[Bug testsuite/15964] poll_map.exp testcase timeout on ARM

["wcohen at redhat dot com" <sourceware-bugzilla@sourceware.org>]

https://sourceware.org/bugzilla/show_bug.cgi?id=15964

--- Comment #4 from William Cohen <wcohen at redhat dot com> ---
More recently an x86_64 fedora39 machine I have running tests seems to hang on
poll_map.exp test also.  It is a Lenovo Thinkpad P51 with 8 logical processors
(4 cores, each hyperthreaded).  Compared to the ARM snowball system used to
originally file the report is pretty different, but both are multiprocessor
machines.  The machine is running a current f39 kernel:

Linux cervelo 6.7.4-200.fc39.x86_64 #1 SMP PREEMPT_DYNAMIC Mon Feb  5 22:21:14
UTC 2024 x86_64 GNU/Linux

When trying to run the poll_map.stp script from the commandline the process
gets hung when cntl-c is hit.  It doesn't exit and print "FAIL" or "SUCCESS":

$  stap -v -g /usr/share/systemtap/testsuite/systemtap.base/poll_map.stp
Pass 1: parsed user script and 731 library scripts using
2991856virt/2743848res/16000shr/2729288data kb, in 7430usr/900sys/8344real ms.
Pass 2: analyzed script: 79 probes, 8 functions, 0 embeds, 2 globals using
3236600virt/2988640res/17424shr/2974032data kb, in 2520usr/180sys/2712real ms.
Pass 3: translated to C into
"/tmp/stapJHWiaf/stap_472aafea38c440a3561a5017fd0961a9_20592_src.c" using
3236600virt/2988768res/17552shr/2974032data kb, in 210usr/100sys/307real ms.
Pass 4: compiled C into "stap_472aafea38c440a3561a5017fd0961a9_20592.ko" in
41040usr/4340sys/9567real ms.
Pass 5: starting run.
^C

Pressing cntl-c multiple times allow getting back to the command prompt:

[wcohen@cervelo ~]$  stap -v -g
/usr/share/systemtap/testsuite/systemtap.base/poll_map.stp
Pass 1: parsed user script and 731 library scripts using
2991856virt/2743668res/15872shr/2729288data kb, in 7490usr/870sys/8374real ms.
Pass 2: analyzed script: 79 probes, 8 functions, 0 embeds, 2 globals using
3236600virt/2988316res/17236shr/2974032data kb, in 2640usr/160sys/2816real ms.
Pass 3: using cached
/home/wcohen/.systemtap/cache/a4/stap_a41f3720dfed8a00a2cb71db2efb5832_20592.c
Pass 4: using cached
/home/wcohen/.systemtap/cache/a4/stap_a41f3720dfed8a00a2cb71db2efb5832_20592.ko
Pass 5: starting run.
^C^C^CToo many interrupts received, exiting.

However noticed in the lsmod that the stap module was still loaded.

Booted with "maxcpus=1" to limit the machine to one processor and the machine
still hung without output.

Tried similar experiment on a x86_64 qemu f39 image on another x86_64 laptop
and output was printed and the test exited. That was successful.

Looks like something specific to lenovo thinkpad p51.  Did a fresh install of
x86_64 f39 on a vm with 4GB memory and 4 CPUs.  it had the same behavior.

-- 
You are receiving this mail because:
You are the assignee for the bug.