This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [patch] Fix stale tp->step_resume_breakpoint
On Tue, 02 Nov 2010 02:05:18 +0100, Pedro Alves wrote:
> On Tuesday 02 November 2010 00:43:01, Jan Kratochvil wrote:
> > A comment is welcome but it seems safe to me.
>
> I think this raises an obvious question, and hints at
> a larger issue: if you find you you need to tuck away step_resume_breakpoint,
> then, how come you don't need to do the same for all the other execution
> command state? (step_range_start, step_range_end, step_frame_id,
> continuations, etc.).
> I'd assume that in the use case you trip on step_resume_breakpoint
> troubles, you'd also be losing thread stepping state (or state
> for any other execution command), thus your thread would end up
> running free, forgetting about the previous command that was
> going on before the infcall. Is that not the case?
Currently I do not have a meaningful reproducer for it.
But I see step_resume_breakpoint on its own does not make much sense without
the associated information so I will try to save more info.
Thanks,
Jan