Multiple breakpoints

Multiple breakpoints

[Nick Roberts]

If I set a breakpoint on TRY_CATCH in gdb_evaluate_expression, I get
multiple breakpoints:

  gdb_evaluate_expression (struct expression *exp, struct value **value)
  {
    volatile struct gdb_exception except;

->  TRY_CATCH (except, RETURN_MASK_ERROR)
      {
        *value = evaluate_expression(exp);
      }

    if (except.reason < 0)
      return 0;
    return 1;
  }

where -> shows the breakpoint location.


Is this to be expected?  The assembler looks like this:


    0x081c04a1 <gdb_evaluate_expression+0>:	push   %ebp
    0x081c04a2 <gdb_evaluate_expression+1>:	mov    %esp,%ebp
    0x081c04a4 <gdb_evaluate_expression+3>:	sub    $0x28,%esp
->  0x081c04a7 <gdb_evaluate_expression+6>:	mov    0x83a55a0,%edx
    0x081c04ad <gdb_evaluate_expression+12>:	movl   $0x2,0x8(%esp)
    0x081c04b5 <gdb_evaluate_expression+20>:	lea    -0x10(%ebp),%eax
    0x081c04b8 <gdb_evaluate_expression+23>:	mov    %eax,0x4(%esp)
    0x081c04bc <gdb_evaluate_expression+27>:	mov    %edx,(%esp)
    0x081c04bf <gdb_evaluate_expression+30>:	call   0x814e660 <exce...
    0x081c04c4 <gdb_evaluate_expression+35>:	mov    %eax,-0x4(%ebp)
    0x081c04c7 <gdb_evaluate_expression+38>:	mov    -0x4(%ebp),%eax
    0x081c04ca <gdb_evaluate_expression+41>:	movl   $0x1,0x4(%esp)
    0x081c04d2 <gdb_evaluate_expression+49>:	mov    %eax,(%esp)
    0x081c04d5 <gdb_evaluate_expression+52>:	call   0x8089058 <__si...
    0x081c04da <gdb_evaluate_expression+57>:	jmp    0x81c04f7 <gdb_...
    0x081c04dc <gdb_evaluate_expression+59>:	mov    0x8(%ebp),%eax
    0x081c04df <gdb_evaluate_expression+62>:	mov    %eax,(%esp)
    0x081c04e2 <gdb_evaluate_expression+65>:	call   0x810fc5e <eval...
    0x081c04e7 <gdb_evaluate_expression+70>:	mov    %eax,%edx
    0x081c04e9 <gdb_evaluate_expression+72>:	mov    0xc(%ebp),%eax
    0x081c04ec <gdb_evaluate_expression+75>:	mov    %edx,(%eax)
->  0x081c04ee <gdb_evaluate_expression+77>:	call   0x814e983 <exce...
    0x081c04f3 <gdb_evaluate_expression+82>:	test   %eax,%eax
    0x081c04f5 <gdb_evaluate_expression+84>:	jne    0x81c04dc <gdb_...
    0x081c04f7 <gdb_evaluate_expression+86>:	call   0x814e96f <exce...
    0x081c04fc <gdb_evaluate_expression+91>:	test   %eax,%eax
    0x081c04fe <gdb_evaluate_expression+93>:	jne    0x81c04ee <gdb_...
    0x081c0500 <gdb_evaluate_expression+95>:	mov    -0x10(%ebp),%ea
    0x081c0503 <gdb_evaluate_expression+98>:	test   %eax,%eax
    0x081c0505 <gdb_evaluate_expression+100>:	jns    0x81c0510 <gdb_...
    0x081c0507 <gdb_evaluate_expression+102>:	movl   $0x0,-0x14(%ebp
    0x081c050e <gdb_evaluate_expression+109>:	jmp    0x81c0517 <gdb_...
    0x081c0510 <gdb_evaluate_expression+111>:	movl   $0x1,-0x14(%ebp)
    0x081c0517 <gdb_evaluate_expression+118>:	mov    -0x14(%ebp),%eax
    0x081c051a <gdb_evaluate_expression+121>:	leave  
    0x081c051b <gdb_evaluate_expression+122>:	ret    
    End of assembler dump.


-- 
Nick                                           http://www.inet.net.nz/~nickrob

Re: Multiple breakpoints

[Andreas Schwab]

Nick Roberts <nickrob@snap.net.nz> writes:

> If I set a breakpoint on TRY_CATCH in gdb_evaluate_expression, I get
> multiple breakpoints:
>
>   gdb_evaluate_expression (struct expression *exp, struct value **value)
>   {
>     volatile struct gdb_exception except;
>
> ->  TRY_CATCH (except, RETURN_MASK_ERROR)
>       {
>         *value = evaluate_expression(exp);
>       }
>
>     if (except.reason < 0)
>       return 0;
>     return 1;
>   }
>
> where -> shows the breakpoint location.
>
>
> Is this to be expected?  The assembler looks like this:

TRY_CATCH expands to a loop, probably the loop condition has been
duplicated by the compiler.

Andreas.

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

Re: Multiple breakpoints

[Daniel Jacobowitz]

On Thu, Apr 03, 2008 at 11:01:10AM +0200, Andreas Schwab wrote:
> TRY_CATCH expands to a loop, probably the loop condition has been
> duplicated by the compiler.

I think we try to set only one breakpoint per containing function,
though.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Multiple breakpoints

[Vladimir Prus]

Daniel Jacobowitz wrote:

> On Thu, Apr 03, 2008 at 11:01:10AM +0200, Andreas Schwab wrote:
>> TRY_CATCH expands to a loop, probably the loop condition has been
>> duplicated by the compiler.
> 
> I think we try to set only one breakpoint per containing function,
> though.

Uh-uh, you keep on telling this :-)
Per *block*, not per function.

- Volodya

Re: Multiple breakpoints

[Daniel Jacobowitz]

On Thu, Apr 03, 2008 at 05:51:00PM +0400, Vladimir Prus wrote:
> Daniel Jacobowitz wrote:
> 
> > On Thu, Apr 03, 2008 at 11:01:10AM +0200, Andreas Schwab wrote:
> >> TRY_CATCH expands to a loop, probably the loop condition has been
> >> duplicated by the compiler.
> > 
> > I think we try to set only one breakpoint per containing function,
> > though.
> 
> Uh-uh, you keep on telling this :-)
> Per *block*, not per function.

Durn.  Well, I would have thought the parts of a for loop were in the
same block, but I can see GCC deciding otherwise.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Multiple breakpoints

[Daniel Jacobowitz]

On Thu, Apr 03, 2008 at 05:51:00PM +0400, Vladimir Prus wrote:
> Daniel Jacobowitz wrote:
> 
> > On Thu, Apr 03, 2008 at 11:01:10AM +0200, Andreas Schwab wrote:
> >> TRY_CATCH expands to a loop, probably the loop condition has been
> >> duplicated by the compiler.
> > 
> > I think we try to set only one breakpoint per containing function,
> > though.
> 
> Uh-uh, you keep on telling this :-)
> Per *block*, not per function.

Durn.  Well, I would have thought the parts of a for loop were in the
same block, but I can see GCC deciding otherwise.

-- 
Daniel Jacobowitz
CodeSourcery

Re: Multiple breakpoints

[Michael Snyder]

On Thu, 2008-04-03 at 10:01 -0400, Daniel Jacobowitz wrote:
> On Thu, Apr 03, 2008 at 05:51:00PM +0400, Vladimir Prus wrote:
> > Daniel Jacobowitz wrote:
> > 
> > > On Thu, Apr 03, 2008 at 11:01:10AM +0200, Andreas Schwab wrote:
> > >> TRY_CATCH expands to a loop, probably the loop condition has been
> > >> duplicated by the compiler.
> > > 
> > > I think we try to set only one breakpoint per containing function,
> > > though.
> > 
> > Uh-uh, you keep on telling this :-)
> > Per *block*, not per function.
> 
> Durn.  Well, I would have thought the parts of a for loop were in the
> same block, but I can see GCC deciding otherwise.

The inside parts probably are, but there has to be loop
block, and there has to be something outside the loop
block to decide whether or not to repeat the loop block.

If that made sense...   ;-)

Re: Multiple breakpoints

[Michael Snyder]

On Thu, 2008-04-03 at 10:01 -0400, Daniel Jacobowitz wrote:
> On Thu, Apr 03, 2008 at 05:51:00PM +0400, Vladimir Prus wrote:
> > Daniel Jacobowitz wrote:
> > 
> > > On Thu, Apr 03, 2008 at 11:01:10AM +0200, Andreas Schwab wrote:
> > >> TRY_CATCH expands to a loop, probably the loop condition has been
> > >> duplicated by the compiler.
> > > 
> > > I think we try to set only one breakpoint per containing function,
> > > though.
> > 
> > Uh-uh, you keep on telling this :-)
> > Per *block*, not per function.
> 
> Durn.  Well, I would have thought the parts of a for loop were in the
> same block, but I can see GCC deciding otherwise.

The inside parts probably are, but there has to be loop
block, and there has to be something outside the loop
block to decide whether or not to repeat the loop block.

If that made sense...   ;-)