blktests failures with v7.0-rc1 kernel

Chaitanya Kulkarni chaitanyak at nvidia.com
Thu Feb 26 07:00:40 PST 2026


On 2/26/26 12:09 AM, Shinichiro Kawasaki wrote:
> Failure description
> ===================
>
> #1: blktrace/002
>
>      The test case blktrace/002 failed due to "BUG: using __this_cpu_read() in
>      preemptible" and a following WARN [2].

I saw this last night it was pretty late, just glad to see simple 
testcase catching bugs.

I've a patch for this I'll send out shortly :-


From 6a285db1bbdbb613a85ac55a395ed2043d4eb11d Mon Sep 17 00:00:00 2001
From: Chaitanya Kulkarni <kch at nvidia.com>
Date: Wed, 25 Feb 2026 23:01:12 -0800
Subject: [PATCH] blktrace: fix __this_cpu_read/write in preemptible context



blktrace/002 (blktrace ftrace corruption with sysfs trace)  [failed]
     runtime  0.367s  ...  0.437s
     something found in dmesg:
     [   81.211018] run blktests blktrace/002 at 2026-02-25 22:24:33
     [   81.239580] null_blk: disk nullb1 created
     [   81.357294] BUG: using __this_cpu_read() in preemptible 
[00000000] code: dd/2516
     [   81.362842] caller is tracing_record_cmdline+0x10/0x40
     [   81.362872] CPU: 16 UID: 0 PID: 2516 Comm: dd Tainted: G         
        N  7.0.0-rc1lblk+ #84 PREEMPT(full)
     [   81.362877] Tainted: [N]=TEST
     [   81.362878] Hardware name: QEMU Standard PC (i440FX + PIIX, 
1996), BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014
     [   81.362881] Call Trace:
     [   81.362884]  <TASK>
     [   81.362886]  dump_stack_lvl+0x8d/0xb0
     ...
     (See '/mnt/sda/blktests/results/nodev/blktrace/002.dmesg' for the 
entire message)

[   81.211018] run blktests blktrace/002 at 2026-02-25 22:24:33
[   81.239580] null_blk: disk nullb1 created
[   81.357294] BUG: using __this_cpu_read() in preemptible [00000000] 
code: dd/2516
[   81.362842] caller is tracing_record_cmdline+0x10/0x40
[   81.362872] CPU: 16 UID: 0 PID: 2516 Comm: dd Tainted: G           
  N  7.0.0-rc1lblk+ #84 PREEMPT(full)
[   81.362877] Tainted: [N]=TEST
[   81.362878] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), 
BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014
[   81.362881] Call Trace:
[   81.362884]  <TASK>
[   81.362886]  dump_stack_lvl+0x8d/0xb0
[   81.362895]  check_preemption_disabled+0xce/0xe0
[   81.362902]  tracing_record_cmdline+0x10/0x40
[   81.362923]  __blk_add_trace+0x307/0x5d0
[   81.362934]  ? lock_acquire+0xe0/0x300
[   81.362940]  ? iov_iter_extract_pages+0x101/0xa30
[   81.362959]  blk_add_trace_bio+0x106/0x1e0
[   81.362968]  submit_bio_noacct_nocheck+0x24b/0x3a0
[   81.362979]  ? lockdep_init_map_type+0x58/0x260
[   81.362988]  submit_bio_wait+0x56/0x90
[   81.363009]  __blkdev_direct_IO_simple+0x16c/0x250
[   81.363026]  ? __pfx_submit_bio_wait_endio+0x10/0x10
[   81.363038]  ? rcu_read_lock_any_held+0x73/0xa0
[   81.363051]  blkdev_read_iter+0xc1/0x140
[   81.363059]  vfs_read+0x20b/0x330
[   81.363083]  ksys_read+0x67/0xe0
[   81.363090]  do_syscall_64+0xbf/0xf00
[   81.363102]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   81.363106] RIP: 0033:0x7f281906029d
[   81.363111] Code: 31 c0 e9 c6 fe ff ff 50 48 8d 3d 66 63 0a 00 e8 59 
ff 01 00 66 0f 1f 84 00 00 00 00 00 80 3d 41 33 0e 00 00 74 17 31 c0 0f 
05 <48> 3d 00 f0 ff ff 77 5b c3 66 2e 0f 1f 84 00 00 00 00 00 48 83 ec
[   81.363113] RSP: 002b:00007ffca127dd48 EFLAGS: 00000246 ORIG_RAX: 
0000000000000000
[   81.363120] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 
00007f281906029d
[   81.363122] RDX: 0000000000001000 RSI: 0000559f8bfae000 RDI: 
0000000000000000
[   81.363123] RBP: 0000000000001000 R08: 0000002863a10a81 R09: 
00007f281915f000
[   81.363124] R10: 00007f2818f77b60 R11: 0000000000000246 R12: 
0000559f8bfae000
[   81.363126] R13: 0000000000000000 R14: 0000000000000000 R15: 
000000000000000a
[   81.363142]  </TASK>
[   81.363157] BUG: using __this_cpu_read() in preemptible [00000000] 
code: dd/2516
[   81.368486] caller is tracing_record_cmdline+0x10/0x40
[   81.368496] CPU: 16 UID: 0 PID: 2516 Comm: dd Tainted: G           
  N  7.0.0-rc1lblk+ #84 PREEMPT(full)
[   81.368501] Tainted: [N]=TEST
[   81.368502] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), 
BIOS rel-1.17.0-0-gb52ca86e094d-prebuilt.qemu.org 04/01/2014
[   81.368503] Call Trace:
[   81.368505]  <TASK>
[   81.368507]  dump_stack_lvl+0x8d/0xb0
[   81.368513]  check_preemption_disabled+0xce/0xe0
[   81.368518]  tracing_record_cmdline+0x10/0x40
[   81.368523]  __blk_add_trace+0x307/0x5d0
[   81.368528]  ? lock_acquire+0xe0/0x300
[   81.368547]  blk_add_trace_bio+0x106/0x1e0
[   81.368556]  blk_mq_submit_bio+0x63c/0xbb0
[   81.368580]  __submit_bio+0xad/0x5c0
[   81.368595]  ? submit_bio_noacct_nocheck+0x2b0/0x3a0
[   81.368600]  submit_bio_noacct_nocheck+0x2b0/0x3a0
[   81.368611]  submit_bio_wait+0x56/0x90
[   81.368622]  __blkdev_direct_IO_simple+0x16c/0x250
[   81.368635]  ? __pfx_submit_bio_wait_endio+0x10/0x10
[   81.368656]  ? rcu_read_lock_any_held+0x73/0xa0
[   81.368664]  blkdev_read_iter+0xc1/0x140
[   81.368672]  vfs_read+0x20b/0x330
[   81.368687]  ksys_read+0x67/0xe0
[   81.368694]  do_syscall_64+0xbf/0xf00
[   81.368702]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
[   81.368705] RIP: 0033:0x7f281906029d
[   81.368708] Code: 31 c0 e9 c6 fe ff ff 50 48 8d 3d 66 63 0a 00 e8 59 
ff 01 00 66 0f 1f 84 00 00 00 00 00 80 3d 41 33 0e 00 00 74 17 31 c0 0f 
05 <48> 3d 00 f0 ff ff 77 5b c3 66 2e 0f 1f 84 00 00 00 00 00 48 83 ec
[   81.368710] RSP: 002b:00007ffca127dd48 EFLAGS: 00000246 ORIG_RAX: 
0000000000000000
[   81.368712] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 
00007f281906029d
[   81.368713] RDX: 0000000000001000 RSI: 0000559f8bfae000 RDI: 
0000000000000000
[   81.368714] RBP: 0000000000001000 R08: 0000002863a10a81 R09: 
00007f281915f000
[   81.368716] R10: 00007f2818f77b60 R11: 0000000000000246 R12: 
0000559f8bfae000
[   81.368717] R13: 0000000000000000 R14: 0000000000000000 R15: 
000000000000000a
[   81.368734]  </TASK>

The same BUG fires from blk_add_trace_plug(), blk_add_trace_unplug(),
and blk_add_trace_rq() paths as well.

Fix by wrapping the tracing_record_cmdline() call with
preempt_disable()/preempt_enable().  This is a best-effort "record
the comm string for this PID" operation and briefly disabling
preemption around it is both safe and correct.

With this patch now blktests for blktrace pass :-

blktests (master) # ./check blktrace
blktrace/001 (blktrace zone management command tracing) [passed]
     runtime  3.652s  ...  3.649s
blktrace/002 (blktrace ftrace corruption with sysfs trace)  [passed]
     runtime  0.437s  ...  0.389s
blktests (master) #


Signed-off-by: Chaitanya Kulkarni <kch at nvidia.com>
---
  kernel/trace/blktrace.c | 2 ++
  1 file changed, 2 insertions(+)

diff --git a/kernel/trace/blktrace.c b/kernel/trace/blktrace.c
index 3b7c102a6eb3..488552036583 100644
--- a/kernel/trace/blktrace.c
+++ b/kernel/trace/blktrace.c
@@ -383,7 +383,9 @@ static void __blk_add_trace(struct blk_trace *bt, 
sector_t sector, int bytes,
      cpu = raw_smp_processor_id();

      if (blk_tracer) {
+        preempt_disable_notrace();
          tracing_record_cmdline(current);
+        preempt_enable_notrace();

          buffer = blk_tr->array_buffer.buffer;
          trace_ctx = tracing_gen_ctx_flags(0);
-- 
2.39.5




More information about the Linux-nvme mailing list