Except for SCSI no device drivers distinguish between physical and
hardware segment limits. Consolidate the two into a single segment
limit.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Commit a9327cac44 added seperate read
and write statistics of in_flight requests. And exported the number
of read and write requests in progress seperately through sysfs.
But Corrado Zoccolo <czoccolo@gmail.com> reported getting strange
output from "iostat -kx 2". Global values for service time and
utilization were garbage. For interval values, utilization was always
100%, and service time is higher than normal.
So this was reverted by commit 0f78ab9899
The problem was in part_round_stats_single(), I missed the following:
if (now == part->stamp)
return;
- if (part->in_flight) {
+ if (part_in_flight(part)) {
__part_stat_add(cpu, part, time_in_queue,
part_in_flight(part) * (now - part->stamp));
__part_stat_add(cpu, part, io_ticks, (now - part->stamp));
With this chunk included, the reported regression gets fixed.
Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
--
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Currently, there is a single in_flight counter measuring the number of
requests in the request_queue. But some monitoring tools would like to
know how many read requests and write requests are in progress. Split the
current in_flight counter into two seperate counters for read and write.
This information is exported as a sysfs attribute, as changing the
currently available stat files would break the existing tools.
Signed-off-by: Nikanth Karthikesan <knikanth@suse.de>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Update scsi_io_completion() such that it only fails requests till the
next error boundary and retry the leftover. This enables block layer
to merge requests with different failfast settings and still behave
correctly on errors. Allow merge of requests of different failfast
settings.
As SCSI is currently the only subsystem which follows failfast status,
there's no need to worry about other block drivers for now.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Niel Lambrechts <niel.lambrechts@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Failfast has characteristics from other attributes. When issuing,
executing and successuflly completing requests, failfast doesn't make
any difference. It only affects how a request is handled on failure.
Allowing requests with different failfast settings to be merged cause
normal IOs to fail prematurely while not allowing has performance
penalties as failfast is used for read aheads which are likely to be
located near in-flight or to-be-issued normal IOs.
This patch introduces the concept of 'mixed merge'. A request is a
mixed merge if it is merge of segments which require different
handling on failure. Currently the only mixable attributes are
failfast ones (or lack thereof).
When a bio with different failfast settings is added to an existing
request or requests of different failfast settings are merged, the
merged request is marked mixed. Each bio carries failfast settings
and the request always tracks failfast state of the first bio. When
the request fails, blk_rq_err_bytes() can be used to determine how
many bytes can be safely failed without crossing into an area which
requires further retrials.
This allows request merging regardless of failfast settings while
keeping the failure handling correct.
This patch only implements mixed merge but doesn't enable it. The
next one will update SCSI to make use of mixed merge.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Niel Lambrechts <niel.lambrechts@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Block layer used to merge requests and bios with different failfast
settings. This caused regular IOs to fail prematurely when they were
merged into failfast requests for readahead.
Niel Lambrechts could trigger the problem semi-reliably on ext4 when
resuming from STR. ext4 uses readahead when reading inodes and
combined with the deterministic extra SATA PHY exception cycle during
resume on the specific configuration, non-readahead inode read would
fail causing ext4 errors. Please read the following thread for
details.
http://lkml.org/lkml/2009/5/23/21
This patch makes block layer reject merging if the failfast settings
don't match. This is correct but likely to lower IO performance by
preventing regular IOs from mingling into surrounding readahead
requests. Changes to allow such mixed merges and handle errors
correctly will be added later.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Niel Lambrechts <niel.lambrechts@gmail.com>
Cc: Theodore Tso <tytso@mit.edu>
Signed-off-by: Jens Axboe <axboe@carl.(none)>
Convert all external users of queue limits to using wrapper functions
instead of poking the request queue variables directly.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Block low level drivers for some reason have been pretty good at
abusing block layer API. Especially struct request's fields tend to
get violated in all possible ways. Make it clear that low level
drivers MUST NOT access or manipulate rq->sector and rq->data_len
directly by prefixing them with double underscores.
This change is also necessary to break build of out-of-tree codes
which assume the previous block API where internal fields can be
manipulated and rq->data_len carries residual count on completion.
[ Impact: hide internal fields, block API change ]
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
struct request has had a few different ways to represent some
properties of a request. ->hard_* represent block layer's view of the
request progress (completion cursor) and the ones without the prefix
are supposed to represent the issue cursor and allowed to be updated
as necessary by the low level drivers. The thing is that as block
layer supports partial completion, the two cursors really aren't
necessary and only cause confusion. In addition, manual management of
request detail from low level drivers is cumbersome and error-prone at
the very least.
Another interesting duplicate fields are rq->[hard_]nr_sectors and
rq->{hard_cur|current}_nr_sectors against rq->data_len and
rq->bio->bi_size. This is more convoluted than the hard_ case.
rq->[hard_]nr_sectors are initialized for requests with bio but
blk_rq_bytes() uses it only for !pc requests. rq->data_len is
initialized for all request but blk_rq_bytes() uses it only for pc
requests. This causes good amount of confusion throughout block layer
and its drivers and determining the request length has been a bit of
black magic which may or may not work depending on circumstances and
what the specific LLD is actually doing.
rq->{hard_cur|current}_nr_sectors represent the number of sectors in
the contiguous data area at the front. This is mainly used by drivers
which transfers data by walking request segment-by-segment. This
value always equals rq->bio->bi_size >> 9. However, data length for
pc requests may not be multiple of 512 bytes and using this field
becomes a bit confusing.
In general, having multiple fields to represent the same property
leads only to confusion and subtle bugs. With recent block low level
driver cleanups, no driver is accessing or manipulating these
duplicate fields directly. Drop all the duplicates. Now rq->sector
means the current sector, rq->data_len the current total length and
rq->bio->bi_size the current segment length. Everything else is
defined in terms of these three and available only through accessors.
* blk_recalc_rq_sectors() is collapsed into blk_update_request() and
now handles pc and fs requests equally other than rq->sector update.
This means that now pc requests can use partial completion too (no
in-kernel user yet tho).
* bio_cur_sectors() is replaced with bio_cur_bytes() as block layer
now uses byte count as the primary data length.
* blk_rq_pos() is now guranteed to be always correct. In-block users
converted.
* blk_rq_bytes() is now guaranteed to be always valid as is
blk_rq_sectors(). In-block users converted.
* blk_rq_sectors() is now guaranteed to equal blk_rq_bytes() >> 9.
More convenient one is used.
* blk_rq_bytes() and blk_rq_cur_bytes() are now inlined and take const
pointer to request.
[ Impact: API cleanup, single way to represent one property of a request ]
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
With recent cleanups, there is no place where low level driver
directly manipulates request fields. This means that the 'hard'
request fields always equal the !hard fields. Convert all
rq->sectors, nr_sectors and current_nr_sectors references to
accessors.
While at it, drop superflous blk_rq_pos() < 0 test in swim.c.
[ Impact: use pos and nr_sectors accessors ]
Signed-off-by: Tejun Heo <tj@kernel.org>
Acked-by: Geert Uytterhoeven <Geert.Uytterhoeven@sonycom.com>
Tested-by: Grant Likely <grant.likely@secretlab.ca>
Acked-by: Grant Likely <grant.likely@secretlab.ca>
Tested-by: Adrian McMenamin <adrian@mcmen.demon.co.uk>
Acked-by: Adrian McMenamin <adrian@mcmen.demon.co.uk>
Acked-by: Mike Miller <mike.miller@hp.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Cc: Bartlomiej Zolnierkiewicz <bzolnier@gmail.com>
Cc: Borislav Petkov <petkovbb@googlemail.com>
Cc: Sergei Shtylyov <sshtylyov@ru.mvista.com>
Cc: Eric Moore <Eric.Moore@lsi.com>
Cc: Alan Stern <stern@rowland.harvard.edu>
Cc: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Pete Zaitcev <zaitcev@redhat.com>
Cc: Stephen Rothwell <sfr@canb.auug.org.au>
Cc: Paul Clements <paul.clements@steeleye.com>
Cc: Tim Waugh <tim@cyberelk.net>
Cc: Jeff Garzik <jgarzik@pobox.com>
Cc: Jeremy Fitzhardinge <jeremy@xensource.com>
Cc: Alex Dubov <oakad@yahoo.com>
Cc: David Woodhouse <dwmw2@infradead.org>
Cc: Martin Schwidefsky <schwidefsky@de.ibm.com>
Cc: Dario Ballabio <ballabio_dario@emc.com>
Cc: David S. Miller <davem@davemloft.net>
Cc: Rusty Russell <rusty@rustcorp.com.au>
Cc: unsik Kim <donari75@gmail.com>
Cc: Laurent Vivier <Laurent@lvivier.info>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This simplifies I/O stat accounting switching code and separates it
completely from I/O scheduler switch code.
Requests are accounted according to the state of their request queue
at the time of the request allocation. There is no need anymore to
flush the request queue when switching I/O accounting state.
Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This forces in_flight to be zero when turning off or on the I/O stat
accounting and stops updating I/O stats in attempt_merge() when
accounting is turned off.
Signed-off-by: Jerome Marchand <jmarchan@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Put a WARN_ON in __blk_put_request if it is about to
leak bio(s). This is a serious bug that can happen in error
handling code paths.
For this to work I have fixed a couple of places in block/ where
request->bio != NULL ownership was not honored. And a small cleanup
at sg_io() while at it.
Signed-off-by: Boaz Harrosh <bharrosh@panasas.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Commit 1e42807918 introduced a bug where we
don't get front/back segment sizes in the bio in blk_recount_segments().
Fix this by tracking the back bio as well as the front bio in
__blk_recalc_rq_segments(), this also cleans up the interface by getting
rid of the segment size pointer passing.
Tested-by: Thomas Gleixner <tglx@linutronix.de>
Tested-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This fixes the bug reported by Nikanth Karthikesan <knikanth@suse.de>:
http://lkml.org/lkml/2008/10/2/203
The root cause of the bug is that blk_phys_contig_segment
miscalculates q->max_segment_size.
blk_phys_contig_segment checks:
req->biotail->bi_size + next_req->bio->bi_size > q->max_segment_size
But blk_recalc_rq_segments might expect that req->biotail and the
previous bio in the req are supposed be merged into one
segment. blk_recalc_rq_segments might also expect that next_req->bio
and the next bio in the next_req are supposed be merged into one
segment. In such case, we merge two requests that can't be merged
here. Later, blk_rq_map_sg gives more segments than it should.
We need to keep track of segment size in blk_recalc_rq_segments and
use it to see if two requests can be merged. This patch implements it
in the similar way that we used to do for hw merging (virtual
merging).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Somewhat incomplete, as we do allow merges of requests and bios
that have different completion CPUs given. This is done on the
assumption that a larger IO is still more beneficial than CPU
locality.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Move stats related fields - stamp, in_flight, dkstats - from disk to
part0 and unify stat handling such that...
* part_stat_*() now updates part0 together if the specified partition
is not part0. ie. part_stat_*() are now essentially all_stat_*().
* {disk|all}_stat_*() are gone.
* part_round_stats() is updated similary. It handles part0 stats
automatically and disk_round_stats() is killed.
* part_{inc|dec}_in_fligh() is implemented which automatically updates
part0 stats for parts other than part0.
* disk_map_sector_rcu() is updated to return part0 if no part matches.
Combined with the above changes, this makes NULL special case
handling in callers unnecessary.
* Separate stats show code paths for disk are collapsed into part
stats show code paths.
* Rename disk_stat_lock/unlock() to part_stat_lock/unlock()
While at it, reposition stat handling macros a bit and add missing
parentheses around macro parameters.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
There are two variants of stat functions - ones prefixed with double
underbars which don't care about preemption and ones without which
disable preemption before manipulating per-cpu counters. It's unclear
whether the underbarred ones assume that preemtion is disabled on
entry as some callers don't do that.
This patch unifies diskstats access by implementing disk_stat_lock()
and disk_stat_unlock() which take care of both RCU (for partition
access) and preemption (for per-cpu counter access). diskstats access
should always be enclosed between the two functions. As such, there's
no need for the versions which disables preemption. They're removed
and double underbars ones are renamed to drop the underbars. As an
extra argument is added, there's no danger of using the old version
unconverted.
disk_stat_lock() uses get_cpu() and returns the cpu index and all
diskstat functions which access per-cpu counters now has @cpu
argument to help RT.
This change adds RCU or preemption operations at some places but also
collapses several preemption ops into one at others. Overall, the
performance difference should be negligible as all involved ops are
very lightweight per-cpu ones.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
disk->part[] is protected by its matching bdev's lock. However,
non-critical accesses like collecting stats and printing out sysfs and
proc information used to be performed without any locking. As
partitions can come and go dynamically, partitions can go away
underneath those non-critical accesses. As some of those accesses are
writes, this theoretically can lead to silent corruption.
This patch fixes the race by using RCU for the partition array and dev
reference counter to hold partitions.
* Rename disk->part[] to disk->__part[] to make sure no one outside
genhd layer proper accesses it directly.
* Use RCU for disk->__part[] dereferencing.
* Implement disk_{get|put}_part() which can be used to get and put
partitions from gendisk respectively.
* Iterators are implemented to help iterate through all partitions
safely.
* Functions which require RCU readlock are marked with _rcu suffix.
* Use disk_put_part() in __blkdev_put() instead of directly putting
the contained kobject.
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
This patch makes the following misc updates in preparation for
disk->part dereference fix and extended block devt support.
* implment part_to_disk()
* fix comment about gendisk->part indexing
* rename get_part() to disk_map_sector()
* don't use n which is always zero while printing disk information in
diskstats_show()
Signed-off-by: Tejun Heo <tj@kernel.org>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Remove hw_segments field from struct bio and struct request. Without virtual
merge accounting they have no purpose.
Signed-off-by: Mikulas Patocka <mpatocka@redhat.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
But blkdev_issue_discard() still emits requests which are interpreted as
soft barriers, because naïve callers might otherwise issue subsequent
writes to those same sectors, which might cross on the queue (if they're
reallocated quickly enough).
Callers still _can_ issue non-barrier discard requests, but they have to
take care of queue ordering for themselves.
Signed-off-by: David Woodhouse <David.Woodhouse@intel.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Some block devices support verifying the integrity of requests by way
of checksums or other protection information that is submitted along
with the I/O.
This patch implements support for generating and verifying integrity
metadata, as well as correctly merging, splitting and cloning bios and
requests that have this extra information attached.
See Documentation/block/data-integrity.txt for more information.
Signed-off-by: Martin K. Petersen <martin.petersen@oracle.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
We can save some atomic ops in the IO path, if we clearly define
the rules of how to modify the queue flags.
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
blk_rq_map_user adjusts bi_size of the last bio. It breaks the rule
that req->data_len (the true data length) is equal to sum(bio). It
broke the scsi command completion code.
commit e97a294ef6 was introduced to fix
the above issue. However, the partial completion code doesn't work
with it. The commit is also a layer violation (scsi mid-layer should
not know about the block layer's padding).
This patch moves the padding adjustment to blk_rq_map_sg (suggested by
James). The padding works like the drain buffer. This patch breaks the
rule that req->data_len is equal to sum(sg), however, the drain buffer
already broke it. So this patch just restores the rule that
req->data_len is equal to sub(bio) without breaking anything new.
Now when a low level driver needs padding, blk_rq_map_user and
blk_rq_map_user_iov guarantee there's enough room for padding.
blk_rq_map_sg can safely extend the last entry of a scatter list.
blk_rq_map_sg must extend the last entry of a scatter list only for a
request that got through bio_copy_user_iov. This patches introduces
new REQ_COPY_USER flag.
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Cc: Tejun Heo <htejun@gmail.com>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
The meaning of rq->data_len was changed to the length of an allocated
buffer from the true data length. It breaks SG_IO friends and
bsg. This patch restores the meaning of rq->data_len to the true data
length and adds rq->extra_len to store an extended length (due to
drain buffer and padding).
This patch also removes the code to update bio in blk_rq_map_user
introduced by the commit 40b01b9bbd.
The commit adjusts bio according to memory alignment
(queue_dma_alignment). However, memory alignment is NOT padding
alignment. This adjustment also breaks SG_IO friends and bsg. Padding
alignment needs to be fixed in a proper way (by a separate patch).
Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Signed-off-by: Jens Axboe <axboe@carl.home.kernel.dk>
Clear drain buffer before chaining if the command in question is a
write.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
Draining shouldn't be done for commands where overflow may indicate
data integrity issues. Add dma_drain_needed callback to
request_queue. Drain buffer is appened iff this function returns
non-zero.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
With padding and draining moved into it, block layer now may extend
requests as directed by queue parameters, so now a request has two
sizes - the original request size and the extended size which matches
the size of area pointed to by bios and later by sgs. The latter size
is what lower layers are primarily interested in when allocating,
filling up DMA tables and setting up the controller.
Both padding and draining extend the data area to accomodate
controller characteristics. As any controller which speaks SCSI can
handle underflows, feeding larger data area is safe.
So, this patch makes the primary data length field, request->data_len,
indicate the size of full data area and add a separate length field,
request->raw_data_len, for the unmodified request size. The latter is
used to report to higher layer (userland) and where the original
request size should be fed to the controller or device.
Signed-off-by: Tejun Heo <htejun@gmail.com>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: Jens Axboe <jens.axboe@oracle.com>
