Watchdog remote CPU read timed out

Watchdog remote CPU read timed out

[Michal Pecio]

Hi Thomas,

This is another warning I'm getting on recent 7.x kernels, apparently
from new code you wrote a few months ago.

Increasing WATCHDOG_DEFAULT_TIMEOUT_NS to 100us makes it go away, at
least for a few minutes (I will try longer uptime later).

With a few debug prints, I found that watchdog_wait_seq() fails every
few seconds, usually on CPU 0 but sporadically on others too.

Example output below. Is this something to worry about / fix?

Thanks,
Michal

[ 4307.502933] clocksource: watchdog_wait_seq done on cpu 0
[ 4307.502946] clocksource: watchdog_wait_seq done on cpu 1
[ 4307.502950] clocksource: watchdog_wait_seq done on cpu 0
[ 4307.502952] clocksource: watchdog_wait_seq done on cpu 1
[ 4307.502954] clocksource: watchdog_wait_seq done on cpu 0
[ 4307.502956] clocksource: watchdog_wait_seq done on cpu 1
[ 4307.502958] clocksource: watchdog_wait_seq done on cpu 0
[ 4307.502960] clocksource: watchdog_wait_seq done on cpu 1
[ 4308.006997] clocksource: watchdog_wait_seq done on cpu 0
[ 4308.007028] clocksource: watchdog_wait_seq done on cpu 0
[ 4308.007031] clocksource: watchdog_wait_seq done on cpu 0
[ 4308.007028] clocksource: WD_CPU_TIMEOUT in watchdog_wait_seq cpu 2
[ 4308.007034] clocksource: watchdog_wait_seq done on cpu 0
[ 4308.502975] clocksource: watchdog_wait_seq done on cpu 0
[ 4308.502991] clocksource: watchdog_wait_seq done on cpu 3
[ 4308.503021] clocksource: WD_CPU_TIMEOUT in watchdog_wait_seq cpu 0
[ 4308.503048] clocksource: watchdog_wait_seq done on cpu 3
[ 4308.503050] clocksource: watchdog_wait_seq done on cpu 3
[ 4308.503051] clocksource: watchdog_wait_seq done on cpu 3

Re: Watchdog remote CPU read timed out

[Thomas Gleixner]

On Sun, Aug 30 2026 at 14:16, Michal Pecio wrote:
> This is another warning I'm getting on recent 7.x kernels, apparently
> from new code you wrote a few months ago.
>
> Increasing WATCHDOG_DEFAULT_TIMEOUT_NS to 100us makes it go away, at
> least for a few minutes (I will try longer uptime later).
>
> With a few debug prints, I found that watchdog_wait_seq() fails every
> few seconds, usually on CPU 0 but sporadically on others too.
>
> Example output below. Is this something to worry about / fix?

Nothing to worry about. The message is mostly informal.

Thanks,

        tglx