Bug#1087981: linux-image-6.1.0-27-amd64: detected stalls in kernel log, system very slow on IO (regression)
Hi Willi,
On Sat, Mar 22, 2025 at 07:47:14PM +0100, Willi Mann wrote:
> Hi Salvatore,
>
> sorry, I somehow missed your mail.
>
> Am 21.03.25 um 21:07 schrieb Salvatore Bonaccorso:
>
> > > While we had no followup after your last email, I wonder now if you
> > > still experience the issues with the most recent 6.1.y kernel
> > > available in Debian (6.1.128-1).
> > >
> > > Can you report back? Can you test on that machine as well newer
> > > kernels ideally from unstable (6.12.16-1) or experimental
> > > (6.13.4-1~exp1)?
> >
> > Any news here?
>
> With 6.1.128-1, there are still backtraces, though I don't experience any
> slowness. I could only retrieve them from dmesg output as the kernel
> logfiles containing them are already rotated. Also, I removed noise
> unrelated to to the crash. See attachment.
>
> I'll install a newer kernel tomorrow and let you know the outcome.
Ack thanks.
Btw, in a earlier log which was more complete from the boot on, I
notice
[...]
[ 2866.062346] sh (1675): drop_caches: 3
[ 6466.150805] sh (2315): drop_caches: 3
[10066.236130] sh (2955): drop_caches: 3
[13666.343713] sh (3732): drop_caches: 3
[17266.432101] sh (4393): drop_caches: 3
[20866.523958] sh (4975): drop_caches: 3
[24466.622511] sh (5717): drop_caches: 3
[26006.697987] rcu: INFO: rcu_preempt detected stalls on CPUs/tasks:
[26016.962347] rcu: 1-...0: (18 ticks this GP) idle=a5ac/1/0x4000000000000000 softirq=94076/94077 fqs=908
[26016.962408] (detected by 0, t=5252 jiffies, g=164937, q=845 ncpus=2)
[26016.962414] Sending NMI from CPU 0 to CPUs 1:
[26024.942628] NMI backtrace for cpu 1
[26024.942634] CPU: 1 PID: 4882 Comm: kworker/u4:2 Tainted: G E 6.1.0-28-amd64 #1 Debian
+6.1.118-1~test1
[...]
before the actual stalls begin. what is doing that and droppying
caches?
thans for your time doing the experiments.
Regards,
Salvatore
Reply to: