linux-stable.git/block, branch v4.19.307

blk-mq: fix IO hang from sbitmap wakeup race

2024-02-23T07:12:51+00:00

[ Upstream commit 5266caaf5660529e3da53004b8b7174cab6374ed ]

In blk_mq_mark_tag_wait(), __add_wait_queue() may be re-ordered
with the following blk_mq_get_driver_tag() in case of getting driver
tag failure.

Then in __sbitmap_queue_wake_up(), waitqueue_active() may not observe
the added waiter in blk_mq_mark_tag_wait() and wake up nothing, meantime
blk_mq_mark_tag_wait() can't get driver tag successfully.

This issue can be reproduced by running the following test in loop, and
fio hang can be observed in < 30min when running it on my test VM
in laptop.

	modprobe -r scsi_debug
	modprobe scsi_debug delay=0 dev_size_mb=4096 max_queue=1 host_max_queue=1 submit_queues=4
	dev=`ls -d /sys/bus/pseudo/drivers/scsi_debug/adapter*/host*/target*/*/block/* | head -1 | xargs basename`
	fio --filename=/dev/"$dev" --direct=1 --rw=randrw --bs=4k --iodepth=1 \
       		--runtime=100 --numjobs=40 --time_based --name=test \
        	--ioengine=libaio

Fix the issue by adding one explicit barrier in blk_mq_mark_tag_wait(), which
is just fine in case of running out of tag.

Cc: Jan Kara 
Cc: Kemeng Shi 
Reported-by: Changhui Zhong 
Signed-off-by: Ming Lei 
Link: https://lore.kernel.org/r/20240112122626.4181044-1-ming.lei@redhat.com
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

block: Remove special-casing of compound pages

2024-02-23T07:12:40+00:00

commit 1b151e2435fc3a9b10c8946c6aebe9f3e1938c55 upstream.

The special casing was originally added in pre-git history; reproducing
the commit log here:

> commit a318a92567d77
> Author: Andrew Morton 
> Date:   Sun Sep 21 01:42:22 2003 -0700
>
>     [PATCH] Speed up direct-io hugetlbpage handling
>
>     This patch short-circuits all the direct-io page dirtying logic for
>     higher-order pages.  Without this, we pointlessly bounce BIOs up to
>     keventd all the time.

In the last twenty years, compound pages have become used for more than
just hugetlb.  Rewrite these functions to operate on folios instead
of pages and remove the special case for hugetlbfs; I don't think
it's needed any more (and if it is, we can put it back in as a call
to folio_test_hugetlb()).

This was found by inspection; as far as I can tell, this bug can lead
to pages used as the destination of a direct I/O read not being marked
as dirty.  If those pages are then reclaimed by the MM without being
dirtied for some other reason, they won't be written out.  Then when
they're faulted back in, they will not contain the data they should.
It'll take a pretty unusual setup to produce this problem with several
races all going the wrong way.

This problem predates the folio work; it could for example have been
triggered by mmaping a THP in tmpfs and using that as the target of an
O_DIRECT read.

Fixes: 800d8c63b2e98 ("shmem: add huge pages support")
Cc:  
Signed-off-by: Matthew Wilcox (Oracle) 
Signed-off-by: Jens Axboe 
Signed-off-by: Greg Kroah-Hartman

blk-throttle: fix lockdep warning of "cgroup_mutex or RCU read lock required!"

2023-12-20T14:38:02+00:00

[ Upstream commit 27b13e209ddca5979847a1b57890e0372c1edcee ]

Inside blkg_for_each_descendant_pre(), both
css_for_each_descendant_pre() and blkg_lookup() requires RCU read lock,
and either cgroup_assert_mutex_or_rcu_locked() or rcu_read_lock_held()
is called.

Fix the warning by adding rcu read lock.

Reported-by: Changhui Zhong 
Signed-off-by: Ming Lei 
Link: https://lore.kernel.org/r/20231117023527.3188627-2-ming.lei@redhat.com
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

block: fix signed int overflow in Amiga partition support

2023-08-30T14:31:46+00:00

[ Upstream commit fc3d092c6bb48d5865fec15ed5b333c12f36288c ]

The Amiga partition parser module uses signed int for partition sector
address and count, which will overflow for disks larger than 1 TB.

Use sector_t as type for sector address and size to allow using disks
up to 2 TB without LBD support, and disks larger than 2 TB with LBD.

