[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <aXyVwEocs-uvd69T@gmail.com>
Date: Fri, 30 Jan 2026 03:33:25 -0800
From: Breno Leitao <leitao@...ian.org>
To: Andrew Morton <akpm@...ux-foundation.org>
Cc: bhe@...hat.com, linux-kernel@...r.kernel.org, kernel-team@...a.com,
kexec@...ts.infradead.org, dyoung@...hat.com, tony.luck@...el.com,
xueshuai@...ux.alibaba.com, vgoyal@...hat.com, zhiquan1.li@...el.com, olja@...a.com
Subject: Re: [PATCH] vmcore_info: expose hardware error recovery statistics
via sysfs
Hello Andrew,
On Thu, Jan 29, 2026 at 02:28:01PM -0800, Andrew Morton wrote:
> On Thu, 29 Jan 2026 05:34:10 -0800 Breno Leitao <leitao@...ian.org> wrote:
>
> > Example output:
> > hwerr_recovery:
> > cpu: 0 (0)
> > memory: 2 (1738148257)
> > pci: 1 (1738147000)
> > cxl: 0 (0)
> > other: 0 (0)
>
> sysfs rules (which are widely ignored) say "one value per file".
>
> As a compromise the above could be squished into a single line. Harder
> for humans to read, but it sounds like that isn't the expected use case.
I'm fine with consolidating into a single line, though it would mean
removing the timestamp of the last recovery event. Since the primary
use case is tracking event counts in a time series, this is acceptable.
My proposal:
# cat /sys/kernel/hwerr_stats
cpu:0 memory:2 pci:1 cxl:0 other:0
I suggest keeping the field names rather than dropping them entirely.
While it would make the output more compact, having explicit names
makes the format more maintainable when adding new recovery types in
the future.
Without names, it would look like:
# cat /sys/kernel/hwerr_stats
0 2 1 0 0
What would you say?
Thanks for reviewing it,
--breno
Powered by blists - more mailing lists