infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andreas Schwab]

I'm getting this assertion failure on ia64-linux when stepping after a
breakpoint with a signal also being delivered.  Both calls to
insert_step_resume_breakpoint_at_sal are coming from the two calls to
insert_step_resume_breakpoint_at_frame in handle_inferior_event.  In both
calls the actual address where the breakpoint is to be set is the same.  I
haven't been able to come up with a small test case, this happens while
debugging Emacs.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andrew Cagney]

Andreas Schwab wrote:
> I'm getting this assertion failure on ia64-linux when stepping after a
> breakpoint with a signal also being delivered.  Both calls to
> insert_step_resume_breakpoint_at_sal are coming from the two calls to
> insert_step_resume_breakpoint_at_frame in handle_inferior_event.  In both
> calls the actual address where the breakpoint is to be set is the same.  I
> haven't been able to come up with a small test case, this happens while
> debugging Emacs.

The sig* tests check this edge case.

For the other architectures I investigated, it was tracked down to 
kernel bugs.  For instance, ptrace(STEP,SIGNAL) was implemented as 
ptrace(CONT,SIGNAL).  Suggest looking upstream, we've (Red Hat) have 
been pushing fixes for other key architectures ;-)

Andrew

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed dot

[Mark Kettenis]

   From: Andreas Schwab <schwab@suse.de>
   Date: Fri, 29 Oct 2004 17:22:57 +0200

   I'm getting this assertion failure on ia64-linux when stepping after a
   breakpoint with a signal also being delivered.  Both calls to
   insert_step_resume_breakpoint_at_sal are coming from the two calls to
   insert_step_resume_breakpoint_at_frame in handle_inferior_event.  In both
   calls the actual address where the breakpoint is to be set is the same.  I
   haven't been able to come up with a small test case, this happens while
   debugging Emacs.

I'm seeing the same problem on NetBSD/amd64 in the regular testsuite
(sigstep.exp).  Haven't found the time yet to look into it though.

Mark

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andreas Schwab]

Andrew Cagney <ac131313@redhat.com> writes:

> Andreas Schwab wrote:
>> I'm getting this assertion failure on ia64-linux when stepping after a
>> breakpoint with a signal also being delivered.  Both calls to
>> insert_step_resume_breakpoint_at_sal are coming from the two calls to
>> insert_step_resume_breakpoint_at_frame in handle_inferior_event.  In both
>> calls the actual address where the breakpoint is to be set is the same.  I
>> haven't been able to come up with a small test case, this happens while
>> debugging Emacs.
>
> The sig* tests check this edge case.

Not this one.  There are only two failures, which are
"gdb.base/sigstep.exp: finish from handleri; leave handler (timeout)" and
"gdb.base/sigstep.exp: finish from handleri; leave signal trampoline", and
neither are not due to internal errors.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andrew Cagney]

Andreas Schwab wrote:
> Andrew Cagney <ac131313@redhat.com> writes:
> 
> 
>>Andreas Schwab wrote:
>>
>>>I'm getting this assertion failure on ia64-linux when stepping after a
>>>breakpoint with a signal also being delivered.  Both calls to
>>>insert_step_resume_breakpoint_at_sal are coming from the two calls to
>>>insert_step_resume_breakpoint_at_frame in handle_inferior_event.  In both
>>>calls the actual address where the breakpoint is to be set is the same.  I
>>>haven't been able to come up with a small test case, this happens while
>>>debugging Emacs.
>>
>>The sig* tests check this edge case.
> 
> 
> Not this one.  There are only two failures, which are
> "gdb.base/sigstep.exp: finish from handleri; leave handler (timeout)" and
> "gdb.base/sigstep.exp: finish from handleri; leave signal trampoline", and
> neither are not due to internal errors.

The "step on breakpoint; ..." tests, based on your description, exercise 
just that senario.  They run to a breakpoint, wait for a signal to be 
come pending, and then try to step et.al.

Sounds like more is going on?

Andrew

Ref 1757, 1738

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andreas Schwab]

Andrew Cagney <cagney@gnu.org> writes:

> The "step on breakpoint; ..." tests,

I can't find any such test in gdb.base.

> based on your description, exercise just that senario.  They run to a
> breakpoint, wait for a signal to be come pending, and then try to step
> et.al.
>
> Sounds like more is going on?

Perhaps the key point is that the signal (SIGIO) is silently passed to the
inferiour by gdb.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andrew Cagney]

Andreas Schwab wrote:
> Andrew Cagney <cagney@gnu.org> writes:
> 
> 
>>The "step on breakpoint; ..." tests,
> 
> 
> I can't find any such test in gdb.base.

sigstep.exp:breakpoint_over_handler, added 25/8.  The above should be in 
your gdb.log.  What's your GDB version?

>>based on your description, exercise just that senario.  They run to a
>>breakpoint, wait for a signal to be come pending, and then try to step
>>et.al.
>>
>>Sounds like more is going on?
> 
> 
> Perhaps the key point is that the signal (SIGIO) is silently passed to the
> inferiour by gdb.

No, the above does that (but with SIGALRM).  Are there multiple signals 
pending?

Andrew

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andrew Cagney]

>>No, the above does that (but with SIGALRM).  Are there multiple signals
>>pending?
> 
> 
> Doesn't seem so.

Time to post (gdb) set debug target 1 ; run.

Andrew

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andreas Schwab]

[-- Attachment #1: Type: text/plain, Size: 447 bytes --]

Andrew Cagney <ac131313@redhat.com> writes:

>>>No, the above does that (but with SIGALRM).  Are there multiple signals
>>>pending?
>> Doesn't seem so.
>
> Time to post (gdb) set debug target 1 ; run.

It's attached.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux AG, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."

[-- Attachment #2: GDB debug dump --]
[-- Type: application/x-bzip2, Size: 28425 bytes --]

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andrew Cagney]

Andreas Schwab wrote:
> Andrew Cagney <ac131313@redhat.com> writes:
> 
> 
>>>>No, the above does that (but with SIGALRM).  Are there multiple signals
>>>>pending?
>>>
>>>Doesn't seem so.
>>
>>Time to post (gdb) set debug target 1 ; run.
> 
> 
> It's attached.
> 
> Andreas.

Andreas,

I've fixed one bug related to back-to-back signals that would cause this 
internal error (testsuite gdb.base/sigrepeat.exp).  Can you re-run your 
senario?  Nick, who encountered similar problems, found his problem 
moved :-/

Andrew

Re: infrun.c:2642: internal-error: insert_step_resume_breakpoint_at_sal: Assertion `step_resume_breakpoint == NULL' failed.

[Andreas Schwab]

Andrew Cagney <cagney@gnu.org> writes:

> I've fixed one bug related to back-to-back signals that would cause this
> internal error (testsuite gdb.base/sigrepeat.exp).  Can you re-run your
> senario?

Seems to work now.

Andreas.

-- 
Andreas Schwab, SuSE Labs, schwab@suse.de
SuSE Linux Products GmbH, Maxfeldstraße 5, 90409 Nürnberg, Germany
Key fingerprint = 58CA 54C7 6D53 942B 1756  01D3 44D5 214B 8276 4ED5
"And now for something completely different."