linux-stable.git/drivers/block/loop.c, branch v4.14.331

loop: Select I/O scheduler 'none' from inside add_disk()

2023-08-11T09:33:56+00:00

commit 2112f5c1330a671fa852051d85cb9eadc05d7eb7 upstream.

We noticed that the user interface of Android devices becomes very slow
under memory pressure. This is because Android uses the zram driver on top
of the loop driver for swapping, because under memory pressure the swap
code alternates reads and writes quickly, because mq-deadline is the
default scheduler for loop devices and because mq-deadline delays writes by
five seconds for such a workload with default settings. Fix this by making
the kernel select I/O scheduler 'none' from inside add_disk() for loop
devices. This default can be overridden at any time from user space,
e.g. via a udev rule. This approach has an advantage compared to changing
the I/O scheduler from userspace from 'mq-deadline' into 'none', namely
that synchronize_rcu() does not get called.

This patch changes the default I/O scheduler for loop devices from
'mq-deadline' into 'none'.

Additionally, this patch reduces the Android boot time on my test setup
with 0.5 seconds compared to configuring the loop I/O scheduler from user
space.

Cc: Christoph Hellwig 
Cc: Ming Lei 
Cc: Tetsuo Handa 
Cc: Martijn Coenen 
Cc: Jaegeuk Kim 
Signed-off-by: Bart Van Assche 
Link: https://lore.kernel.org/r/20210805174200.3250718-3-bvanassche@acm.org
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

loop: Check for overflow while configuring loop

2022-09-05T08:25:05+00:00

commit c490a0b5a4f36da3918181a8acdc6991d967c5f3 upstream.

The userspace can configure a loop using an ioctl call, wherein
a configuration of type loop_config is passed (see lo_ioctl()'s
case on line 1550 of drivers/block/loop.c). This proceeds to call
loop_configure() which in turn calls loop_set_status_from_info()
(see line 1050 of loop.c), passing &config->info which is of type
loop_info64*. This function then sets the appropriate values, like
the offset.

loop_device has lo_offset of type loff_t (see line 52 of loop.c),
which is typdef-chained to long long, whereas loop_info64 has
lo_offset of type __u64 (see line 56 of include/uapi/linux/loop.h).

The function directly copies offset from info to the device as
follows (See line 980 of loop.c):
	lo->lo_offset = info->lo_offset;

This results in an overflow, which triggers a warning in iomap_iter()
due to a call to iomap_iter_done() which has:
	WARN_ON_ONCE(iter->iomap.offset > iter->pos);

Thus, check for negative value during loop_set_status_from_info().

Bug report: https://syzkaller.appspot.com/bug?id=c620fe14aac810396d3c3edc9ad73848bf69a29e

Reported-and-tested-by: syzbot+a8e049cd3abd342936b6@syzkaller.appspotmail.com
Cc: stable@vger.kernel.org
Reviewed-by: Matthew Wilcox (Oracle) 
Signed-off-by: Siddh Raman Pant 
Reviewed-by: Christoph Hellwig 
Link: https://lore.kernel.org/r/20220823160810.181275-1-code@siddh.me
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

loop: use sysfs_emit() in the sysfs xxx show()

2022-04-20T07:08:21+00:00

[ Upstream commit b27824d31f09ea7b4a6ba2c1b18bd328df3e8bed ]

sprintf does not know the PAGE_SIZE maximum of the temporary buffer
used for outputting sysfs content and it's possible to overrun the
PAGE_SIZE buffer length.

Use a generic sysfs_emit function that knows the size of the
temporary buffer and ensures that no overrun is done for offset
attribute in
loop_attr_[offset|sizelimit|autoclear|partscan|dio]_show() callbacks.

Signed-off-by: Chaitanya Kulkarni 
Reviewed-by: Himanshu Madhani 
Link: https://lore.kernel.org/r/20220215213310.7264-2-kch@nvidia.com
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

loop: replace kill_bdev with invalidate_bdev

2020-06-30T19:38:02+00:00

commit f4bd34b139a3fa2808c4205f12714c65e1548c6c upstream.

