Question about watchdog

[valdis.kletnieks at vt.edu]

On Tue, 03 Jul 2018 10:28:48 +0800, you said:

> yes ! but there are days when people have to do a real-time work with no
> delay in kernel space ,i think i can feed the dog
> a percpu variable "nmi_touch" looks suspicious，thanks！

1) The Linux kernel community convention is t use 'reply all' so that a copy
goes back to the list, so that all may read, learn, and comment...

2) If you're hanging a CPU with something for long enough to trigger a watchdog
event, you've *already* *broken* real-time processing on that CPU.

Read that second sentence again, and repeat until it sinks in.

If the watchdog is firing for a CPU because you've been running a tight loop
on it for *that* long, *that CPU is by definition not able to do realtime in a Linux
context*.

-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 486 bytes
Desc: not available
URL: <http://lists.kernelnewbies.org/pipermail/kernelnewbies/attachments/20180703/43a8bcf9/attachment.sig>