[Lazarus] Post-Mortem Debugging
Mark Morgan Lloyd
markMLl.lazarus at telemetry.co.uk
Thu Feb 26 21:22:04 CET 2015
Martin Frb wrote:
> On 26/02/15 19:52, Mark Morgan Lloyd wrote:
>> I've just forced a core dump of a misbehaving multithreaded program
>> and am looking at it using gdb to good effect.
>>
>> Can Lazarus be used as a post-mortem debugger, reminiscent of products
>> from Logitech et al?
>>
> probably not without changes.
>
> I havent used gdb for this either, so I am not aware which commands you
> can send to gdb.
>
> I guess that the code used for "attach" can be copied/modified to load a
> core dump.
>
> then you might be able do use the stack, and watches. but I do not know
In this case all I did was (before running program) ulimit -Hc
unlimited followed by (while looking at stuck program) gcore <pid>
and then gdb <binary> core.<pid> after which I could use bt etc. In
principle, something similar should work for a program which had failed
with a bus error or segment violation.
I'm converting somebody else's code from single- to multithreaded. If
run in the context of Lazarus's own debugging it's fine, but if running
"unattended" it locks up with a race condition about 75% of the time. So
being able to look at a snapshot is really extremely useful, even if I
can't take full control.
--
Mark Morgan Lloyd
markMLl .AT. telemetry.co .DOT. uk
[Opinions above are the author's, not those of his employers or colleagues]
More information about the Lazarus
mailing list