trivial PATCH to dwarf2read.c:dwarf_attr_name

trivial PATCH to dwarf2read.c:dwarf_attr_name

[Jason Merrill]

[-- Attachment #1: Type: text/plain, Size: 238 bytes --]

I'm checking this in under the obvious rule; DW_AT_MIPS_linkage_name is
used on all targets.

2003-08-26  Jason Merrill  <jason@redhat.com>

	* dwarf2read.c (dwarf_attr_name): Move DW_AT_MIPS_linkage_name
	case out of #ifdef MIPS block.


[-- Warning: decoded text below may be mangled, UTF-8 assumed --]
[-- Attachment #2: Type: text/x-patch, Size: 801 bytes --]

*** dwarf2read.c.~1~	2003-08-15 19:38:15.000000000 -0400
--- dwarf2read.c	2003-08-15 19:35:38.000000000 -0400
*************** dwarf_attr_name (register unsigned attr)
*** 6049,6057 ****
        return "DW_AT_MIPS_loop_unroll_factor";
      case DW_AT_MIPS_software_pipeline_depth:
        return "DW_AT_MIPS_software_pipeline_depth";
      case DW_AT_MIPS_linkage_name:
        return "DW_AT_MIPS_linkage_name";
- #endif
  
      case DW_AT_sf_names:
        return "DW_AT_sf_names";
--- 6049,6057 ----
        return "DW_AT_MIPS_loop_unroll_factor";
      case DW_AT_MIPS_software_pipeline_depth:
        return "DW_AT_MIPS_software_pipeline_depth";
+ #endif
      case DW_AT_MIPS_linkage_name:
        return "DW_AT_MIPS_linkage_name";
  
      case DW_AT_sf_names:
        return "DW_AT_sf_names";

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Andrew Cagney]

> I'm checking this in under the obvious rule; DW_AT_MIPS_linkage_name is
> used on all targets.
> 
> 2003-08-26  Jason Merrill  <jason@redhat.com>
> 
> 	* dwarf2read.c (dwarf_attr_name): Move DW_AT_MIPS_linkage_name
> 	case out of #ifdef MIPS block.
> 

Jason, exactly what prompted the change?  In particular which GCC?

Andrew

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Jason Merrill]

On Tue, 26 Aug 2003 23:56:42 -0400, Andrew Cagney <ac131313@redhat.com> wrote:

>> I'm checking this in under the obvious rule; DW_AT_MIPS_linkage_name is
>> used on all targets.
>> 2003-08-26  Jason Merrill  <jason@redhat.com>
>> 	* dwarf2read.c (dwarf_attr_name): Move DW_AT_MIPS_linkage_name
>> 	case out of #ifdef MIPS block.
>>
>
> Jason, exactly what prompted the change?  In particular which GCC?

I don't remember.  I think it's been in my local tree for years.  It's
quite likely that I just noticed it.

Jason

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Andrew Cagney]

> On Tue, 26 Aug 2003 23:56:42 -0400, Andrew Cagney <ac131313@redhat.com> wrote:
> 
> 
>>> I'm checking this in under the obvious rule; DW_AT_MIPS_linkage_name is
>>> used on all targets.
>>> 2003-08-26  Jason Merrill  <jason@redhat.com>
>>> 	* dwarf2read.c (dwarf_attr_name): Move DW_AT_MIPS_linkage_name
>>> 	case out of #ifdef MIPS block.
>>>
> 
>>
>> Jason, exactly what prompted the change?  In particular which GCC?
> 
> 
> I don't remember.  I think it's been in my local tree for years.  It's
> quite likely that I just noticed it.

Can you please find out.  There must be a stronger rationale then "it's 
been in my local tree for years".

Andrew

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Daniel Jacobowitz]

On Wed, Aug 27, 2003 at 10:46:42AM -0400, Andrew Cagney wrote:
> >On Tue, 26 Aug 2003 23:56:42 -0400, Andrew Cagney <ac131313@redhat.com> 
> >wrote:
> >
> >
> >>>I'm checking this in under the obvious rule; DW_AT_MIPS_linkage_name is
> >>>used on all targets.
> >>>2003-08-26  Jason Merrill  <jason@redhat.com>
> >>>	* dwarf2read.c (dwarf_attr_name): Move DW_AT_MIPS_linkage_name
> >>>	case out of #ifdef MIPS block.
> >>>
> >
> >>
> >>Jason, exactly what prompted the change?  In particular which GCC?
> >
> >
> >I don't remember.  I think it's been in my local tree for years.  It's
> >quite likely that I just noticed it.
> 
> Can you please find out.  There must be a stronger rationale then "it's 
> been in my local tree for years".

Well, the patch is certainly right.  GCC generates
DW_AT_MIPS_linkage_name on all targets, and has for as far back as I
have GCC sources to check it.

We just don't use dwarf_attr_name much, so it didn't matter.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Andrew Cagney]

> Can you please find out.  There must be a stronger rationale then "it's 
>> been in my local tree for years".
> 
> 
> Well, the patch is certainly right.  GCC generates
> DW_AT_MIPS_linkage_name on all targets, and has for as far back as I
> have GCC sources to check it.

Can you please add a comment explaining exactly when/where this occures 
then?  Someone looking that that code is going to assume that 
DW_AT_MIPS_linkage_name is outside the #ifdef MIPS because of a 
host/target cross platform problem and _not_ because GCC uses it for all 
architectures.

Andrew

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Daniel Jacobowitz]

On Wed, Aug 27, 2003 at 11:57:41AM -0400, Andrew Cagney wrote:
> >Can you please find out.  There must be a stronger rationale then "it's 
> >>been in my local tree for years".
> >
> >
> >Well, the patch is certainly right.  GCC generates
> >DW_AT_MIPS_linkage_name on all targets, and has for as far back as I
> >have GCC sources to check it.
> 
> Can you please add a comment explaining exactly when/where this occures 
> then?  Someone looking that that code is going to assume that 
> DW_AT_MIPS_linkage_name is outside the #ifdef MIPS because of a 
> host/target cross platform problem and _not_ because GCC uses it for all 
> architectures.

Actually, I have no idea where the #ifdef MIPS came from, but it should
be removed.  There's no reason to recognize extensions only on a
specific platform, and their other uses are not protected.

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Andrew Cagney]

> On Wed, Aug 27, 2003 at 11:57:41AM -0400, Andrew Cagney wrote:
> 
>> >Can you please find out.  There must be a stronger rationale then "it's 
> 
>> >>been in my local tree for years".
> 
>> >
>> >
>> >Well, the patch is certainly right.  GCC generates
>> >DW_AT_MIPS_linkage_name on all targets, and has for as far back as I
>> >have GCC sources to check it.
> 
>> 
>> Can you please add a comment explaining exactly when/where this occures 
>> then?  Someone looking that that code is going to assume that 
>> DW_AT_MIPS_linkage_name is outside the #ifdef MIPS because of a 
>> host/target cross platform problem and _not_ because GCC uses it for all 
>> architectures.
> 
> 
> Actually, I have no idea where the #ifdef MIPS came from, but it should
> be removed.  There's no reason to recognize extensions only on a
> specific platform, and their other uses are not protected.

That's a separate problem.  That specific attribute should have gained a 
comment explaining how/why it is used by non-MIPS platforms.  Otherwize 
no one knows what's about to hit them.

Andrew

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Daniel Jacobowitz]

On Wed, Aug 27, 2003 at 12:14:48PM -0400, Andrew Cagney wrote:
> >On Wed, Aug 27, 2003 at 11:57:41AM -0400, Andrew Cagney wrote:
> >
> >>>Can you please find out.  There must be a stronger rationale then "it's 
> >
> >>>>been in my local tree for years".
> >
> >>>
> >>>
> >>>Well, the patch is certainly right.  GCC generates
> >>>DW_AT_MIPS_linkage_name on all targets, and has for as far back as I
> >>>have GCC sources to check it.
> >
> >>
> >>Can you please add a comment explaining exactly when/where this occures 
> >>then?  Someone looking that that code is going to assume that 
> >>DW_AT_MIPS_linkage_name is outside the #ifdef MIPS because of a 
> >>host/target cross platform problem and _not_ because GCC uses it for all 
> >>architectures.
> >
> >
> >Actually, I have no idea where the #ifdef MIPS came from, but it should
> >be removed.  There's no reason to recognize extensions only on a
> >specific platform, and their other uses are not protected.
> 
> That's a separate problem.  That specific attribute should have gained a 
> comment explaining how/why it is used by non-MIPS platforms.  Otherwize 
> no one knows what's about to hit them.

I have no idea what you mean by "hit".  Care to expand?

-- 
Daniel Jacobowitz
MontaVista Software                         Debian GNU/Linux Developer

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Jason Merrill]

On Wed, 27 Aug 2003 10:46:42 -0400, Andrew Cagney <ac131313@redhat.com> wrote:

>> On Tue, 26 Aug 2003 23:56:42 -0400, Andrew Cagney <ac131313@redhat.com> wrote:
>>
>>>> I'm checking this in under the obvious rule; DW_AT_MIPS_linkage_name is
>>>> used on all targets.
>>>> 2003-08-26  Jason Merrill  <jason@redhat.com>
>>>> 	* dwarf2read.c (dwarf_attr_name): Move DW_AT_MIPS_linkage_name
>>>> 	case out of #ifdef MIPS block.

>>> Jason, exactly what prompted the change?  In particular which GCC?
>> I don't remember.  I think it's been in my local tree for years.  It's
>> quite likely that I just noticed it.
>
> Can you please find out.  There must be a stronger rationale then "it's
> been in my local tree for years".

The only rationale I have is that GCC emits DW_AT_MIPS_linkage_name on all
targets, so gdb should recognize it on all targets.

This isn't something that would affect normal operation--dwarf_attr_name is
only used by dump_die, which is only called when there's some sort of
internal error.  I probably just noticed this in a grep for
DW_AT_MIPS_linkage_name while I was working on something else.

Jason

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Andrew Cagney]

>> That's a separate problem.  That specific attribute should have gained a 
>> comment explaining how/why it is used by non-MIPS platforms.  Otherwize 
>> no one knows what's about to hit them.
> 
> 
> I have no idea what you mean by "hit".  Care to expand?

What would happen if MIPS was deleted.  Presumably this attribute would 
be deleted along with the rest of the code (it looks mips specific and 
there's no commentary to suggest otherwize).

Andrew

Re: trivial PATCH to dwarf2read.c:dwarf_attr_name

[Andrew Cagney]

> The only rationale I have is that GCC emits DW_AT_MIPS_linkage_name on all
> targets, so gdb should recognize it on all targets.
> 
> This isn't something that would affect normal operation--dwarf_attr_name is
> only used by dump_die, which is only called when there's some sort of
> internal error.  I probably just noticed this in a grep for
> DW_AT_MIPS_linkage_name while I was working on something else.

Ok, can you add this as a comment then?

Andrew