Closed tinloaf closed 9 months ago
Thanks for the detailed problem report. I should be able to reproduce it using your directions.
I wasn't able to duplicate it with your instructions. However, the traceback pointed me to the error in my code.
I've fixed it in the mainline.
The crash is gone, thanks a lot!
With the recent
main
, myseergdb
crashes when I try to debug a segfault that I recorded viarr
. I am usingrr
version 5.7.0.Note: I can only reproduce this when debugging an
rr
trace, not when 'directly' debugging usingseergdb
. I don't think that this is really dependent onrr
, but I suspect the crucial fact is that inrr
mode,seergdb
already stops once beforemain()
and the 'Locals' tab is populated. If I don't open the 'Locals' tab at this point (see step 5 below), the crash disappears.Reproduction
I can reproduce it like this:
Create a segfaulting file, e.g. this
mnwe.cpp
file:g++ -g -O0 mnwe.cpp -o mnwe
rr
:rr record --output-trace-dir=rrrun ./mnwe
(you should see the segfault)seergdb
with this trace:seergdb --rr rrrun
seergbd
stops beforemain()
, make sure that the 'Locals' tab is openF8
to run the traceAt this point, my seergbd crashes.
Backtrace
This is the backtrace I see when running the above under
gdb
: