Re: tpm2_createprim blocked for more than 122 seconds

[Jarkko Sakkinen <jarkko@kernel.org>]

On Tue, Aug 16, 2022 at 09:29:29AM -0700, Anatol Pomozov wrote:
> Hi
> 
> On Tue, Aug 16, 2022 at 7:27 AM Jarkko Sakkinen <jarkko@kernel.org> wrote:
> >
> > On Mon, Aug 15, 2022 at 06:07:00PM -0700, Anatol Pomozov wrote:
> > > Hello folks,
> > >
> > > I am running a tool that performs some basic operations (seal/unseal)
> > > with TPM. Most of my machines work fine. Though once in a while I see
> > > the following problem with my tool:
> > >
> > > [7955220.676259] INFO: task tpm2_createprim:258741 blocked for more
> > > than 122 seconds.
> > > [7955220.683837]       Not tainted 5.10.89-t3.el7.twitter.x86_64 #1
> > > [7955220.689848] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs"
> > > disables this message.
> > > [7955220.697846] task:tpm2_createprim state:D stack:    0 pid:258741
> > > ppid:258724 flags:0x10000080
> > > [7955220.706454] Call Trace:
> > > [7955220.709088]  __schedule+0x266/0x840
> > > [7955220.712756]  schedule+0x46/0xb0
> > > [7955220.716073]  schedule_preempt_disabled+0xa/0x10
> > > [7955220.720780]  __mutex_lock.constprop.0+0x123/0x440
> > > [7955220.725660]  ? handle_mm_fault+0x168b/0x1ae0
> > > [7955220.730105]  tpm_try_get_ops+0x3c/0x90
> > > [7955220.734035]  tpm_common_write+0xfb/0x1c0
> > > [7955220.738141]  vfs_write+0xc0/0x260
> > > [7955220.741636]  ksys_write+0x4f/0xc0
> > > [7955220.745134]  do_syscall_64+0x33/0x40
> > > [7955220.748886]  entry_SYSCALL_64_after_hwframe+0x44/0xa9
> > > [7955220.754111] RIP: 0033:0x7fbd13d446e0
> > > [7955220.757864] RSP: 002b:00007ffc77d004f8 EFLAGS: 00000246 ORIG_RAX:
> > > 0000000000000001
> > > [7955220.765603] RAX: ffffffffffffffda RBX: 0000000000000000 RCX:
> > > 00007fbd13d446e0
> > > [7955220.772910] RDX: 0000000000000043 RSI: 00005626520d33e8 RDI:
> > > 0000000000000003
> > > [7955220.780215] RBP: 00005626520d22f0 R08: 00007ffc77d004c4 R09:
> > > 00007ffc77d006b0
> > > [7955220.787519] R10: 00007ffc77cfff60 R11: 0000000000000246 R12:
> > > 00005626520d33e8
> > > [7955220.794827] R13: 0000000000000043 R14: 0000000000000000 R15:
> > > 00007ffc77d00780
> > > [7955290.222553] tpm tpm0: Operation Timed out
> > > [7955292.226513] tpm tpm0: tpm_try_transmit: send(): error -62
> > > [8311409.491438] tpm tpm0: tpm_try_transmit: send(): error -62
> > > [8311412.625426] tpm tpm0: tpm_try_transmit: send(): error -62
> > >
> > >
> > >
> > > The kernel is 5.10.89; CentOS7. I do not know how to reproduce the
> > > problem reliably. I know if I reboot the host it gets "fixed".
> > >
> > > Is it a known problem? If not what would be the best way to debug it?
> >
> > We really cannot give feedback on non-mainline issues.
> 
> The kernel that is used here is 5.10.89 compiled from the stable repo.
> So it is as mainline as a production kernel can be.
> 
> I also saw the same issues with 5.10.113 (also compiled from stable repo).

I'm sorry got the wrong impression from your original email.

It's not really a problem that can be fixed though.
Primary key creation just takes enough time to trigger
this warning.

For pure interest, do you get the same result with O_NONBLOCK?

BR, Jarkko