Qemu on Zynq Hanging

[Chris Johns]

On 11/3/21 5:17 am, Joel Sherrill wrote:
> On Wed, Mar 10, 2021 at 11:42 AM Chris Johns <chrisj at rtems.org
> <mailto:chrisj at rtems.org>> wrote:
> 
>     On 11/3/21 1:22 am, Joel Sherrill wrote:
>     > I see qemu hanging for days testing zynq. This is on the Xeon Centos
>     computer I
>     > use for batch and automated testing.
>     >
>     > This is ps but top shows six qemu processes having around 2231:52 in CPU time
>     > used and using 100% CPU. Luckily this machine has more cores than this.
>     >
>     >   Any ideas?
>     Looks like a bug in the tester not timing out these tests. It maybe timing out
>     but the kill may not be stopping qemu or the testing is missing the timeout
>     event. The fact qemu is not finishing by it's self is a contributing factor. It
>     would require some debugging to find out which it is.
> 
>     Tests not finishing should be a handled state no matter the reason.
> 
>     > [joel at devel ~]$ ps -efl | grep test
>     > 2 S joel      4423 30108 98  80   0 - 448725 poll_s Mar08 ?       1-13:11:11
>     > qemu-system-arm -no-reboot -nographic -net none -serial null -serial mon:stdio
>     > -M xilinx-zynq-a9 -m 256M -kernel
>     > ./arm-rtems6/c/xilinx_zynq_a9_qemu/testsuites/tmtests/tm12.exe
>     > 0 R joel      6493 30568  0  80   0 - 28204 -      08:20 pts/1    00:00:00
>     grep
>     > --color=auto test
>     > 2 S joel     15440 30108 97  80   0 - 448715 poll_s Mar08 ?       1-14:29:27
>     > qemu-system-arm -no-reboot -nographic -net none -serial null -serial mon:stdio
>     > -M xilinx-zynq-a9 -m 256M -kernel
>     > ./arm-rtems6/c/xilinx_zynq_a9_qemu/testsuites/psxtests/psxsysconf.exe
>     > 2 S joel     15981 30108 97  80   0 - 448707 poll_s Mar08 ?       1-14:26:10
>     > qemu-system-arm -no-reboot -nographic -net none -serial null -serial mon:stdio
>     > -M xilinx-zynq-a9 -m 256M -kernel
>     > ./arm-rtems6/c/xilinx_zynq_a9_qemu/testsuites/psxtmtests/psxtmbarrier01.exe
>     > 2 S joel     19281 30108 97  80   0 - 497737 poll_s Mar08 ?       1-14:12:14
>     > qemu-system-arm -no-reboot -nographic -net none -serial null -serial mon:stdio
>     > -M xilinx-zynq-a9 -m 256M -kernel
>     > ./arm-rtems6/c/xilinx_zynq_a9_qemu/testsuites/psxtmtests/psxtmsem04.exe
>     > 2 S joel     24394 30108 97  80   0 - 448708 poll_s Mar08 ?       1-13:52:24
>     > qemu-system-arm -no-reboot -nographic -net none -serial null -serial mon:stdio
>     > -M xilinx-zynq-a9 -m 256M -kernel
>     > ./arm-rtems6/c/xilinx_zynq_a9_qemu/testsuites/sptests/sp48.exe
>     > 2 S joel     28252 30108 98  80   0 - 497719 poll_s Mar08 ?       1-13:38:38
>     > qemu-system-arm -no-reboot -nographic -net none -serial null -serial mon:stdio
>     > -M xilinx-zynq-a9 -m 256M -kernel
>     > ./arm-rtems6/c/xilinx_zynq_a9_qemu/testsuites/sptests/spfatal04.exe
>     > 0 S joel     30108  6096  0  80   0 - 967025 futex_ Mar08 ?       00:00:22
>     > python /home/joel/rtems-cron-6/tools/6/bin/rtems-test
>     > --rtems-tools=/home/joel/rtems-cron-6/tools/6 --rtems-bsp=xilinx_zynq_a9_qemu
>     > --log=run.log --mail --mail-from=joel at rtems.org <mailto:joel at rtems.org>
>     <mailto:joel at rtems.org <mailto:joel at rtems.org>>
>     > --mail-to=build at rtems.org <mailto:build at rtems.org> <mailto:build at rtems.org
>     <mailto:build at rtems.org>> .
> 
>     Can you kill these tests by hand? If so what was the kill command?
> 
> 
> Yep. Just use kill PID. I realized that -9 wasn't needed over the few times this 
> has happened:
> 
> ps -efl | grep test | cut -c10-19 | xargs -e kill

Great. The tester's kill should work.

It will need some debugging. The tester has a --trace option that should help.

Chris