Re: Problems with rcutorture on ppc64le: allmodconfig(2) and other failures

From: Paul E. McKenney
Date: Wed Jan 26 2022 - 20:20:13 EST

Next message: Wei Yang: "[PATCH] mm/memory_hotplug: build zonelist for managed_zone"
Previous message: Tim Chen: "Re: [PATCH v2 2/2] sched/fair: Scan cluster before scanning LLC in wake-up path"
In reply to: Zhouyi Zhou: "Re: Problems with rcutorture on ppc64le: allmodconfig(2) and other failures"
Next in thread: Zhouyi Zhou: "Re: Problems with rcutorture on ppc64le: allmodconfig(2) and other failures"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

On Thu, Jan 27, 2022 at 08:57:02AM +0800, Zhouyi Zhou wrote:
> Hi Paul
>
> Thank you for your perfect patch!
>
> After apply your patch, I invoke
> "tools/testing/selftests/rcutorture/bin/torture.sh --duration 10" on
> next-20220125. we succeed in allmodconfig!

Very good!

> Following are the test logs (154.223.142.244 is my own internet cloud
> server maintained by myself, I believe it is safe to access the text
> content via web, but not so sure)
>
> 1. torture log
> http://154.223.142.244/logs/20220127/2022.01.26-19.43.04.log (there
> are many Chinese words "Thursday" in it because I forget to set
> terminal locale to English, I am sorry), I excerpt some content here:

Not a problem, Google Translate decided that this was the corresponding
English:

Thursday January 27th, 2022

I really do need to get my browser to do the Chinese characters. My
vim displays them just fine. :-/

> SUCCESSES:
> allmodconfig(0)
> tools/testing/selftests/rcutorture/res/2022.01.26-19.43.04-torture/allmodconfig
> FAILURES:
> rcutorture(2) /home/zzy/Program/linux-next/linux-next/tools/testing/selftests/rcutorture/res/2022.01.26-19.43.04-torture/results-rcutorture
> Summary: Warnings: 1 Call Traces: 1
> Summary: Warnings: 1 Call Traces: 1
> Summary: Warnings: 1 Call Traces: 1
> Summary: Warnings: 1 Call Traces: 1
> Summary: Warnings: 1 Call Traces: 1
>
> 2. Make.exitcode for allmodeconfig
> http://154.223.142.244/logs/20220127/allmodconfig/Make.exitcode

Zero, which is good. ;-)

> 3. Make.out for allmodeconfig
> http://154.223.142.244/logs/20220127/allmodconfig/Make.out
>
> 4. result of invoking: find 2022.01.26-19.43.04-torture -name
> console.log.diags|xargs -I{} echo "echo {}; cat {}"|xargs -I{} bash -c
> "{}":
> http://154.223.142.244/logs/20220127/consoles.logs.diags
> All of the console.log.diags complain about one thing:
> 2022.01.26-19.43.04-torture/results-refscale-clock/NOPREEMPT/console.log.diags
> [ 0.017721] WARNING: CPU: 0 PID: 0 at
> kernel/trace/trace_events.c:417 trace_event_raw_init+0xea/0x580
> [ 0.017754] Call Trace:
> 2022.01.26-19.43.04-torture/results-rcuscale-kvfree-kasan/TREE/console.log.diags
> [ 0.102911] WARNING: CPU: 0 PID: 0 at
> kernel/trace/trace_events.c:417 trace_event_raw_init+0x113/0x6e0
> [ 0.102992] Call Trace:
> 2022.01.26-19.43.04-torture/results-rcuscale-tasks-tracing-kasan/TREE/console.log.diags
> [ 0.060924] WARNING: CPU: 0 PID: 0 at
> kernel/trace/trace_events.c:417 trace_event_raw_init+0x113/0x6e0
> [ 0.061006] Call Trace:
> I am glad to study these warnings further but they don't seem to be
> related to this patch.

There was at least one RCU bug in this -next, fixed by this commit:

da123016ca8c ("rcu-tasks: Fix computation of CPU-to-list shift counts")

But this warning looks like something else. For one thing, you would
have had to be running on a system with exactly three CPUs to trigger
the RCU bug given the usual rcutorture scenarios.

May I add your Tested-by to this commit?

Thanx, Paul

> Thanks again
> Sincerely
> Zhouyi
>
> On Wed, Jan 26, 2022 at 1:14 PM Paul E. McKenney <paulmck@xxxxxxxxxx> wrote:
> >
> > On Tue, Jan 25, 2022 at 08:50:22PM -0800, Paul E. McKenney wrote:
> > > On Wed, Jan 26, 2022 at 12:09:12PM +0800, Zhouyi Zhou wrote:
> > > > Great Discovery !!
> > > > Hi Paul,
> > > > I track the log
> > > > make[2]: *** No rule to make target
> > > > 'arch/x86/kvm//home/zzy/Program/linux-next/linux-next/tools/testing/selftests/rcutorture/kvm_main.o',
> > >
> > > That does look like a very odd pathname, now that you mention it.
> > >
> > > > needed by 'arch/x86/kvm/kvm.o
> > > > into the file virt/kvm/Makefile.kvm:
> > > > 1# SPDX-License-Identifier: GPL-2.0
> > > > 2#
> > > > 3# Makefile for Kernel-based Virtual Machine module
> > > > 4#
> > > > 5
> > > > 6KVM ?= ../../../virt/kvm
> > > > 7
> > > > 8kvm-y := $(KVM)/kvm_main.o $(KVM)/eventfd.o $(KVM)/binary_stats.o
> > > > 9kvm-$(CONFIG_KVM_VFIO) += $(KVM)/vfio.o
> > > > 10kvm-$(CONFIG_KVM_MMIO) += $(KVM)/coalesced_mmio.o
> > > > 11kvm-$(CONFIG_KVM_ASYNC_PF) += $(KVM)/async_pf.o
> > > > 12kvm-$(CONFIG_HAVE_KVM_IRQ_ROUTING) += $(KVM)/irqchip.o
> > > > 13kvm-$(CONFIG_HAVE_KVM_DIRTY_RING) += $(KVM)/dirty_ring.o
> > > > 14kvm-$(CONFIG_HAVE_KVM_PFNCACHE) += $(KVM)/pfncache.o
> > > > line 6's ?= indicates to set the KDIR variable only if it's not
> > > > set/doesn't have a value.
> > > >
> > > > I guess we have set the variable KVM somewhere before make, I will
> > > > continue tracking it.
> > >
> > > But if you build an allmodconfig kernel by hand afterwards, it works?
> > >
> > > I wonder if I need to use a different environment variable in my scripts.
> > > For example, from tools/testing/selftests/rcutorture/bin/kvm.sh:
> > >
> > > KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > > PATH=${KVM}/bin:$PATH; export PATH
> > >
> > > Maybe like the (untested) patch shown below.
> >
> > A somewhat more official version below. Thank you for following up
> > on this!
> >
> > Thanx, Paul
> >
> > ------------------------------------------------------------------------
> >
> > commit 6b5cdbe80b01ce604fdf0f3ef6717a6466aca041
> > Author: Paul E. McKenney <paulmck@xxxxxxxxxx>
> > Date: Tue Jan 25 21:08:55 2022 -0800
> >
> > torture: Change KVM environment variable to RCUTORTURE
> >
> > The torture-test scripting's long-standing use of KVM as the environment
> > variable tracking the pathname of the rcutorture directory now conflicts
> > with allmodconfig builds due to the virt/kvm/Makefile.kvm file's use
> > of this as a makefile variable. This commit therefore changes the
> > torture-test scripting from KVM to RCUTORTURE, avoiding the name conflict.
> >
> > Reported-by: Zhouyi Zhou <zhouzhouyi@xxxxxxxxx>
> > Signed-off-by: Paul E. McKenney <paulmck@xxxxxxxxxx>
> >
> > diff --git a/tools/testing/selftests/rcutorture/bin/kvm-again.sh b/tools/testing/selftests/rcutorture/bin/kvm-again.sh
> > index 5a0023d183dac..0941f1ddab658 100755
> > --- a/tools/testing/selftests/rcutorture/bin/kvm-again.sh
> > +++ b/tools/testing/selftests/rcutorture/bin/kvm-again.sh
> > @@ -47,8 +47,8 @@ else
> > exit 1
> > fi
> >
> > -KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > -PATH=${KVM}/bin:$PATH; export PATH
> > +RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
> > +PATH=${RCUTORTURE}/bin:$PATH; export PATH
> > . functions.sh
> >
> > dryrun=
> > diff --git a/tools/testing/selftests/rcutorture/bin/kvm-check-branches.sh b/tools/testing/selftests/rcutorture/bin/kvm-check-branches.sh
> > index 370406bbfeedc..f17000a2ccf1f 100755
> > --- a/tools/testing/selftests/rcutorture/bin/kvm-check-branches.sh
> > +++ b/tools/testing/selftests/rcutorture/bin/kvm-check-branches.sh
> > @@ -49,8 +49,8 @@ fi
> > mkdir $resdir/$ds
> > echo Results directory: $resdir/$ds
> >
> > -KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > -PATH=${KVM}/bin:$PATH; export PATH
> > +RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
> > +PATH=${RCUTORTURE}/bin:$PATH; export PATH
> > . functions.sh
> > echo Using all `identify_qemu_vcpus` CPUs.
> >
> > diff --git a/tools/testing/selftests/rcutorture/bin/kvm-end-run-stats.sh b/tools/testing/selftests/rcutorture/bin/kvm-end-run-stats.sh
> > index e4a00779b8c69..ee886b40a5d2c 100755
> > --- a/tools/testing/selftests/rcutorture/bin/kvm-end-run-stats.sh
> > +++ b/tools/testing/selftests/rcutorture/bin/kvm-end-run-stats.sh
> > @@ -22,8 +22,8 @@ T=${TMPDIR-/tmp}/kvm-end-run-stats.sh.$$
> > trap 'rm -rf $T' 0
> > mkdir $T
> >
> > -KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > -PATH=${KVM}/bin:$PATH; export PATH
> > +RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
> > +PATH=${RCUTORTURE}/bin:$PATH; export PATH
> > . functions.sh
> > default_starttime="`get_starttime`"
> > starttime="${2-default_starttime}"
> > diff --git a/tools/testing/selftests/rcutorture/bin/kvm-remote.sh b/tools/testing/selftests/rcutorture/bin/kvm-remote.sh
> > index 29b068a55b466..8c4c1e4792d02 100755
> > --- a/tools/testing/selftests/rcutorture/bin/kvm-remote.sh
> > +++ b/tools/testing/selftests/rcutorture/bin/kvm-remote.sh
> > @@ -19,8 +19,8 @@ then
> > exit 1
> > fi
> >
> > -KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > -PATH=${KVM}/bin:$PATH; export PATH
> > +RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
> > +PATH=${RCUTORTURE}/bin:$PATH; export PATH
> > . functions.sh
> >
> > starttime="`get_starttime`"
> > @@ -108,8 +108,8 @@ else
> > cat $T/kvm-again.sh.out | tee -a "$oldrun/remote-log"
> > exit 2
> > fi
> > - cp -a "$rundir" "$KVM/res/"
> > - oldrun="$KVM/res/$ds"
> > + cp -a "$rundir" "$RCUTORTURE/res/"
> > + oldrun="$RCUTORTURE/res/$ds"
> > fi
> > echo | tee -a "$oldrun/remote-log"
> > echo " ----" kvm-again.sh output: "(`date`)" | tee -a "$oldrun/remote-log"
> > diff --git a/tools/testing/selftests/rcutorture/bin/kvm.sh b/tools/testing/selftests/rcutorture/bin/kvm.sh
> > index 348ad177a5acc..55b2c15332827 100755
> > --- a/tools/testing/selftests/rcutorture/bin/kvm.sh
