linux-kernel - Re: [bug report, 6.1.52] /proc/loadavg shows incorrect values

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <CAN2Y7hzCk4FoV_aZzDEXbn0nfqLTeYKbWxfKo7WS6225y1gbcQ@mail.gmail.com>
Date: Mon, 24 Mar 2025 10:31:02 +0800
From: ying chen <yc1082463@...il.com>
To: Steven Rostedt <rostedt@...dmis.org>
Cc: "mingo@...hat.com" <mingo@...hat.com>, peterz@...radead.org, juri.lelli@...hat.com, 
	vincent.guittot@...aro.org, linux-kernel@...r.kernel.org, 
	dietmar.eggemann@....com, bsegall@...gle.com, mgorman@...e.de, 
	bristot@...hat.com, vschneid@...hat.com
Subject: Re: [bug report, 6.1.52] /proc/loadavg shows incorrect values

On Mon, Mar 24, 2025 at 2:04 AM Steven Rostedt <rostedt@...dmis.org> wrote:
>
> On Sun, 23 Mar 2025 20:45:51 +0800
> ying chen <yc1082463@...il.com> wrote:
>
> > Hello everyone. Have you ever encountered a similar situation?
> >
> > On Tue, Mar 18, 2025 at 9:54 PM ying chen <yc1082463@...il.com> wrote:
> > >
> > > Hello all,
> > >
> > > In our production environment, "cat /proc/loadavg" shows incorrect
> > > huge values. The kernel version is 6.1.52. So far, at least four such
> > > cases have been found. It seems to be a kernel bug.
> > >
> > > ~$ cat /proc/loadavg
> > > 4294967392.49 4294967395.80 4294967395.83 87/16100 2341720
> > >
> > > top output is below:
> > >
> > > top - 21:12:13 up 191 days, 20:50,  1 user,  load average:
> > > 4294967397.45, 4294967396.82, 4294967396.15
>
>   4294967397 = 0x100000065
>
> Which looks like some calculation overflowed.
>
> 191 day uptime is quite long (I reboot to update my kernel every
> month). Perhaps there's something there that caused an overflow.
>
> Interestingly in 5.14, some values were converted from long to int. Not
> sure if there was anything there that could have caused this.
>
> Just something to look at.
>
> -- Steve
Do you mean the patch "sched: Make multiple runqueue task counters 32-bit"?
I've also noticed this patch, but I'm not sure if this problem was
introduced by this patch.
>
> > > Tasks: 2388 total,   3 running, 1208 sleeping,   0 stopped,   0 zombie
> > > %Cpu(s): 27.9 us,  6.7 sy,  0.0 ni, 57.3 id,  0.5 wa,  1.7 hi,  5.8 si,  0.0 st
> > > KiB Mem : 99966995+total, 56704217+free, 22655678+used, 20607096+buff/cache
> > > KiB Swap:        0 total,        0 free,        0 used. 68817177+avail Mem
>