Ping 4.0 regressions (again)

[Roger Sayle]

On Sun, 20 Mar 2005, Andrew Pinski wrote:
> http://gcc.gnu.org/ml/gcc-patches/2005-03/msg01871.html
> [PATCH] PR middle-end/20539: More tree type safety issues, C and C++
> front-end

The middle-end bits are approved, but the C front-end change needs
approval :)


> http://gcc.gnu.org/ml/gcc-patches/2005-03/msg01762.html
> PR PCH/20568
> [PATCH]: Fix 4 current darwin pch debug regressions

I'll approve all the non C frontend bits (which are independent of
the final c-opts.c change).  This should allow a C front-end maintainer
(in addition to a global write maintainer) to approve the final bit.


> http://gcc.gnu.org/ml/gcc-patches/2005-03/msg01529.html
> [PATCH] Fix PR middle-end/20263, Incorrect asm for global register vars

This is OK for mainline and the 4.0 branch.


> http://gcc.gnu.org/ml/gcc-patches/2005-03/msg00621.html
> [patch] for PR 20249, ICE with -fprofile-arcs on some targets (PPC
> included)

This is OK for mainline and the 4.0 branch, but it should also include
a new testcase derived from the variables.i attached to the PR.  I'd
also be interested in knowing if there's any performance impact of this
change.  It might be that we need to check the modes match when used,
rather than prevent them being recorded.


> http://gcc.gnu.org/ml/gcc-patches/2005-02/msg01592.html
> [PATCH]: PR 20203: unrecognizable insn building binutils, middle-end,
> expanding memset

This is OK for mainline and the 4.0 branch.  Again it needs a testcase
derived from the small example attached to the bugzilla PR.


> http://gcc.gnu.org/ml/gcc-patches/2004-11/msg00138.html
> [PATCH] PR17913:[4.0 Regression] ICE jumping into statement expression
> 	This patch has been partly approved for more than 3 months now,
>       getting the C front-end part approved would be nice.

Perhaps we should deprecate the C front-end in 4.1 and removed it in 4.2
:-)


> http://gcc.gnu.org/ml/gcc-patches/2004-11/msg02664.html
> [PATCH] [PR18071] -Winline and -fno-default-inline

The tree-inline.c bit is fine, the c-decl.c change...



Many thanks to Andrew and the patch contributors.

Roger
--