This is the mail archive of the
gdb-patches@sources.redhat.com
mailing list for the GDB project.
Re: [PATCH/RFC] Improve the "thread_step_needed" logic
Daniel Jacobowitz wrote:
>
> On Fri, Jun 15, 2001 at 04:35:09PM -0700, Michael Snyder wrote:
> > This is a fairly significant change, that both simplifies and
> > improves the logic for deciding when a single thread should be
> > stepped (to get past a breakpoint), rather than stepping all threads.
> > In a nutshell, this replaces the state variable "thread_step_needed"
> > with the following logic in resume():
>
> <sigh>
>
> While you're fixing this, could you please address my issue from last
> week with this code?
I'm pretty sure I did!
Checking "step" at this point in resume() is not
> correct, since it will have been cleared if SOFTWARE_SINGLE_STEP_P ().
But I don't just check "step", I check
"(step || singlestep_breakpoints_inserted_p)".
That was specifically to address your issue.
Do you think it fails to do so?
>
> > --- 879,913 ----
> > if (should_resume)
> > {
> > ptid_t resume_ptid;
> > +
> > + resume_ptid = RESUME_ALL; /* Default */
> >
> > ! if ((step || singlestep_breakpoints_inserted_p) &&
> > ! !breakpoints_inserted && breakpoint_here_p (read_pc ()))
>
> > +
> > + #ifdef CANNOT_STEP_BREAKPOINT
> > + /* Most targets can step a breakpoint instruction, thus executing it
> > + normally. But if this one cannot, just continue and we will hit
> > + it anyway. */
> > + if (step && breakpoints_inserted && breakpoint_here_p (read_pc ()))
> > + step = 0;
> > + #endif
>
> Specifically, those bits.
>
> --
> Daniel Jacobowitz Debian GNU/Linux Developer
> Monta Vista Software Debian Security Team