Re: [PATCH 0/6] locking/rtmutex: Avoid PI state recursion through sched_submit_work()

[Peter Zijlstra <peterz@infradead.org>]

On Wed, Aug 16, 2023 at 10:58:26AM +0200, Sebastian Andrzej Siewior wrote:
> On 2023-08-15 18:15:57 [+0200], Peter Zijlstra wrote:
> > N/m - 0day found a problem. Futex-PI trips the rt_mutex_schedule()
> > assertion for not passing through rt_mutex_pre_schedule().
> > 
> > I'll go try and untangle that...
> Is this the same as

Not the same -- this is namespace_lock(), right? That's a regular rwsem
afaict and that *should* be good. Clearly I messed something up.

Thanks!

> | ------------[ cut here ]------------
> | WARNING: CPU: 3 PID: 995 at kernel/sched/core.c:7155 rt_mutex_schedule+0x52/0x60
> | Modules linked in:
> | CPU: 3 PID: 995 Comm: mount Tainted: G        W          6.5.0-rc6-rt2+ #228
> | Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
> | RIP: 0010:rt_mutex_schedule+0x52/0x60
> | Code: 00 00 00 e8 b0 80 ff ff 31 ff e8 a9 99 cb 00 bf 01 00 00 00 e8 3f 6d ff ff 48 8b 03 a9 08 00 08 00 75 db 5b e9 6f 82 cc 00 90 <0f> 0b 90 eb c6 66 0f 1f 84 00 00 0
> | RSP: 0018:ffffc90001043e28 EFLAGS: 00010246
> | RAX: ffff888107ab8040 RBX: 0000000000000202 RCX: 0000000000000000
> | RDX: 0000000000000000 RSI: ffffffff82460bfd RDI: 00000000ffffffff
> | RBP: ffffffff827e8fe0 R08: 0000000000000001 R09: 0000000000000001
> | R10: 0000000000000000 R11: 0000000000000001 R12: ffffffff827e8fe8
> | R13: 0000000000000002 R14: 0000000000000000 R15: ffff888107ab8040
> | FS:  00007fb36281e840(0000) GS:ffff88817b4c0000(0000) knlGS:0000000000000000
> | CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> | CR2: 00007fb3629702a0 CR3: 0000000108210000 CR4: 00000000003506e0
> | Call Trace:
> |  <TASK>
> |  ? __warn+0x90/0x200
> |  ? rt_mutex_schedule+0x52/0x60
> |  ? report_bug+0x1c5/0x1f0
> |  ? handle_bug+0x3b/0x70
> |  ? exc_invalid_op+0x13/0x60
> |  ? asm_exc_invalid_op+0x16/0x20
> |  ? rt_mutex_schedule+0x52/0x60
> |  rwbase_write_lock+0xc1/0x230
> |  do_lock_mount+0x4c/0x200
> |  ? __x86_return_thunk+0x5/0x10
> |  path_mount+0x8a7/0xb00
> |  __x64_sys_mount+0xf1/0x140
> |  do_syscall_64+0x44/0x90
> |  entry_SYSCALL_64_after_hwframe+0x74/0xde
> 
> Sebastian