BMC/Intel-BMC/linux.git/block, branch dev-5.14-intel

block: don't call rq_qos_ops->done_bio if the bio isn't tracked

2021-10-09T13:02:41+00:00

[ Upstream commit a647a524a46736786c95cdb553a070322ca096e3 ] rq_qos framework is only applied on request based driver, so: 1) rq_qos_done_bio() needn't to be called for bio based driver 2) rq_qos_done_bio() needn't to be called for bio which isn't tracked, such as bios ended from error handling code. Especially in bio_endio(): 1) request queue is referred via bio->bi_bdev->bd_disk->queue, which may be gone since request queue refcount may not be held in above two cases 2) q->rq_qos may be freed in blk_cleanup_queue() when calling into __rq_qos_done_bio() Fix the potential kernel panic by not calling rq_qos_ops->done_bio if the bio isn't tracked. This way is safe because both ioc_rqos_done_bio() and blkcg_iolatency_done_bio() are nop if the bio isn't tracked. Reported-by: Yu Kuai Cc: tj@kernel.org Signed-off-by: Ming Lei Reviewed-by: Christoph Hellwig Acked-by: Tejun Heo Link: https://lore.kernel.org/r/20210924110704.1541818-1-ming.lei@redhat.com Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

Revert "block, bfq: honor already-setup queue merges"

2021-10-07T05:53:14+00:00

[ Upstream commit ebc69e897e17373fbe1daaff1debaa77583a5284 ] This reverts commit 2d52c58b9c9bdae0ca3df6a1eab5745ab3f7d80b. We have had several folks complain that this causes hangs for them, which is especially problematic as the commit has also hit stable already. As no resolution seems to be forthcoming right now, revert the patch. Link: https://bugzilla.kernel.org/show_bug.cgi?id=214503 Fixes: 2d52c58b9c9b ("block, bfq: honor already-setup queue merges") Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

blk-cgroup: fix UAF by grabbing blkcg lock before destroying blkg pd

2021-09-30T08:13:05+00:00

[ Upstream commit 858560b27645e7e97aca37ee8f232cccd658fbd2 ] KASAN reports a use-after-free report when doing fuzz test: [693354.104835] ================================================================== [693354.105094] BUG: KASAN: use-after-free in bfq_io_set_weight_legacy+0xd3/0x160 [693354.105336] Read of size 4 at addr ffff888be0a35664 by task sh/1453338 [693354.105607] CPU: 41 PID: 1453338 Comm: sh Kdump: loaded Not tainted 4.18.0-147 [693354.105610] Hardware name: Huawei 2288H V5/BC11SPSCB0, BIOS 0.81 07/02/2018 [693354.105612] Call Trace: [693354.105621] dump_stack+0xf1/0x19b [693354.105626] ? show_regs_print_info+0x5/0x5 [693354.105634] ? printk+0x9c/0xc3 [693354.105638] ? cpumask_weight+0x1f/0x1f [693354.105648] print_address_description+0x70/0x360 [693354.105654] kasan_report+0x1b2/0x330 [693354.105659] ? bfq_io_set_weight_legacy+0xd3/0x160 [693354.105665] ? bfq_io_set_weight_legacy+0xd3/0x160 [693354.105670] bfq_io_set_weight_legacy+0xd3/0x160 [693354.105675] ? bfq_cpd_init+0x20/0x20 [693354.105683] cgroup_file_write+0x3aa/0x510 [693354.105693] ? ___slab_alloc+0x507/0x540 [693354.105698] ? cgroup_file_poll+0x60/0x60 [693354.105702] ? 0xffffffff89600000 [693354.105708] ? usercopy_abort+0x90/0x90 [693354.105716] ? mutex_lock+0xef/0x180 [693354.105726] kernfs_fop_write+0x1ab/0x280 [693354.105732] ? cgroup_file_poll+0x60/0x60 [693354.105738] vfs_write+0xe7/0x230 [693354.105744] ksys_write+0xb0/0x140 [693354.105749] ? __ia32_sys_read+0x50/0x50 [693354.105760] do_syscall_64+0x112/0x370 [693354.105766] ? syscall_return_slowpath+0x260/0x260 [693354.105772] ? do_page_fault+0x9b/0x270 [693354.105779] ? prepare_exit_to_usermode+0xf9/0x1a0 [693354.105784] ? enter_from_user_mode+0x30/0x30 [693354.105793] entry_SYSCALL_64_after_hwframe+0x65/0xca [693354.105875] Allocated by task 1453337: [693354.106001] kasan_kmalloc+0xa0/0xd0 [693354.106006] kmem_cache_alloc_node_trace+0x108/0x220 [693354.106010] bfq_pd_alloc+0x96/0x120 [693354.106015] blkcg_activate_policy+0x1b7/0x2b0 [693354.106020] bfq_create_group_hierarchy+0x1e/0x80 [693354.106026] bfq_init_queue+0x678/0x8c0 [693354.106031] blk_mq_init_sched+0x1f8/0x460 [693354.106037] elevator_switch_mq+0xe1/0x240 [693354.106041] elevator_switch+0x25/0x40 [693354.106045] elv_iosched_store+0x1a1/0x230 [693354.106049] queue_attr_store+0x78/0xb0 [693354.106053] kernfs_fop_write+0x1ab/0x280 [693354.106056] vfs_write+0xe7/0x230 [693354.106060] ksys_write+0xb0/0x140 [693354.106064] do_syscall_64+0x112/0x370 [693354.106069] entry_SYSCALL_64_after_hwframe+0x65/0xca [693354.106114] Freed by task 1453336: [693354.106225] __kasan_slab_free+0x130/0x180 [693354.106229] kfree+0x90/0x1b0 [693354.106233] blkcg_deactivate_policy+0x12c/0x220 [693354.106238] bfq_exit_queue+0xf5/0x110 [693354.106241] blk_mq_exit_sched+0x104/0x130 [693354.106245] __elevator_exit+0x45/0x60 [693354.106249] elevator_switch_mq+0xd6/0x240 [693354.106253] elevator_switch+0x25/0x40 [693354.106257] elv_iosched_store+0x1a1/0x230 [693354.106261] queue_attr_store+0x78/0xb0 [693354.106264] kernfs_fop_write+0x1ab/0x280 [693354.106268] vfs_write+0xe7/0x230 [693354.106271] ksys_write+0xb0/0x140 [693354.106275] do_syscall_64+0x112/0x370 [693354.106280] entry_SYSCALL_64_after_hwframe+0x65/0xca [693354.106329] The buggy address belongs to the object at ffff888be0a35580 which belongs to the cache kmalloc-1k of size 1024 [693354.106736] The buggy address is located 228 bytes inside of 1024-byte region [ffff888be0a35580, ffff888be0a35980) [693354.107114] The buggy address belongs to the page: [693354.107273] page:ffffea002f828c00 count:1 mapcount:0 mapping:ffff888107c17080 index:0x0 compound_mapcount: 0 [693354.107606] flags: 0x17ffffc0008100(slab|head) [693354.107760] raw: 0017ffffc0008100 ffffea002fcbc808 ffffea0030bd3a08 ffff888107c17080 [693354.108020] raw: 0000000000000000 00000000001c001c 00000001ffffffff 0000000000000000 [693354.108278] page dumped because: kasan: bad access detected [693354.108511] Memory state around the buggy address: [693354.108671] ffff888be0a35500: fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc fc [693354.116396] ffff888be0a35580: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [693354.124473] >ffff888be0a35600: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [693354.132421] ^ [693354.140284] ffff888be0a35680: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [693354.147912] ffff888be0a35700: fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb fb [693354.155281] ================================================================== blkgs are protected by both queue and blkcg locks and holding either should stabilize them. However, the path of destroying blkg policy data is only protected by queue lock in blkcg_activate_policy()/blkcg_deactivate_policy(). Other tasks can get the blkg policy data before the blkg policy data is destroyed, and use it after destroyed, which will result in a use-after-free. CPU0 CPU1 blkcg_deactivate_policy spin_lock_irq(&q->queue_lock) bfq_io_set_weight_legacy spin_lock_irq(&blkcg->lock) blkg_to_bfqg(blkg) pd_to_bfqg(blkg->pd[pol->plid]) ^^^^^^blkg->pd[pol->plid] != NULL bfqg != NULL pol->pd_free_fn(blkg->pd[pol->plid]) pd_to_bfqg(blkg->pd[pol->plid]) bfqg_put(bfqg) kfree(bfqg) blkg->pd[pol->plid] = NULL spin_unlock_irq(q->queue_lock); bfq_group_set_weight(bfqg, val, 0) bfqg->entity.new_weight ^^^^^^trigger uaf here spin_unlock_irq(&blkcg->lock); Fix by grabbing the matching blkcg lock before trying to destroy blkg policy data. Suggested-by: Tejun Heo Signed-off-by: Li Jinlin Acked-by: Tejun Heo Link: https://lore.kernel.org/r/20210914042605.3260596-1-lijinlin3@huawei.com Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