> > +++ b/tools/testing/selftests/rcutorture/bin/kvm.sh
> > @@ -25,15 +25,15 @@ LANG=en_US.UTF-8; export LANG
> >
> > dur=$((30*60))
> > dryrun=""
> > -KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > -PATH=${KVM}/bin:$PATH; export PATH
> > +RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
> > +PATH=${RCUTORTURE}/bin:$PATH; export PATH
> > . functions.sh
> >
> > TORTURE_ALLOTED_CPUS="`identify_qemu_vcpus`"
> > TORTURE_DEFCONFIG=defconfig
> > TORTURE_BOOT_IMAGE=""
> > TORTURE_BUILDONLY=
> > -TORTURE_INITRD="$KVM/initrd"; export TORTURE_INITRD
> > +TORTURE_INITRD="$RCUTORTURE/initrd"; export TORTURE_INITRD
> > TORTURE_KCONFIG_ARG=""
> > TORTURE_KCONFIG_GDB_ARG=""
> > TORTURE_BOOT_GDB_ARG=""
> > @@ -262,7 +262,7 @@ else
> > exit 1
> > fi
> >
> > -CONFIGFRAG=${KVM}/configs/${TORTURE_SUITE}; export CONFIGFRAG
> > +CONFIGFRAG=${RCUTORTURE}/configs/${TORTURE_SUITE}; export CONFIGFRAG
> >
> > defaultconfigs="`tr '\012' ' ' < $CONFIGFRAG/CFLIST`"
> > if test -z "$configs"
> > @@ -272,7 +272,7 @@ fi
> >
> > if test -z "$resdir"
> > then
> > - resdir=$KVM/res
> > + resdir=$RCUTORTURE/res
> > fi
> >
> > # Create a file of test-name/#cpus pairs, sorted by decreasing #cpus.
> > @@ -386,7 +386,7 @@ END {
> > # Generate a script to execute the tests in appropriate batches.
> > cat << ___EOF___ > $T/script
> > CONFIGFRAG="$CONFIGFRAG"; export CONFIGFRAG
> > -KVM="$KVM"; export KVM
> > +RCUTORTURE="$RCUTORTURE"; export RCUTORTURE
> > PATH="$PATH"; export PATH
> > TORTURE_ALLOTED_CPUS="$TORTURE_ALLOTED_CPUS"; export TORTURE_ALLOTED_CPUS
> > TORTURE_BOOT_IMAGE="$TORTURE_BOOT_IMAGE"; export TORTURE_BOOT_IMAGE
> > @@ -569,7 +569,7 @@ ___EOF___
> > awk < $T/cfgcpu.pack \
> > -v TORTURE_BUILDONLY="$TORTURE_BUILDONLY" \
> > -v CONFIGDIR="$CONFIGFRAG/" \
> > - -v KVM="$KVM" \
> > + -v RCUTORTURE="$RCUTORTURE" \
> > -v ncpus=$cpus \
> > -v jitter="$jitter" \
> > -v rd=$resdir/$ds/ \
> > diff --git a/tools/testing/selftests/rcutorture/bin/torture.sh b/tools/testing/selftests/rcutorture/bin/torture.sh
> > index d1cb60085d8f2..e00e60efb2310 100755
> > --- a/tools/testing/selftests/rcutorture/bin/torture.sh
> > +++ b/tools/testing/selftests/rcutorture/bin/torture.sh
> > @@ -13,8 +13,8 @@
> > scriptname=$0
> > args="$*"
> >
> > -KVM="`pwd`/tools/testing/selftests/rcutorture"; export KVM
> > -PATH=${KVM}/bin:$PATH; export PATH
> > +RCUTORTURE="`pwd`/tools/testing/selftests/rcutorture"; export RCUTORTURE
> > +PATH=${RCUTORTURE}/bin:$PATH; export PATH
> > . functions.sh
> >
> > TORTURE_ALLOTED_CPUS="`identify_qemu_vcpus`"

Next message: Wei Yang: "[PATCH] mm/memory_hotplug: build zonelist for managed_zone"
Previous message: Tim Chen: "Re: [PATCH v2 2/2] sched/fair: Scan cluster before scanning LLC in wake-up path"
In reply to: Zhouyi Zhou: "Re: Problems with rcutorture on ppc64le: allmodconfig(2) and other failures"
Next in thread: Zhouyi Zhou: "Re: Problems with rcutorture on ppc64le: allmodconfig(2) and other failures"
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]