On Wed, Dec 10, 2014 at 07:57:46AM +0100, Marek Polacek wrote:
> I don't really know this code, but this typo looks obvious enough.
> Using if (p->high != NULL_TREE) ... else if (p->high != NULL_TREE)
> couldn't be possibly desired, so use Q in the else branch, as in
> the code slightly above.
>
>
I don't really know this code, but this typo looks obvious enough.
Using if (p->high != NULL_TREE) ... else if (p->high != NULL_TREE)
couldn't be possibly desired, so use Q in the else branch, as in
the code slightly above.
Bootstrapped/regtested on x86_64-linux and ppc64-linux, ok for trunk?
201
Sorry for all the trouble. I messed up with the Changelog by mistake. I'll be
careful now onwards.
Prachi
-Original Message-
From: Jeff Law [mailto:l...@redhat.com]
Sent: Wednesday, December 10, 2014 4:05 AM
To: Matthew Fortune; Prachi Godbole
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [P
On Wed, Dec 10, 2014 at 6:58 AM, Jeff Law wrote:
> On 12/05/14 05:15, Bin Cheng wrote:
>>
>> Hi,
>> Though PR62178 is hidden by recent cost change in aarch64 backend, the
>> ivopt
>> issue still exists.
>>
>> Current candidate selecting algorithm tends to select fewer candidates
>> given
>> below
2014-12-09 23:08 GMT+08:00 Ian Lance Taylor :
> On Mon, Oct 27, 2014 at 8:19 PM, Kito Cheng wrote:
>>
>> This patch update `Bit operations` section in libgcc.text, most bit
>> operation function is take an unsigned integer instead of signed
>> integer in libgcc/libgcc2.c [1], and it seem more make
On Wed, Dec 10, 2014 at 10:18 AM, Andrew Pinski wrote:
> Hi,
> As mentioned in
> https://gcc.gnu.org/ml/gcc-patches/2014-12/msg00609.html, the
> load/store pair peepholes currently accept volatile mem which can
> cause wrong code as the architecture does not define which part of the
> pair happe
2014-12-09 21:16 GMT+08:00 Richard Biener :
> On Tue, Dec 9, 2014 at 6:36 AM, Chung-Ju Wu wrote:
>> Hi, all,
>>
>> In the discussion thread last year:
>> https://gcc.gnu.org/ml/gcc-patches/2013-05/msg01334.html
>>
>> I extended the script ./contrib/download_prerequisites so that it can
>> downlo
In this testcase, while we're tentatively parsing a function declarator
we came across __FUNCTION__, tried to predeclare it, and failed because
we were in function parameter scope; then when we try again later it
fails again because we cached the error_mark_node. So don't try the
first time.
We should check whether the template we're trying to specialize is the
right kind of template.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit af5860b1a4bac653093a74d97a00b162c2b1add7
Author: Jason Merrill
Date: Tue Dec 9 12:53:26 2014 -0500
PR c++/64129
* decl.c (grokdeclara
Hi,
As mentioned in
https://gcc.gnu.org/ml/gcc-patches/2014-12/msg00609.html, the
load/store pair peepholes currently accept volatile mem which can
cause wrong code as the architecture does not define which part of the
pair happens first.
This patch disables the peephole for volatile mem and add
Two defect reports that protect against dangerous uses of std::regex
involving dangling references to temporaries:
http://cplusplus.github.io/LWG/lwg-defects.html#2329
http://cplusplus.github.io/LWG/lwg-defects.html#2332
Tested x86_64-linux & powerpc64-linux, committed to trunk.
commit 616ac2600b
This patch to the Go compiler fixes a crash when using field tracking.
The code was failing to lower a function call added by the field
tracking code. When called in a type equality function for a type
that was not otherwise used, this could cause a crash as the type was
never lowered. This patch
On 12/04/2014 05:16 PM, Zhenqiang Chen wrote:
> +static rtx
> +cc_in_cond (rtx cond)
> +{
> + if ((HAVE_cbranchcc4) && cond
Silly parens around the HAVE_cbranchcc4.
> + && (GET_MODE_CLASS (GET_MODE (XEXP (cond, 0))) == MODE_CC))
More silly parens around the ==.
> + /* Skip it if the inst
Thanks to Geoff Romer for suggesting this change, which might be
needed if we change std::function to only be able to call const
operator() overloads for C++17.
Tested x86_64-linux and powerpc64-linux, committed to trunk.
commit d4eb711ad60cf2925d8f8f6c622918c648a51e13
Author: Jonathan Wakely
D
Hi,
This mentions the SH changes for GCC 5 that have happened so far.
Committed.
Cheers,
Oleg
Index: htdocs/gcc-5/changes.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-5/changes.html,v
retrieving revision 1.49
diff -u -r1.49 changes
Hi,
This documents the new SH FPSCR built-in functions.
Tested with 'make info dvi pdf', committed as r218551.
Cheers,
Oleg
gcc/ChangeLog:
PR target/53513
* doc/extend.texi (__builtin_sh_get_fpscr, __builtin_sh_get_fpscr):
Document it.
Index: gcc/doc/extend.texi
=
On 30/11/14 20:48 +, Jonathan Wakely wrote:
I think we also need this to make __numpunct_cache and
__moneypunct_cache exception-safe. If we set _M_allocated=true at the
start of the _M_cache functions and then an allocation throws we will
delete[] the memory allocated in _M_cache, but then th
On Wed, Dec 10, 2014 at 12:54 AM, Jakub Jelinek wrote:
> On Wed, Dec 10, 2014 at 12:33:52AM +0300, Evgeny Stupachenko wrote:
>> --- /dev/null
>> +++ b/gcc/testsuite/gcc.target/i386/blend.c
>> @@ -0,0 +1,61 @@
>> +/* Test correctness of size 3 store groups permutation. */
>> +/* { dg-do run } */
>
On 09/12/14 20:34 +0100, Marc Glisse wrote:
On Tue, 9 Dec 2014, Jonathan Wakely wrote:
On 08/12/14 23:53 +0100, François Dumont wrote:
After having installed all necessary locales on my system I end
up with 4 failures. Here is a patch to fix them all.
Did you discover why only you are seei
On 12/09/14 12:21, Wilco Dijkstra wrote:
This fixes a bug in register preferencing. When live range splitting creates a
new register from
another, it copies most fields except for the register preferences. The
preference GENERAL_REGS is
used as reg_pref[i].prefclass is initialized with GENERAL_
Hi Jason,
> I ran the tramp3d benchmark over 500 iterations before and after the
change and couldn't see any measurable difference in runtime. The
binary with my
> change was 0.4% smaller.
I'm going to go ahead and check it in; if a performance hit shows up on
the automated testing we can revi
On 12/09/14 14:10, Aldy Hernandez wrote:
I am tired of dumping entire DIEs just to see what type they are. With
this patch, we get:
(gdb) print context_die
$5 = >
I know it's past the end of stage1, but this debugging aid can help in
fixing bugs in stage >= 3.
I am committing this to the [deb
On 12/05/14 05:15, Bin Cheng wrote:
Hi,
Though PR62178 is hidden by recent cost change in aarch64 backend, the ivopt
issue still exists.
Current candidate selecting algorithm tends to select fewer candidates given
below reasons:
1) to better handle loops with many induction uses but the best
On Sun, 2014-11-30 at 20:08 +0100, Oleg Endo wrote:
> Hi,
>
> In libgcc/lib1funcs.S the usage of the __SHMEDIA__ macro seems to
> mismatch the setup in sh/sh-c.c (sh_cpu_cpp_builtins). __SHMEDIA__ is
> always defined when building for SH5 / SH64, even when the subtarget is
> SHcompact. Thus, whe
On 09 Dec 14:59, Richard Biener wrote:
> On Mon, 8 Dec 2014, Ilya Verbin wrote:
> > Unfortunately, this fix was not general enough.
> > There might be cases when mixed object files get into lto-wrapper, ie some
> > of
> > them contain only LTO sections, some contain only offload sections, and some
On 12/09/14 15:33, Matthew Fortune wrote:
Hi Prachi,
I'm afraid you updated the wrong Changelog with this commit. GCC changes
are recorded in gcc/ChangeLog.
I'm not sure what the correct procedure is for fixing this.
Jeff: Should a mistake like this be fixed by removing the entry
from the top
Hi Prachi,
I'm afraid you updated the wrong Changelog with this commit. GCC changes
are recorded in gcc/ChangeLog.
I'm not sure what the correct procedure is for fixing this.
Jeff: Should a mistake like this be fixed by removing the entry
from the top level and adding one at the appropriate loca
On Tue, 9 Dec 2014, Uros Bizjak wrote:
> Attached patch checks the return value and sets ptr to NULL in this case.
>
> 2014-12-09 Uros Bizjak
>
> * directives.c (cpp_define_formatted): Check return value of
> vasprintf and in case of error set ptr to NULL.
>
> Bootstrapped on x86_64-
On Wed, Dec 10, 2014 at 12:33:52AM +0300, Evgeny Stupachenko wrote:
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/blend.c
> @@ -0,0 +1,61 @@
> +/* Test correctness of size 3 store groups permutation. */
> +/* { dg-do run } */
> +/* { dg-options "-O3" } */
> +
> +#define N 50
> +
> +enum nu
On 12/08/2014 06:25 AM, David Edelsohn wrote:
> Okay. I don't know of another option that preserves backward
> compatibility. If RTH looks at this series of patches, he may have a
> better suggestion.
No, this looks like what I'd do.
Indeed, I added this hook back when the i386 port gained dire
I noticed while debugging another issue that at some point I'd broken
the source locations that toyvm.c sets up into the .toy source files,
by using the filename, rather than the filesystem path.
Also fix a segfault due to fclose (NULL) when the .toy file is not
found.
Committed to trunk as r2185
I've added the reproducer to the patch.
is it ok?
ChangeLog:
2014-12-10 Evgeny Stupachenko
gcc/testsuite
* gcc.target/i386/blend.c: New.
gcc/
* config/i386/i386.c (expand_vec_perm_pblendv): Gen new rtx for
expand_vec_perm_1 target.
diff --git a/gcc/config/i386/i386.
I am tired of dumping entire DIEs just to see what type they are. With
this patch, we get:
(gdb) print context_die
$5 = DW_TAG_compile_unit>>
I know it's past the end of stage1, but this debugging aid can help in
fixing bugs in stage >= 3.
I am committing this to the [debug-early] branch,
On 12/9/2014 3:00 PM, Jeff Law wrote:
Thoughts on using both patches to ensure the we don't have extraneous
entries in the GEN set and that we're adding more stuff to the KILL
set for sibcalls?
Maybe we should add all stores to the KILL set
for a sibling call and not worry about whether
they ar
2014-12-09 22:57 GMT+03:00 Jeff Law :
> On 12/09/14 01:24, Ilya Enkovich wrote:
>>
>> On 24 Nov 17:02, Ilya Enkovich wrote:
>>>
>>>
>>> Right. This works for both top level and multilib checks because failing
>>> test is used and CC is usually not set when it's called by the top level
>>> configur
Committed to trunk as r218538.
Index: gcc/jit/ChangeLog
===
--- gcc/jit/ChangeLog (revision 218537)
+++ gcc/jit/ChangeLog (revision 218538)
@@ -1,5 +1,12 @@
2014-12-09 David Malcolm
+ PR jit/63854
+ * docs/internals/index.rst (Ru
On 12/09/14 01:24, Ilya Enkovich wrote:
On 24 Nov 17:02, Ilya Enkovich wrote:
Right. This works for both top level and multilib checks because failing test
is used and CC is usually not set when it's called by the top level configure.
If we configure with CC=... then it may go wrong. I lef
On Dec 9, 2014, at 10:39 AM, Jeff Law wrote:
> Also, a test for the GCC testsuite would be good. I have no idea what
> license covers the LLVM testsuite. But given a good analysis of the problem
> we may be able to write a suitable test independent of the LLVM test.
So, the usual engineerings
On Dec 9, 2014, at 3:17 AM, Kyrill Tkachov wrote:
> In the gcc and g++ testsuite we already catch such cases and mark them as
> UNSUPPORTED. In libstdc++.exp there is no such functionality.