block: flush the integrity workqueue in blk_integrity_unregister

2021-09-30T08:13:05+00:00

[ Upstream commit 3df49967f6f1d2121b0c27c381ca1c8386b1dab9 ] When the integrity profile is unregistered there can still be integrity reads queued up which could see a NULL verify_fn as shown by the race window below: CPU0 CPU1 process_one_work nvme_validate_ns bio_integrity_verify_fn nvme_update_ns_info nvme_update_disk_info blk_integrity_unregister ---set queue->integrity as 0 bio_integrity_process --access bi->profile->verify_fn(bi is a pointer of queue->integity) Before calling blk_integrity_unregister in nvme_update_disk_info, we must make sure that there is no work item in the kintegrityd_wq. Just call blk_flush_integrity to flush the work queue so the bug can be resolved. Signed-off-by: Lihong Kou [hch: split up and shortened the changelog] Signed-off-by: Christoph Hellwig Reviewed-by: Sagi Grimberg Link: https://lore.kernel.org/r/20210914070657.87677-3-hch@lst.de Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

block: check if a profile is actually registered in blk_integrity_unregister

2021-09-30T08:13:05+00:00

[ Upstream commit 783a40a1b3ac7f3714d2776fa8ac8cce3535e4f6 ] While clearing the profile itself is harmless, we really should not clear the stable writes flag if it wasn't set due to a registered integrity profile. Reported-by: Lihong Kou Signed-off-by: Christoph Hellwig Reviewed-by: Sagi Grimberg Link: https://lore.kernel.org/r/20210914070657.87677-2-hch@lst.de Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

blk-mq: avoid to iterate over stale request

2021-09-30T08:13:04+00:00

[ Upstream commit 67f3b2f822b7e71cfc9b42dbd9f3144fa2933e0b ] blk-mq can't run allocating driver tag and updating ->rqs[tag] atomically, meantime blk-mq doesn't clear ->rqs[tag] after the driver tag is released. So there is chance to iterating over one stale request just after the tag is allocated and before updating ->rqs[tag]. scsi_host_busy_iter() calls scsi_host_check_in_flight() to count scsi in-flight requests after scsi host is blocked, so no new scsi command can be marked as SCMD_STATE_INFLIGHT. However, driver tag allocation still can be run by blk-mq core. One request is marked as SCMD_STATE_INFLIGHT, but this request may have been kept in another slot of ->rqs[], meantime the slot can be allocated out but ->rqs[] isn't updated yet. Then this in-flight request is counted twice as SCMD_STATE_INFLIGHT. This way causes trouble in handling scsi error. Fixes the issue by not iterating over stale request. Cc: linux-scsi@vger.kernel.org Cc: "Martin K. Petersen" Reported-by: luojiaxing Signed-off-by: Ming Lei Link: https://lore.kernel.org/r/20210906065003.439019-1-ming.lei@redhat.com Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

blk-mq: allow 4x BLK_MAX_REQUEST_COUNT at blk_plug for multiple_queues

2021-09-26T12:10:25+00:00

[ Upstream commit 7f2a6a69f7ced6db8220298e0497cf60482a9d4b ] Limiting number of request to BLK_MAX_REQUEST_COUNT at blk_plug hurts performance for large md arrays. [1] shows resync speed of md array drops for md array with more than 16 HDDs. Fix this by allowing more request at plug queue. The multiple_queue flag is used to only apply higher limit to multiple queue cases. [1] https://lore.kernel.org/linux-raid/CAFDAVznS71BXW8Jxv6k9dXc2iR3ysX3iZRBww_rzA8WifBFxGg@mail.gmail.com/ Tested-by: Marcin Wanat Signed-off-by: Song Liu Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

blk-throttle: fix UAF by deleteing timer in blk_throtl_exit()

2021-09-26T12:10:24+00:00

[ Upstream commit 884f0e84f1e3195b801319c8ec3d5774e9bf2710 ] The pending timer has been set up in blk_throtl_init(). However, the timer is not deleted in blk_throtl_exit(). This means that the timer handler may still be running after freeing the timer, which would result in a use-after-free. Fix by calling del_timer_sync() to delete the timer in blk_throtl_exit(). Signed-off-by: Li Jinlin Link: https://lore.kernel.org/r/20210907121242.2885564-1-lijinlin3@huawei.com Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