When a filesystem is mounted on a loop device and on a loop ioctl
LOOP_SET_STATUS64, because of kill_bdev, buffer_head mappings are getting
destroyed.
kill_bdev
  truncate_inode_pages
    truncate_inode_pages_range
      do_invalidatepage
        block_invalidatepage
          discard_buffer  -->clear BH_Mapped flag

sb_bread
  __bread_gfp
  bh = __getblk_gfp
  -->discard_buffer clear BH_Mapped flag
  __bread_slow
    submit_bh
      submit_bh_wbc
        BUG_ON(!buffer_mapped(bh))  --> hit this BUG_ON

Fixes: 5db470e229e2 ("loop: drop caches if offset or block_size are changed")
Signed-off-by: Zheng Bin 
Reviewed-by: Christoph Hellwig 
Reviewed-by: Bart Van Assche 
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

loop: fix no-unmap write-zeroes request behavior

2019-12-31T11:37:18+00:00

[ Upstream commit efcfec579f6139528c9e6925eca2bc4a36da65c6 ]

Currently, if the loop device receives a WRITE_ZEROES request, it asks
the underlying filesystem to punch out the range.  This behavior is
correct if unmapping is allowed.  However, a NOUNMAP request means that
the caller doesn't want us to free the storage backing the range, so
punching out the range is incorrect behavior.

To satisfy a NOUNMAP | WRITE_ZEROES request, loop should ask the
underlying filesystem to FALLOC_FL_ZERO_RANGE, which is (according to
the fallocate documentation) required to ensure that the entire range is
backed by real storage, which suffices for our purposes.

Fixes: 19372e2769179dd ("loop: implement REQ_OP_WRITE_ZEROES")
Signed-off-by: Darrick J. Wong 
Reviewed-by: Christoph Hellwig 
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

loop: Add LOOP_SET_DIRECT_IO to compat ioctl

2019-10-29T08:17:04+00:00

[ Upstream commit fdbe4eeeb1aac219b14f10c0ed31ae5d1123e9b8 ]

Enabling Direct I/O with loop devices helps reducing memory usage by
avoiding double caching.  32 bit applications running on 64 bits systems
are currently not able to request direct I/O because is missing from the
lo_compat_ioctl.

This patch fixes the compatibility issue mentioned above by exporting
LOOP_SET_DIRECT_IO as additional lo_compat_ioctl() entry.
The input argument for this ioctl is a single long converted to a 1-bit
boolean, so compatibility is preserved.

Cc: Jens Axboe 
Signed-off-by: Alessio Balsini 
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

loop: set PF_MEMALLOC_NOIO for the worker thread

2019-08-16T08:13:46+00:00

commit d0a255e795ab976481565f6ac178314b34fbf891 upstream.

A deadlock with this stacktrace was observed.

The loop thread does a GFP_KERNEL allocation, it calls into dm-bufio
shrinker and the shrinker depends on I/O completion in the dm-bufio
subsystem.

In order to fix the deadlock (and other similar ones), we set the flag
PF_MEMALLOC_NOIO at loop thread entry.

PID: 474    TASK: ffff8813e11f4600  CPU: 10  COMMAND: "kswapd0"
   #0 [ffff8813dedfb938] __schedule at ffffffff8173f405
   #1 [ffff8813dedfb990] schedule at ffffffff8173fa27
   #2 [ffff8813dedfb9b0] schedule_timeout at ffffffff81742fec
   #3 [ffff8813dedfba60] io_schedule_timeout at ffffffff8173f186
   #4 [ffff8813dedfbaa0] bit_wait_io at ffffffff8174034f
   #5 [ffff8813dedfbac0] __wait_on_bit at ffffffff8173fec8
   #6 [ffff8813dedfbb10] out_of_line_wait_on_bit at ffffffff8173ff81
   #7 [ffff8813dedfbb90] __make_buffer_clean at ffffffffa038736f [dm_bufio]
   #8 [ffff8813dedfbbb0] __try_evict_buffer at ffffffffa0387bb8 [dm_bufio]
   #9 [ffff8813dedfbbd0] dm_bufio_shrink_scan at ffffffffa0387cc3 [dm_bufio]
  #10 [ffff8813dedfbc40] shrink_slab at ffffffff811a87ce
  #11 [ffff8813dedfbd30] shrink_zone at ffffffff811ad778
  #12 [ffff8813dedfbdc0] kswapd at ffffffff811ae92f
  #13 [ffff8813dedfbec0] kthread at ffffffff810a8428
  #14 [ffff8813dedfbf50] ret_from_fork at ffffffff81745242

  PID: 14127  TASK: ffff881455749c00  CPU: 11  COMMAND: "loop1"
   #0 [ffff88272f5af228] __schedule at ffffffff8173f405
   #1 [ffff88272f5af280] schedule at ffffffff8173fa27
   #2 [ffff88272f5af2a0] schedule_preempt_disabled at ffffffff8173fd5e
   #3 [ffff88272f5af2b0] __mutex_lock_slowpath at ffffffff81741fb5
   #4 [ffff88272f5af330] mutex_lock at ffffffff81742133
   #5 [ffff88272f5af350] dm_bufio_shrink_count at ffffffffa03865f9 [dm_bufio]
   #6 [ffff88272f5af380] shrink_slab at ffffffff811a86bd
   #7 [ffff88272f5af470] shrink_zone at ffffffff811ad778
   #8 [ffff88272f5af500] do_try_to_free_pages at ffffffff811adb34
   #9 [ffff88272f5af590] try_to_free_pages at ffffffff811adef8
  #10 [ffff88272f5af610] __alloc_pages_nodemask at ffffffff811a09c3
  #11 [ffff88272f5af710] alloc_pages_current at ffffffff811e8b71
  #12 [ffff88272f5af760] new_slab at ffffffff811f4523
  #13 [ffff88272f5af7b0] __slab_alloc at ffffffff8173a1b5
  #14 [ffff88272f5af880] kmem_cache_alloc at ffffffff811f484b
  #15 [ffff88272f5af8d0] do_blockdev_direct_IO at ffffffff812535b3
  #16 [ffff88272f5afb00] __blockdev_direct_IO at ffffffff81255dc3
  #17 [ffff88272f5afb30] xfs_vm_direct_IO at ffffffffa01fe3fc [xfs]
  #18 [ffff88272f5afb90] generic_file_read_iter at ffffffff81198994
  #19 [ffff88272f5afc50] __dta_xfs_file_read_iter_2398 at ffffffffa020c970 [xfs]
  #20 [ffff88272f5afcc0] lo_rw_aio at ffffffffa0377042 [loop]
  #21 [ffff88272f5afd70] loop_queue_work at ffffffffa0377c3b [loop]
  #22 [ffff88272f5afe60] kthread_worker_fn at ffffffff810a8a0c
  #23 [ffff88272f5afec0] kthread at ffffffff810a8428
  #24 [ffff88272f5aff50] ret_from_fork at ffffffff81745242

Signed-off-by: Mikulas Patocka 
Cc: stable@vger.kernel.org
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

Revert "block/loop: Use global lock for ioctl() operation."

2019-05-02T07:40:33+00:00

This reverts commit 57da9a9742200f391d1cf93fea389f7ddc25ec9a which is
commit 310ca162d779efee8a2dc3731439680f3e9c1e86 upstream.

Jan Kara has reported seeing problems with this patch applied, as has
Salvatore Bonaccorso, so let's drop it for now.

Reported-by: Salvatore Bonaccorso 
Reported-by: Jan Kara 
Cc: Tetsuo Handa 
Cc: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

Revert "loop: Fold __loop_release into loop_release"

2019-03-05T16:57:58+00:00

This reverts commit f1e81ba8a3fa56dcc48828869b392b29559a0ac3 which is
commit 967d1dc144b50ad005e5eecdfadfbcfb399ffff6 upstream.

It does not work properly in the 4.14.y tree and causes more problems
than it fixes, so revert it.

Reported-by: Thomas Lindroth 
Reported-by: Jan Kara 
Cc: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

Revert "loop: Get rid of loop_index_mutex"

2019-03-05T16:57:57+00:00

This reverts commit c1e63df4f30c3918476ac9bc594355b0e9629893 which is
commit 0a42e99b58a208839626465af194cfe640ef9493 upstream.

It does not work properly in the 4.14.y tree and causes more problems
than it fixes, so revert it.

Reported-by: Thomas Lindroth 
Reported-by: Jan Kara 
Cc: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman