* Possible scriping bug in rcutorture/kvm.sh
@ 2025-04-10 13:59 Joel Fernandes
2025-04-10 14:31 ` Paul E. McKenney
0 siblings, 1 reply; 3+ messages in thread
From: Joel Fernandes @ 2025-04-10 13:59 UTC (permalink / raw)
To: Paul E. McKenney, rcu
Hello,
Currently with my ARM machine of 288 CPUs, I am only seeing 56 of them used at a
time. Any ideas if this is a bug or I am doing something wrong?
I started a batch of 36 SRCU-P runs, which is exactly 1 batch:
$ sudo tools/testing/selftests/rcutorture/bin/kvm.sh --duration 120 --allcpus
--configs "10*SRCU-P" --dryrun batches
1 SRCU-P 8
1 SRCU-P.2 8
1 SRCU-P.3 8
1 SRCU-P.4 8
1 SRCU-P.5 8
1 SRCU-P.6 8
1 SRCU-P.7 8
1 SRCU-P.8 8
1 SRCU-P.9 8
1 SRCU-P.10 8
But, when I run it via:
$ sudo tools/testing/selftests/rcutorture/bin/kvm.sh --duration 240 --cpus 288
--configs "36*SRCU-P"
ps shows me only 7 qemu instances of 8 CPUs each. However, weirdly it is only 1
thread for each instance (thus its not even 56 CPUs, but rather 7 or so).
1496915 root 20 0 1684276 256996 66816 R 90.9 0.2 8:38.06
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
1496913 root 20 0 1684276 256996 66816 R 89.7 0.2 8:50.53
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
1496929 root 20 0 1684276 256996 66816 R 89.3 0.2 8:44.53
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
1496938 root 20 0 1684276 256996 66816 R 82.8 0.2 8:46.03
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
1496922 root 20 0 1684276 256996 66816 R 82.1 0.2 8:57.83
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
1496908 root 20 0 1684276 256996 66816 R 81.5 0.2 8:50.59
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
1496920 root 20 0 1684276 256996 66816 R 63.6 0.2 8:51.46
qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
-machine virt,gic-version=host -cpu host+
Thanks.
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Possible scriping bug in rcutorture/kvm.sh
2025-04-10 13:59 Possible scriping bug in rcutorture/kvm.sh Joel Fernandes
@ 2025-04-10 14:31 ` Paul E. McKenney
2025-04-10 14:46 ` Joel Fernandes
0 siblings, 1 reply; 3+ messages in thread
From: Paul E. McKenney @ 2025-04-10 14:31 UTC (permalink / raw)
To: Joel Fernandes; +Cc: rcu
On Thu, Apr 10, 2025 at 09:59:07AM -0400, Joel Fernandes wrote:
> Hello,
> Currently with my ARM machine of 288 CPUs, I am only seeing 56 of them used at a
> time. Any ideas if this is a bug or I am doing something wrong?
I regularly run with 80 CPUs, but perhaps the script's check for the
number of CPUs is broken on ARM? It currently relies on this:
getconf _NPROCESSORS_ONLN
(See identify_qemu_vcpus() in functions.sh.)
What should it be doing instead on ARM?
Huh. 56 does not go evenly into 288, either. Search me!
Thanx, Paul
> I started a batch of 36 SRCU-P runs, which is exactly 1 batch:
>
> $ sudo tools/testing/selftests/rcutorture/bin/kvm.sh --duration 120 --allcpus
> --configs "10*SRCU-P" --dryrun batches
> 1 SRCU-P 8
>
>
> 1 SRCU-P.2 8
>
> 1 SRCU-P.3 8
>
> 1 SRCU-P.4 8
>
> 1 SRCU-P.5 8
>
> 1 SRCU-P.6 8
>
> 1 SRCU-P.7 8
>
> 1 SRCU-P.8 8
>
> 1 SRCU-P.9 8
>
> 1 SRCU-P.10 8
>
> But, when I run it via:
> $ sudo tools/testing/selftests/rcutorture/bin/kvm.sh --duration 240 --cpus 288
> --configs "36*SRCU-P"
>
> ps shows me only 7 qemu instances of 8 CPUs each. However, weirdly it is only 1
> thread for each instance (thus its not even 56 CPUs, but rather 7 or so).
