Delayed logging adds some serialisation to the log force process to
ensure that it does not deference a bad commit context structure
when determining if a CIL push is necessary or not. It does this by
grabing the CIL context lock exclusively, then dropping it before
pushing the CIL if necessary. This causes serialisation of all log
forces and pushes regardless of whether a force is necessary or not.
As a result fsync heavy workloads (like dbench) can be significantly
slower with delayed logging than without.
To avoid this penalty, copy the current sequence from the context to
the CIL structure when they are swapped. This allows us to do
unlocked checks on the current sequence without having to worry
about dereferencing context structures that may have already been
freed. Hence we can remove the CIL context locking in the forcing
code and only call into the push code if the current context matches
the sequence we need to force.
By passing the sequence into the push code, we can check the
sequence again once we have the CIL lock held exclusive and abort if
the sequence has already been pushed. This avoids a lock round-trip
and unnecessary CIL pushes when we have racing push calls.
The result is that the regression in dbench performance goes away -
this change improves dbench performance on a ramdisk from ~2100MB/s
to ~2500MB/s. This compares favourably to not using delayed logging
which retuns ~2500MB/s for the same workload.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
When we need to cover the log, we issue dummy transactions to ensure
the current log tail is on disk. Unfortunately we currently use the
root inode in the dummy transaction, and the act of committing the
transaction dirties the inode at the VFS level.
As a result, the VFS writeback of the dirty inode will prevent the
filesystem from idling long enough for the log covering state
machine to complete. The state machine gets stuck in a loop issuing
new dummy transactions to cover the log and never makes progress.
To avoid this problem, the dummy transactions should not cause
externally visible state changes. To ensure this occurs, make sure
that dummy transactions log an unchanging field in the superblock as
it's state is never propagated outside the filesystem. This allows
the log covering state machine to complete successfully and the
filesystem now correctly enters a fully idle state about 90s after
the last modification was made.
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Because of delayed updates to sb_icount field in the super block, it
is possible to allocate over maxicount number of inodes. This
causes the arithmetic to calculate a negative number of free inodes
in user commands like df or stat -f.
Since maxicount is a somewhat arbitrary number, a slight over
allocation is not critical but user commands should be displayed as
0 or greater and never go negative. To do this the value in the
stats buffer f_ffree is capped to never go negative.
[ Modified to use max_t as per Christoph's comment. ]
Signed-off-by: Stu Brodsky <sbrodsky@sgi.com>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
During data integrity (WB_SYNC_ALL) writeback, wbc->nr_to_write will
go negative on inodes with more than 1024 dirty pages due to
implementation details of write_cache_pages(). Currently XFS will
abort page clustering in writeback once nr_to_write drops below
zero, and so for data integrity writeback we will do very
inefficient page at a time allocation and IO submission for inodes
with large numbers of dirty pages.
Fix this by only aborting the page clustering code when
wbc->nr_to_write is negative and the sync mode is WB_SYNC_NONE.
Cc: <stable@kernel.org>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
I noticed XFS writeback in 2.6.36-rc1 was much slower than it should have
been. Enabling writeback tracing showed:
flush-253:16-8516 [007] 1342952.351608: wbc_writepage: bdi 253:16: towrt=1024 skip=0 mode=0 kupd=0 bgrd=1 reclm=0 cyclic=1 more=0 older=0x0 start=0x0 end=0x0
flush-253:16-8516 [007] 1342952.351654: wbc_writepage: bdi 253:16: towrt=1023 skip=0 mode=0 kupd=0 bgrd=1 reclm=0 cyclic=1 more=0 older=0x0 start=0x0 end=0x0
flush-253:16-8516 [000] 1342952.369520: wbc_writepage: bdi 253:16: towrt=0 skip=0 mode=0 kupd=0 bgrd=1 reclm=0 cyclic=1 more=0 older=0x0 start=0x0 end=0x0
flush-253:16-8516 [000] 1342952.369542: wbc_writepage: bdi 253:16: towrt=-1 skip=0 mode=0 kupd=0 bgrd=1 reclm=0 cyclic=1 more=0 older=0x0 start=0x0 end=0x0
flush-253:16-8516 [000] 1342952.369549: wbc_writepage: bdi 253:16: towrt=-2 skip=0 mode=0 kupd=0 bgrd=1 reclm=0 cyclic=1 more=0 older=0x0 start=0x0 end=0x0
Writeback is not terminating in background writeback if ->writepage is
returning with wbc->nr_to_write == 0, resulting in sub-optimal single page
writeback on XFS.
Fix the write_cache_pages loop to terminate correctly when this situation
occurs and so prevent this sub-optimal background writeback pattern. This
improves sustained sequential buffered write performance from around
250MB/s to 750MB/s for a 100GB file on an XFS filesystem on my 8p test VM.
Cc:<stable@kernel.org>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Wu Fengguang <fengguang.wu@intel.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Commit 7124fe0a5b ("xfs: validate untrusted inode
numbers during lookup") changes the inode lookup code to do btree lookups for
untrusted inode numbers. This change made an invalid assumption about the
alignment of inodes and hence incorrectly calculated the first inode in the
cluster. As a result, some inode numbers were being incorrectly considered
invalid when they were actually valid.
The issue was not picked up by the xfstests suite because it always runs fsr
and dump (the two utilities that utilise the bulkstat interface) on cache hot
inodes and hence the lookup code in the cold cache path was not sufficiently
exercised to uncover this intermittent problem.
Fix the issue by relaxing the btree lookup criteria and then checking if the
record returned contains the inode number we are lookup for. If it we get an
incorrect record, then the inode number is invalid.
Cc: <stable@kernel.org>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
Under heavy load parallel metadata loads (e.g. dbench), we can fail
to mark all the inodes in a cluster being freed as XFS_ISTALE as we
skip inodes we cannot get the XFS_ILOCK_EXCL or the flush lock on.
When this happens and the inode cluster buffer has already been
marked stale and freed, inode reclaim can try to write the inode out
as it is dirty and not marked stale. This can result in writing th
metadata to an freed extent, or in the case it has already
been overwritten trigger a magic number check failure and return an
EUCLEAN error such as:
Filesystem "ram0": inode 0x442ba1 background reclaim flush failed with 117
Fix this by ensuring that we hoover up all in memory inodes in the
cluster and mark them XFS_ISTALE when freeing the cluster.
Cc: <stable@kernel.org>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
When we commit a transaction using delayed logging, we need to
unlock the items in the transaciton before we unlock the CIL context
and allow it to be checkpointed. If we unlock them after we release
the CIl context lock, the CIL can checkpoint and complete before
we free the log items. This breaks stale buffer item unlock and
unpin processing as there is an implicit assumption that the unlock
will occur before the unpin.
Also, some log items need to store the LSN of the transaction commit
in the item (inodes and EFIs) and so can race with other transaction
completions if we don't prevent the CIL from checkpointing before
the unlock occurs.
Cc: <stable@kernel.org>
Signed-off-by: Dave Chinner <dchinner@redhat.com>
Reviewed-by: Christoph Hellwig <hch@lst.de>
* git://git.kernel.org/pub/scm/linux/kernel/git/davem/net-2.6: (27 commits)
netfilter: fix CONFIG_COMPAT support
isdn/avm: fix build when PCMCIA is not enabled
header: fix broken headers for user space
e1000e: don't check for alternate MAC addr on parts that don't support it
e1000e: disable ASPM L1 on 82573
ll_temac: Fix poll implementation
netxen: fix a race in netxen_nic_get_stats()
qlnic: fix a race in qlcnic_get_stats()
irda: fix a race in irlan_eth_xmit()
net: sh_eth: remove unused variable
netxen: update version 4.0.74
netxen: fix inconsistent lock state
vlan: Match underlying dev carrier on vlan add
ibmveth: Fix opps during MTU change on an active device
ehea: Fix synchronization between HW and SW send queue
bnx2x: Update bnx2x version to 1.52.53-4
bnx2x: Fix PHY locking problem
rds: fix a leak of kernel memory
netlink: fix compat recvmsg
netfilter: fix userspace header warning
...
* 'for-upstream/pvhvm' of git://xenbits.xensource.com/people/ianc/linux-2.6:
xen: pvhvm: make it clearer that XEN_UNPLUG_* define bits in a bitfield
xen: pvhvm: rename xen_emul_unplug=ignore to =unnnecessary
xen: pvhvm: allow user to request no emulated device unplug
* 'rc-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/mmarek/kbuild-2.6:
setlocalversion: Ignote SCMs above the linux source tree
makefile: not need to regenerate kernel.release file when make kernelrelease
fixes for using make 3.82
kconfig: fix segfault when detecting recursive dependency
kconfig: fix savedefconfig with choice marked optional
* 'drm-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (33 commits)
drm/radeon/kms: fix typo in radeon_compute_pll_gain
drm/radeon/kms: try to detect tv vs monitor for underscan
drm/radeon/kms: fix sideport detection on newer rs880 boards
drm/radeon: fix passing wrong type to gem object create.
drm/radeon/kms: set encoder type to DVI for HDMI on evergreen
drm/radeon/kms: add back missing break in info ioctl
drm/radeon/kms: don't enable MSIs on AGP boards
drm/radeon/kms: fix agp mode setup on cards that use pcie bridges
drm: move dereference below check
drm: fix end of loop test
drm/radeon/kms: rework radeon_dp_detect() logic
drm/radeon/kms: add missing asic callback assignment for evergreen
drm/radeon/kms/DCE3+: switch pads to ddc mode when going i2c
drm/radeon/kms/pm: bail early if nothing's changing
drm/radeon/kms/atom: clean up dig atom handling
drm/radeon/kms: DCE3/4 transmitter fixes
drm/radeon/kms: rework encoder handling
drm/radeon/kms: DCE3/4 AdjustPixelPll updates
drm/radeon: Fix stack data leak
drm/radeon/kms: fix GTT/VRAM overlapping test
...
As copy_*_user() calls access_ok() it should not be called explicitly.
Signed-off-by: Kulikov Vasiliy <segooon@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
There is a path which still holds its mutex here.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This path needs a mutex_unlock(). This is stuff from the bkl to mutex
transition.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Acked-by: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
copy_to_user() returns the number of bytes remaining but we want to return
a negative error code on errors.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Cc: "Michael H. Warfield" <mhw@wittsend.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We don't pass NULL tty pointers to the close function, and anyway we
already dereferenced it at this point. This check can be removed.
Signed-off-by: Dan Carpenter <error27@gmail.com>
Cc: "Michael H. Warfield" <mhw@wittsend.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Device addresses are usually printed in hex.
Signed-off-by: Tony Luck <tony.luck@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
It's currently stalled and the original submitter recommended that it
just be dropped at this point in time due.
Cc: Alan Cox <alan@linux.intel.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Each net_device in a system will automatically managed as a possible
batman_if and holds different informations like a buffer with a prepared
originator messages. To reduce the memory usage, the packet_buff will
only be allocated when the interface is really added/enabled for
batman-adv.
The function to update the hw address information inside the packet_buff
just assumes that the packet_buff is always initialised and thus the
kernel will just oops when we try to change the hw address of a not
already fully enabled interface.
We must always check if the packet_buff is allocated before we try to
change information inside of it.
Reported-by: Tim Glaremin <Tim.Glaremin@web.de>
Reported-by: Kazuki Shimada <zukky@bb.banban.jp>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
dev_put allows a device to be freed when all its references are dropped.
After that we are not allowed to access that information anymore. Access
to the data structure of a net_device must be surrounded a dev_hold
and ended using dev_put.
batman-adv adds a device to its own management structure in
hardif_add_interface and will release it in hardif_remove_interface.
Thus it must hold a reference all the time between those functions to
prevent any access to the already released net_device structure.
Reported-by: Tim Glaremin <Tim.Glaremin@web.de>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
We try to get all events for all net_devices to be able to add special
sysfs folders for the batman-adv configuration. This also includes such
events like NETDEV_POST_INIT which has no valid kobject according to
v2.6.32-rc3-13-g7ffbe3f. This would create an oops in that situation.
It is enough to create the batman_if only on NETDEV_REGISTER events
because we will also receive those events for devices which already
existed when we registered the notifier call.
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Earlier batman-adv versions would only create a batman_if struct after
a corresponding interface had been activated by a user. Now each
existing system interface has a batman_if struct and has to be checked
by verifying the IF_ACTIVE flag.
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
When receiving an batman icmp echo request or in case of a time-to-live
exceeded batman would reply with the mac address of the outgoing
interface which might be a secondary interface. Because secondary
interfaces are not globally known this might lead to confusion.
Now, replies are sent with the mac address of the primary interface.
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
If a batman icmp packet had to be routed over a secondary interface
at the first hop, the mac address of that secondary interface would
be written in the 'orig' field of the icmp packet. A node which is
more than one hop away is not aware of the mac address because
secondary interfaces are not flooded through the whole mesh and
therefore can't send a reply.
This patch always sends the mac address of the primary interface
in the 'orig' field of the icmp packet.
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
The orig_hash_lock spinlock always has to be locked with IRQs being
disabled to avoid deadlocks between code that is being executed in
IRQ context and code that is being executed in non-IRQ context.
Reported-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Marek Lindner <lindner_marek@yahoo.de>
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Cc: stable <stable@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Greg Kroah-Hartman merged Linus 2.6.36 tree in
e9563355ac with his staging tree.
Different parts of the merge conflicts were resolved incorrectly and may
result in an abnormal behavior.
Signed-off-by: Sven Eckelmann <sven.eckelmann@gmx.de>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix a compile warning by initializaing lblk. Since FTL_Get_Block_Index()
returns BAD_BLOCK if it doesn't find the logical block number, lblk
number is initizalized to BAD_BLOCK.
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Fix a compile warning by removing an unused variable int i.
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This is the first patch of a patchset that removes all compilations
warnings in staging/spectra.
These patches are a delta from a previous patchset and it assumes that
these three patches all already applied:
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Last patch has a style problem. Sending the correct one. Sorry for the noise
Since BKL was removed from block ioctl handling code, locked_ioctl doesn't
exist anymore.
Using ioctl instead and doing the locking manually.
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
REQ_TYPE_LINUX_BLOCK and REQ_LB_OP_FLUSH doesn't exist anymore. Using
the new REQ_FLUSH flag instead
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
This patch is the first one of a patchset that allows
stagin/spectra driver to compile in linux-next.
blk_queue_ordered doesn't receive a prepare flush function anymore
Signed-off-by: Javier Martinez Canillas <martinez.javier@gmail.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
In the error path, _request_firmware sets
firmware_p to NULL rather than *firmware_p,
which leads to passing a freed firmware
struct to drivers when the firmware file
cannot be found. Fix this.
Broken by commit f8a4bd3456.
Reported-by: Wey-Yi Guy <wey-yi.w.guy@intel.com>
Signed-off-by: Johannes Berg <johannes.berg@intel.com>
Acked-by: Dmitry Torokhov <dmitry.torokhov@gmail.com>
Signed-off-by: Pekka Enberg <penberg@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
Break the kobject namespace defs into their own header to avoid a header file
inclusion ordering problem between linux/sysfs.h and linux/kobject.h.
This fixes the build breakage on older versions of gcc.
Signed-off-by: David Howells <dhowells@redhat.com>
Cc: Eric Biederman <ebiederm@xmission.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit f3c5c1bfd4
(netfilter: xtables: make ip_tables reentrant) forgot to
also compute the jumpstack size in the compat handlers.
Result is that "iptables -I INPUT -j userchain" turns into -j DROP.
Reported by Sebastian Roesner on #netfilter, closes
http://bugzilla.netfilter.org/show_bug.cgi?id=669.
Note: arptables change is compile-tested only.
Signed-off-by: Florian Westphal <fw@strlen.de>
Acked-by: Eric Dumazet <eric.dumazet@gmail.com>
Tested-by: Mikael Pettersson <mikpe@it.uu.se>
Signed-off-by: David S. Miller <davem@davemloft.net>
The machines I have appear to provide their return value in the arguments
structure, not the output structure. Rework the driver to use that again
in order to get rfkill working again.
Signed-off-by: Matthew Garrett <mjg@redhat.com>
Don't ask how ACPI_TOSHIBA got enabled on in desktop system's .config -
I don't know. But it has silently been there until I tried 2.6.36-rc2,
where it broke the build because I don't have LED support turned on.
Attached patch fixes things up.
(I had to change BACKLIGHT_CLASS_DEVICE to "depends" because otherwise
I get unsightly core dumps out of scripts/kconfig/conf).
jon
--
toshiba: make sure we pull in LED support
The Toshiba extras driver uses the LED module, so make sure we have it
configure in.
Signed-off-by: Jonathan Corbet <corbet@lwn.net>
Signed-off-by: Matthew Garrett <mjg@redhat.com>
dev_name always dereferences its argument, so it should not be called if
the argument is NULL. The function indeed later tests the argument for
being NULL.
The semantic match that finds this problem is as follows:
(http://coccinelle.lip6.fr/)
// <smpl>
@@
expression dev,E;
@@
*dev_name(dev)
... when != dev = E
(
*dev == NULL
|
*dev != NULL
)
// </smpl>
Signed-off-by: Julia Lawall <julia@diku.dk>
Signed-off-by: David Howells <dhowells@redhat.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
It's possible for a cifsSesInfo struct to have a NULL password, so we
need to check for that prior to running strncmp on it.
Signed-off-by: Jeff Layton <jlayton@redhat.com>
Signed-off-by: Steve French <sfrench@us.ibm.com>
In synchronous mode the SSI_SRCCR values are ignored. Instead
SSI_STCCR must be used for both receiving and transmitting.
Signed-off-by: Sascha Hauer <s.hauer@pengutronix.de>
Acked-by: Liam Girdwood <lrg@slimlogic.co.uk>
Signed-off-by: Mark Brown <broonie@opensource.wolfsonmicro.com>
