UBI device attributes (correspond to files in '/<sysfs>/class/ubi/ubiX') */
|
|
|
|
|
static struct device_attribute dev_eraseblock_size =
|
|
|
|
|
__ATTR(eraseblock_size, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_avail_eraseblocks =
|
|
|
|
|
__ATTR(avail_eraseblocks, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_total_eraseblocks =
|
|
|
|
|
__ATTR(total_eraseblocks, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_volumes_count =
|
|
|
|
|
__ATTR(volumes_count, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_max_ec =
|
|
|
|
|
__ATTR(max_ec, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_reserved_for_bad =
|
|
|
|
|
__ATTR(reserved_for_bad, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_bad_peb_count =
|
|
|
|
|
__ATTR(bad_peb_count, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_max_vol_count =
|
|
|
|
|
__ATTR(max_vol_count, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_min_io_size =
|
|
|
|
|
__ATTR(min_io_size, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
static struct device_attribute dev_bgt_enabled =
|
|
|
|
|
__ATTR(bgt_enabled, S_IRUGO, dev_attribute_show, NULL);
|
|
|
|
|
|
2007-12-17 08:37:26 -07:00
|
|
|
|
/**
|
|
|
|
|
* ubi_get_device - get UBI device.
|
|
|
|
|
* @ubi_num: UBI device number
|
|
|
|
|
*
|
|
|
|
|
* This function returns UBI device description object for UBI device number
|
|
|
|
|
* @ubi_num, or %NULL if the device does not exist. This function increases the
|
|
|
|
|
* device reference count to prevent removal of the device. In other words, the
|
|
|
|
|
* device cannot be removed if its reference count is not zero.
|
|
|
|
|
*/
|
|
|
|
|
struct ubi_device *ubi_get_device(int ubi_num)
|
|
|
|
|
{
|
|
|
|
|
struct ubi_device *ubi;
|
|
|
|
|
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
ubi = ubi_devices[ubi_num];
|
|
|
|
|
if (ubi) {
|
|
|
|
|
ubi_assert(ubi->ref_count >= 0);
|
|
|
|
|
ubi->ref_count += 1;
|
|
|
|
|
get_device(&ubi->dev);
|
|
|
|
|
}
|
|
|
|
|
spin_unlock(&ubi_devices_lock);
|
|
|
|
|
|
|
|
|
|
return ubi;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* ubi_put_device - drop an UBI device reference.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*/
|
|
|
|
|
void ubi_put_device(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
ubi->ref_count -= 1;
|
|
|
|
|
put_device(&ubi->dev);
|
|
|
|
|
spin_unlock(&ubi_devices_lock);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* ubi_get_by_major - get UBI device description object by character device
|
|
|
|
|
* major number.
|
|
|
|
|
* @major: major number
|
|
|
|
|
*
|
|
|
|
|
* This function is similar to 'ubi_get_device()', but it searches the device
|
|
|
|
|
* by its major number.
|
|
|
|
|
*/
|
|
|
|
|
struct ubi_device *ubi_get_by_major(int major)
|
|
|
|
|
{
|
|
|
|
|
int i;
|
|
|
|
|
struct ubi_device *ubi;
|
|
|
|
|
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
for (i = 0; i < UBI_MAX_DEVICES; i++) {
|
|
|
|
|
ubi = ubi_devices[i];
|
|
|
|
|
if (ubi && MAJOR(ubi->cdev.dev) == major) {
|
|
|
|
|
ubi_assert(ubi->ref_count >= 0);
|
|
|
|
|
ubi->ref_count += 1;
|
|
|
|
|
get_device(&ubi->dev);
|
|
|
|
|
spin_unlock(&ubi_devices_lock);
|
|
|
|
|
return ubi;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
spin_unlock(&ubi_devices_lock);
|
|
|
|
|
|
|
|
|
|
return NULL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* ubi_major2num - get UBI device number by character device major number.
|
|
|
|
|
* @major: major number
|
|
|
|
|
*
|
|
|
|
|
* This function searches UBI device number object by its major number. If UBI
|
2007-12-17 11:33:20 -07:00
|
|
|
|
* device was not found, this function returns -ENODEV, otherwise the UBI device
|
2007-12-17 08:37:26 -07:00
|
|
|
|
* number is returned.
|
|
|
|
|
*/
|
|
|
|
|
int ubi_major2num(int major)
|
|
|
|
|
{
|
|
|
|
|
int i, ubi_num = -ENODEV;
|
|
|
|
|
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
for (i = 0; i < UBI_MAX_DEVICES; i++) {
|
|
|
|
|
struct ubi_device *ubi = ubi_devices[i];
|
|
|
|
|
|
|
|
|
|
if (ubi && MAJOR(ubi->cdev.dev) == major) {
|
|
|
|
|
ubi_num = ubi->ubi_num;
|
|
|
|
|
break;
|
|
|
|
|
}
|
|
|
|
|
}
|
|
|
|
|
spin_unlock(&ubi_devices_lock);
|
|
|
|
|
|
|
|
|
|
return ubi_num;
|
|
|
|
|
}
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
/* "Show" method for files in '/<sysfs>/class/ubi/ubiX/' */
|
|
|
|
|
static ssize_t dev_attribute_show(struct device *dev,
|
|
|
|
|
struct device_attribute *attr, char *buf)
|
|
|
|
|
{
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ssize_t ret;
|
|
|
|
|
struct ubi_device *ubi;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
2007-12-17 08:37:26 -07:00
|
|
|
|
/*
|
|
|
|
|
* The below code looks weird, but it actually makes sense. We get the
|
|
|
|
|
* UBI device reference from the contained 'struct ubi_device'. But it
|
|
|
|
|
* is unclear if the device was removed or not yet. Indeed, if the
|
|
|
|
|
* device was removed before we increased its reference count,
|
|
|
|
|
* 'ubi_get_device()' will return -ENODEV and we fail.
|
|
|
|
|
*
|
|
|
|
|
* Remember, 'struct ubi_device' is freed in the release function, so
|
|
|
|
|
* we still can use 'ubi->ubi_num'.
|
|
|
|
|
*/
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi = container_of(dev, struct ubi_device, dev);
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ubi = ubi_get_device(ubi->ubi_num);
|
|
|
|
|
if (!ubi)
|
|
|
|
|
return -ENODEV;
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
if (attr == &dev_eraseblock_size)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->leb_size);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_avail_eraseblocks)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->avail_pebs);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_total_eraseblocks)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->good_peb_count);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_volumes_count)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->vol_count);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_max_ec)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->max_ec);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_reserved_for_bad)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->beb_rsvd_pebs);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_bad_peb_count)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->bad_peb_count);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_max_vol_count)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->vtbl_slots);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_min_io_size)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->min_io_size);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else if (attr == &dev_bgt_enabled)
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ret = sprintf(buf, "%d\n", ubi->thread_enabled);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
else
|
|
|
|
|
BUG();
|
|
|
|
|
|
2007-12-17 08:37:26 -07:00
|
|
|
|
ubi_put_device(ubi);
|
|
|
|
|
return ret;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Fake "release" method for UBI devices */
|
|
|
|
|
static void dev_release(struct device *dev) { }
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* ubi_sysfs_init - initialize sysfs for an UBI device.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*
|
|
|
|
|
* This function returns zero in case of success and a negative error code in
|
|
|
|
|
* case of failure.
