linux/block
Tejun Heo 404ced25b4 block: __elv_next_request() shouldn't call into the elevator if bypassing
commit 556ee818c0 upstream.

request_queue bypassing is used to suppress higher-level function of a
request_queue so that they can be switched, reconfigured and shut
down.  A request_queue does the followings while bypassing.

* bypasses elevator and io_cq association and queues requests directly
  to the FIFO dispatch queue.

* bypasses block cgroup request_list lookup and always uses the root
  request_list.

Once confirmed to be bypassing, specific elevator and block cgroup
policy implementations can assume that nothing is in flight for them
and perform various operations which would be dangerous otherwise.

Such confirmation is acheived by short-circuiting all new requests
directly to the dispatch queue and waiting for all the requests which
were issued before to finish.  Unfortunately, while the request
allocating and draining sides were properly handled, we forgot to
actually plug the request dispatch path.  Even after bypassing mode is
confirmed, if the attached driver tries to fetch a request and the
dispatch queue is empty, __elv_next_request() would invoke the current
elevator's elevator_dispatch_fn() callback.  As all in-flight requests
were drained, the elevator wouldn't contain any request but once
bypass is confirmed we don't even know whether the elevator is even
there.  It might be in the process of being switched and half torn
down.

Frank Mayhar reports that this actually happened while switching
elevators, leading to an oops.

Let's fix it by making __elv_next_request() avoid invoking the
elevator_dispatch_fn() callback if the queue is bypassing.  It already
avoids invoking the callback if the queue is dying.  As a dying queue
is guaranteed to be bypassing, we can simply replace blk_queue_dying()
check with blk_queue_bypass().

Reported-by: Frank Mayhar <fmayhar@google.com>
References: http://lkml.kernel.org/g/1390319905.20232.38.camel@bobble.lax.corp.google.com
Tested-by: Frank Mayhar <fmayhar@google.com>
Signed-off-by: Jens Axboe <axboe@kernel.dk>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
2014-02-22 12:41:28 -08:00
..
partitions partitions/efi.c: replace useless kzalloc's by kmalloc's 2013-04-30 08:34:25 +02:00
blk-cgroup.c blkcg: fix "scheduling while atomic" in blk_queue_bypass_start 2013-04-09 15:01:21 +02:00
blk-cgroup.h Update of blkg_stat and blkg_rwstat may happen in bh context. While u64_stats_fetch_retry is only preempt_disable on 32bit UP system. This is not enough to avoid preemption by bh and may read strange 64 bit value. 2013-12-11 22:36:27 -08:00
blk-core.c elevator: Fix a race in elevator switching and md device initialization 2013-12-08 07:29:27 -08:00
blk-exec.c Merge branch 'for-3.9/core' of git://git.kernel.dk/linux-block 2013-02-28 12:52:24 -08:00
blk-flush.c Block: blk-flush: Fixed indent code style 2013-03-22 12:22:51 -06:00
blk-integrity.c scatterlist: introduce sg_unmark_end 2013-03-20 15:43:04 +10:30
blk-ioc.c hlist: drop the node parameter from iterators 2013-02-27 19:10:24 -08:00
blk-iopoll.c tree-wide: fix assorted typos all over the place 2009-12-04 15:39:55 +01:00
blk-lib.c block: account iowait time when waiting for completion of IO request 2013-02-15 16:45:07 +01:00
blk-map.c block: re-use existing 'reading' variable instead of checking direction again 2011-12-21 15:27:24 +01:00
blk-merge.c scatterlist: introduce sg_unmark_end 2013-03-20 15:43:04 +10:30
blk-settings.c block: properly stack underlying max_segment_size to DM device 2013-11-29 11:11:51 -08:00
blk-softirq.c sched, block: Unify cache detection 2012-01-27 13:28:48 +01:00
blk-sysfs.c block: avoid using uninitialized value in from queue_var_store 2013-04-03 21:53:57 +02:00
blk-tag.c block/blk-tag.c: Remove useless kfree 2012-09-12 22:25:12 +02:00
blk-throttle.c block: Rename queue dead flag 2012-12-06 14:30:58 +01:00
blk-timeout.c block: fix race between request completion and timeout handling 2013-11-29 11:11:50 -08:00
blk.h block: __elv_next_request() shouldn't call into the elevator if bypassing 2014-02-22 12:41:28 -08:00
bsg-lib.c bsg: Remove unused function bsg_goose_queue() 2012-12-06 14:33:02 +01:00
bsg.c hlist: drop the node parameter from iterators 2013-02-27 19:10:24 -08:00
cfq-iosched.c cfq: explicitly use 64bit divide operation for 64bit arguments 2013-10-01 09:17:48 -07:00
compat_ioctl.c block: Add BLKROTATIONAL ioctl 2012-01-11 16:29:31 +01:00
deadline-iosched.c elevator: Fix a race in elevator switching 2013-08-20 08:43:03 -07:00
elevator.c elevator: acquire q->sysfs_lock in elevator_change() 2013-12-08 07:29:27 -08:00
genhd.c block: do not pass disk names as format strings 2013-07-13 11:42:26 -07:00
ioctl.c Merge branch 'for-3.7/core' of git://git.kernel.dk/linux-block 2012-10-11 09:04:23 +09:00
Kconfig block: don't select PERCPU_RWSEM 2013-02-22 10:42:45 +01:00
Kconfig.iosched blkcg: make CONFIG_BLK_CGROUP bool 2012-03-06 21:27:21 +01:00
Makefile separate partition format handling from generic code 2012-01-03 22:54:06 -05:00
noop-iosched.c elevator: Fix a race in elevator switching 2013-08-20 08:43:03 -07:00
partition-generic.c Revert "loop: cleanup partitions when detaching loop device" 2013-04-08 10:12:11 +02:00
scsi_ioctl.c aio: don't include aio.h in sched.h 2013-05-07 20:16:25 -07:00