Unreviewed stage2 patches

Unreviewed stage2 patches

[Jakub Jelinek]

Hi!

I have still 4 stage2-ish patches I'd like to see in GCC 4.3.
The 2 new extensions and diagnostic change are meant for glibc 2.7, so that
we can switch the remaining checking function-like macros to inline
functions, plus one new optimization (although I plan to rewrite it
using a propagator for 4.4, already the simple pass has many hits during
gcc bootstrap and so it would help 4.3).

__artificial__ attribute (except the builtins.c hunk which is addressed
by the diagnostic changes below):
http://gcc.gnu.org/ml/gcc-patches/2007-08/msg02300.html

redundant zero store elimination optimization (simplistic version,
but nevertheless is able to trigger many times during gcc bootstrap):
http://gcc.gnu.org/ml/gcc-patches/2007-09/msg00663.html

diagnostic changes to print virtual call stack:
http://gcc.gnu.org/ml/gcc-patches/2007-09/msg00979.html

__error_decl__ and __warning_decl__ attributes:
http://gcc.gnu.org/ml/gcc-patches/2007-09/msg00876.html

Thanks.

	Jakub

Re: Unreviewed stage2 patches

[Jason Merrill]

Jakub Jelinek wrote:
> __error_decl__ and __warning_decl__ attributes:
> http://gcc.gnu.org/ml/gcc-patches/2007-09/msg00876.html

Why "_decl"?

Jason

Re: Unreviewed stage2 patches

[Jakub Jelinek]

On Fri, Sep 21, 2007 at 01:13:34AM -0400, Jason Merrill wrote:
> Jakub Jelinek wrote:
> >__error_decl__ and __warning_decl__ attributes:
> >http://gcc.gnu.org/ml/gcc-patches/2007-09/msg00876.html
> 
> Why "_decl"?

Because glibc already uses for such things __warndecl and __errordecl
macros, except that they result only into link time warnings or
link time errors without gcc help.

I don't care about the names though, can be anything else,
all I care is about the semantics.
So __error_fn__, __error_function__, __error_call__, __error_on_call__
or even just __error__ are ok with me.

	Jakub

error_decl, warning_decl (wa Re: Unreviewed stage2 patches)

[Jason Merrill]

Jakub Jelinek wrote:
> On Fri, Sep 21, 2007 at 01:13:34AM -0400, Jason Merrill wrote:
>> Jakub Jelinek wrote:
>>> __error_decl__ and __warning_decl__ attributes:
>>> http://gcc.gnu.org/ml/gcc-patches/2007-09/msg00876.html
>> Why "_decl"?
> 
> Because glibc already uses for such things __warndecl and __errordecl
> macros, except that they result only into link time warnings or
> link time errors without gcc help.

Ah.  But there the name is a directive rather than an attribute.

> I don't care about the names though, can be anything else,
> all I care is about the semantics.
> So __error_fn__, __error_function__, __error_call__, __error_on_call__
> or even just __error__ are ok with me.

Any of those would be fine with me.  With that change the patch is OK.

Jason

Re: error_decl, warning_decl (wa Re: Unreviewed stage2 patches)

[Gabriel Dos Reis]

On Fri, 21 Sep 2007, Jason Merrill wrote:

| > I don't care about the names though, can be anything else,
| > all I care is about the semantics.
| > So __error_fn__, __error_function__, __error_call__, __error_on_call__
| > or even just __error__ are ok with me.
| 
| Any of those would be fine with me.  With that change the patch is OK.

I prefer the simpler __error__.

-- Gaby