block: genhd: don't call blkdev_show() with major_names_lock held

2021-09-26T12:10:24+00:00

[ Upstream commit dfbb3409b27fa42b96f5727a80d3ceb6a8663991 ] If CONFIG_BLK_DEV_LOOP && CONFIG_MTD (at least; there might be other combinations), lockdep complains circular locking dependency at __loop_clr_fd(), for major_names_lock serves as a locking dependency aggregating hub across multiple block modules. ====================================================== WARNING: possible circular locking dependency detected 5.14.0+ #757 Tainted: G E ------------------------------------------------------ systemd-udevd/7568 is trying to acquire lock: ffff88800f334d48 ((wq_completion)loop0){+.+.}-{0:0}, at: flush_workqueue+0x70/0x560 but task is already holding lock: ffff888014a7d4a0 (&lo->lo_mutex){+.+.}-{3:3}, at: __loop_clr_fd+0x4d/0x400 [loop] which lock already depends on the new lock. the existing dependency chain (in reverse order) is: -> #6 (&lo->lo_mutex){+.+.}-{3:3}: lock_acquire+0xbe/0x1f0 __mutex_lock_common+0xb6/0xe10 mutex_lock_killable_nested+0x17/0x20 lo_open+0x23/0x50 [loop] blkdev_get_by_dev+0x199/0x540 blkdev_open+0x58/0x90 do_dentry_open+0x144/0x3a0 path_openat+0xa57/0xda0 do_filp_open+0x9f/0x140 do_sys_openat2+0x71/0x150 __x64_sys_openat+0x78/0xa0 do_syscall_64+0x3d/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae -> #5 (&disk->open_mutex){+.+.}-{3:3}: lock_acquire+0xbe/0x1f0 __mutex_lock_common+0xb6/0xe10 mutex_lock_nested+0x17/0x20 bd_register_pending_holders+0x20/0x100 device_add_disk+0x1ae/0x390 loop_add+0x29c/0x2d0 [loop] blk_request_module+0x5a/0xb0 blkdev_get_no_open+0x27/0xa0 blkdev_get_by_dev+0x5f/0x540 blkdev_open+0x58/0x90 do_dentry_open+0x144/0x3a0 path_openat+0xa57/0xda0 do_filp_open+0x9f/0x140 do_sys_openat2+0x71/0x150 __x64_sys_openat+0x78/0xa0 do_syscall_64+0x3d/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae -> #4 (major_names_lock){+.+.}-{3:3}: lock_acquire+0xbe/0x1f0 __mutex_lock_common+0xb6/0xe10 mutex_lock_nested+0x17/0x20 blkdev_show+0x19/0x80 devinfo_show+0x52/0x60 seq_read_iter+0x2d5/0x3e0 proc_reg_read_iter+0x41/0x80 vfs_read+0x2ac/0x330 ksys_read+0x6b/0xd0 do_syscall_64+0x3d/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae -> #3 (&p->lock){+.+.}-{3:3}: lock_acquire+0xbe/0x1f0 __mutex_lock_common+0xb6/0xe10 mutex_lock_nested+0x17/0x20 seq_read_iter+0x37/0x3e0 generic_file_splice_read+0xf3/0x170 splice_direct_to_actor+0x14e/0x350 do_splice_direct+0x84/0xd0 do_sendfile+0x263/0x430 __se_sys_sendfile64+0x96/0xc0 do_syscall_64+0x3d/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae -> #2 (sb_writers#3){.+.+}-{0:0}: lock_acquire+0xbe/0x1f0 lo_write_bvec+0x96/0x280 [loop] loop_process_work+0xa68/0xc10 [loop] process_one_work+0x293/0x480 worker_thread+0x23d/0x4b0 kthread+0x163/0x180 ret_from_fork+0x1f/0x30 -> #1 ((work_completion)(&lo->rootcg_work)){+.+.}-{0:0}: lock_acquire+0xbe/0x1f0 process_one_work+0x280/0x480 worker_thread+0x23d/0x4b0 kthread+0x163/0x180 ret_from_fork+0x1f/0x30 -> #0 ((wq_completion)loop0){+.+.}-{0:0}: validate_chain+0x1f0d/0x33e0 __lock_acquire+0x92d/0x1030 lock_acquire+0xbe/0x1f0 flush_workqueue+0x8c/0x560 drain_workqueue+0x80/0x140 destroy_workqueue+0x47/0x4f0 __loop_clr_fd+0xb4/0x400 [loop] blkdev_put+0x14a/0x1d0 blkdev_close+0x1c/0x20 __fput+0xfd/0x220 task_work_run+0x69/0xc0 exit_to_user_mode_prepare+0x1ce/0x1f0 syscall_exit_to_user_mode+0x26/0x60 do_syscall_64+0x4c/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae other info that might help us debug this: Chain exists of: (wq_completion)loop0 --> &disk->open_mutex --> &lo->lo_mutex Possible unsafe locking scenario: CPU0 CPU1 ---- ---- lock(&lo->lo_mutex); lock(&disk->open_mutex); lock(&lo->lo_mutex); lock((wq_completion)loop0); *** DEADLOCK *** 2 locks held by systemd-udevd/7568: #0: ffff888012554128 (&disk->open_mutex){+.+.}-{3:3}, at: blkdev_put+0x4c/0x1d0 #1: ffff888014a7d4a0 (&lo->lo_mutex){+.+.}-{3:3}, at: __loop_clr_fd+0x4d/0x400 [loop] stack backtrace: CPU: 0 PID: 7568 Comm: systemd-udevd Tainted: G E 5.14.0+ #757 Hardware name: VMware, Inc. VMware Virtual Platform/440BX Desktop Reference Platform, BIOS 6.00 02/27/2020 Call Trace: dump_stack_lvl+0x79/0xbf print_circular_bug+0x5d6/0x5e0 ? stack_trace_save+0x42/0x60 ? save_trace+0x3d/0x2d0 check_noncircular+0x10b/0x120 validate_chain+0x1f0d/0x33e0 ? __lock_acquire+0x953/0x1030 ? __lock_acquire+0x953/0x1030 __lock_acquire+0x92d/0x1030 ? flush_workqueue+0x70/0x560 lock_acquire+0xbe/0x1f0 ? flush_workqueue+0x70/0x560 flush_workqueue+0x8c/0x560 ? flush_workqueue+0x70/0x560 ? sched_clock_cpu+0xe/0x1a0 ? drain_workqueue+0x41/0x140 drain_workqueue+0x80/0x140 destroy_workqueue+0x47/0x4f0 ? blk_mq_freeze_queue_wait+0xac/0xd0 __loop_clr_fd+0xb4/0x400 [loop] ? __mutex_unlock_slowpath+0x35/0x230 blkdev_put+0x14a/0x1d0 blkdev_close+0x1c/0x20 __fput+0xfd/0x220 task_work_run+0x69/0xc0 exit_to_user_mode_prepare+0x1ce/0x1f0 syscall_exit_to_user_mode+0x26/0x60 do_syscall_64+0x4c/0xb0 entry_SYSCALL_64_after_hwframe+0x44/0xae RIP: 0033:0x7f0fd4c661f7 Code: 00 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 03 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 41 c3 48 83 ec 18 89 7c 24 0c e8 13 fc ff ff RSP: 002b:00007ffd1c9e9fd8 EFLAGS: 00000246 ORIG_RAX: 0000000000000003 RAX: 0000000000000000 RBX: 00007f0fd46be6c8 RCX: 00007f0fd4c661f7 RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000000006 RBP: 0000000000000006 R08: 000055fff1eaf400 R09: 0000000000000000 R10: 00007f0fd46be6c8 R11: 0000000000000246 R12: 0000000000000000 R13: 0000000000000000 R14: 0000000000002f08 R15: 00007ffd1c9ea050 Commit 1c500ad706383f1a ("loop: reduce the loop_ctl_mutex scope") is for breaking "loop_ctl_mutex => &lo->lo_mutex" dependency chain. But enabling a different block module results in forming circular locking dependency due to shared major_names_lock mutex. The simplest fix is to call probe function without holding major_names_lock [1], but Christoph Hellwig does not like such idea. Therefore, instead of holding major_names_lock in blkdev_show(), introduce a different lock for blkdev_show() in order to break "sb_writers#$N => &p->lock => major_names_lock" dependency chain. Link: https://lkml.kernel.org/r/b2af8a5b-3c1b-204e-7f56-bea0b15848d6@i-love.sakura.ne.jp [1] Signed-off-by: Tetsuo Handa Link: https://lore.kernel.org/r/18a02da2-0bf3-550e-b071-2b4ab13c49f0@i-love.sakura.ne.jp Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin

block, bfq: honor already-setup queue merges

2021-09-22T10:39:27+00:00

[ Upstream commit 2d52c58b9c9bdae0ca3df6a1eab5745ab3f7d80b ] The function bfq_setup_merge prepares the merging between two bfq_queues, say bfqq and new_bfqq. To this goal, it assigns bfqq->new_bfqq = new_bfqq. Then, each time some I/O for bfqq arrives, the process that generated that I/O is disassociated from bfqq and associated with new_bfqq (merging is actually a redirection). In this respect, bfq_setup_merge increases new_bfqq->ref in advance, adding the number of processes that are expected to be associated with new_bfqq. Unfortunately, the stable-merging mechanism interferes with this setup. After bfqq->new_bfqq has been set by bfq_setup_merge, and before all the expected processes have been associated with bfqq->new_bfqq, bfqq may happen to be stably merged with a different queue than the current bfqq->new_bfqq. In this case, bfqq->new_bfqq gets changed. So, some of the processes that have been already accounted for in the ref counter of the previous new_bfqq will not be associated with that queue. This creates an unbalance, because those references will never be decremented. This commit fixes this issue by reestablishing the previous, natural behaviour: once bfqq->new_bfqq has been set, it will not be changed until all expected redirections have occurred. Signed-off-by: Davide Zini Signed-off-by: Paolo Valente Link: https://lore.kernel.org/r/20210802141352.74353-2-paolo.valente@linaro.org Signed-off-by: Jens Axboe Signed-off-by: Sasha Levin