Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

[Nick Clifton]

Hi Kelvin,

> Attached test.c is code extracted from elflink.c to demonstrate the problem.
> You can use attached comp.sh to compile the test program. The test program
> must be compile inside the bfd directory of the binutils-2.15 package. The
> embedded data is the data extract from the gcc-3.4 build which caused
> ld-2.15 to crash. BTW, the test program also shows a possible work around of
> the problem, i.e. instead of doing the calculation, simply comparing the
> values inside elf_sort_symbol. (You may also use the attached diff file to
> patch elflink.c.) The patch seems working fine or at least survived the gcc
> build cycle.

Thanks very much for discovering this problem and producing a 
workaround.  I have applied the patch you suggested, together with this 
ChangeLog entry:

bfd/ChangeLog
2004-07-13  Kelvin Lee  <Kelvin.Lee@lansa.com.au>

	* elflink.c (elf_sort_symbol): Restructure code to avoid bug in
	Solairs hosted versions of gcc.

Cheers
   Nick

Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

[Eric Botcazou]

> Thanks very much for discovering this problem and producing a
> workaround.  I have applied the patch you suggested, together with this
> ChangeLog entry:
>
> bfd/ChangeLog
> 2004-07-13  Kelvin Lee  <Kelvin.Lee@lansa.com.au>
>
> 	* elflink.c (elf_sort_symbol): Restructure code to avoid bug in
> 	Solairs hosted versions of gcc.

I'm a bit confused by this patch:
- is there really a bug "in Solaris hosted versions of gcc" involved here?  
If so, it would be nice to produce a testcase and file a PR with GCC 
bugzilla (cc-ing ebotcazou@gcc.gnu.org would be even nicer).
- is the bug reported by Kevin not a mere duplicate of
  http://sources.redhat.com/ml/binutils/2004-06/msg00089.html
which has been fixed by
  http://sources.redhat.com/ml/binutils/2004-06/msg00114.html
in which case the patch is superfluous and the ChangeLog entry a bit 
misleading?

-- 
Eric Botcazou

Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

[Nick Clifton]

Hi Eric,

>>bfd/ChangeLog
>>2004-07-13  Kelvin Lee  <Kelvin.Lee@lansa.com.au>
>>
>>	* elflink.c (elf_sort_symbol): Restructure code to avoid bug in
>>	Solairs hosted versions of gcc.
> 
> 
> I'm a bit confused by this patch:
> - is there really a bug "in Solaris hosted versions of gcc" involved here?  

Maybe...  It involves arithmetic comparisons of pointers to separately 
allocated objects, which is a bit dubious in the first place.

> If so, it would be nice to produce a testcase and file a PR with GCC 
> bugzilla (cc-ing ebotcazou@gcc.gnu.org would be even nicer).

Indeed.

> - is the bug reported by Kevin not a mere duplicate of
>   http://sources.redhat.com/ml/binutils/2004-06/msg00089.html

Yes.

> which has been fixed by
>   http://sources.redhat.com/ml/binutils/2004-06/msg00114.html

Doh!  Yes - I missed this.  Very sorry.  I will revert Kelvin's patch.

Cheers
   Nick

Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

[Eric Botcazou]

> Hi Eric,

Hi Nick,

> Maybe...  It involves arithmetic comparisons of pointers to separately
> allocated objects, which is a bit dubious in the first place.

Note that it doesn't compare pointers directly, but first performs a 
substraction on pointers which is strictly specified by C99 6.5.6 §9: "When 
two pointers are substracted, both shall point to elements of the same array 
object, or one past the last element of the array object; the result is the 
difference of the subscripts of the two array elements."

So my understanding is that there isn't any GCC bug here.

> Doh!  Yes - I missed this.  Very sorry.  I will revert Kelvin's patch.

Thanks!

-- 
Eric Botcazou

Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

[Nick Clifton]

Hi Eric,
>>Hi Eric,
> 
> 
> Hi Nick,
> 
> 
>>Maybe...  It involves arithmetic comparisons of pointers to separately
>>allocated objects, which is a bit dubious in the first place.
> 
> 
> Note that it doesn't compare pointers directly, but first performs a 
> substraction on pointers which is strictly specified by C99 6.5.6 §9: "When 
> two pointers are substracted, both shall point to elements of the same array 
> object, or one past the last element of the array object; the result is the 
> difference of the subscripts of the two array elements."
> 
> So my understanding is that there isn't any GCC bug here.
> 
> 
>>Doh!  Yes - I missed this.  Very sorry.  I will revert Kelvin's patch.
> 
> 
> Thanks!
> 

Re: binutils-2.15 - work around for gcc optimization problem on sparc -sun-solaris2.7

[Nick Clifton]

Hi Eric,

   [oops - pressed send by mistake.  here is the reply I meant to write...]


>>Maybe...  It involves arithmetic comparisons of pointers to separately
>>allocated objects, which is a bit dubious in the first place.

> Note that it doesn't compare pointers directly, but first performs a 
> substraction on pointers which is strictly specified by C99 6.5.6 §9: "When 
> two pointers are substracted, both shall point to elements of the same array 
> object, or one past the last element of the array object; the result is the 
> difference of the subscripts of the two array elements."

Which is OK if the two pointers a re both pointing to elements of the 
same array.  If however they are pointing to malloc()'ed structures for 
example then the semantics are not defined.  So the problem with 
elf_sort_symbol() was that it was comparing pointers to entities that 
were not part of a single array...

Cheers
   Nick