LinuxPPC-Dev Archive on lore.kernel.org
 help / color / mirror / Atom feed
* [BUG] Interrupt timing warnings in pSeries guest on PowerNV host
@ 2026-08-21 16:38 Timothy Pearson
  2026-08-27 17:21 ` Shrikanth Hegde
  0 siblings, 1 reply; 2+ messages in thread
From: Timothy Pearson @ 2026-08-21 16:38 UTC (permalink / raw)
  To: linuxppc-dev; +Cc: Ritesh Harjani

On a POWER9/PowerNV host, when running a pSeries guest using QEMU, sporadic interrupt timing warnings are observed.  This is on kernel 6.12 (Debian Stable) on both host and guest, system load is moderate, the host CPUs are not pegged at 100% at any time.  The guest has only one virtual CPU core assigned.

[11622.031047] hrtimer: interrupt took 12354614 ns
[46723.139964] hrtimer: interrupt took 8522695 ns

This *may* be related to the dropped / late interrupts I have reported in earlier messages, but it could also be a separate issue.  QEMU version is 7.2.22.


^ permalink raw reply	[flat|nested] 2+ messages in thread

* Re: [BUG] Interrupt timing warnings in pSeries guest on PowerNV host
  2026-08-21 16:38 [BUG] Interrupt timing warnings in pSeries guest on PowerNV host Timothy Pearson
@ 2026-08-27 17:21 ` Shrikanth Hegde
  0 siblings, 0 replies; 2+ messages in thread
From: Shrikanth Hegde @ 2026-08-27 17:21 UTC (permalink / raw)
  To: Timothy Pearson, linuxppc-dev; +Cc: Ritesh Harjani

Hi Timothy,

On 8/21/26 10:08 PM, Timothy Pearson wrote:
> On a POWER9/PowerNV host, when running a pSeries guest using QEMU, sporadic interrupt timing warnings are observed.  This is on kernel 6.12 (Debian Stable) on both host and guest, system load is moderate, the host CPUs are not pegged at 100% at any time.  The guest has only one virtual CPU core assigned.
> 
> [11622.031047] hrtimer: interrupt took 12354614 ns
> [46723.139964] hrtimer: interrupt took 8522695 ns
> 

         /*
          * The next timer was already expired due to:
          * - tracing
          * - long lasting callbacks
          * - being scheduled away when running in a VM

This message used to come if the timer expiry was in past and it was tried 3 times and still
can't catch up.  Very likely it was due to guest being scheduled away.

This behavior has changed now and this particular trace has been removed.
https://lore.kernel.org/all/20260224163430.870639266@kernel.org/

> This *may* be related to the dropped / late interrupts I have reported in earlier messages, but it could also be a separate issue.  QEMU version is 7.2.22.



^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2026-08-27 17:22 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2026-08-21 16:38 [BUG] Interrupt timing warnings in pSeries guest on PowerNV host Timothy Pearson
2026-08-27 17:21 ` Shrikanth Hegde

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox