On 09/20/2011 09:51 PM, Joseph S. Myers wrote:
Some target libraries have
# Subdir rules rely on $(FLAGS_TO_PASS)
FLAGS_TO_PASS = $(AM_MAKEFLAGS)
in their Makefile.am. This was introduced to libstdc++-v3 by
http://gcc.gnu.org/ml/libstdc++/2002-06/msg00098.html to fix a
problem described in
Hi
This patch fixed PR 49452, thus decreased 30+ dejaGNU failures for target arm.
Tested with buildit and arm regression testing on qemu.
OK for google/gcc-4_6?
thanks
Carrot
ChangeLog
2011-09-21 Guozhi Wei car...@google.com
Backport r178995 from trunk.
2011-09-20 Wei
Hi!
I haven't done any SPARC related work in the last 5 years and did
a very bad job as SPARC maintainer, I don't even have any bootable
SPARC boxes anymore, fortunately we have lots of other
SPARC maintainers, Eric does a good job at it and DaveM works on new
SPARC stuff. I should have done
Hi,
starting with r177691 store_bit_field is used for source operands
which cannot be covered exactly with word mode chunks. So far
store_bit_field does not seem to handle that correctly. With the
attached patch extract_bit_field is used for the remaining bits of the
source on big endian
On Mon, Sep 19, 2011 at 07:10:11PM +0200, Uros Bizjak wrote:
2011-09-19 Uros Bizjak ubiz...@gmail.com
* config/i386/i386.md (maxmin): New code iterator.
* config/i386/sse.md (maxmin:codemode3): Macroize expander
from umaxmin:codemode3 and smaxmin:codemode3 using maxmin
On Wed, Sep 21, 2011 at 9:45 AM, Jakub Jelinek ja...@redhat.com wrote:
* config/i386/i386.md (maxmin): New code iterator.
* config/i386/sse.md (maxmin:codemode3): Macroize expander
from umaxmin:codemode3 and smaxmin:codemode3 using maxmin
code iterator.
Committed in ARM/embedded-4_6-branch.
2011-09-20 Jiangning Liu jiangning@arm.com
Tune branch cost for armv7-m
* config/arm/arm-cores.def (cortex-m3): Use cortex-v7m.
(cortex-m4) : Likewise.
* config/arm/arm.c (arm_cortex_v7m_branch_cost): New.
Hi,
The following test cases are to check predictive commoning optimization by
detecting loop unrolling times. Originally by default the max-unroll-times
is 8. If max-unroll-times is specified to be less than the expected
unrolling times, those cases would fail.
The fix is to explicitly turn on
While fiddling around with the VIS intrinsic builtins I noticed that
none of them have TREE_READONLY set, so the resulting code can be
terrible.
We can't currently do this for alignaddr and faligndata because we
don't model the way those instructions use the %gsr register.
Committed to trunk.
Committed in ARM/embedded-4_6-branch.
2011-09-21 Jiangning Liu jiangning@arm.com
Tune loop unrolling for cortex-m
* config/arm/arm-cores.def (cortex-m0): Change to new tune
cortex_v6m.
(cortex-m1): Likewise.
* config/arm/arm-protos.h
于 2011/9/21 10:56, Cesar Strauss 写道:
Please let me present an opposing view.
On the one hand, according to comment 4 of [1], by using
--enable-fully-dynamic-string, all other users will miss a very good
optimization. On the other hand, these users of -static-libstdc++ are
mixing shared and
Hi there,
I've prepared a patch which implements couple of intrinsics for
generation of MULX 32 and 64 bit wide (part of BMI2 extensions).
Tests are added as well
ChangeLog entry:
2011-09-21 Kirill Yukhin kirill.yuk...@intel.com
* config/i386/bmi2intrin.h (_mulx_u64): New.
Hi,
this patch adds some info on changes in IPA and LTO. Not sure if I missed
something important. Martin, perhaps you could add some example on new
devirutalization stuff?
Suggestions for less lame explanations are welcome :)
Honza
Index: changes.html
As proposed in PR45099, avr-gcc will now just print a warning instead of an
error when a fixed register is needed to pass a parameter to a function.
That way the user can inspect the source, and there are situation imaginable
where clobbering a fixed register is not a problem.
Committed as trunk
Backport mainline r178975 to ARM/embedded-4_6-branch.
Committed.
2011-09-21 Terry Guo terry@arm.com
Backport r178975 from mainline
2011-09-19 Ian Bolton ian.bol...@arm.com
* gcc.dg/20100906-1.c: Use -fno-short-enums option for
target arm_eabi.
On Wed, Sep 21, 2011 at 11:17 AM, Kirill Yukhin kirill.yuk...@gmail.com wrote:
Hi there,
I've prepared a patch which implements couple of intrinsics for
generation of MULX 32 and 64 bit wide (part of BMI2 extensions).
Tests are added as well
ChangeLog entry:
2011-09-21 Kirill Yukhin
Hi,
now that c++/50454 is fixed we can do this, replace the legacy typedefs.
Tested x86_64-linux multilib, committed.
Thanks,
Paolo.
///
2011-09-21 Paolo Carlini paolo.carl...@oracle.com
* src/limits.cc: Replace everywhere __int128_t - __int128, and
Hi!
For vcond{,u} etc. we currently generate vpandn+vpand+vpor
sequence but SSE4.1+ has instructions for at least some modes
to handle those 3 in one instruction (haven't benchmarked how much
faster/slower it is though).
Bootstrapped/regtested on x86_64-linux and i686-linux, tested
on
On Wed, Sep 21, 2011 at 1:37 PM, Jakub Jelinek ja...@redhat.com wrote:
For vcond{,u} etc. we currently generate vpandn+vpand+vpor
sequence but SSE4.1+ has instructions for at least some modes
to handle those 3 in one instruction (haven't benchmarked how much
faster/slower it is though).
Comments at #else and #endif are now wrong. Also, please avoid -dp in the
tests.
Thanks for inputs. Updated patch is attached (updated comment +
updated tests without -dp).
New tests still passing
Is it OK?
Thanks, K
bmi2.mulx-intrin-2.gcc.patch
Description: Binary data
On Wed, Sep 21, 2011 at 2:07 PM, Kirill Yukhin kirill.yuk...@gmail.com wrote:
Comments at #else and #endif are now wrong. Also, please avoid -dp in the
tests.
Thanks for inputs. Updated patch is attached (updated comment +
updated tests without -dp).
New tests still passing
OK for
Dinar Temirbulatov dtemirbula...@gmail.com writes:
@@ -14696,7 +14696,11 @@ mips_avoid_hazard (rtx after, rtx insn,
clobber hi and lo. */
if (*hilo_delay 2 reg_set_p (lo_reg, pattern))
nops = 2 - *hilo_delay;
- else if (*delayed_reg != 0 reg_referenced_p (*delayed_reg,
On Wed, Sep 21, 2011 at 02:02:44PM +0200, Uros Bizjak wrote:
+ rtx (*gen) (rtx, rtx, rtx, rtx) = NULL;
+
op_false = force_reg (mode, op_false);
+ switch (mode)
+ {
+ case V4SFmode: if (TARGET_SSE4_1) gen = gen_sse4_1_blendvps; break;
+ case V2DFmode:
On 9/21/2011 5:49 AM, Georg-Johann Lay wrote:
As proposed in PR45099, avr-gcc will now just print a warning instead of an
error when a fixed register is needed to pass a parameter to a function.
Where's the proposal in the PR? I see a problem report that was
addressed four months ago by
the problem here is that we consider param.foo to be a memory reference,
while it is not. The following patch should fix it, but because it changes
the behaviour of inline heuristics, I will run bechmarks tonight before
committing it.
Thanks in advance. Can you add PR
Hello,
This revised patch adds support to tree-ssa-reassoc for intial
normalizing of bitwise operation. So the reassociation pass is able
to have better folding result.
Also it has a final denormalzing of bitwise-operations after
reassociation pass is completed to restore optimized writing.
On 09/21/2011 04:37 AM, Jakub Jelinek wrote:
op_false = force_reg (mode, op_false);
+ switch (mode)
+ {
+ case V4SFmode: if (TARGET_SSE4_1) gen = gen_sse4_1_blendvps; break;
+ case V2DFmode: if (TARGET_SSE4_1) gen = gen_sse4_1_blendvpd; break;
+ case V16QImode: if
On Wed, Sep 21, 2011 at 07:25:43AM -0700, Richard Henderson wrote:
On 09/21/2011 04:37 AM, Jakub Jelinek wrote:
op_false = force_reg (mode, op_false);
+ switch (mode)
+ {
+ case V4SFmode: if (TARGET_SSE4_1) gen = gen_sse4_1_blendvps; break;
+ case V2DFmode: if
On 09/15/11 21:42, Richard Henderson wrote:
On 09/14/2011 06:12 PM, Bernd Schmidt wrote:
+ unsigned int orig_regno = REGNO (reg);
+ int nregs = hard_regno_nregs[orig_regno][GET_MODE (reg)];
+ while (nregs-- 0)
The rest of the file seems to use targetm.dwarf_register_span.
This probably
Eric,
I am not sure to understand the sentence:
There is apparently an ACATS failure on x86-64/Darwin, but I've installed the
testcase as gnat.dg/opt19.adb in the tree.
The only failure I had was pr50433, now fixed by revision 179046. With it,
all the ADA tests pass (including
On 21 Sep 2011, at 15:58, Dominique Dhumieres wrote:
I am not sure to understand the sentence:
There is apparently an ACATS failure on x86-64/Darwin, but I've
installed the
testcase as gnat.dg/opt19.adb in the tree.
The only failure I had was pr50433, now fixed by revision 179046.
With
On 20 September 2011 09:38, Richard Sandiford
richard.sandif...@linaro.org wrote:
My fix for PR49030 had the unintended side-effect of forcing libcalls
to be used for some DImode comparisons. The problem (which I should
have noticed at the time, sorry) is that arm_comparison_operator is
used
Hi,
basing on [class.friend] (11.3/3) in C++11 I think Jon is right that
cv-qualification should be ok in a friend declaration. I'm not sure,
however, if we want to somehow allow that also in c++98 mode... Tested
x86_64-linux.
Ok?
Thanks,
Paolo.
///
/cp
2011-09-21 Paolo
On 09/21/2011 07:40 AM, Jakub Jelinek wrote:
* config/i386/i386.c (ix86_expand_sse_movcc): Use
blendvps, blendvpd and pblendvb if possible.
* gcc.dg/vect/vect-cond-7.c: New test.
* gcc.target/i386/sse4_1-cond-1.c: New test.
* gcc.target/i386/avx-cond-1.c: New
On Wed, Sep 21, 2011 at 5:16 AM, Uros Bizjak ubiz...@gmail.com wrote:
On Wed, Sep 21, 2011 at 2:07 PM, Kirill Yukhin kirill.yuk...@gmail.com
wrote:
Comments at #else and #endif are now wrong. Also, please avoid -dp in the
tests.
Thanks for inputs. Updated patch is attached (updated
On Tue, Sep 20, 2011 at 7:51 PM, Iyer, Balaji V balaji.v.i...@intel.com wrote:
Hello Everyone,
This patch is for the Cilkplus branch. In C and C++ compilers,
at optimization -O and -O2, when there are inlined functions inside a
function that calls a spawn it was giving an
This patch improves IOR, XOR, AND for HI and SI.
There is room for improvement in the current implementation:
- better analysis if a scratch is actually needed
- add clobber operands to some insns
- use 8-bit scratch instead of reload the constant
- reuse value in scratch register if known
-
Thank you!
K
On Wed, Sep 21, 2011 at 7:57 PM, H.J. Lu hjl.to...@gmail.com wrote:
On Wed, Sep 21, 2011 at 5:16 AM, Uros Bizjak ubiz...@gmail.com wrote:
On Wed, Sep 21, 2011 at 2:07 PM, Kirill Yukhin kirill.yuk...@gmail.com
wrote:
Comments at #else and #endif are now wrong. Also, please
Nathan Froyd wrote:
On 9/21/2011 5:49 AM, Georg-Johann Lay wrote:
As proposed in PR45099, avr-gcc will now just print a warning instead
of an
error when a fixed register is needed to pass a parameter to a function.
Where's the proposal in the PR? I see a problem report that was
addressed
On 09/21/2011 11:12 AM, Paolo Carlini wrote:
basing on [class.friend] (11.3/3) in C++11 I think Jon is right that
cv-qualification should be ok in a friend declaration. I'm not sure,
however, if we want to somehow allow that also in c++98 mode...
I think we do.
Jason
On 09/21/2011 07:55 AM, Bernd Schmidt wrote:
On 09/15/11 21:42, Richard Henderson wrote:
On 09/14/2011 06:12 PM, Bernd Schmidt wrote:
+ unsigned int orig_regno = REGNO (reg);
+ int nregs = hard_regno_nregs[orig_regno][GET_MODE (reg)];
+ while (nregs-- 0)
The rest of the file seems to
ping
rename subject line due to
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg01189.html
references
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg00310.html
http://gcc.gnu.org/ml/gcc-patches/2011-08/msg00216.html
http://gcc.gnu.org/ml/gcc-patches/2011-07/msg00281.html
Hello!
Predicates for xop_pcmov were wrong. Also, immediate operand can leak
in through vcond patterns, so a fixup is needed in
ix86_expand_sse_movcc for TARGET_XOP.
2011-09-21 Uros Bizjak ubiz...@gmail.com
PR target/50464
* config/i386/sse.md (xop_pcmov_modeavxsizesuffix):
On Wed, Sep 21, 2011 at 02:24, Guozhi Wei car...@google.com wrote:
2011-09-21 Guozhi Wei car...@google.com
Backport r178995 from trunk.
2011-09-20 Wei Guozhi car...@google.com
PR rtl-optimization/49452
* postreload.c (reload_combine):
The Go language spec clarified that a lookup on nil map should work and
return a zero value. This patch to the Go frontend and the Go library
implements that. Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Committed to mainline. (This patch requires
rebuilding the entire Go
Hi,
2011/9/20 Jason Merrill ja...@redhat.com:
On 09/17/2011 09:44 AM, Fabien Chêne wrote:
I tried various things without success, and I ended up hacking
supplement_binding_1 to handle those ENUMERAL_TYPEs.
I am all ear for another solution ...
Your solution seems reasonable to me, but it
... with the ChangeLog
gcc/ChangeLog
2011-09-21 Fabien Chêne fab...@gcc.gnu.org
PR c++/26256
* dbxout.c (dbxout_type_fields): Ignore using declarations.
gcc/testsuite/ChangeLog
2011-09-21 Fabien Chêne fab...@gcc.gnu.org
PR c++/26256
*
On 09/21/2011 01:59 PM, Fabien Chêne wrote:
if (!DECL_DEPENDENT_P (field))
- continue;
+ {
+ tree using_decl = USING_DECL_DECLS (field);
+ if ((TREE_CODE (using_decl) == FIELD_DECL
+ || TREE_CODE (using_decl) == TYPE_DECL)
+
On 09/21/2011 06:55 PM, Jason Merrill wrote:
I think we do.
Agreed. I'm finishing testing the below, Ok if it passes?
Thanks,
Paolo.
//
/cp
2011-09-21 Jonathan Wakely jwakely@gmail.com
Paolo Carlini paolo.carl...@oracle.com
PR c++/50344
Jason Merrill ja...@redhat.com writes:
My point was more that the name of the function is confusing;
Sorry about that.
if what you get back is another virtual location, that's not what I
would consider a def point.
For tokens that are not function-like macro arguments, I think the
http://codereview.appspot.com/5050045/diff/1/gcc/cp/pph-streamer-in.c
File gcc/cp/pph-streamer-in.c (right):
http://codereview.appspot.com/5050045/diff/1/gcc/cp/pph-streamer-in.c#newcode2024
gcc/cp/pph-streamer-in.c:2024: /* Read the language-independent
bitfields for expr. */
s/expr/EXPR/
I've gotten tired of the false positives in the current testsuite for
the cxxbitfield* tests. The tests are highly dependent on assembly
output, and currently they only work on x86-64/x86-32 (mostly).
I have rewritten the tests to be platform agnostic, by using the
GDB/thread testing
On Sep 21, 2011, at 1:22 AM, Jiangning Liu wrote:
The fix is to explicitly turn on loop unroll and set max-unroll-times to 8,
which is larger than the unrolling times being detected in the cases.
Sounds reasonable to me. Ok, though, do watch for any comments by people that
know more than I.
http://codereview.appspot.com/5090041/diff/1/gcc/cp/pph-streamer-in.c
File gcc/cp/pph-streamer-in.c (right):
http://codereview.appspot.com/5090041/diff/1/gcc/cp/pph-streamer-in.c#newcode2146
gcc/cp/pph-streamer-in.c:2146: pph_read_namespace_chain (pph_stream
*stream, tree enclosing_namespace)
On Tue, May 3, 2011 at 16:20, Jason Merrill ja...@redhat.com wrote:
On 04/29/2011 10:59 AM, Diego Novillo wrote:
This patch from Le-Chun Wu adds a new warning flag -Wreal-conversion
that warns about implicit type conversions from real (double or float)
values to integral values.
Why do you
This patch to the Go frontend fixes a couple of cases with shift
expressions. A shift of a floating point constant (which would have to
have integral value) is still a constant, and as such must be lowered to
a constant value. This patch fixes the general case of shifting a
non-numeric value,
Hi all,
here is a patch which adds checks for the ELEMENTAL attribute in
procedure pointer assignments and dummy procedures. For details see
the PR and the c.l.f. thread mentioned therein. For the PURE
attribute, we already had a check, which I moved and reformulated (so
that it is applicable
On Wed, Sep 21, 2011 at 10:27:34PM +0200, Janus Weil wrote:
Hi all,
here is a patch which adds checks for the ELEMENTAL attribute in
procedure pointer assignments and dummy procedures. For details see
the PR and the c.l.f. thread mentioned therein. For the PURE
attribute, we already had a
From: Michael Meissner meiss...@linux.vnet.ibm.com
Date: Thu, 15 Sep 2011 14:16:45 -0400
On the powerpc, we have an issue with Spec 2006 and calculix when FMAs are
generated and -ffast-math is used, where line 307 of rubber.f is:
tt=datan2(dsqrt(1.d0-cn*cn),cn)/3.d0
The FNMSUB
Hello!
Compile tests do not need to check assembler for sse support.
2011-09-21 Uros Bizjak ubiz...@gmail.com
* g++.dg/other/i386-3.C: Remove duplicated dg-options.
* g++.dg/other/i386-4.C: Remove dg-requre-effective-target directive.
* g++.dg/other/i386-8.C: Ditto.
This small patch improves the handling of one case of unexpected EOF in
the Go frontend parser. Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Committed to mainline.
Ian
diff -r 5a1ab3177e67 go/parse.cc
--- a/go/parse.cc Wed Sep 21 13:20:32 2011 -0700
+++ b/go/parse.cc Wed Sep
Similar to various previous such patches, this patch adds dg-skip-if
markers to various gcc.target/i386 tests to skip them if the multilib
options with which the testsuite is run contain -march= options
conflicting with those in the testcases (recall that multilib options
go *after* those from
Hi,
this patch fixes two rather embarassing typos that together caused that benefits
are pretty much ignored by the inliner heuristics.
Also to make benefits go right for large function I remived caping logic
in do_estimate_edge_time. It is no longer needed since we no longe propagate
the
Hi,
this patch extends handling of non-SSA arguments to bultin_constant_p and
execution predicates.
Bootstrapped/regtested x86_64-linux, will commit it shortly.
Honza
* ipa-inline-analysis.c (set_cond_stmt_execution_predicate): Allow
handled components in parameter of
Hi,
ipa-inline-analysis use is_gimple_min_invariant that in turn require
current_function_decl
to be set to the corresponding function or all addresses of automatic vars are
considered
non-invariant.
Bootstrapped/regtested x86_64-linux, will commit it shortly.
* ipa-inline-analsis.c
I increased the default timeout for the libgo testsuite, simply because
the tests were occasionally failing on my old laptop. The tests seem
reasonably reliable these days, so the timeout is just a fallback. Most
of the tests finish much more quickly. Bootstrapped and ran libgo
testsuite on
Richard Sandiford richard.sandif...@linaro.org wrote on 30/08/2011
03:10:50 PM:
From: Richard Sandiford richard.sandif...@linaro.org
To: gcc-patches@gcc.gnu.org
Cc: Ayal Zaks/Haifa/IBM@IBMIL
Date: 30/08/2011 03:10 PM
Subject: [3/4] SMS: Record moves in the partial schedule
This patch adds
A few minor nits:
Thanks for the hints.
The attached patch and changelog below should do now.
BTW, have you done your paper work with FSF?
Yep. Took a while but it's done now.
Cheers,
Oleg
2011-09-21 Oleg Endo oleg.e...@t-online.de
* config/sh/sh.c (andcosts):
Fix for image directory location. The generated epub is ok, but missing
images. There's another tool that may be used to generate epubs, called
db2epub.py, but it isn't cooperating. So we will stick with the status
quo for now.
-benjamin2011-09-21 Benjamin Kosnik b...@redhat.com
*
Unfortunately, we can't use these for comparisons on vectors as the
compiler wants them. Instead of producing a vector of comparison
results, these instructions produce a small bitmask of those results
in a destination integer register. Therefore we have to provide
them using unspecs and
Oleg Endo oleg.e...@t-online.de wrote:
* config/sh/sh.c (andcosts): Renamed to and_xor_ior_costs.
Added AND special case. Adapted comments.
(sh_rtx_costs): Added XOR and IOR case.
Thanks! I've just applied your patch as revision 179073.
Regards,
kaz
Hello!
During performance tuning, it's useful to know if the static profile
is accurate. This patch compares static profiles to instrumentation
profiles, and uses degree of overlap to measure its accuracy. Degree
of overlap is a number between 0 and 1, representing the correlation
between two
On Fri, 16 Sep 2011, David Miller wrote:
I've been meaning to toss something like this together for a while.
If we were going to do this, I wanted to get it out of the way before
adding VIS2 and VIS3 support.
While revisiting VIS, *please* consider fixing a big usability
problem: the pack
From: Hans-Peter Nilsson h...@bitrange.com
Date: Wed, 21 Sep 2011 21:27:08 -0400 (EDT)
While revisiting VIS, *please* consider fixing a big usability
problem: the pack and aligndata builtins don't take GSR in
account; it has unknown state and might be changed as a
side-effect of a previous
On Wed, 21 Sep 2011, David Miller wrote:
From: Hans-Peter Nilsson h...@bitrange.com
Date: Wed, 21 Sep 2011 21:27:08 -0400 (EDT)
While revisiting VIS, *please* consider fixing a big usability
problem: the pack and aligndata builtins don't take GSR in
account; it has unknown state and might
Hi Mike,
OK. I will wait 24 more hours. If no objections by then, I will get it
checked into trunk.
Thanks,
-Jiangning
-Original Message-
From: Mike Stump [mailto:mikest...@comcast.net]
Sent: Thursday, September 22, 2011 3:10 AM
To: Jiangning Liu
Cc: gcc-patches@gcc.gnu.org;
This patch updates the Go library to the r60.1 release. Bootstrapped
and ran Go testsuite on x86_64-unknown-linux-gnu. Committed to
mainline.
Ian
diff -r dc1239c722f8 libgo/MERGE
--- a/libgo/MERGE Wed Sep 21 21:00:54 2011 -0700
+++ b/libgo/MERGE Wed Sep 21 21:35:15 2011 -0700
@@ -1,4 +1,4 @@
77 matches
Mail list logo
