On Wed, Mar 4, 2015 at 8:50 PM, Steve Ellcey wrote:
> While examining some MIPS code I noticed that GCC did not seem to be
> fully honoring the aligned attribute on some local variables. I submitted
> PR middle-end/65315 to record the bug and I think I now understand it and
> have a fix. The pro
On Thu, Mar 5, 2015 at 1:54 AM, Jan Hubicka wrote:
>> >
>> > It gets passed the valueize callback now which returns NULL_TREE for
>> > SSA names we can't follow.
>>
>> Btw, for match-and-simplify I had to use that as default for fold_stmt
>> _exactly_ because of the call to fold_stmt from replace_
On Thu, Mar 5, 2015 at 8:10 AM, Bernd Edlinger
wrote:
> Hi,
>
> on ARM we have a code quality regression, because of the strict volatile
> bitfields handing. The reason is that the current implementation directly
> jumps to store_fixed_bit_field_1 which emits a sequence of and/or/shift
> expressi
On Wed, Mar 4, 2015 at 3:13 PM, Bernd Edlinger
wrote:
> bounced... again, without html.
>
>
> Hi Richard,
>
> while working on another bug in the area of -fstrict-volatile-bitfields
> I became aware of another example where -fstrict-volatile-bitfields may
> generate
> wrong code. This is re
Hi,
On Thu, 5 Mar 2015 09:52:54, Richard Biener wrote:
>
> On Thu, Mar 5, 2015 at 8:10 AM, Bernd Edlinger
> wrote:
>> Hi,
>>
>> on ARM we have a code quality regression, because of the strict volatile
>> bitfields handing. The reason is that the current implementation directly
>> jumps to store_f
Hi Eric,
Following this commit (r221088) testing dump-ada-spec-3.C with
make -k check-g++ RUNTESTFLAGS="dg.exp=other/dump-ada-spec-3.C
generates a lot of *.ads files in the gcc/testsuite/g++ directory
which are not cleaned up after completion.
Any idea about how to do the cleaning?
TIA
Domini
On Thu, Mar 5, 2015 at 10:03 AM, Bernd Edlinger
wrote:
> Hi,
>
> On Thu, 5 Mar 2015 09:52:54, Richard Biener wrote:
>>
>> On Thu, Mar 5, 2015 at 8:10 AM, Bernd Edlinger
>> wrote:
>>> Hi,
>>>
>>> on ARM we have a code quality regression, because of the strict volatile
>>> bitfields handing. The re
Note: this is stage1 material.
Currently loop2_invariant pass hoist instructions out of loop by creating a new
temporary for the destination register of that instruction and leaving there a
mov from new temporary to old register as shown below:
loop header
start of loop body
//stuff
(set (reg 1
Hi,
with -march=z10 we use much higher values for peel and unroll limits.
This makes the loop in the testcase to disappear on tree level
already. With the patch these values are set back to the default
values making the testcase to pass again.
Committed to mainline
Bye,
-Andreas-
2015-03-05
Hi,
the initializer value in that testcase ends up in literal pool. As
described in the testcase the optimization does currently not work in
that situation.
Committed to mainline.
Bye,
-Andreas-
2015-03-05 Andreas Krebbel
* gcc.dg/tree-ssa/ssa-dom-cse-2.c:
diff --git a/gcc/testsui
Hello,
Tests gcc.target/arm/macro_defs0.c and gcc.target/arm/macro_defs1.c fail
in multilib which forces -marm as pointed out in this message:
https://gcc.gnu.org/ml/gcc-patches/2015-02/msg00483.html .
This patch will cause these tests to be classified as unsupported rather
than FAIL.
Ok f
Hello,
This patch adds scatter support for vectorizer (for AVX512F
instructions). Please have a look. Is it ok for stage 1?
2015-03-05 Andrey Turetskiy
* config/i386/i386-builtin-types.def
(VOID_PFLOAT_HI_V8DI_V16SF_INT): New.
(VOID_PDOUBLE_QI_V16SI_V8DF_INT): Ditto.
Hello,
This is a backport for gcc-4_8-branch of the patch " [PATCH][ARM]
__ARM_FP & __ARM_NEON_FP defined when -march=armv7-m" posted in:
https://gcc.gnu.org/ml/gcc-patches/2014-12/msg00250.html
arm-none-linux-gnueabi/hf tested without new regressions.
OK for gcc-4_8-branch?
Kind regards,
M
Hi,
On Thu, 5 Mar 2015 10:00:26, Richard Biener wrote:
>
> On Wed, Mar 4, 2015 at 3:13 PM, Bernd Edlinger
> wrote:
>> bounced... again, without html.
>>
>>
>> Hi Richard,
>>
>> while working on another bug in the area of -fstrict-volatile-bitfields
>> I became aware of another example where -fstr
On Thu, Mar 5, 2015 at 10:53 AM, Thomas Preud'homme
wrote:
> Note: this is stage1 material.
>
> Currently loop2_invariant pass hoist instructions out of loop by creating a
> new temporary for the destination register of that instruction and leaving
> there a mov from new temporary to old registe
On Thu, Mar 5, 2015 at 12:00 PM, Bernd Edlinger
wrote:
> Hi,
>
> On Thu, 5 Mar 2015 10:00:26, Richard Biener wrote:
>>
>> On Wed, Mar 4, 2015 at 3:13 PM, Bernd Edlinger
>> wrote:
>>> bounced... again, without html.
>>>
>>>
>>> Hi Richard,
>>>
>>> while working on another bug in the area of -fstri
S390: Hotpatching fixes.
* Properly align function labels with -mhotpatch and add test cases.
* Include the nops after the function label in the area covered by cfi and
debug information.
* Correct a typo in the documentation.
* Fix formatting in the generated 6-byte-NOP and adapt the test
Hi,
The expanding of widen-sum pattern always fails. The vectorizer expects
the operands to have the same size, while the current implementation of
widen-sum pattern dose not conform to this.
This patch implements the widen-sum pattern with vpadal. Change the
vaddw pattern to anonymous. Add
On 04/03/15 17:20 +, Jonathan Wakely wrote:
To fix the non-portable 22_locale/conversions/string/2.cc test I
changed it to use char16_t and char32_t where I can reliably create an
invalid sequence that causes a conversion error. That revealed some
more problems in the Unicode conversion utili
On 05/03/15 13:34, Xingxing Pan wrote:
Hi,
Hi Xingxing,
Thanks for improving this! Some comments inline.
The expanding of widen-sum pattern always fails. The vectorizer expects
the operands to have the same size, while the current implementation of
widen-sum pattern dose not conform to this.
Hello.
This is patch that prevents merge operation for ICF on variables types which
are not compatible.
Regression tests were run on x86_64-linux-pc.
Ready for trunk?
Thanks,
Martin
>From b92ec230162b99ff11d4e5688f63ae978e75af12 Mon Sep 17 00:00:00 2001
From: mliska
Date: Thu, 5 Mar 2015 13:41
This removes obsolete code in c-ada-spec.c that valgrind rightfully complains
about and should fix the PR in the process.
Tested on x86_64-linux-gnu, applied on the mainline and 4.9 branch.
2015-03-05 Eric Botcazou
PR ada/65319
* c-ada-spec.c (print_destructor): Remove obsol
Hi Xingxing,
I'm a little confused by your reasons for adding testcases marked XFAIL.
On Thu, Mar 05, 2015 at 01:34:25PM +, Xingxing Pan wrote:
> +/* { dg-final { scan-tree-dump-times "pattern recognized.*w\\\+" 1 "vect" {
> xfail *-*-* } } } */
> +/* { dg-final { cleanup-tree-dump "vect" }
On Thu, Mar 05, 2015 at 02:53:44PM +0100, Martin Liška wrote:
> --- a/gcc/ipa-icf.c
> +++ b/gcc/ipa-icf.c
> @@ -1501,6 +1501,11 @@ sem_variable::equals (sem_item *item,
>if (DECL_INITIAL (item->decl) == error_mark_node && in_lto_p)
> dyn_cast (item->node)->get_constructor ();
>
> + /* A
On 03/05/2015 03:29 PM, Marek Polacek wrote:
On Thu, Mar 05, 2015 at 02:53:44PM +0100, Martin Liška wrote:
--- a/gcc/ipa-icf.c
+++ b/gcc/ipa-icf.c
@@ -1501,6 +1501,11 @@ sem_variable::equals (sem_item *item,
if (DECL_INITIAL (item->decl) == error_mark_node && in_lto_p)
dyn_cast (item->
On Wed, Mar 4, 2015 at 3:26 PM, H.J. Lu wrote:
> Protected symbol means that it can't be pre-emptied. It
> doesn't mean its address won't be external. This is true
> for pointer to protected function. With copy relocation,
> address of protected data defined in the shared library may
> also be
This patch fixes the shift costs for Cortex-A53 so they are more accurate -
immediate shifts use
SBFM/UBFM which takes 2 cycles, register controlled shifts take 1 cycle.
Bootstrap and regression
OK.
ChangeLog:
2015-03-05 Wilco Dijkstra
* gcc/config/arm/aarch-cost-tables.h (cortexa53
On 03/03/15 15:58, Alex Velenko wrote:
On 19/02/15 17:26, Richard Henderson wrote:
On 02/19/2015 09:08 AM, Alex Velenko wrote:
Your suggestion seem to fix gcc.target/arm/long-calls-1.c, but has
to be
thoroughly tested.
Before you do complete testing, please also delete the TREE_STATIC test.
This test should be run rather than just compiled.
Bootstrapped/regtested on x86_64-linux, applying to trunk.
2015-03-05 Marek Polacek
* c-c++-common/ubsan/bounds-6.c: Use dg-do run.
diff --git gcc/testsuite/c-c++-common/ubsan/bounds-6.c
gcc/testsuite/c-c++-common/ubsan/bounds-6.c
i
Hi,
On Thu, 5 Mar 2015 12:24:56, Richard Biener wrote:
>
> On Thu, Mar 5, 2015 at 12:00 PM, Bernd Edlinger
> wrote:
>> Hi,
>>
>> On Thu, 5 Mar 2015 10:00:26, Richard Biener wrote:
>>>
>>> On Wed, Mar 4, 2015 at 3:13 PM, Bernd Edlinger
>> Maybe one thing is missing from strict_volatile_bitfield_p,
diff --git a/gcc/config/arm/arm.c b/gcc/config/arm/arm.c
index 7bf5b4d..777230e 100644
--- a/gcc/config/arm/arm.c
+++ b/gcc/config/arm/arm.c
@@ -6392,14 +6392,8 @@ arm_set_default_type_attributes (tree type)
static bool
arm_function_in_section_p (tree decl, section *section)
{
- /* We can
On Thu, 2015-03-05 at 07:37 +0100, Bert Wesarg wrote:
> Hi David,
>
> while reading the very good tutorial at
>
> https://gcc.gnu.org/onlinedocs/jit/intro/tutorial03.html
>
> I noticed that the calls to gcc_jit_block_end_with_conditional()
> misses the on_true and on_false parameters.
Good cat
On Thu, Mar 5, 2015 at 4:05 PM, Bernd Edlinger
wrote:
> Hi,
>
> On Thu, 5 Mar 2015 12:24:56, Richard Biener wrote:
>>
>> On Thu, Mar 5, 2015 at 12:00 PM, Bernd Edlinger
>> wrote:
>>> Hi,
>>>
>>> On Thu, 5 Mar 2015 10:00:26, Richard Biener wrote:
On Wed, Mar 4, 2015 at 3:13 PM, Bernd Edl
I am currently testing the following patch to reduce peak memory
usage of the out-of-SSA phase for the testcase in the PR. The
issue is (as usual) big live and SSA conflict graph memory use.
This side tackles live info and frees livein before computing
the conflict graph (which only needs liveout
On 03/05/15 01:47, Richard Biener wrote:
On Thu, Mar 5, 2015 at 1:54 AM, Jan Hubicka wrote:
It gets passed the valueize callback now which returns NULL_TREE for
SSA names we can't follow.
Btw, for match-and-simplify I had to use that as default for fold_stmt
_exactly_ because of the call to
On 03/04/15 12:50, Steve Ellcey wrote:
While examining some MIPS code I noticed that GCC did not seem to be
fully honoring the aligned attribute on some local variables. I submitted
PR middle-end/65315 to record the bug and I think I now understand it and
have a fix. The problem was that expan
On Thu, 2015-03-05 at 09:36 +0100, Richard Biener wrote:
> >
> > I have verified the fix on the MIPS test case in PR 65315 and am doing a
> > regression test now. OK to checkin if there are no regressions?
>
> It looks like large_align vars are dynamically allocated and thus they
> should be sor
On 27/02/15 09:41, Richard Earnshaw wrote:
On 19/02/15 12:19, Matthew Wahab wrote:
The LEGITIMIZE_RELOAD_ADDRESS macro is only needed for reload. Since the
ARM backend no longer supports reload, this macro is not needed and this
patch removes it.
gcc/
2015-02-19 Matthew Wahab
* config/
On 03/05/15 07:37, Martin Liška wrote:
From 3f35d9ec57880409cde384bb7b9e8dbaae5231ef Mon Sep 17 00:00:00 2001
From: mliska
Date: Thu, 5 Mar 2015 13:41:07 +0100
Subject: [PATCH] Fix PR ipa/65318.
gcc/ChangeLog:
2015-03-05 Martin Liska
PR ipa/65318
* ipa-icf.c (sem_variable::
Add h8300-*-linux target for h8300 linux kernel and userland.
h8300-*-elf is some difference of standard elf.
h8300-*-linux is compatible of standard elf rules.
Thanks.
diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index cfacea1..fc5101c 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,12
Hello, gentle maintainer.
This is a message from the Translation Project robot.
A revised PO file for textual domain 'gcc' has been submitted
by the German team of translators. The file is available at:
http://translationproject.org/latest/gcc/de.po
(This file, 'gcc-5.1-b20150208.de.po', h
> gcc/ChangeLog:
>
> 2015-03-05 Martin Liska
>
> PR ipa/65318
> * ipa-icf.c (sem_variable::equals): Compare variables types.
>
> gcc/testsuite/ChangeLog:
>
> 2015-03-05 Martin Liska
>
> * gcc.dg/ipa/pr65318.c: New test.
OK,
Honza
> > gcc/ChangeLog:
> >
> > 2015-03-05 Martin Liska
> >
> > PR ipa/65318
> > * ipa-icf.c (sem_variable::equals): Compare variables types.
> >
> > gcc/testsuite/ChangeLog:
> >
> > 2015-03-05 Martin Liska
> >
> > * gcc.dg/ipa/pr65318.c: New test.
>
> OK,
Though actually I think
On Thu, Mar 05, 2015 at 06:39:10AM -0800, H.J. Lu wrote:
> On Wed, Mar 4, 2015 at 3:26 PM, H.J. Lu wrote:
> > Protected symbol means that it can't be pre-emptied. It
> > doesn't mean its address won't be external. This is true
> > for pointer to protected function. With copy relocation,
> > add
On 05/03/15 16:34, Matthew Wahab wrote:
thumb_legitimize_reload_address was added by
https://gcc.gnu.org/ml/gcc-patches/2005-08/msg01140.html to fix PR
23436. It replaces sequences like
mov r3, r9
mov r2, r10
ldr r0, [r3, r2]
with
mov r3, r9
On Thu, Mar 5, 2015 at 9:32 AM, Rich Felker wrote:
> On Thu, Mar 05, 2015 at 06:39:10AM -0800, H.J. Lu wrote:
>> On Wed, Mar 4, 2015 at 3:26 PM, H.J. Lu wrote:
>> > Protected symbol means that it can't be pre-emptied. It
>> > doesn't mean its address won't be external. This is true
>> > for poi
Hi,
this patch sovles the incorrect folding. The very same unification (ignoring
signedness by checking that memory representation is the same) is done by
constant pool.
Some of the other uses of ctor_for_folding therefore already uses
VIEW_CONVERT_EXPR, I suppose as a partial fix for past bugs. T
On March 5, 2015 7:08:16 PM CET, Jan Hubicka wrote:
>Hi,
>this patch sovles the incorrect folding. The very same unification
>(ignoring
>signedness by checking that memory representation is the same) is done
>by
>constant pool.
>
>Some of the other uses of ctor_for_folding therefore already uses
>
On March 5, 2015 7:08:16 PM CET, Jan Hubicka wrote:
>Hi,
>this patch sovles the incorrect folding. The very same unification
>(ignoring
>signedness by checking that memory representation is the same) is done
>by
>constant pool.
>
>Some of the other uses of ctor_for_folding therefore already uses
>
> >Index: gimple-fold.c
> >===
> >--- gimple-fold.c(revision 221170)
> >+++ gimple-fold.c(working copy)
> >@@ -263,7 +263,16 @@ get_symbol_constant_value (tree sym)
> > {
> > val = canonicalize_constructor_val (unshar
On Mar 4, 2015, Richard Biener wrote:
> Compile-time was slightly faster with the patch, 45s vs. 47s,
> but the machine wasn't completely un-loaded. var-tracking parts:
> unpatched:
> variable tracking : 0.63 ( 1%) usr 0.03 ( 1%) sys 0.82 (
> 2%) wall 28641 kB ( 2%) ggc
> var-
The following patch fixes a bad code generation for avx512f-kandnw-1.c
reported in
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64342
The patch was bootstrapped and tested on x86-64.
Committed as rev. 221223.
2015-03-05 Vladimir Makarov
PR target/64342
* lra-assigns.c
On March 5, 2015 8:26:42 PM CET, Alexandre Oliva wrote:
>On Mar 4, 2015, Richard Biener wrote:
>
>> Compile-time was slightly faster with the patch, 45s vs. 47s,
>> but the machine wasn't completely un-loaded. var-tracking parts:
>
>> unpatched:
>
>> variable tracking : 0.63 ( 1%) usr
ubsan's code to determine whether we're dealing with a flexible array member
didn't check for a COMPONENT_REF, but it should, since flexible array members
can only occur in a structure. Consequently, we didn't instrument stuff we
should instrument.
Bootstrapped/regtested on x86_64-linux, if I hea
This arranges to build a powerpc64le-linux compiler without -m32
support by default. Bootstrapped and regression tested on Ubuntu
powerpc64le-linux without --disable-multilib, and on powerpc64-linux
and powerpc-linux. OK for mainline and branches?
This part of the config.gcc patch does most of t
On Jan 30, 2015, at 8:27 AM, Mike Stump wrote:
> On Jan 30, 2015, at 7:49 AM, Joseph Myers wrote:
>> Use error_at, and %u directly in the format.
>
> Done.
Ping?
Index: ada/gcc-interface/trans.c
===
--- ada/gcc-interface/trans.c
While looking at PR65284, I was confused by the gimple we generate for
returns of empty classes by value:
class obj {
public:
obj(int);
};
obj funky(){
return obj(555);
}
For the above snippet, we generate:
obj funky() ()
{
struct obj D.2248;
struct obj D.2246;
obj::obj (&D.22
Attached is a scaled down version of the test for the bug.
It fixes the scan-tree-dump-times string to match what GCC
5 prints and moves the result checking out of the test
function and into main to prevent it from getting optimized
away (as observed in comment #8 on the bug).
The patch also adds
This patch by Chris Manghane fixes a bug for cases like this:
switch x := v.(type) {
case *x:
in which the type name in the case happens to be the same as the
variable name in the type switch. This is rather confusing code, but
it should work. This is http://golang.org/issue/10047 . Boot
Protected data symbol means that it can't be pre-emptied. It doesn't mean
its address won't be external. This is true for pointer to protected
function. With copy relocation, address of protected data defined in the
shared library may also be external. We only know that for sure at
run-time. T
On 03/05/2015 04:12 PM, Mike Stump wrote:
Ping?
Just commenting on the documentation part:
Index: doc/extend.texi
===
--- doc/extend.texi (revision 220084)
+++ doc/extend.texi (working copy)
@@ -17881,6 +17881,18 @@ voi
Hello Maintainers:
Please help check this patch when you have time.
I have to leave Sunrus, the mail address (gang.c...@sunrus.com.cn) will
be closed soon (Sunrus will be closed soon because of money, I guess).
I change my new email address (xili_gchen_5...@hotmail.com) to continue
communicatin
On Wed, Mar 04, 2015 at 03:26:10PM -0800, H.J. Lu wrote:
> Protected symbol means that it can't be pre-emptied. It
> doesn't mean its address won't be external. This is true
> for pointer to protected function. With copy relocation,
> address of protected data defined in the shared library may
>
-Original Message-
From: Michael Eager [mailto:ea...@eagerm.com]
Sent: Thursday, February 26, 2015 4:29 AM
To: Ajit Kumar Agarwal; GCC Patches
Cc: Vinod Kathail; Shail Aditya Gupta; Vidhumouli Hunsigida; Nagaraju Mekala
Subject: Re: [Patch,microblaze]: Optimized usage of pcmp conditional
On 03/05/2015 06:25 PM, Aldy Hernandez wrote:
+ tree ret = TREE_OPERAND (*expr_p, 0);
+ if (ret && (TREE_CODE (ret) == INIT_EXPR
+ || TREE_CODE (ret) == MODIFY_EXPR)
+ && TREE_CODE (TREE_OPERAND (ret, 0)) == RESULT_DECL
+ && is_gimple_lvalue (TREE
Hi,
This patch is the first part fixing memset-inline-{4,5,6,8,9}.c failures on
cortex-a9. GCC/arm doesn't generate any tuning information in assembly, it
can't tell whether we are compiling for cortex-a9 tune if the compiler is
configured so by default.
This patch introduces a new (target depende
Hi,
This patch is the second part fixing memset-inline-{4,5,6,8,9}.c failures on
cortex-a9. It adds a function checking CPU tuning information in dejagnu,
it also uses that function to skip related testcase when we are compiling
for cortex-a9 tune.
Build and test on arm-none-eabi. Is it OK?
gcc
67 matches
Mail list logo