This bug was reported originally in 2012, and the fix was created by
the RDB author, Joanne Dow . A patch had been
discussed and reviewed on linux-m68k at that time but never officially
submitted. This patch differs from Joanne's patch only in its use of
sector_t instead of unsigned int. No checking for overflows is done
(see patch 3 of this series for that).

Reported-by: Martin Steigerwald 
Closes: https://bugzilla.kernel.org/show_bug.cgi?id=43511
Fixes: 1da177e4c3f4 ("Linux-2.6.12-rc2")
Message-ID: <201206192146.09327.Martin@lichtvoll.de>
Cc:  # 5.2
Signed-off-by: Michael Schmitz 
Tested-by: Martin Steigerwald 
Reviewed-by: Geert Uytterhoeven 
Reviewed-by: Christoph Hellwig 
Link: https://lore.kernel.org/r/20230620201725.7020-2-schmitzmic@gmail.com
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

block: bio-integrity: Copy flags when bio_integrity_payload is cloned

2023-03-11T15:31:33+00:00

[ Upstream commit b6a4bdcda430e3ca43bbb9cb1d4d4d34ebe15c40 ]

Make sure to copy the flags when a bio_integrity_payload is cloned.
Otherwise per-I/O properties such as IP checksum flag will not be
passed down to the HBA driver. Since the integrity buffer is owned by
the original bio, the BIP_BLOCK_INTEGRITY flag needs to be masked off
to avoid a double free in the completion path.

Fixes: aae7df50190a ("block: Integrity checksum flag")
Fixes: b1f01388574c ("block: Relocate bio integrity flags")
Reported-by: Saurav Kashyap 
Tested-by: Saurav Kashyap 
Signed-off-by: Martin K. Petersen 
Reviewed-by: Christoph Hellwig 
Reviewed-by: Chaitanya Kulkarni 
Link: https://lore.kernel.org/r/20230215171801.21062-1-martin.petersen@oracle.com
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

blk-mq: remove stale comment for blk_mq_sched_mark_restart_hctx

2023-03-11T15:31:33+00:00

[ Upstream commit c31e76bcc379182fe67a82c618493b7b8868c672 ]

Commit 97889f9ac24f8 ("blk-mq: remove synchronize_rcu() from
blk_mq_del_queue_tag_set()") remove handle of TAG_SHARED in restart,
then shared_hctx_restart counted for how many hardware queues are marked
for restart is removed too.
Remove the stale comment that we still count hardware queues need restart.

Fixes: 97889f9ac24f ("blk-mq: remove synchronize_rcu() from blk_mq_del_queue_tag_set()")
Reviewed-by: Christoph Hellwig 
Signed-off-by: Kemeng Shi 
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

block: fix and cleanup bio_check_ro

2023-02-06T06:49:41+00:00

commit 57e95e4670d1126c103305bcf34a9442f49f6d6a upstream.

Don't use a WARN_ON when printing a potentially user triggered
condition.  Also don't print the partno when the block device name
already includes it, and use the %pg specifier to simplify printing
the block device name.

Signed-off-by: Christoph Hellwig 
Reviewed-by: Chaitanya Kulkarni 
Reviewed-by: Johannes Thumshirn 
Link: https://lore.kernel.org/r/20220304180105.409765-2-hch@lst.de
Signed-off-by: Jens Axboe 
Signed-off-by: Fedor Pchelkin 
Signed-off-by: Greg Kroah-Hartman

blk-mq: fix possible memleak when register 'hctx' failed

2023-01-18T10:30:37+00:00

[ Upstream commit 4b7a21c57b14fbcd0e1729150189e5933f5088e9 ]

There's issue as follows when do fault injection test:
unreferenced object 0xffff888132a9f400 (size 512):
  comm "insmod", pid 308021, jiffies 4324277909 (age 509.733s)
  hex dump (first 32 bytes):
    00 00 00 00 00 00 00 00 08 f4 a9 32 81 88 ff ff  ...........2....
    08 f4 a9 32 81 88 ff ff 00 00 00 00 00 00 00 00  ...2............
  backtrace:
    [<00000000e8952bb4>] kmalloc_node_trace+0x22/0xa0
    [<00000000f9980e0f>] blk_mq_alloc_and_init_hctx+0x3f1/0x7e0
    [<000000002e719efa>] blk_mq_realloc_hw_ctxs+0x1e6/0x230
    [<000000004f1fda40>] blk_mq_init_allocated_queue+0x27e/0x910
    [<00000000287123ec>] __blk_mq_alloc_disk+0x67/0xf0
    [<00000000a2a34657>] 0xffffffffa2ad310f
    [<00000000b173f718>] 0xffffffffa2af824a
    [<0000000095a1dabb>] do_one_initcall+0x87/0x2a0
    [<00000000f32fdf93>] do_init_module+0xdf/0x320
    [<00000000cbe8541e>] load_module+0x3006/0x3390
    [<0000000069ed1bdb>] __do_sys_finit_module+0x113/0x1b0
    [<00000000a1a29ae8>] do_syscall_64+0x35/0x80
    [<000000009cd878b0>] entry_SYSCALL_64_after_hwframe+0x46/0xb0

Fault injection context as follows:
 kobject_add
 blk_mq_register_hctx
 blk_mq_sysfs_register
 blk_register_queue
 device_add_disk
 null_add_dev.part.0 [null_blk]

As 'blk_mq_register_hctx' may already add some objects when failed halfway,
but there isn't do fallback, caller don't know which objects add failed.
To solve above issue just do fallback when add objects failed halfway in
'blk_mq_register_hctx'.

Signed-off-by: Ye Bin 
Reviewed-by: Ming Lei 
Link: https://lore.kernel.org/r/20221117022940.873959-1-yebin@huaweicloud.com
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin

block: unhash blkdev part inode when the part is deleted

2023-01-18T10:29:59+00:00

v5.11 changes the blkdev lookup mechanism completely since commit
22ae8ce8b892 ("block: simplify bdev/disk lookup in blkdev_get"),
and small part of the change is to unhash part bdev inode when
deleting partition. Turns out this kind of change does fix one
nasty issue in case of BLOCK_EXT_MAJOR:

1) when one partition is deleted & closed, disk_put_part() is always
called before bdput(bdev), see blkdev_put(); so the part's devt can
be freed & re-used before the inode is dropped

2) then new partition with same devt can be created just before the
inode in 1) is dropped, then the old inode/bdev structurein 1) is
re-used for this new partition, this way causes use-after-free and
kernel panic.

It isn't possible to backport the whole big patchset of "merge struct
block_device and struct hd_struct v4" for addressing this issue.

https://lore.kernel.org/linux-block/20201128161510.347752-1-hch@lst.de/

So fixes it by unhashing part bdev in delete_partition(), and this way
is actually aligned with v5.11+'s behavior.

Backported from the following 5.10.y commit:

5f2f77560591 ("block: unhash blkdev part inode when the part is deleted")

Reported-by: Shiwei Cui 
Tested-by: Shiwei Cui 
Cc: Christoph Hellwig 
Cc: Jan Kara 
Signed-off-by: Ming Lei 
Signed-off-by: Greg Kroah-Hartman

block: sed-opal: kmalloc the cmd/resp buffers

2022-11-25T16:40:23+00:00

[ Upstream commit f829230dd51974c1f4478900ed30bb77ba530b40 ]

In accordance with [1] the DMA-able memory buffers must be
cacheline-aligned otherwise the cache writing-back and invalidation
performed during the mapping may cause the adjacent data being lost. It's
specifically required for the DMA-noncoherent platforms [2]. Seeing the
opal_dev.{cmd,resp} buffers are implicitly used for DMAs in the NVME and
SCSI/SD drivers in framework of the nvme_sec_submit() and sd_sec_submit()
methods respectively they must be cacheline-aligned to prevent the denoted
problem. One of the option to guarantee that is to kmalloc the buffers
[2]. Let's explicitly allocate them then instead of embedding into the
opal_dev structure instance.

Note this fix was inspired by the commit c94b7f9bab22 ("nvme-hwmon:
kmalloc the NVME SMART log buffer").

[1] Documentation/core-api/dma-api.rst
[2] Documentation/core-api/dma-api-howto.rst

Fixes: 455a7b238cd6 ("block: Add Sed-opal library")
Signed-off-by: Serge Semin 
Reviewed-by: Christoph Hellwig 
Link: https://lore.kernel.org/r/20221107203944.31686-1-Sergey.Semin@baikalelectronics.ru
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin