Stuart McRobert | 1 Dec 01:09 2003
Picon
Picon

Re: PATCH: Re: sunsite.doc.ic.ac.uk mirror out of date

Hi,

You asked me to drop you a note once we had our new server
cluster up and running again.

	www.sunsite.org.uk
	ftp.sunsite.org.uk
	nfs.sunsite.org.uk

Are now taking the strain (old names include, sunsite.doc.ic.ac.uk,
sunsite.org.uk, src.doc.ic.ac.uk, etc).

If you could do a quick check of our gnu mirror, currently from

	ftp://ftp.fu-berlin.de/unix/gnu/

is all okay, and add an entry back into the mirrors.html list
that would be great.

Best wishes

	Stuart

> From: Gerald Pfeifer <pfeifer <at> dbai.tuwien.ac.at>
> To: Stuart McRobert <sm <at> doc.ic.ac.uk>
> Cc: gcc <at> gcc.gnu.org, gcc-patches <at> gcc.gnu.org,
>         william.gallafent <at> virgin.net
> In-Reply-To: <E19MjKy-0006HH-00 <at> duck.doc.ic.ac.uk>
> Subject: PATCH: Re: sunsite.doc.ic.ac.uk mirror out of date
> 
(Continue reading)

Roger Sayle | 1 Dec 03:20 2003

[PATCH] PR opt/12628: Call reg_scan before jump bypassing


The following patch is my proposed fix for PR optimization/12628
which is an ICE-on-valid caused by an obscure combination of compiler
flags.  The jump bypassing pass calls init_alias_analysis assuming
that register information is currently correct.  Unfortunately,
with the options "-O2 -fno-expensive-optimizations -fno-rerun-loop-opt"
reg_scan isn't called after either GCSE nor loop.  This then leads
to the ICE in init_alias_analysis.

The proposed (and least invasive) solution is for toplev.c's
rest_of_handle_jump_bypass to call reg_scan itself as it needs
aliasing information.  This is consistent with most other uses
of reg_scan in toplev.c where it's called immediately before
the pass that requires it, for example, the CSE passes.  Alas
this policy is not entirely consistent, with GCSE relying on
reg_scan being called at the end of addressof.  It would be
more consistent to move reg_scan from the end of addressof to
the beginning of GCSE, but this could potentially expose a
similar failure caused by another obscure set of compiler flags.

Longer term, it would be nice if, like the CFG, we maintained
a set of flags indicating whether these data structures were
valid following an optimization pass.  This would save us from
recalculating information that hasn't changed.

Unfortunately, calling reg_scan performs a traversal of the
RTL insn stream, so we can expect a minor slow-down from this
change :<.  It isn't clear whether GCSE or loop preserve the
"reg_scan" information, so this change is safe even when the
user hasn't specified wierd command line options.  However,
(Continue reading)

Jim Wilson | 1 Dec 03:49 2003

Re: Unreviewed patches

Janis Johnson wrote:
> As the author of these tests, this looks good to me, but I don't have
> authority to approve it.  See also this small fix to the dg-compat
> tests:
> 
>  http://gcc.gnu.org/ml/gcc-patches/2003-10/msg01114.html 

OK.

You should have the authority to approve patches to the compat testsuite 
soon.
--

-- 
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com

Jim Wilson | 1 Dec 03:59 2003

Re: Nonlocal gotos: wrong frame pointer handling

Andreas Krebbel wrote:
> 	* builtins.c (expand_builtin_longjmp): Added two memory clobbers.

OK.
--

-- 
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com

Kelley Cook | 1 Dec 04:00 2003
Picon

Re: fastjar.info not installed any longer?

Jim Wilson wrote:

 > OK.  Plus you will need the gcc/doc/install.texi change Joseph Myers
 > asked for.
 >
 > This was accidentally not sent to the gcc-patches list.  Looks like
 > maybe the cc line got a little confused.

Reposted to gcc-patches with changes.  Committed.

Kelley Cook
fastjar/
2003-12-01  Kelley Cook  <kcook <at> gcc.gnu.org>

	* Makefile.am: Define AM_MAKINFOFLAGS.  Remove Automake 1.4 hack.
	* Makefile.in: Regenerate with automake 1.7.6 & autoconf 2.57 tools.
	* aclocal.m4, config.h.in, configure: Likewise.
	* install-sh, missing, mkinstalldirs, stamp-h.in: Remove.

--- fastjar/Makefile.am	3 Nov 2003 13:21:00 -0000	1.13
+++ fastjar/Makefile.am	1 Dec 2003 02:49:39 -0000
 <at>  <at>  -57,6 +57,7  <at>  <at>  grepjar_LDADD = $(ZLIBS) $(LIBIBERTY)
 grepjar_DEPENDENCIES = $(ZDEPS) $(LIBIBERTY)

 AM_CFLAGS =  <at> fastjar_warn_cflags <at> 
+AM_MAKEINFOFLAGS = -I $(srcdir)/../gcc/doc/include

 TEXINFO_TEX = ../gcc/doc/include/texinfo.tex
(Continue reading)

Gabriel Dos Reis | 1 Dec 04:25 2003
Picon

DOC PATCH to c-tree.texi: CATCH_ALL_TYPES is gone


This is a follow-up to

  http://gcc.gnu.org/ml/gcc/2003-11/msg01516.html

Applied to mainline and 3_3-branch.

-- Gaby

Index: ChangeLog
===================================================================
RCS file: /cvs/gcc/gcc/gcc/ChangeLog,v
retrieving revision 2.1870
diff -p -r2.1870 ChangeLog
*** ChangeLog	30 Nov 2003 23:43:01 -0000	2.1870
--- ChangeLog	1 Dec 2003 03:19:14 -0000
***************
*** 1,3 ****
--- 1,7 ----
+ 2003-12-01 Gabriel Dos Reis <gdr <at> integrable-solutions.net>
+ 
+ 	* doc/c-tree.texi (Function Bodies): Update HANDLER documentation.
+ 
  2003-12-01  Alan Modra  <amodra <at> bigpond.net.au>

  	* config/rs6000/linux64.h (SUBSUBTARGET_OVERRIDE_OPTIONS): Test
Index: doc/c-tree.texi
===================================================================
RCS file: /cvs/gcc/gcc/gcc/doc/c-tree.texi,v
retrieving revision 1.47
(Continue reading)

Jim Wilson | 1 Dec 04:25 2003

Re: avoid unnecessary register saves for setjmp

Richard Kenner wrote:
> Well, you certainly need to save all the caller-saved registers if there
> are any nonlocal gotos (which you have in the Ada exception case), but,
> as I understand this change, it still does.

This is handled by a different piece of code that I did not touch. 
reload() has a check for current_function_nonlocal_label, and if true, 
it marks call-saved registers as live, so that they will be saved and 
restored.
--

-- 
Jim Wilson, GNU Tools Support, http://www.SpecifixInc.com

Daniel Berlin | 1 Dec 04:40 2003

[PATCH]: Updated patch to output dwarf2 namespace debug info

This one works and passes regtests.
Okay to commit?

2003-11-31  Daniel Berlin  <dberlin <at> dberlin.org>

	* dwarf2out.c (force_out_decl): New function.
	(gen_namespace_die): New function.
	(scope_die_for): Handle namespaces.
	(class_scope_p): Ditto.
	(gen_decl_die): Ditto.
	(dwarf2out_decl): Ditto.
	* tree.h: Move anonymous_namespace_name declaration to here.
	* tree.c: And the variable to here.
	* cp/decl.c: From here.
	* cp/cp-tree.h: And here.

Index: dwarf2out.c
===================================================================
RCS file: /cvs/gcc/gcc/gcc/dwarf2out.c,v
retrieving revision 1.465
diff -u -3 -p -r1.465 dwarf2out.c
--- dwarf2out.c	21 Nov 2003 21:18:45 -0000	1.465
+++ dwarf2out.c	1 Dec 2003 03:29:21 -0000
 <at>  <at>  -3740,7 +3740,7  <at>  <at>  static void push_decl_scope (tree);
 static void pop_decl_scope (void);
 static dw_die_ref scope_die_for (tree, dw_die_ref);
 static inline int local_scope_p (dw_die_ref);
-static inline int class_scope_p (dw_die_ref);
+static inline int class_or_namespace_scope_p (dw_die_ref);
 static void add_type_attribute (dw_die_ref, tree, int, int, dw_die_ref);
(Continue reading)

Feng Wang | 1 Dec 05:02 2003
Picon

[gfortran, patch] Old patches needed to apply.

Hi, Paul
   I think the two patches you said ok have not applied 
yet. Please help me apply. I have no CVS access writing.
Thanks.

The first Changelog entry:

2003-12-01  Feng Wang  <fengwang <at> nudt.edu.cn>

       * gfortran.fortran-
torture/excute/intrinsic_fraction_exponent.f90: Fix an 
error about checking the equal of floating point 
numbers and a return type error of EXPONENT intrinsic 
function.
       * gfortran.fortran-
torture/excute/intrinsic_scale.f90: Fix an error about 
checking the equal of floating point numbers and delete 
a variable.

The second:

2003-12-01  Feng Wang  <fengwang <at> nudt.edu.cn>

        * io.c (gfc_match_format): Check if missing 
format label.

  Feng Wang

(Continue reading)

Geoff Keating | 1 Dec 05:26 2003

Re: [PATCH] PR opt/12628: Call reg_scan before jump bypassing

Roger Sayle <roger <at> eyesopen.com> writes:

> The following patch is my proposed fix for PR optimization/12628
...
> Unfortunately, calling reg_scan performs a traversal of the
> RTL insn stream, so we can expect a minor slow-down from this
> change :<.  It isn't clear whether GCSE or loop preserve the
> "reg_scan" information, so this change is safe even when the
> user hasn't specified wierd command line options.  However,
> to try and combat this, the patch below contains some clean-ups
> to reg_scan and reg_scan_update.  I've also moved reg_scan's
> call to allocate_reg_info (which looks like it can do a
> significant amount of work) into the TV_REG_SCAN timevar so
> we can get a more accurate picture of how much reg_scan costs.
> 
> 
> My next patch addresses PR/12322 and will speed-up RTL's GCSE
> pass, so hopefully we'll still have a net performance win.
> 
> 
> This patch has been tested on i686-pc-linux-gnu with a complete
> "make bootstrap", all languages except treelang, and regression
> tested with a top-level "make -k check" with no new failures.
> 
> Ok for mainline?

This is OK.  Thanks for the clear explanation of the patch, it helped
a lot.

I expect that your clean-ups have no effect on reg_scan's performance,
(Continue reading)


Gmane