Hi Sebastian.
Thanks for the report.
+cc mukesh
On 9/22/26 8:36 PM, Sebastian Andrzej Siewior wrote:
Hi,
I've seen similar reports such as
https://lore.kernel.org/all/[email protected]/
Yes, we had taken a look at them. There is work being done, but patches
haven't been posted since we were burning cycles to fix the functional issues
even without lockdep on generic entry.
but nothing on recent one. I am not sure if this is the same issue or
What do you mean nothing on recent one? You don't see the lockdep warning?
something different. This happens during boot:
| ------------[ cut here ]------------
| DEBUG_LOCKS_WARN_ON(lockdep_hardirq_context())
| WARNING: kernel/locking/lockdep.c:4448 at
lockdep_hardirqs_on_prepare+0x22c/0x2e0, CPU#0: swapper/0/0
| Modules linked in:
| CPU: 0 UID: 0 PID: 0 Comm: swapper/0 Not tainted 7.3.0-rc4 #6 PREEMPT(lazy)
| Hardware name: IBM pSeries (emulated by qemu) Power11 (architected) 0x821200
0xf000007 of:SLOF,HEAD pSeries
| NIP: c0000000001ec72c LR: c0000000001ec728 CTR: c000000000e1ab70
| REGS: c0000000017531e0 TRAP: 0700 Not tainted (7.3.0-rc4)
| MSR: 8000000002021033 <SF,VEC,ME,IR,DR,RI,LE> CR: 48028228 XER: 00000000
| CFAR: c00000000011ee84 IRQMASK: 3
| GPR00: c0000000001ec728 c000000001753480 c000000001248000 000000000000002e
| GPR04: c0000000015ee8d8 c0000000015ee9c8 0000000000000001 0000000000000080
| GPR08: c0000000015ee7b8 0000000001000003 c000000001581400 0000000048028228
| GPR12: c000000001788228 c000000002340000 c0000000005bbd50 c000000000ea30f0
| GPR16: c0000000016cec98 c0000000011a1518 c000000003327658 0000000000000001
| GPR20: c0000000016c9620 c0000000015e3c80 c0000000016c9670 0000000000000000
| GPR24: 0000000000000000 0000000000000001 0000000000000000 c000000001788228
| GPR28: c000000001380580 c000000000027644 0000000000000000 c000000001387160
| NIP [c0000000001ec72c] lockdep_hardirqs_on_prepare+0x22c/0x2e0
| LR [c0000000001ec728] lockdep_hardirqs_on_prepare+0x228/0x2e0
| Call Trace:
| [c000000001753480] [c0000000001ec728] lockdep_hardirqs_on_prepare+0x228/0x2e0
(unreliable)
| [c000000001753500] [c000000000331c90] trace_hardirqs_on+0x90/0x130
| [c000000001753540] [c000000000e257f4] irqentry_exit+0x554/0x10b0
| [c000000001753600] [c000000000027644] timer_interrupt+0x254/0x5c0
| [c0000000017536c0] [c00000000003a2c0] __replay_soft_interrupts+0x140/0x390
| [c000000001753880] [c00000000003a614] arch_local_irq_restore+0x104/0x1f0
| [c0000000017538b0] [c0000000001eed9c] lock_acquire+0x19c/0x440
| [c0000000017539b0] [c000000000541024] __kmalloc_cache_noprof+0x114/0x630
| [c000000001753a90] [c0000000004970c8] shrinker_alloc+0x38/0xe0
| [c000000001753ac0] [c0000000005bf374] sget_fc+0x364/0x940
| [c000000001753c00] [c0000000005bfd78] get_tree_nodev+0x38/0xf0
| [c000000001753c40] [c000000000649060] nullfs_fs_get_tree+0x20/0x40
| [c000000001753c60] [c0000000005ba01c] vfs_get_tree+0x3c/0x110
| [c000000001753ce0] [c0000000006043a0] fc_mount+0x20/0xc0
| [c000000001753d20] [c000000000604544] vfs_kern_mount.part.0+0x104/0x120
| [c000000001753d80] [c0000000006045bc] kern_mount+0x2c/0x60
| [c000000001753da0] [c000000001296f20] mnt_init+0x3c8/0x50c
| [c000000001753e80] [c000000001296610] vfs_caches_init+0x100/0x144
| [c000000001753f30] [c00000000125442c] start_kernel+0xa4c/0xabc
| [c000000001753fe0] [c00000000000d688] start_here_common+0x1c/0x20
| Code: 4182ff80 3d220054 3929106c 81290000 2c090000 4082ff6c 3c82ffe7 3c62ffe7
38846858 3863dfd8 4bf32645 60000000 <0fe00000> 4bffff4c 60000000 60000000
| irq event stamp: 472
| hardirqs last enabled at (471): [<c000000000e39034>]
_raw_spin_unlock_irqrestore+0xa4/0xb0
| hardirqs last disabled at (472): [<c00000000003a5b8>]
arch_local_irq_restore+0xa8/0x1f0
| softirqs last enabled at (10): [<c00000000012db60>]
handle_softirqs+0x670/0x6d0
| softirqs last disabled at (3): [<c000000000016f88>]
do_softirq_own_stack+0x38/0x50
| ---[ end trace 0000000000000000 ]---
Sebastian
IIRC, it is tricky to fix since softreplay is a unique thing on its own
and tracking right context gets tricky.
Mukesh had spent some time on it. Lets wait to hear from him.
Plus we had some more lockdep warnings too. There is effort ongoing to fix
them. Hopefully soon.