>
> 1496915 root 20 0 1684276 256996 66816 R 90.9 0.2 8:38.06
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
> 1496913 root 20 0 1684276 256996 66816 R 89.7 0.2 8:50.53
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
> 1496929 root 20 0 1684276 256996 66816 R 89.3 0.2 8:44.53
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
> 1496938 root 20 0 1684276 256996 66816 R 82.8 0.2 8:46.03
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
> 1496922 root 20 0 1684276 256996 66816 R 82.1 0.2 8:57.83
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
> 1496908 root 20 0 1684276 256996 66816 R 81.5 0.2 8:50.59
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
> 1496920 root 20 0 1684276 256996 66816 R 63.6 0.2 8:51.46
> qemu-system-aarch64 -enable-kvm -display none -no-reboot -smp 8 -net none
> -machine virt,gic-version=host -cpu host+
>
> Thanks.
>
>
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: Possible scriping bug in rcutorture/kvm.sh
2025-04-10 14:31 ` Paul E. McKenney
@ 2025-04-10 14:46 ` Joel Fernandes
0 siblings, 0 replies; 3+ messages in thread
From: Joel Fernandes @ 2025-04-10 14:46 UTC (permalink / raw)
To: paulmck; +Cc: rcu
On 4/10/2025 10:31 AM, Paul E. McKenney wrote:
> On Thu, Apr 10, 2025 at 09:59:07AM -0400, Joel Fernandes wrote:
>> Hello,
>> Currently with my ARM machine of 288 CPUs, I am only seeing 56 of them used at a
>> time. Any ideas if this is a bug or I am doing something wrong?
>
> I regularly run with 80 CPUs, but perhaps the script's check for the
> number of CPUs is broken on ARM? It currently relies on this:
>
> getconf _NPROCESSORS_ONLN
>
> (See identify_qemu_vcpus() in functions.sh.)
>
> What should it be doing instead on ARM?
I am guessing nothing different because it still gives 288:
$ getconf _NPROCESSORS_ONLN
288
>
> Huh. 56 does not go evenly into 288, either. Search me!
I think its failing because its only running the first SRCU-P for me and failing
for the rest in the batch because:
I see this for the rest of the runs: "Initial build failed, not running KVM":
I think its because this line is failing for ARM because it does not have a
bzImage (it instead has Image):
if test "$base_resdir" != "$resdir" && test -f $base_resdir/bzImage && test -f
$base_resdir/vmlinux
I'll try this and see if it works (sorry my thunderbird is wrapping the diff..):
diff --git a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
index ad79784e552d..957800c9ffba 100755
--- a/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
+++ b/tools/testing/selftests/rcutorture/bin/kvm-test-1-run.sh
@@ -73,7 +73,7 @@ config_override_param "$config_dir/CFcommon.$(uname -m)" KcList \
cp $T/KcList $resdir/ConfigFragment
base_resdir=`echo $resdir | sed -e 's/\.[0-9]\+$//'`
-if test "$base_resdir" != "$resdir" && test -f $base_resdir/bzImage && test -f
$base_resdir/vmlinux
+if test "$base_resdir" != "$resdir" && (test -f $base_resdir/bzImage || test -f
$base_resdir/Image) && test -f $base_resdir/vmlinux
then
# Rerunning previous test, so use that test's kernel.
QEMU="`identify_qemu $base_resdir/vmlinux`"
Thanks.
^ permalink raw reply related [flat|nested] 3+ messages in thread
end of thread, other threads:[~2025-04-10 14:46 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2025-04-10 13:59 Possible scriping bug in rcutorture/kvm.sh Joel Fernandes
2025-04-10 14:31 ` Paul E. McKenney
2025-04-10 14:46 ` Joel Fernandes
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox