* [PATCH] tests/qtest/libqtest: Do not overwrite child coredump
@ 2020-07-07 3:19 Philippe Mathieu-Daudé
2020-07-07 9:03 ` Thomas Huth
0 siblings, 1 reply; 4+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-07-07 3:19 UTC (permalink / raw)
To: qemu-devel
Cc: Laurent Vivier, Thomas Huth, Richard Henderson, Markus Armbruster,
Philippe Mathieu-Daudé, Yury Kotov, Paolo Bonzini
We are interested by the coredump of the child, not the qtest
parent. If the child generated a coredump, simply call
exit(EXIT_FAILURE) in the parent to avoid overwriting the
child coredump.
Fixes: 71a268a5fd ("tests/libqtest: Improve kill_qemu()")
Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
---
tests/qtest/libqtest.c | 7 ++++++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/tests/qtest/libqtest.c b/tests/qtest/libqtest.c
index 49075b55a1..bd85d01145 100644
--- a/tests/qtest/libqtest.c
+++ b/tests/qtest/libqtest.c
@@ -173,7 +173,12 @@ static void kill_qemu(QTestState *s)
fprintf(stderr, "%s:%d: kill_qemu() detected QEMU death "
"from signal %d (%s)%s\n",
__FILE__, __LINE__, sig, signame, dump);
- abort();
+ if (WCOREDUMP(wstatus)) {
+ /* Preserve child coredump */
+ exit(1);
+ } else {
+ abort();
+ }
}
}
--
2.21.3
^ permalink raw reply related [flat|nested] 4+ messages in thread* Re: [PATCH] tests/qtest/libqtest: Do not overwrite child coredump
2020-07-07 3:19 [PATCH] tests/qtest/libqtest: Do not overwrite child coredump Philippe Mathieu-Daudé
@ 2020-07-07 9:03 ` Thomas Huth
2020-07-07 9:29 ` Paolo Bonzini
0 siblings, 1 reply; 4+ messages in thread
From: Thomas Huth @ 2020-07-07 9:03 UTC (permalink / raw)
To: Philippe Mathieu-Daudé, qemu-devel
Cc: Laurent Vivier, Richard Henderson, Markus Armbruster, Yury Kotov,
Paolo Bonzini
On 07/07/2020 05.19, Philippe Mathieu-Daudé wrote:
> We are interested by the coredump of the child, not the qtest
> parent. If the child generated a coredump, simply call
> exit(EXIT_FAILURE) in the parent to avoid overwriting the
> child coredump.
>
> Fixes: 71a268a5fd ("tests/libqtest: Improve kill_qemu()")
> Signed-off-by: Philippe Mathieu-Daudé <f4bug@amsat.org>
> ---
> tests/qtest/libqtest.c | 7 ++++++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/tests/qtest/libqtest.c b/tests/qtest/libqtest.c
> index 49075b55a1..bd85d01145 100644
> --- a/tests/qtest/libqtest.c
> +++ b/tests/qtest/libqtest.c
> @@ -173,7 +173,12 @@ static void kill_qemu(QTestState *s)
> fprintf(stderr, "%s:%d: kill_qemu() detected QEMU death "
> "from signal %d (%s)%s\n",
> __FILE__, __LINE__, sig, signame, dump);
> - abort();
> + if (WCOREDUMP(wstatus)) {
> + /* Preserve child coredump */
> + exit(1);
> + } else {
> + abort();
> + }
> }
> }
Would it maybe rather make sense to always use exit(1) unconditionally here?
Thomas
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] tests/qtest/libqtest: Do not overwrite child coredump
2020-07-07 9:03 ` Thomas Huth
@ 2020-07-07 9:29 ` Paolo Bonzini
2020-07-07 13:32 ` Philippe Mathieu-Daudé
0 siblings, 1 reply; 4+ messages in thread
From: Paolo Bonzini @ 2020-07-07 9:29 UTC (permalink / raw)
To: Thomas Huth, Philippe Mathieu-Daudé, qemu-devel
Cc: Laurent Vivier, Yury Kotov, Richard Henderson, Markus Armbruster
On 07/07/20 11:03, Thomas Huth wrote:
>> +++ b/tests/qtest/libqtest.c
>> @@ -173,7 +173,12 @@ static void kill_qemu(QTestState *s)
>> fprintf(stderr, "%s:%d: kill_qemu() detected QEMU death "
>> "from signal %d (%s)%s\n",
>> __FILE__, __LINE__, sig, signame, dump);
>> - abort();
>> + if (WCOREDUMP(wstatus)) {
>> + /* Preserve child coredump */
>> + exit(1);
>> + } else {
>> + abort();
>> + }
>> }
>> }
> Would it maybe rather make sense to always use exit(1) unconditionally here?
But why is it a problem to overwrite the child core dump? Aren't both
stashed away if you use the core.PID name as is common?
Paolo
^ permalink raw reply [flat|nested] 4+ messages in thread* Re: [PATCH] tests/qtest/libqtest: Do not overwrite child coredump
2020-07-07 9:29 ` Paolo Bonzini
@ 2020-07-07 13:32 ` Philippe Mathieu-Daudé
0 siblings, 0 replies; 4+ messages in thread
From: Philippe Mathieu-Daudé @ 2020-07-07 13:32 UTC (permalink / raw)
To: Paolo Bonzini, Thomas Huth, qemu-devel
Cc: Laurent Vivier, Yury Kotov, Richard Henderson, Markus Armbruster
On 7/7/20 11:29 AM, Paolo Bonzini wrote:
> On 07/07/20 11:03, Thomas Huth wrote:
>>> +++ b/tests/qtest/libqtest.c
>>> @@ -173,7 +173,12 @@ static void kill_qemu(QTestState *s)
>>> fprintf(stderr, "%s:%d: kill_qemu() detected QEMU death "
>>> "from signal %d (%s)%s\n",
>>> __FILE__, __LINE__, sig, signame, dump);
>>> - abort();
>>> + if (WCOREDUMP(wstatus)) {
>>> + /* Preserve child coredump */
>>> + exit(1);
>>> + } else {
>>> + abort();
>>> + }
>>> }
>>> }
>> Would it maybe rather make sense to always use exit(1) unconditionally here?
>
> But why is it a problem to overwrite the child core dump? Aren't both
> stashed away if you use the core.PID name as is common?
I'm not sure what you mean. Without this patch, the coredump I get
is qtest parent, coredumpctl list the child but the coredump is
unavailable. With this patch I don't get the uninteresting (for my
uses) qtest parent but the child coredump. I'm using Fedora 30,
I don't remember changing the coredumpctl default config. Travis-CI
is based on Ubuntu.
>
> Paolo
>
>
^ permalink raw reply [flat|nested] 4+ messages in thread
end of thread, other threads:[~2020-07-07 13:33 UTC | newest]
Thread overview: 4+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2020-07-07 3:19 [PATCH] tests/qtest/libqtest: Do not overwrite child coredump Philippe Mathieu-Daudé
2020-07-07 9:03 ` Thomas Huth
2020-07-07 9:29 ` Paolo Bonzini
2020-07-07 13:32 ` Philippe Mathieu-Daudé
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).