Re: [PATCH] core_pattern: add CPU specifier

[Oleksandr Natalenko]

Hello.

On čtvrtek 8. září 2022 0:00:43 CEST Eric W. Biederman wrote:
> Oleg Nesterov <oleg@xxxxxxxxxx> writes:
> 
> > On 09/07, Oleksandr Natalenko wrote:
> >>
> >> The advantage of having CPU recorded in the file name is that
> >> in case of multiple cores one can summarise them with a simple
> >> ls+grep without invoking a fully-featured debugger to find out
> >> whether the segfaults happened on the same CPU.
> >
> > Besides, if you only need to gather the statistics about the faulting
> > CPU(s), you do not even need to actually dump the the core. For example,
> > something like
> >
> > 	#!/usr/bin/sh
> >
> > 	echo $* >> path/to/coredump-stat.txt
> >
> > and
> > 	echo '| path-to-script-above %C' >/proc/sys/kernel/core_pattern
> >
> > can help.
> 
> So I am confused.  I thought someone had modified print_fatal_signal
> to print this information.  Looking at the code now I don't see it,
> but perhaps that is in linux-next somewhere.

That's a different story that gets solved here: [1] [2].

[1] https://git.kernel.org/pub/scm/linux/kernel/git/tip/tip.git/commit/?h=x86/cpu&id=c926087eb38520b268515ae1a842db6db62554cc
[2] https://lore.kernel.org/lkml/20220811024903.178925-1-ira.weiny@xxxxxxxxx/

> That would seem to be the really obvious place to put this and much
> closer to the original fault so we ware more likely to record the
> cpu on which things actually happened on.
> 
> If we don't care about the core dump just getting the information in
> syslog where it can be analyzed seems like the thing to do.
> 
> For a developers box putting it in core pattern makes sense, isn't a
> hinderance to use.  For anyone else's box the information needs to come
> out in a way that allows automated tools to look for a pattern.
> Requiring someone to take an extra step to print the information seems
> a hinderance to automated tools doing the looking.
> 
> Eric
> 
> 


-- 
Oleksandr Natalenko (post-factum)
Principal Software Maintenance Engineer