https://bugs.kde.org/show_bug.cgi?id=378494
Milian Wolff changed:
What|Removed |Added
Status|CONFIRMED |RESOLVED
Latest Commit|
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #9 from Milian Wolff ---
heaptrack already depends on pthread (for the timer thread). I also asked a
colleague and he confirmed this is safe to do. I'll push it - thanks for the
report and helpful test case!
--
You are receiving this mail
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #8 from Maksim Golov ---
Confirm that it works for my case.
One thing to note, the proposed patch introduces a dependency on pthreads,
which is probably quite safe in practice, and I'd expect CMake can probe for
pthreads being available. I
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #7 from Milian Wolff ---
Reading up a bit, I think this patch should actually quite safe to do. Can you
confirm, considering you seem to have more knowledge around this idiom? If so,
then I'd push it to master in the hope that it's not break
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #6 from Milian Wolff ---
Indeed, that work-arounds this issue:
diff --git a/src/track/libheaptrack.cpp b/src/track/libheaptrack.cpp
index b21b00b..6df2101 100644
--- a/src/track/libheaptrack.cpp
+++ b/src/track/libheaptrack.cpp
@@ -30,6 +30
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #5 from Maksim Golov ---
What the code in the reproduce does is to ensure that all signals are handled
by a single dedicated thread only, and all other threads have them disabled.
This ensures that the signals are handled in well-defined man
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #4 from Maksim Golov ---
I was not aware of timer thread, must have missed it in my reading of heaptrack
code. Likely it has the signals enabled and gets to handle the signal instead
of the signal thread.
Not sure how this can be addressed,
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #3 from Milian Wolff ---
ah, I can reproduce the issue with your test code even without heaptrack, when
I slightly modify it:
at the very start of main, add:
auto t = std::thread([]() { while(true)
{std::this_thread::sleep_for(std::chrono:
https://bugs.kde.org/show_bug.cgi?id=378494
--- Comment #2 from Milian Wolff ---
What's odd is that it does not happen when I disable the timer thread. Stracing
the signals syscalls also doesn't reveal anything obvious. I still have no clue
what's going on here.
--
You are receiving this mail b
https://bugs.kde.org/show_bug.cgi?id=378494
Milian Wolff changed:
What|Removed |Added
Ever confirmed|0 |1
Status|UNCONFIRMED
10 matches
Mail list logo