|
|
|
|
|
*/
|
|
|
|
|
static int ubi_sysfs_init(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
int err;
|
|
|
|
|
|
|
|
|
|
ubi->dev.release = dev_release;
|
2007-12-15 09:13:56 -07:00
|
|
|
|
ubi->dev.devt = ubi->cdev.dev;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi->dev.class = ubi_class;
|
|
|
|
|
sprintf(&ubi->dev.bus_id[0], UBI_NAME_STR"%d", ubi->ubi_num);
|
|
|
|
|
err = device_register(&ubi->dev);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
|
|
|
|
err = device_create_file(&ubi->dev, &dev_eraseblock_size);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_avail_eraseblocks);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_total_eraseblocks);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_volumes_count);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_max_ec);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_reserved_for_bad);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_bad_peb_count);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_max_vol_count);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_min_io_size);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
return err;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = device_create_file(&ubi->dev, &dev_bgt_enabled);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* ubi_sysfs_close - close sysfs for an UBI device.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*/
|
|
|
|
|
static void ubi_sysfs_close(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_bgt_enabled);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_min_io_size);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_max_vol_count);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_bad_peb_count);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_reserved_for_bad);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_max_ec);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_volumes_count);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_total_eraseblocks);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_avail_eraseblocks);
|
|
|
|
|
device_remove_file(&ubi->dev, &dev_eraseblock_size);
|
|
|
|
|
device_unregister(&ubi->dev);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* kill_volumes - destroy all volumes.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*/
|
|
|
|
|
static void kill_volumes(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
int i;
|
|
|
|
|
|
|
|
|
|
for (i = 0; i < ubi->vtbl_slots; i++)
|
|
|
|
|
if (ubi->volumes[i])
|
2007-12-16 11:00:38 -07:00
|
|
|
|
ubi_free_volume(ubi, ubi->volumes[i]);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* uif_init - initialize user interfaces for an UBI device.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*
|
|
|
|
|
* This function returns zero in case of success and a negative error code in
|
|
|
|
|
* case of failure.
|
|
|
|
|
*/
|
|
|
|
|
static int uif_init(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
int i, err;
|
|
|
|
|
dev_t dev;
|
|
|
|
|
|
2007-12-17 03:46:48 -07:00
|
|
|
|
mutex_init(&ubi->volumes_mutex);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
spin_lock_init(&ubi->volumes_lock);
|
|
|
|
|
|
|
|
|
|
sprintf(ubi->ubi_name, UBI_NAME_STR "%d", ubi->ubi_num);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Major numbers for the UBI character devices are allocated
|
|
|
|
|
* dynamically. Major numbers of volume character devices are
|
|
|
|
|
* equivalent to ones of the corresponding UBI character device. Minor
|
|
|
|
|
* numbers of UBI character devices are 0, while minor numbers of
|
|
|
|
|
* volume character devices start from 1. Thus, we allocate one major
|
|
|
|
|
* number and ubi->vtbl_slots + 1 minor numbers.
|
|
|
|
|
*/
|
|
|
|
|
err = alloc_chrdev_region(&dev, 0, ubi->vtbl_slots + 1, ubi->ubi_name);
|
|
|
|
|
if (err) {
|
|
|
|
|
ubi_err("cannot register UBI character devices");
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
2007-12-15 09:13:56 -07:00
|
|
|
|
ubi_assert(MINOR(dev) == 0);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
cdev_init(&ubi->cdev, &ubi_cdev_operations);
|
2007-12-15 09:13:56 -07:00
|
|
|
|
dbg_msg("%s major is %u", ubi->ubi_name, MAJOR(dev));
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi->cdev.owner = THIS_MODULE;
|
|
|
|
|
|
|
|
|
|
err = cdev_add(&ubi->cdev, dev, 1);
|
|
|
|
|
if (err) {
|
2007-12-15 10:56:51 -07:00
|
|
|
|
ubi_err("cannot add character device");
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
goto out_unreg;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
err = ubi_sysfs_init(ubi);
|
|
|
|
|
if (err)
|
2007-12-17 06:48:49 -07:00
|
|
|
|
goto out_sysfs;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
|
|
|
|
for (i = 0; i < ubi->vtbl_slots; i++)
|
|
|
|
|
if (ubi->volumes[i]) {
|
2007-12-16 11:00:38 -07:00
|
|
|
|
err = ubi_add_volume(ubi, ubi->volumes[i]);
|
2007-12-15 10:56:51 -07:00
|
|
|
|
if (err) {
|
|
|
|
|
ubi_err("cannot add volume %d", i);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
goto out_volumes;
|
2007-12-15 10:56:51 -07:00
|
|
|
|
}
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
out_volumes:
|
|
|
|
|
kill_volumes(ubi);
|
2007-12-17 06:48:49 -07:00
|
|
|
|
out_sysfs:
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi_sysfs_close(ubi);
|
|
|
|
|
cdev_del(&ubi->cdev);
|
|
|
|
|
out_unreg:
|
2007-12-15 09:13:56 -07:00
|
|
|
|
unregister_chrdev_region(ubi->cdev.dev, ubi->vtbl_slots + 1);
|
2007-12-15 10:56:51 -07:00
|
|
|
|
ubi_err("cannot initialize UBI %s, error %d", ubi->ubi_name, err);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* uif_close - close user interfaces for an UBI device.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*/
|
|
|
|
|
static void uif_close(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
kill_volumes(ubi);
|
|
|
|
|
ubi_sysfs_close(ubi);
|
|
|
|
|
cdev_del(&ubi->cdev);
|
2007-12-15 09:13:56 -07:00
|
|
|
|
unregister_chrdev_region(ubi->cdev.dev, ubi->vtbl_slots + 1);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* attach_by_scanning - attach an MTD device using scanning method.
|
|
|
|
|
* @ubi: UBI device descriptor
|
|
|
|
|
*
|
|
|
|
|
* This function returns zero in case of success and a negative error code in
|
|
|
|
|
* case of failure.
|
|
|
|
|
*
|
|
|
|
|
* Note, currently this is the only method to attach UBI devices. Hopefully in
|
|
|
|
|
* the future we'll have more scalable attaching methods and avoid full media
|
|
|
|
|
* scanning. But even in this case scanning will be needed as a fall-back
|
|
|
|
|
* attaching method if there are some on-flash table corruptions.
|
|
|
|
|
*/
|
|
|
|
|
static int attach_by_scanning(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
int err;
|
|
|
|
|
struct ubi_scan_info *si;
|
|
|
|
|
|
|
|
|
|
si = ubi_scan(ubi);
|
|
|
|
|
if (IS_ERR(si))
|
|
|
|
|
return PTR_ERR(si);
|
|
|
|
|
|
|
|
|
|
ubi->bad_peb_count = si->bad_peb_count;
|
|
|
|
|
ubi->good_peb_count = ubi->peb_count - ubi->bad_peb_count;
|
|
|
|
|
ubi->max_ec = si->max_ec;
|
|
|
|
|
ubi->mean_ec = si->mean_ec;
|
|
|
|
|
|
|
|
|
|
err = ubi_read_volume_table(ubi, si);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_si;
|
|
|
|
|
|
|
|
|
|
err = ubi_wl_init_scan(ubi, si);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_vtbl;
|
|
|
|
|
|
|
|
|
|
err = ubi_eba_init_scan(ubi, si);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_wl;
|
|
|
|
|
|
|
|
|
|
ubi_scan_destroy_si(si);
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
out_wl:
|
|
|
|
|
ubi_wl_close(ubi);
|
|
|
|
|
out_vtbl:
|
2007-06-15 04:01:22 -06:00
|
|
|
|
vfree(ubi->vtbl);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
out_si:
|
|
|
|
|
ubi_scan_destroy_si(si);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* io_init - initialize I/O unit for a given UBI device.
|
|
|
|
|
* @ubi: UBI device description object
|
|
|
|
|
*
|
|
|
|
|
* If @ubi->vid_hdr_offset or @ubi->leb_start is zero, default offsets are
|
|
|
|
|
* assumed:
|
|
|
|
|
* o EC header is always at offset zero - this cannot be changed;
|
|
|
|
|
* o VID header starts just after the EC header at the closest address
|
2007-12-17 11:33:20 -07:00
|
|
|
|
* aligned to @io->hdrs_min_io_size;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
* o data starts just after the VID header at the closest address aligned to
|
2007-12-17 11:33:20 -07:00
|
|
|
|
* @io->min_io_size
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
*
|
|
|
|
|
* This function returns zero in case of success and a negative error code in
|
|
|
|
|
* case of failure.
|
|
|
|
|
*/
|
|
|
|
|
static int io_init(struct ubi_device *ubi)
|
|
|
|
|
{
|
|
|
|
|
if (ubi->mtd->numeraseregions != 0) {
|
|
|
|
|
/*
|
|
|
|
|
* Some flashes have several erase regions. Different regions
|
|
|
|
|
* may have different eraseblock size and other
|
|
|
|
|
* characteristics. It looks like mostly multi-region flashes
|
|
|
|
|
* have one "main" region and one or more small regions to
|
|
|
|
|
* store boot loader code or boot parameters or whatever. I
|
|
|
|
|
* guess we should just pick the largest region. But this is
|
|
|
|
|
* not implemented.
|
|
|
|
|
*/
|
|
|
|
|
ubi_err("multiple regions, not implemented");
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
if (ubi->vid_hdr_offset < 0 || ubi->leb_start < ubi->vid_hdr_offset)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
/*
|
|
|
|
|
* Note, in this implementation we support MTD devices with 0x7FFFFFFF
|
|
|
|
|
* physical eraseblocks maximum.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
ubi->peb_size = ubi->mtd->erasesize;
|
|
|
|
|
ubi->peb_count = ubi->mtd->size / ubi->mtd->erasesize;
|
|
|
|
|
ubi->flash_size = ubi->mtd->size;
|
|
|
|
|
|
|
|
|
|
if (ubi->mtd->block_isbad && ubi->mtd->block_markbad)
|
|
|
|
|
ubi->bad_allowed = 1;
|
|
|
|
|
|
|
|
|
|
ubi->min_io_size = ubi->mtd->writesize;
|
|
|
|
|
ubi->hdrs_min_io_size = ubi->mtd->writesize >> ubi->mtd->subpage_sft;
|
|
|
|
|
|
|
|
|
|
/* Make sure minimal I/O unit is power of 2 */
|
2007-06-11 23:01:05 -06:00
|
|
|
|
if (!is_power_of_2(ubi->min_io_size)) {
|
2007-12-15 10:56:51 -07:00
|
|
|
|
ubi_err("min. I/O unit (%d) is not power of 2",
|
|
|
|
|
ubi->min_io_size);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ubi_assert(ubi->hdrs_min_io_size > 0);
|
|
|
|
|
ubi_assert(ubi->hdrs_min_io_size <= ubi->min_io_size);
|
|
|
|
|
ubi_assert(ubi->min_io_size % ubi->hdrs_min_io_size == 0);
|
|
|
|
|
|
|
|
|
|
/* Calculate default aligned sizes of EC and VID headers */
|
|
|
|
|
ubi->ec_hdr_alsize = ALIGN(UBI_EC_HDR_SIZE, ubi->hdrs_min_io_size);
|
|
|
|
|
ubi->vid_hdr_alsize = ALIGN(UBI_VID_HDR_SIZE, ubi->hdrs_min_io_size);
|
|
|
|
|
|
|
|
|
|
dbg_msg("min_io_size %d", ubi->min_io_size);
|
|
|
|
|
dbg_msg("hdrs_min_io_size %d", ubi->hdrs_min_io_size);
|
|
|
|
|
dbg_msg("ec_hdr_alsize %d", ubi->ec_hdr_alsize);
|
|
|
|
|
dbg_msg("vid_hdr_alsize %d", ubi->vid_hdr_alsize);
|
|
|
|
|
|
|
|
|
|
if (ubi->vid_hdr_offset == 0)
|
|
|
|
|
/* Default offset */
|
|
|
|
|
ubi->vid_hdr_offset = ubi->vid_hdr_aloffset =
|
|
|
|
|
ubi->ec_hdr_alsize;
|
|
|
|
|
else {
|
|
|
|
|
ubi->vid_hdr_aloffset = ubi->vid_hdr_offset &
|
|
|
|
|
~(ubi->hdrs_min_io_size - 1);
|
|
|
|
|
ubi->vid_hdr_shift = ubi->vid_hdr_offset -
|
|
|
|
|
ubi->vid_hdr_aloffset;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Similar for the data offset */
|
|
|
|
|
if (ubi->leb_start == 0) {
|
|
|
|
|
ubi->leb_start = ubi->vid_hdr_offset + ubi->vid_hdr_alsize;
|
|
|
|
|
ubi->leb_start = ALIGN(ubi->leb_start, ubi->min_io_size);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
dbg_msg("vid_hdr_offset %d", ubi->vid_hdr_offset);
|
|
|
|
|
dbg_msg("vid_hdr_aloffset %d", ubi->vid_hdr_aloffset);
|
|
|
|
|
dbg_msg("vid_hdr_shift %d", ubi->vid_hdr_shift);
|
|
|
|
|
dbg_msg("leb_start %d", ubi->leb_start);
|
|
|
|
|
|
|
|
|
|
/* The shift must be aligned to 32-bit boundary */
|
|
|
|
|
if (ubi->vid_hdr_shift % 4) {
|
|
|
|
|
ubi_err("unaligned VID header shift %d",
|
|
|
|
|
ubi->vid_hdr_shift);
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* Check sanity */
|
|
|
|
|
if (ubi->vid_hdr_offset < UBI_EC_HDR_SIZE ||
|
|
|
|
|
ubi->leb_start < ubi->vid_hdr_offset + UBI_VID_HDR_SIZE ||
|
|
|
|
|
ubi->leb_start > ubi->peb_size - UBI_VID_HDR_SIZE ||
|
|
|
|
|
ubi->leb_start % ubi->min_io_size) {
|
|
|
|
|
ubi_err("bad VID header (%d) or data offsets (%d)",
|
|
|
|
|
ubi->vid_hdr_offset, ubi->leb_start);
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* It may happen that EC and VID headers are situated in one minimal
|
|
|
|
|
* I/O unit. In this case we can only accept this UBI image in
|
|
|
|
|
* read-only mode.
|
|
|
|
|
*/
|
|
|
|
|
if (ubi->vid_hdr_offset + UBI_VID_HDR_SIZE <= ubi->hdrs_min_io_size) {
|
|
|
|
|
ubi_warn("EC and VID headers are in the same minimal I/O unit, "
|
|
|
|
|
"switch to read-only mode");
|
|
|
|
|
ubi->ro_mode = 1;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ubi->leb_size = ubi->peb_size - ubi->leb_start;
|
|
|
|
|
|
|
|
|
|
if (!(ubi->mtd->flags & MTD_WRITEABLE)) {
|
|
|
|
|
ubi_msg("MTD device %d is write-protected, attach in "
|
|
|
|
|
"read-only mode", ubi->mtd->index);
|
|
|
|
|
ubi->ro_mode = 1;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
dbg_msg("leb_size %d", ubi->leb_size);
|
|
|
|
|
dbg_msg("ro_mode %d", ubi->ro_mode);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Note, ideally, we have to initialize ubi->bad_peb_count here. But
|
|
|
|
|
* unfortunately, MTD does not provide this information. We should loop
|
|
|
|
|
* over all physical eraseblocks and invoke mtd->block_is_bad() for
|
|
|
|
|
* each physical eraseblock. So, we skip ubi->bad_peb_count
|
|
|
|
|
* uninitialized and initialize it after scanning.
|
|
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2007-12-17 11:33:20 -07:00
|
|
|
|
* ubi_attach_mtd_dev - attach an MTD device.
|
|
|
|
|
* @mtd_dev: MTD device description object
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
* @vid_hdr_offset: VID header offset
|
|
|
|
|
* @data_offset: data offset
|
|
|
|
|
*
|
|
|
|
|
* This function attaches an MTD device to UBI. It first treats @mtd_dev as the
|
|
|
|
|
* MTD device name, and tries to open it by this name. If it is unable to open,
|
|
|
|
|
* it tries to convert @mtd_dev to an integer and open the MTD device by its
|
2007-12-17 11:33:20 -07:00
|
|
|
|
* number. Returns new UBI device's number in case of success and a negative
|
|
|
|
|
* error code in case of failure.
|
|
|
|
|
*
|
|
|
|
|
* Note, the invocations of this function has to be serialized by the
|
|
|
|
|
* @ubi_devices_mutex.
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
*/
|
2007-12-17 11:33:20 -07:00
|
|
|
|
int ubi_attach_mtd_dev(struct mtd_info *mtd, int vid_hdr_offset,
|
|
|
|
|
int data_offset)
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
{
|
|
|
|
|
struct ubi_device *ubi;
|
|
|
|
|
int i, err;
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
/*
|
|
|
|
|
* Check if we already have the same MTD device attached.
|
|
|
|
|
*
|
|
|
|
|
* Note, this function assumes that UBI devices creations and deletions
|
|
|
|
|
* are serialized, so it does not take the &ubi_devices_lock.
|
|
|
|
|
*/
|
2007-12-16 04:01:03 -07:00
|
|
|
|
for (i = 0; i < UBI_MAX_DEVICES; i++)
|
|
|
|
|
ubi = ubi_devices[i];
|
2007-12-17 11:33:20 -07:00
|
|
|
|
if (ubi && mtd->index == ubi->mtd->index) {
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi_err("mtd%d is already attached to ubi%d",
|
|
|
|
|
mtd->index, i);
|
2007-12-17 11:33:20 -07:00
|
|
|
|
return -EINVAL;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
2007-12-16 04:01:03 -07:00
|
|
|
|
/* Search for an empty slot in the @ubi_devices array */
|
|
|
|
|
for (i = 0; i < UBI_MAX_DEVICES; i++)
|
2007-12-17 11:33:20 -07:00
|
|
|
|
if (!ubi_devices[i])
|
2007-12-16 04:01:03 -07:00
|
|
|
|
break;
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
if (i == UBI_MAX_DEVICES) {
|
2007-12-16 04:01:03 -07:00
|
|
|
|
ubi_err("only %d UBI devices may be created", UBI_MAX_DEVICES);
|
2007-12-17 11:33:20 -07:00
|
|
|
|
return -ENFILE;
|
2007-12-16 04:01:03 -07:00
|
|
|
|
}
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
ubi = kzalloc(sizeof(struct ubi_device), GFP_KERNEL);
|
|
|
|
|
if (!ubi)
|
|
|
|
|
return -ENOMEM;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
ubi->mtd = mtd;
|
|
|
|
|
ubi->ubi_num = i;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi->vid_hdr_offset = vid_hdr_offset;
|
|
|
|
|
ubi->leb_start = data_offset;
|
2007-12-17 11:33:20 -07:00
|
|
|
|
|
|
|
|
|
dbg_msg("attaching mtd%d to ubi%d: VID header offset %d data offset %d",
|
|
|
|
|
mtd->index, ubi->ubi_num, vid_hdr_offset, data_offset);
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = io_init(ubi);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_free;
|
|
|
|
|
|
2007-08-29 05:51:52 -06:00
|
|
|
|
mutex_init(&ubi->buf_mutex);
|
|
|
|
|
ubi->peb_buf1 = vmalloc(ubi->peb_size);
|
|
|
|
|
if (!ubi->peb_buf1)
|
|
|
|
|
goto out_free;
|
|
|
|
|
|
|
|
|
|
ubi->peb_buf2 = vmalloc(ubi->peb_size);
|
|
|
|
|
if (!ubi->peb_buf2)
|
|
|
|
|
goto out_free;
|
|
|
|
|
|
|
|
|
|
#ifdef CONFIG_MTD_UBI_DEBUG
|
|
|
|
|
mutex_init(&ubi->dbg_buf_mutex);
|
|
|
|
|
ubi->dbg_peb_buf = vmalloc(ubi->peb_size);
|
|
|
|
|
if (!ubi->dbg_peb_buf)
|
|
|
|
|
goto out_free;
|
|
|
|
|
#endif
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
err = attach_by_scanning(ubi);
|
|
|
|
|
if (err) {
|
|
|
|
|
dbg_err("failed to attach by scanning, error %d", err);
|
|
|
|
|
goto out_free;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
err = uif_init(ubi);
|
|
|
|
|
if (err)
|
|
|
|
|
goto out_detach;
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
ubi->bgt_thread = kthread_create(ubi_thread, ubi, ubi->bgt_name);
|
|
|
|
|
if (IS_ERR(ubi->bgt_thread)) {
|
|
|
|
|
err = PTR_ERR(ubi->bgt_thread);
|
|
|
|
|
ubi_err("cannot spawn \"%s\", error %d", ubi->bgt_name,
|
|
|
|
|
err);
|
|
|
|
|
goto out_uif;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
ubi_msg("attached mtd%d to ubi%d", mtd->index, ubi->ubi_num);
|
|
|
|
|
ubi_msg("MTD device name: \"%s\"", mtd->name);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi_msg("MTD device size: %llu MiB", ubi->flash_size >> 20);
|
|
|
|
|
ubi_msg("physical eraseblock size: %d bytes (%d KiB)",
|
|
|
|
|
ubi->peb_size, ubi->peb_size >> 10);
|
|
|
|
|
ubi_msg("logical eraseblock size: %d bytes", ubi->leb_size);
|
|
|
|
|
ubi_msg("number of good PEBs: %d", ubi->good_peb_count);
|
|
|
|
|
ubi_msg("number of bad PEBs: %d", ubi->bad_peb_count);
|
|
|
|
|
ubi_msg("smallest flash I/O unit: %d", ubi->min_io_size);
|
|
|
|
|
ubi_msg("VID header offset: %d (aligned %d)",
|
|
|
|
|
ubi->vid_hdr_offset, ubi->vid_hdr_aloffset);
|
|
|
|
|
ubi_msg("data offset: %d", ubi->leb_start);
|
|
|
|
|
ubi_msg("max. allowed volumes: %d", ubi->vtbl_slots);
|
|
|
|
|
ubi_msg("wear-leveling threshold: %d", CONFIG_MTD_UBI_WL_THRESHOLD);
|
|
|
|
|
ubi_msg("number of internal volumes: %d", UBI_INT_VOL_COUNT);
|
|
|
|
|
ubi_msg("number of user volumes: %d",
|
|
|
|
|
ubi->vol_count - UBI_INT_VOL_COUNT);
|
|
|
|
|
ubi_msg("available PEBs: %d", ubi->avail_pebs);
|
|
|
|
|
ubi_msg("total number of reserved PEBs: %d", ubi->rsvd_pebs);
|
|
|
|
|
ubi_msg("number of PEBs reserved for bad PEB handling: %d",
|
|
|
|
|
ubi->beb_rsvd_pebs);
|
|
|
|
|
ubi_msg("max/mean erase counter: %d/%d", ubi->max_ec, ubi->mean_ec);
|
|
|
|
|
|
|
|
|
|
/* Enable the background thread */
|
|
|
|
|
if (!DBG_DISABLE_BGT) {
|
|
|
|
|
ubi->thread_enabled = 1;
|
|
|
|
|
wake_up_process(ubi->bgt_thread);
|
|
|
|
|
}
|
|
|
|
|
|
2007-12-16 04:01:03 -07:00
|
|
|
|
ubi_devices[ubi->ubi_num] = ubi;
|
2007-12-17 11:33:20 -07:00
|
|
|
|
return ubi->ubi_num;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
out_uif:
|
|
|
|
|
uif_close(ubi);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
out_detach:
|
|
|
|
|
ubi_eba_close(ubi);
|
|
|
|
|
ubi_wl_close(ubi);
|
2007-06-15 04:01:22 -06:00
|
|
|
|
vfree(ubi->vtbl);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
out_free:
|
2007-08-29 05:51:52 -06:00
|
|
|
|
vfree(ubi->peb_buf1);
|
|
|
|
|
vfree(ubi->peb_buf2);
|
|
|
|
|
#ifdef CONFIG_MTD_UBI_DEBUG
|
|
|
|
|
vfree(ubi->dbg_peb_buf);
|
|
|
|
|
#endif
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
kfree(ubi);
|
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
2007-12-17 11:33:20 -07:00
|
|
|
|
* ubi_detach_mtd_dev - detach an MTD device.
|
|
|
|
|
* @ubi_num: UBI device number to detach from
|
|
|
|
|
* @anyway: detach MTD even if device reference count is not zero
|
|
|
|
|
*
|
|
|
|
|
* This function destroys an UBI device number @ubi_num and detaches the
|
|
|
|
|
* underlying MTD device. Returns zero in case of success and %-EBUSY if the
|
|
|
|
|
* UBI device is busy and cannot be destroyed, and %-EINVAL if it does not
|
|
|
|
|
* exist.
|
|
|
|
|
*
|
|
|
|
|
* Note, the invocations of this function has to be serialized by the
|
|
|
|
|
* @ubi_devices_mutex.
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
*/
|
2007-12-17 11:33:20 -07:00
|
|
|
|
int ubi_detach_mtd_dev(int ubi_num, int anyway)
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
{
|
2007-12-17 11:33:20 -07:00
|
|
|
|
struct ubi_device *ubi;
|
|
|
|
|
|
|
|
|
|
if (ubi_num < 0 || ubi_num >= UBI_MAX_DEVICES)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
ubi = ubi_devices[ubi_num];
|
|
|
|
|
if (!ubi) {
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (ubi->ref_count) {
|
|
|
|
|
if (!anyway) {
|
|
|
|
|
spin_lock(&ubi_devices_lock);
|
|
|
|
|
return -EBUSY;
|
|
|
|
|
}
|
|
|
|
|
/* This may only happen if there is a bug */
|
|
|
|
|
ubi_err("%s reference count %d, destroy anyway",
|
|
|
|
|
ubi->ubi_name, ubi->ref_count);
|
|
|
|
|
}
|
|
|
|
|
ubi_devices[ubi->ubi_num] = NULL;
|
|
|
|
|
spin_unlock(&ubi_devices_lock);
|
|
|
|
|
|
|
|
|
|
dbg_msg("detaching mtd%d from ubi%d", ubi->mtd->index, ubi->ubi_num);
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Before freeing anything, we have to stop the background thread to
|
|
|
|
|
* prevent it from doing anything on this device while we are freeing.
|
|
|
|
|
*/
|
|
|
|
|
if (ubi->bgt_thread)
|
|
|
|
|
kthread_stop(ubi->bgt_thread);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
|
|
|
|
uif_close(ubi);
|
|
|
|
|
ubi_eba_close(ubi);
|
|
|
|
|
ubi_wl_close(ubi);
|
2007-05-06 07:12:54 -06:00
|
|
|
|
vfree(ubi->vtbl);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
put_mtd_device(ubi->mtd);
|
2007-08-29 05:51:52 -06:00
|
|
|
|
vfree(ubi->peb_buf1);
|
|
|
|
|
vfree(ubi->peb_buf2);
|
|
|
|
|
#ifdef CONFIG_MTD_UBI_DEBUG
|
|
|
|
|
vfree(ubi->dbg_peb_buf);
|
|
|
|
|
#endif
|
2007-12-17 11:33:20 -07:00
|
|
|
|
ubi_msg("mtd%d is detached from ubi%d", ubi->mtd->index, ubi->ubi_num);
|
|
|
|
|
kfree(ubi);
|
|
|
|
|
return 0;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
2007-12-16 03:32:51 -07:00
|
|
|
|
/**
|
|
|
|
|
* ltree_entry_ctor - lock tree entries slab cache constructor.
|
|
|
|
|
* @obj: the lock-tree entry to construct
|
|
|
|
|
* @cache: the lock tree entry slab cache
|
|
|
|
|
* @flags: constructor flags
|
|
|
|
|
*/
|
|
|
|
|
static void ltree_entry_ctor(struct kmem_cache *cache, void *obj)
|
|
|
|
|
{
|
|
|
|
|
struct ubi_ltree_entry *le = obj;
|
|
|
|
|
|
|
|
|
|
le->users = 0;
|
|
|
|
|
init_rwsem(&le->mutex);
|
|
|
|
|
}
|
|
|
|
|
|
2007-12-17 11:33:20 -07:00
|
|
|
|
/**
|
|
|
|
|
* find_mtd_device - open an MTD device by its name or number.
|
|
|
|
|
* @mtd_dev: name or number of the device
|
|
|
|
|
*
|
|
|
|
|
* This function tries to open and MTD device with name @mtd_dev, and if it
|
|
|
|
|
* fails, then it tries to interpret the @mtd_dev string as an ASCII-coded
|
|
|
|
|
* integer and open an MTD device with this number. Returns MTD device
|
|
|
|
|
* description object in case of success and a negative error code in case of
|
|
|
|
|
* failure.
|
|
|
|
|
*/
|
|
|
|
|
static struct mtd_info * __init open_mtd_device(const char *mtd_dev)
|
|
|
|
|
{
|
|
|
|
|
struct mtd_info *mtd;
|
|
|
|
|
|
|
|
|
|
mtd = get_mtd_device_nm(mtd_dev);
|
|
|
|
|
if (IS_ERR(mtd)) {
|
|
|
|
|
int mtd_num;
|
|
|
|
|
char *endp;
|
|
|
|
|
|
|
|
|
|
if (PTR_ERR(mtd) != -ENODEV)
|
|
|
|
|
return mtd;
|
|
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* Probably this is not MTD device name but MTD device number -
|
|
|
|
|
* check this out.
|
|
|
|
|
*/
|
|
|
|
|
mtd_num = simple_strtoul(mtd_dev, &endp, 0);
|
|
|
|
|
if (*endp != '\0' || mtd_dev == endp) {
|
|
|
|
|
ubi_err("incorrect MTD device: \"%s\"", mtd_dev);
|
|
|
|
|
return ERR_PTR(-ENODEV);
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
mtd = get_mtd_device(NULL, mtd_num);
|
|
|
|
|
if (IS_ERR(mtd))
|
|
|
|
|
return mtd;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return mtd;
|
|
|
|
|
}
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
static int __init ubi_init(void)
|
|
|
|
|
{
|
|
|
|
|
int err, i, k;
|
|
|
|
|
|
|
|
|
|
/* Ensure that EC and VID headers have correct size */
|
|
|
|
|
BUILD_BUG_ON(sizeof(struct ubi_ec_hdr) != 64);
|
|
|
|
|
BUILD_BUG_ON(sizeof(struct ubi_vid_hdr) != 64);
|
|
|
|
|
|
|
|
|
|
if (mtd_devs > UBI_MAX_DEVICES) {
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: too many MTD devices, "
|
|
|
|
|
"maximum is %d\n", UBI_MAX_DEVICES);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
2007-12-16 07:59:31 -07:00
|
|
|
|
/* Create base sysfs directory and sysfs files */
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
ubi_class = class_create(THIS_MODULE, UBI_NAME_STR);
|
2007-12-16 07:59:31 -07:00
|
|
|
|
if (IS_ERR(ubi_class)) {
|
|
|
|
|
err = PTR_ERR(ubi_class);
|
|
|
|
|
printk(KERN_ERR "UBI error: cannot create UBI class\n");
|
|
|
|
|
goto out;
|
|
|
|
|
}
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
|
|
|
|
err = class_create_file(ubi_class, &ubi_version);
|
2007-12-16 07:59:31 -07:00
|
|
|
|
if (err) {
|
|
|
|
|
printk(KERN_ERR "UBI error: cannot create sysfs file\n");
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
goto out_class;
|
2007-12-16 07:59:31 -07:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
err = misc_register(&ubi_ctrl_cdev);
|
|
|
|
|
if (err) {
|
|
|
|
|
printk(KERN_ERR "UBI error: cannot register device\n");
|
|
|
|
|
goto out_version;
|
|
|
|
|
}
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
2007-12-16 03:32:51 -07:00
|
|
|
|
ubi_ltree_slab = kmem_cache_create("ubi_ltree_slab",
|
|
|
|
|
sizeof(struct ubi_ltree_entry), 0,
|
|
|
|
|
0, <ree_entry_ctor);
|
|
|
|
|
if (!ubi_ltree_slab)
|
2007-12-16 07:59:31 -07:00
|
|
|
|
goto out_dev_unreg;
|
2007-12-16 03:32:51 -07:00
|
|
|
|
|
2007-12-16 03:49:01 -07:00
|
|
|
|
ubi_wl_entry_slab = kmem_cache_create("ubi_wl_entry_slab",
|
|
|
|
|
sizeof(struct ubi_wl_entry),
|
|
|
|
|
0, 0, NULL);
|
|
|
|
|
if (!ubi_wl_entry_slab)
|
|
|
|
|
goto out_ltree;
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
/* Attach MTD devices */
|
|
|
|
|
for (i = 0; i < mtd_devs; i++) {
|
|
|
|
|
struct mtd_dev_param *p = &mtd_dev_param[i];
|
2007-12-17 11:33:20 -07:00
|
|
|
|
struct mtd_info *mtd;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
|
|
|
|
cond_resched();
|
2007-12-17 11:33:20 -07:00
|
|
|
|
|
|
|
|
|
mtd = open_mtd_device(p->name);
|
|
|
|
|
if (IS_ERR(mtd)) {
|
|
|
|
|
err = PTR_ERR(mtd);
|
|
|
|
|
goto out_detach;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
mutex_lock(&ubi_devices_mutex);
|
|
|
|
|
err = ubi_attach_mtd_dev(mtd, p->vid_hdr_offs, p->data_offs);
|
|
|
|
|
mutex_unlock(&ubi_devices_mutex);
|
|
|
|
|
if (err < 0) {
|
|
|
|
|
put_mtd_device(mtd);
|
2007-12-16 07:59:31 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: cannot attach %s\n",
|
|
|
|
|
p->name);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
goto out_detach;
|
2007-12-16 07:59:31 -07:00
|
|
|
|
}
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return 0;
|
|
|
|
|
|
|
|
|
|
out_detach:
|
|
|
|
|
for (k = 0; k < i; k++)
|
2007-12-17 11:33:20 -07:00
|
|
|
|
if (ubi_devices[k]) {
|
|
|
|
|
mutex_lock(&ubi_devices_mutex);
|
|
|
|
|
ubi_detach_mtd_dev(ubi_devices[k]->ubi_num, 1);
|
|
|
|
|
mutex_unlock(&ubi_devices_mutex);
|
|
|
|
|
}
|
2007-12-16 03:49:01 -07:00
|
|
|
|
kmem_cache_destroy(ubi_wl_entry_slab);
|
|
|
|
|
out_ltree:
|
2007-12-16 03:32:51 -07:00
|
|
|
|
kmem_cache_destroy(ubi_ltree_slab);
|
2007-12-16 07:59:31 -07:00
|
|
|
|
out_dev_unreg:
|
|
|
|
|
misc_deregister(&ubi_ctrl_cdev);
|
2007-12-16 03:32:51 -07:00
|
|
|
|
out_version:
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
class_remove_file(ubi_class, &ubi_version);
|
|
|
|
|
out_class:
|
|
|
|
|
class_destroy(ubi_class);
|
2007-12-16 07:59:31 -07:00
|
|
|
|
out:
|
|
|
|
|
printk(KERN_ERR "UBI error: cannot initialize UBI, error %d\n", err);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return err;
|
|
|
|
|
}
|
|
|
|
|
module_init(ubi_init);
|
|
|
|
|
|
|
|
|
|
static void __exit ubi_exit(void)
|
|
|
|
|
{
|
2007-12-16 04:01:03 -07:00
|
|
|
|
int i;
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
2007-12-16 04:01:03 -07:00
|
|
|
|
for (i = 0; i < UBI_MAX_DEVICES; i++)
|
2007-12-17 11:33:20 -07:00
|
|
|
|
if (ubi_devices[i]) {
|
|
|
|
|
mutex_lock(&ubi_devices_mutex);
|
|
|
|
|
ubi_detach_mtd_dev(ubi_devices[i]->ubi_num, 1);
|
|
|
|
|
mutex_unlock(&ubi_devices_mutex);
|
|
|
|
|
}
|
2007-12-16 03:49:01 -07:00
|
|
|
|
kmem_cache_destroy(ubi_wl_entry_slab);
|
2007-12-16 03:32:51 -07:00
|
|
|
|
kmem_cache_destroy(ubi_ltree_slab);
|
2007-12-16 07:59:31 -07:00
|
|
|
|
misc_deregister(&ubi_ctrl_cdev);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
class_remove_file(ubi_class, &ubi_version);
|
|
|
|
|
class_destroy(ubi_class);
|
|
|
|
|
}
|
|
|
|
|
module_exit(ubi_exit);
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* bytes_str_to_int - convert a string representing number of bytes to an
|
|
|
|
|
* integer.
|
|
|
|
|
* @str: the string to convert
|
|
|
|
|
*
|
|
|
|
|
* This function returns positive resulting integer in case of success and a
|
|
|
|
|
* negative error code in case of failure.
|
|
|
|
|
*/
|
|
|
|
|
static int __init bytes_str_to_int(const char *str)
|
|
|
|
|
{
|
|
|
|
|
char *endp;
|
|
|
|
|
unsigned long result;
|
|
|
|
|
|
|
|
|
|
result = simple_strtoul(str, &endp, 0);
|
|
|
|
|
if (str == endp || result < 0) {
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: incorrect bytes count: \"%s\"\n",
|
|
|
|
|
str);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
switch (*endp) {
|
|
|
|
|
case 'G':
|
|
|
|
|
result *= 1024;
|
|
|
|
|
case 'M':
|
|
|
|
|
result *= 1024;
|
|
|
|
|
case 'K':
|
|
|
|
|
case 'k':
|
|
|
|
|
result *= 1024;
|
|
|
|
|
if (endp[1] == 'i' && (endp[2] == '\0' ||
|
|
|
|
|
endp[2] == 'B' || endp[2] == 'b'))
|
|
|
|
|
endp += 2;
|
|
|
|
|
case '\0':
|
|
|
|
|
break;
|
|
|
|
|
default:
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: incorrect bytes count: \"%s\"\n",
|
|
|
|
|
str);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
return result;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/**
|
|
|
|
|
* ubi_mtd_param_parse - parse the 'mtd=' UBI parameter.
|
|
|
|
|
* @val: the parameter value to parse
|
|
|
|
|
* @kp: not used
|
|
|
|
|
*
|
|
|
|
|
* This function returns zero in case of success and a negative error code in
|
|
|
|
|
* case of error.
|
|
|
|
|
*/
|
|
|
|
|
static int __init ubi_mtd_param_parse(const char *val, struct kernel_param *kp)
|
|
|
|
|
{
|
|
|
|
|
int i, len;
|
|
|
|
|
struct mtd_dev_param *p;
|
|
|
|
|
char buf[MTD_PARAM_LEN_MAX];
|
|
|
|
|
char *pbuf = &buf[0];
|
|
|
|
|
char *tokens[3] = {NULL, NULL, NULL};
|
|
|
|
|
|
2007-12-16 07:46:57 -07:00
|
|
|
|
if (!val)
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
if (mtd_devs == UBI_MAX_DEVICES) {
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: too many parameters, max. is %d\n",
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
UBI_MAX_DEVICES);
|
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
len = strnlen(val, MTD_PARAM_LEN_MAX);
|
|
|
|
|
if (len == MTD_PARAM_LEN_MAX) {
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: parameter \"%s\" is too long, "
|
|
|
|
|
"max. is %d\n", val, MTD_PARAM_LEN_MAX);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
if (len == 0) {
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_WARNING "UBI warning: empty 'mtd=' parameter - "
|
|
|
|
|
"ignored\n");
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
strcpy(buf, val);
|
|
|
|
|
|
|
|
|
|
/* Get rid of the final newline */
|
|
|
|
|
if (buf[len - 1] == '\n')
|
2007-07-11 07:03:29 -06:00
|
|
|
|
buf[len - 1] = '\0';
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
|
|
|
|
|
for (i = 0; i < 3; i++)
|
|
|
|
|
tokens[i] = strsep(&pbuf, ",");
|
|
|
|
|
|
|
|
|
|
if (pbuf) {
|
2007-12-19 08:03:42 -07:00
|
|
|
|
printk(KERN_ERR "UBI error: too many arguments at \"%s\"\n",
|
|
|
|
|
val);
|
UBI: Unsorted Block Images
UBI (Latin: "where?") manages multiple logical volumes on a single
flash device, specifically supporting NAND flash devices. UBI provides
a flexible partitioning concept which still allows for wear-levelling
across the whole flash device.
In a sense, UBI may be compared to the Logical Volume Manager
(LVM). Whereas LVM maps logical sector numbers to physical HDD sector
numbers, UBI maps logical eraseblocks to physical eraseblocks.
More information may be found at
http://www.linux-mtd.infradead.org/doc/ubi.html
Partitioning/Re-partitioning
An UBI volume occupies a certain number of erase blocks. This is
limited by a configured maximum volume size, which could also be
viewed as the partition size. Each individual UBI volume's size can
be changed independently of the other UBI volumes, provided that the
sum of all volume sizes doesn't exceed a certain limit.
UBI supports dynamic volumes and static volumes. Static volumes are
read-only and their contents are protected by CRC check sums.
Bad eraseblocks handling
UBI transparently handles bad eraseblocks. When a physical
eraseblock becomes bad, it is substituted by a good physical
eraseblock, and the user does not even notice this.
Scrubbing
On a NAND flash bit flips can occur on any write operation,
sometimes also on read. If bit flips persist on the device, at first
they can still be corrected by ECC, but once they accumulate,
correction will become impossible. Thus it is best to actively scrub
the affected eraseblock, by first copying it to a free eraseblock
and then erasing the original. The UBI layer performs this type of
scrubbing under the covers, transparently to the UBI volume users.
Erase Counts
UBI maintains an erase count header per eraseblock. This frees
higher-level layers (like file systems) from doing this and allows
for centralized erase count management instead. The erase counts are
used by the wear-levelling algorithm in the UBI layer. The algorithm
itself is exchangeable.
Booting from NAND
For booting directly from NAND flash the hardware must at least be
capable of fetching and executing a small portion of the NAND
flash. Some NAND flash controllers have this kind of support. They
usually limit the window to a few kilobytes in erase block 0. This
"initial program loader" (IPL) must then contain sufficient logic to
load and execute the next boot phase.
Due to bad eraseblocks, which may be randomly scattered over the
flash device, it is problematic to store the "secondary program
loader" (SPL) statically. Also, due to bit-flips it may become
corrupted over time. UBI allows to solve this problem gracefully by
storing the SPL in a small static UBI volume.
UBI volumes vs. static partitions
UBI volumes are still very similar to static MTD partitions:
* both consist of eraseblocks (logical eraseblocks in case of UBI
volumes, and physical eraseblocks in case of static partitions;
* both support three basic operations - read, write, erase.
But UBI volumes have the following advantages over traditional
static MTD partitions:
* there are no eraseblock wear-leveling constraints in case of UBI
volumes, so the user should not care about this;
* there are no bit-flips and bad eraseblocks in case of UBI volumes.
So, UBI volumes may be considered as flash devices with relaxed
restrictions.
Where can it be found?
Documentation, kernel code and applications can be found in the MTD
gits.
What are the applications for?
The applications help to create binary flash images for two purposes: pfi
files (partial flash images) for in-system update of UBI volumes, and plain
binary images, with or without OOB data in case of NAND, for a manufacturing
step. Furthermore some tools are/and will be created that allow flash content
analysis after a system has crashed..
Who did UBI?
The original ideas, where UBI is based on, were developed by Andreas
Arnez, Frank Haverkamp and Thomas Gleixner. Josh W. Boyer and some others
were involved too. The implementation of the kernel layer was done by Artem
B. Bityutskiy. The user-space applications and tools were written by Oliver
Lohmann with contributions from Frank Haverkamp, Andreas Arnez, and Artem.
Joern Engel contributed a patch which modifies JFFS2 so that it can be run on
a UBI volume. Thomas Gleixner did modifications to the NAND layer. Alexander
Schmidt made some testing work as well as core functionality improvements.
Signed-off-by: Artem B. Bityutskiy <dedekind@linutronix.de>
Signed-off-by: Frank Haverkamp <haver@vnet.ibm.com>
2006-06-27 02:22:22 -06:00
|
|
|
|
return -EINVAL;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
p = &mtd_dev_param[mtd_devs];
|
|
|
|
|
strcpy(&p->name[0], tokens[0]);
|
|
|
|
|
|
|
|
|
|
if (tokens[1])
|
|
|
|
|
p->vid_hdr_offs = bytes_str_to_int(tokens[1]);
|
|
|
|
|
if (tokens[2])
|
|
|
|
|
p->data_offs = bytes_str_to_int(tokens[2]);
|
|
|
|
|
|
|
|
|
|
if (p->vid_hdr_offs < 0)
|
|
|
|
|
return p->vid_hdr_offs;
|
|
|
|
|
if (p->data_offs < 0)
|
|
|
|
|
return p->data_offs;
|
|
|
|
|
|
|
|
|
|
mtd_devs += 1;
|
|
|
|
|
return 0;
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
module_param_call(mtd, ubi_mtd_param_parse, NULL, NULL, 000);
|
|
|
|
|
MODULE_PARM_DESC(mtd, "MTD devices to attach. Parameter format: "
|
|
|
|
|
"mtd=<name|num>[,<vid_hdr_offs>,<data_offs>]. "
|
|
|
|
|
"Multiple \"mtd\" parameters may be specified.\n"
|
|
|
|
|
"MTD devices may be specified by their number or name. "
|
|
|
|
|
"Optional \"vid_hdr_offs\" and \"data_offs\" parameters "
|
|
|
|
|
"specify UBI VID header position and data starting "
|
|
|
|
|
"position to be used by UBI.\n"
|
|
|
|
|
"Example: mtd=content,1984,2048 mtd=4 - attach MTD device"
|
|
|
|
|
"with name content using VID header offset 1984 and data "
|
|
|
|
|
"start 2048, and MTD device number 4 using default "
|
|
|
|
|
"offsets");
|
|
|
|
|
|
|
|
|
|
MODULE_VERSION(__stringify(UBI_VERSION));
|
|
|
|
|
MODULE_DESCRIPTION("UBI - Unsorted Block Images");
|
|
|
|
|
MODULE_AUTHOR("Artem Bityutskiy");
|
|
|
|
|
MODULE_LICENSE("GPL");
|