Hi!
This patch adds support for OpenMP 3.1 atomics to C++ FE.
Tested on x86_64-linux, committed to gomp-3_1-branch.
2011-04-27 Jakub Jelinek ja...@redhat.com
* parser.c (cp_parser_omp_atomic): Handle parsing OpenMP 3.1 atomics.
Adjust finish_omp_atomic caller.
*
This fixes TARGET_MEM_REF handling in indirect_refs_may_alias_p,
unifying handling of it and MEM_REFs where possible.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2011-04-27 Richard Guenther rguent...@suse.de
* tree-ssa-alias.c
Hi Guys,
I am applying the patch below to remove the duplicate register classes
from the FRV backend. This was causing problems with the IRA
initialization code.
Cheers
Nick
gcc/ChangeLog
2011-04-27 Nick Clifton ni...@redhat.com
* config/frv/frv.h (enum reg_class): Delete
Richard Henderson schrieb:
Why are you adding optimize to all these insns? None of them will
be matched unless combine is run, which implies optimization.
Here is a patch without optimize in the insn conditions.
The optimize condition is still present in the insv expander because I
do not
On Wed, 27 Apr 2011, Thomas Schwinge wrote:
One (minor) additional idea: for clarity, shouldn't it now be possible to
remove any #undef GNU_USER_TARGET_OS_CPP_BUILTINS, #undef
GNU_USER_DYNAMIC_LINKER, #undef GNU_USER_LINK_EMULATION, etc. from
*gnu.h? I don't expect you to test this; I'll
On Tue, 26 Apr 2011, Cary Coutant wrote:
@@ -1856,6 +1871,8 @@ set_debug_level (enum debug_info_type type, int
extended, const char *arg,
else
opts-x_debug_info_level = (enum debug_info_levels) argval;
}
+
+ generate_debug_line_table = debug_info_level =
On Wed, 27 Apr 2011, Kaz Kojima wrote:
Joseph S. Myers jos...@codesourcery.com wrote:
I think you should add a testcase to gcc.c-torture/compile, unless there
is already one that this patch fixes.
Ah, indeed. How about the attached testcase?
Yes, that testcase looks like what I had in
This fixes PR48772, we can't update stmts with virtual defs during
VN elimination as that may destroy SSA names we use in the VN lattice.
So we have to defer it, similar to removing stmts.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2011-04-27 Richard
On Thu, Jul 29, 2010 at 7:14 AM, H.J. Lu hjl.to...@gmail.com wrote:
On Wed, Jul 28, 2010 at 8:50 PM, H.J. Lu hjl.to...@gmail.com wrote:
On Wed, Jul 28, 2010 at 6:07 PM, H.J. Lu hjl.to...@gmail.com wrote:
On Tue, Jul 27, 2010 at 1:20 PM, Xinliang David Li davi...@google.com
wrote:
On Tue, Jul
On Wed, Apr 20, 2011 at 10:08:21AM -0700, Nathan Froyd wrote:
This patch changes most of the uses of build_function_type in the Ada to
use build_function_type_list. There are a handful of
build_function_type calls left; replacing those will have to wait until
we get a
Hello,
spu_expand_epilogue used to emit a barrier after the return jump; this seems
to confuse the CFG layer after some recent changes, causing internal errors.
Since other backends don't emit a barrier here either, I've removed this
from the SPU backend as well.
Tested on spu-elf, committed to
Hello,
the SPU __multi3 and __udivmodti4 routines were using code that violates
the aliasing rules for type-punning between TImode and vector types.
This now causes errors when building those files. The patch below
replaces this code by type-punning via unions.
Tested on spu-elf, committed to
Hello,
since recent changes to haifa-sched.c, the SPU back-end runs into
internal errors due to its use of the sched_emit_insn routine to
emit instruction prefetch instructions during final scheduling.
Now that scheduled instructions aren't placed into the instruction
stream directly,
Hello Ralf,
* Rainer Orth wrote on Tue, Apr 26, 2011 at 05:28:19PM CEST:
it's been a week since I answered your questions on this patch. Could
you please have a look?
Sorry for the delay. I'm practically AFK until the weekend or maybe
next weekend, whenever I have connectivity again after
On Wed, Apr 27, 2011 at 5:16 AM, Jan Hubicka hubi...@ucw.cz wrote:
Hi,
I don't really have testcase for the HP nor AIX ICE, however I can reproduce
same ICE when I hack x86 to
not use ctors/dtors. This patch fixes it - the problem is that ipa-prop
ignore newly added functions
(the global
As preparation for a fix for PR48764 this removes the need for PTA
to allocate and preserve new decls. We still need to allocate
UIDs from the decl namespace to have something to point to from
points-to sets, but that can be done with allocating them directly.
Bootstrapped and tested on
Hello!
Mode can't be passed to match_parallel predicates, so instead of
calling avx_{vpermilp,vperm2f128}_parallel functions via predicates,
we can call them from insn condition directly.
2011-04-27 Uros Bizjak ubiz...@gmail.com
* config/i386/predicates.md (avx_vpermilp_*_operand):
On Apr 22, 2011, at 3:54 PM, Mike Stump wrote:
On Apr 22, 2011, at 8:12 AM, Nicola Pero wrote:
This patch fixes a building annoyance that I had when building on a new
machine (an x86_64 gnu/linux box).
The building failed. It was down to two problems:
* due to how I got a copy of the GCC
Ping?
On Apr 22, 2011, at 3:01 PM, Mike Stump wrote:
On Apr 22, 2011, at 3:28 AM, Eric Botcazou wrote:
This patch fixes up hot/cold partitioning on ports that don't have long
conditional branches. I'll note that the entire file has lots of other
jump optimizations that are suspect.
Do you
This may have caused:
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48791
Oops, yes, it is mine. The insertion hook at expansion time is incorrectly
called
after function is expanded, not before.
ipa-prop should deregister itself earlier, but that can be done incrementally.
I am testing
Hi!
This patch adds Fortran support for OpenMP 3.1 atomics.
Tested on x86_64-linux, committed to gomp-3_1-branch.
2011-04-27 Jakub Jelinek ja...@redhat.com
* gfortran.h (gfc_statement): Add ST_OMP_END_ATOMIC.
(gfc_omp_atomic_op): New enum typedef.
(struct gfc_code):
This patch adds the google release branches to the 'distribution
branches' section of svn.html.
Committed to wwwdocs.
Diego.
Index: htdocs/svn.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/svn.html,v
retrieving revision 1.160
diff -u
Hi Guys,
I am applying the patch below to the mainline and 4.6 branch sources.
It fixes a few GCC testsuite failures for the RX target, specifically:
gcc.dg/20020312-2.c
gcc.dg/pr28796-2.c
gcc.dg/torture/builtin-cproj-1.c
gcc.dg/torture/type-generic-1.c
This moves PTA changed flag handling during solving from using
an sbitmap and an explicitly managed change_count to a bitmap.
Way easier to get right and also a requirement for pending work
that adds variables on-the-fly during solving (which would
require growing the sbitmap).
Bootstrapped on
Hi!
#pragma omp atomic write RHS can be arbitrary expression (not mentioning
the LHS), so it is wrong to parse it just as unary expression.
Fixed thusly, tested on x86_64-linux, committed to gomp-3_1-branch.
2011-04-27 Jakub Jelinek ja...@redhat.com
* c-parser.c (c_parser_omp_atomic):
Continuing the toplevel cleanups separating the cases disabling
different subdirectories - on the basis that eventually this
information should come from the subdirectories, and that as
previously discussed much of the default disabling of ${libgcj}
doesn't actually make sense - this patch
Honza, any more comments?
Thanks,
David
On Mon, Apr 25, 2011 at 9:35 AM, Xinliang David Li davi...@google.com wrote:
Is the patch ok?
Thanks,
David
On Fri, Apr 22, 2011 at 12:48 PM, Jan Hubicka hubi...@ucw.cz wrote:
Please review the new patch which only implements cfg checksum.
The
This regression was introduced by the fix for PR 35112, but it turns out
that fixing it is a simple matter of moving one line up in a function so
that we are committed to the tentative parse before we call
cp_parser_lookup_name.
Tested x86_64-pc-linux-gnu, applying to trunk and 4.6. Will
Please don't forget about changelogs..
Index: tree.c
===
--- tree.c(revision 172802)
+++ tree.c(working copy)
@@ -8513,14 +8513,12 @@ dump_tree_statistics (void)
The crc bits was already reviewed, right?
- else if
Hi,
this patch finishes the basic predicate infrastructure of ipa-inline by adding
predicates to edges.
The predicates are combined, so we can work out very simple cases where
function call becmes
dead after inlining. In tramp3d we get couple hundred cases like that, so it
seems to make
On 21/04/11 16:58, Joseph S. Myers wrote:
You need to add
%inherit GCC_4.7.0 GCC_4.6.0
GCC_4.7.0 {
}
Ok, done. Is this OK now?
Andrew
2011-04-27 Andrew Stubbs a...@codesourcery.com
gcc/
* config/arm/arm.c (arm_init_libfuncs): Change __gnu_f2h_ieee to
__aeabi_f2h, __gnu_f2h_alternative
Hello,
PR middle-end/43085 is a crash of cc1plus when building libstdc++ during
make profiledbootstrap on the 4.5 branch. The crash is caused by a
mis-compile of the add_method routine with -fprofile-use due to a
dataflow bug during the final if-conversion pass. For more detailed
analysis see
This patch, relative to a tree with
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02123.html applied,
continues the cleanup of toplevel cases relating to disabling Java or
Java libraries by arranging for Java to be disabled (via
unsupported_languages) on targets not supporting libffi, and removing
...for a validator never implemented. Andrea Bocci did some work
around this in 2003, but we never completed the effort.
validate-empty.png, validate-error.png, and validate-ok.png are
the three images.
Gerald
- else if (entry-checksum != checksum)
+ else if (entry-lineno_checksum != lineno_checksum
+ || entry-cfg_checksum != cfg_checksum)
{
error (coverage mismatch for function %u while reading
execution counters,
On Wed, Apr 27, 2011 at 10:03 AM, Jan Hubicka hubi...@ucw.cz wrote:
- else if (entry-checksum != checksum)
+ else if (entry-lineno_checksum != lineno_checksum
+ || entry-cfg_checksum != cfg_checksum)
{
error (coverage mismatch for function
On Apr 27, 2011, at 8:50 AM, Joseph S. Myers wrote:
Continuing the toplevel cleanups separating the cases disabling
different subdirectories
I've audited the two patches for darwin and they seem safe. Took me a second
to figure out why though.
+ generate_debug_line_table = debug_info_level = DINFO_LEVEL_NORMAL;
set_debug_level should not use global state; this needs to check
opts-x_debug_info_level (not the global debug_info_level) and set
opts-x_generate_debug_line_table.
Oops, missed that. Thanks!
-cary
Hi please review the trivial patch below. It reduces race conditions
in value profiling. Another trivial change (to initialize
function_list struct) is also included.
Bootstrapped and regression tested on x86-64/linux.
Thanks,
David
2011-04-27 Xinliang David Li davi...@google.com
*
Add a flag -fpph-dump-tree to dump the namespace trees read and written by PPH.
This dump could use some clarification, but it can wait for another day.
Replace PPH calls to pushdecl_with_scope with pushdecl_into_namespace. It
turns out that pushdecl_with_scope always inserts into the current
Hi,
On Wed, 27 Apr 2011, Jan Hubicka wrote:
*** false_predicate (void)
*** 163,168
--- 166,195
}
+ /* Return true if P is (false). */
+
+ static inline bool
+ true_predicate_p (struct predicate *p)
Comment doesn't match function.
+ {
+ return
This discussion continues in the thread
[patch] Split Parse Timevar (rev 2) (issue4433076)
which has a new uploaded patch.
On 4/23/11, Jason Merrill ja...@redhat.com wrote:
On 04/22/2011 06:41 PM, Lawrence Crowl wrote:
On 4/21/11, Jason Merrillja...@redhat.com wrote:
On 04/21/2011 07:17 PM,
On Wed, Apr 27, 2011 at 6:27 PM, Ulrich Weigand uweig...@de.ibm.com wrote:
Hello,
PR middle-end/43085 is a crash of cc1plus when building libstdc++ during
make profiledbootstrap on the 4.5 branch. The crash is caused by a
mis-compile of the add_method routine with -fprofile-use due to a
commit 44417f1103b0823f950f539a38a5539b8ff74330
Author: Ollie Wild a...@google.com
Date: Mon Apr 25 21:55:44 2011 -0400
When using GRTE (libc) and linking statically, instead of just -lc add:
-lc -lnss_borg -lnss_cache -lnss_dns -lnss_files -lresolv
to the link by
Ping^2. This patch
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00919.html is still pending
review (except insofar as it relates to GNU Hurd).
--
Joseph S. Myers
jos...@codesourcery.com
On Wed, Apr 27, 2011 at 15:13, Ollie Wild a...@google.com wrote:
ChangeLog:
2011-04-27 Ollie Wild a...@google.com
* config.gcc: Add GRTE detection.
* config/i386/linux.h (LINUX_GRTE_EXTRA_SPECS): New macro.
(SUBTARGET_EXTRA_SPECS): Add LINUX_GRTE_EXTRA_SPECS.
On Wed, Apr 27, 2011 at 08:53, Janne Blomqvist
blomqvist.ja...@gmail.com wrote:
On Wed, Apr 27, 2011 at 07:09, Jerry DeLisle jvdeli...@frontier.com wrote:
On 04/25/2011 07:36 AM, Janne Blomqvist wrote:
On Mon, Apr 25, 2011 at 14:44, Jerry DeLislejvdeli...@frontier.com
wrote:
On 04/25/2011
Hello,
this patch adds the ability to gimple-fold to operate for truth and/or
operations
on folded binary-and/or optimized truth trees - as done by fold-const. As fold
converts trivial operations like (A B) to (A B) != 0, in most cases further
folding of truth and/or trees wasn't done.
folding
set_debug_level should not use global state; this needs to check
opts-x_debug_info_level (not the global debug_info_level) and set
opts-x_generate_debug_line_table.
Oops, missed that. Thanks!
I've uploaded the revised patch to
http://codereview.appspot.com/4440072 -- should I also post it
On Wed, Apr 27, 2011 at 10:03 PM, Kai Tietz ktiet...@googlemail.com wrote:
Hello,
this patch adds the ability to gimple-fold to operate for truth and/or
operations
on folded binary-and/or optimized truth trees - as done by fold-const. As
fold
converts trivial operations like (A B) to (A
On Wed, Apr 27, 2011 at 10:26 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Wed, Apr 27, 2011 at 10:03 PM, Kai Tietz ktiet...@googlemail.com wrote:
Hello,
this patch adds the ability to gimple-fold to operate for truth and/or
operations
on folded binary-and/or optimized truth
First, I'd be glad if someone could review my patch at
http://gcc.gnu.org/ml/fortran/2011-04/msg00271.html
* * *
Secondly, the attached patch fixes a segfault after printing an error
about the invalid function-return type. It has been committed as Rev.
173048. As it is a 4.6/4.7 regression,
I moved some of your ChangeLog entries to different ChangeLog files.
In general, each sub-directory has its own ChangeLog file, so please
use those.
This is not completely standard, unfortunately. The rule of thumb is
to check whether the directory you are modifying has a ChangeLog file.
If so,
On Wednesday 27 April 2011 01:06:26 Steve Kargl wrote:
Index: primary.c
===
--- primary.c (revision 172974)
+++ primary.c (working copy)
@@ -541,6 +541,17 @@ match_real_constant (gfc_expr **result,
goto
On Tuesday 26 April 2011 23:41:17 Tobias Burnus wrote:
This patch fixes three issues, (a) and (b) are 4.6/4.7 regressions.
a) PR48112: Due to incomplete resolution, there was an ICE when writing
the module file. Solution: Back-out the patch which introduced the
incomplete resolution - and
On Wed, Apr 27, 2011 at 4:05 PM, Ville Voutilainen
ville.voutilai...@gmail.com wrote:
On 27 April 2011 10:14, Ville Voutilainen ville.voutilai...@gmail.com wrote:
The -fpermissive because there are conversions from literals to char*,
and c++ nowadays doesn't like that. I'm not at the build
On 28 April 2011 00:56, Gabriel Dos Reis g...@integrable-solutions.net wrote:
Seems to fall over itself in a bunch of places. C++ bootstrap smells
like a red herring
to me at this point.
Hmm, do you mean GCC fails to bootstrap with C++ a compiler before your mods?
I did make distclean,
On Wed, Apr 27, 2011 at 4:58 PM, Ville Voutilainen
ville.voutilai...@gmail.com wrote:
On 28 April 2011 00:56, Gabriel Dos Reis g...@integrable-solutions.net
wrote:
Seems to fall over itself in a bunch of places. C++ bootstrap smells
like a red herring
to me at this point.
Hmm, do you mean
On Wed, Mar 23, 2011 at 7:17 AM, Richard Guenther rguent...@suse.de wrote:
For quite some releases we'd wanted some (memory) CSE optimization
during early opts. Currently we rely heavily on SRA which doesn't
always fulfill the make code only smaller promises which is
what early opts should
On Wednesday 27 April 2011 23:10:14 Steve Kargl wrote:
On Wed, Apr 27, 2011 at 10:54:37PM +0200, Mikael Morin wrote:
On Wednesday 27 April 2011 01:06:26 Steve Kargl wrote:
It's an 'if -- else if' construct. If gfc_notify_std == FAILURE, then
the error message is issues and the function
On Wed, Apr 27, 2011 at 19:23, Xinliang David Li davi...@google.com wrote:
This can happen when the module does not have function bodies.
Ok after testing?
Thanks,
David
2011-04-27 Xinliang David Li davi...@google.com
* value-prof.c (del_node_map): Do not free
null
Ok -- false alarm ...
David
On Wed, Apr 27, 2011 at 5:30 PM, Nathan Froyd froy...@codesourcery.com wrote:
On Wed, Apr 27, 2011 at 04:23:42PM -0700, Xinliang David Li wrote:
This can happen when the module does not have function bodies.
- VEC_free (cgraph_node_ptr, heap, cgraph_node_map);
This patch adds new parameters to control peeling when profile
feedback information is available.
For google/main.
Tested:
bootstrapped on x86_64.
2011-04-27 Sharad Singhai sing...@google.com
* gcc/params.def: Add new parameters to control peeling.
*
On Wed, Apr 27, 2011 at 20:52, Sharad Singhai sing...@google.com wrote:
This patch adds new parameters to control peeling when profile
feedback information is available.
For google/main.
Tested:
bootstrapped on x86_64.
Looks OK with a couple of minor nits below.
2011-04-27 Sharad
See the following emails for a few inlined patches
to /trunk/gcc/doc/extend.texi (revision 172911):
[1] Docs: extend.texi: Add missing semicolon for consistency
[2] Docs: extend.texi: Remove trailing blanks from lines
[3] Docs: extend.texi: Rearrange nodes; no text was removed or added
---
trunk/gcc/doc/extend.texi |2 +-
1 files changed, 1 insertions(+), 1 deletions(-)
diff --git a/trunk/gcc/doc/extend.texi b/trunk/gcc/doc/extend.texi
index eddff95..c154958 100644
--- a/trunk/gcc/doc/extend.texi
+++ b/trunk/gcc/doc/extend.texi
@@ -3997,7 +3997,7 @@
@smallexample
sed -i s/[ $(printf '\t')]\{1,\}\$// trunk/gcc/doc/extend.texi
---
trunk/gcc/doc/extend.texi | 82 ++--
1 files changed, 41 insertions(+), 41 deletions(-)
diff --git a/trunk/gcc/doc/extend.texi b/trunk/gcc/doc/extend.texi
index c154958..cdbf69f 100644
The rearrangement performed in the previous patch left the text in
somewhat of an inconsistent state in terms of the flow of concepts;
this patch improves the flow of concepts to something more sane by
editing the introductions to these nodes:
Attribute Syntax
Type Attributes
Variable
Richard,
Your recent patch broke --enable-build-with-cxx because internal-fn.h
was using good 'ol C-style name lookup trick that behaves badly with
C++ linkage rules. Fixed thusly. Applied as obvious.
-- Gaby
2011-04-27 Gabriel Dos Reis g...@integrable-solutions.net
*
In the discussion of 48760, Gaby suggested that we allow
list-initialization of built-in complex numbers. This made a lot of
sense to me, so I've gone ahead and implemented it.
Basically it works as though complex were an aggregate except for one
thing: for backwards compatibility, we never
Hi,
This patch causes cxxabi_tweaks.h to be installed in freestanding
mode. We need this because libsupc++ installs cxxabi.h, which
includes cxx_tweaks.h.
-Doug
2011-04-27 Doug Kwan dougk...@google.com
* include/Makefile.am (install-freestanding-headers): Also install
When we expand a default argument, we need to rebuild any TARGET_EXPRs
so that we don't run into issues with invalid tree sharing. But
copy_tree_r isn't prepared to deal with the STATEMENT_LIST created by
array initialization.
Since I already introduced VEC_INIT_EXPR in the constexpr work,
Andrew Pinski pins...@gmail.com writes:
| On Wed, Apr 27, 2011 at 6:30 PM, Gabriel Dos Reis g...@cs.tamu.edu wrote:
|
| Richard,
|
| Your recent patch broke --enable-build-with-cxx because internal-fn.h
| was using good 'ol C-style name lookup trick that behaves badly with
| C++ linkage
Jason Merrill ja...@redhat.com writes:
| In the discussion of 48760, Gaby suggested that we allow
| list-initialization of built-in complex numbers. This made a lot of
| sense to me, so I've gone ahead and implemented it.
|
| Basically it works as though complex were an aggregate except for one
On 04/27/2011 09:36 PM, Jason Merrill wrote:
A a = { 1, 2 };
a.c is initialized to 1+0i rather than 1+2i as it has been previously,
This should have said a.c is initialized to 1+0i, as it has been
previously, rather than 1+2i, as it would be if it were an aggregate.
Jason
On Wed, Apr 27, 2011 at 6:48 PM, Gabriel Dos Reis g...@cs.tamu.edu wrote:
A local `extern' declaration does not give the entity an external linkage
-- irrespective of the linkage of the function enclosing the declaration.
It just makes the name locally available for name lookup
purpose. A
Hi Diego,
Thanks for the quick feedback. Here is a an updated version of the patch.
2011-04-27 Sharad Singhai sing...@google.com
ChangeLog.google-main
* params.def: Add new parameters to control peeling.
* tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use
Andrew Pinski pins...@gmail.com writes:
| On Wed, Apr 27, 2011 at 6:48 PM, Gabriel Dos Reis g...@cs.tamu.edu wrote:
| A local `extern' declaration does not give the entity an external linkage
| -- irrespective of the linkage of the function enclosing the declaration.
| It just makes the name
Hi, this patch tries to solve the problem of the LR register not
being used in leaf functions on ARM.
Looking at the dumps, it shows that register 14 (lr) conflicts with all
allocnos throughout the entire leaf procedure. A little digging shows
that lr is present in the OBJECT_CONFLICT_HARD_REGS()
On 4/27/11, Jason Merrill ja...@redhat.com wrote:
On 04/27/2011 09:36 PM, Jason Merrill wrote:
A a = { 1, 2 };
a.c is initialized to 1+0i rather than 1+2i as it has been previously,
This should have said a.c is initialized to 1+0i, as it has been
previously, rather than 1+2i, as it would be
On 04/27/2011 04:50 AM, Paolo Carlini wrote:
Tested x86_64-linux. Ok for mainline?
Yes.
(not sure if it qualifies for the
branch or we want to reconsider it in a couple of weeks?!?)
The latter, I think.
Jason
81 matches
Mail list logo