> I'm not very happy that it had to be copied, but I couldn't find a way to
> include the gcc definition
Move tempdir handling out from jit-playback.c into new
jit-tempdir.{c|h} files.
This simplifies jit-playback.c, and may help fix PR jit/64206.
Committed to trunk as r218533.
gcc/jit/ChangeLog:
PR jit/64206
* Make-lang.in (jit_OBJS): Add jit/jit-tempdir.o.
* jit-common.h (
On 12/08/14 14:50, John David Anglin wrote:
On 12/8/2014 3:49 PM, Jeff Law wrote:
I think the terminology and variable names certainly makes this
tougher to follow than it should.
I certainly agree. When I first looked at the code, I thought it
was completely backwards.
Me too.
Thoughts on us
On 12/09/14 12:43, Richard Biener wrote:
This seems to be indicating that we have with no edge from the latch
block to the header block. I'd like to know better how we got into
that
state.
It Also returns null for loops with multiple latches. So the patch looks OK for
me.
Ah, OK.
Jeff
On 12/09/14 01:27, Ilya Enkovich wrote:
2014-12-09 Ilya Enkovich
* config/i386/linux-common.h (LIBMPX_WRAPPERSSPEC): New.
(CHKP_SPEC): Add wrappers library.
* c-family/c.opt (static-libmpxwrappers): New.
libmpx/
2014-12-09 Ilya Enkovich
* Makefile.am (SU
On December 9, 2014 7:39:48 PM CET, Jeff Law wrote:
>On 12/09/14 10:38, Sebastian Pop wrote:
>> Richard Biener wrote:
>>> On Mon, Dec 8, 2014 at 10:49 PM, Steve Ellcey
>wrote:
expected? Should this test also check flag_thread_jumps? Or
>should
that be getting checked somewhere else?
>
On Tue, 9 Dec 2014, Jonathan Wakely wrote:
On 08/12/14 23:53 +0100, François Dumont wrote:
After having installed all necessary locales on my system I end up with
4 failures. Here is a patch to fix them all.
Did you discover why only you are seeing failures?
Not just him, anyone with a d
On Tue, 9 Dec 2014, Marek Polacek wrote:
The issue here is that TYPE_OVERFLOW_TRAPS, TYPE_OVERFLOW_UNDEFINED,
and TYPE_OVERFLOW_WRAPS macros work on integral types only, yet we
pass e.g. pointer_type/real_type to them. This patch adds proper
checking for these macros and adds missing guards to
This fixes a bug in register preferencing. When live range splitting creates a
new register from
another, it copies most fields except for the register preferences. The
preference GENERAL_REGS is
used as reg_pref[i].prefclass is initialized with GENERAL_REGS in
allocate_reg_info () and
resize_re
On Wed, 2014-11-19 at 13:38 -0500, David Malcolm wrote:
> On Wed, 2014-11-19 at 09:57 -0700, Jeff Law wrote:
> > On 11/19/14 03:46, David Malcolm wrote:
> > > This commit updates jit.exp so that if RUN_UNDER_VALGRIND is present
> > > in the environment, all of the built client code using libgccjit.
On 12/09/14 12:07, Segher Boessenkool wrote:
On Tue, Dec 09, 2014 at 05:49:18PM +0800, Zhenqiang Chen wrote:
Do you need to verify SETA and SETB satisfy single_set? Or has that
already been done elsewhere?
A is NEXT_INSN (insn)
B is prev_nonnote_nondebug_insn (insn),
For I1 -> I2 -> B; I2 ->
On Tue, Dec 09, 2014 at 05:49:18PM +0800, Zhenqiang Chen wrote:
> > Do you need to verify SETA and SETB satisfy single_set? Or has that
> > already been done elsewhere?
>
> A is NEXT_INSN (insn)
> B is prev_nonnote_nondebug_insn (insn),
>
> For I1 -> I2 -> B; I2 -> A;
> LOG_LINK can make sure I1
On Mon, Dec 8, 2014 at 9:18 AM, Segher Boessenkool
wrote:
> Peter tells me it was an artifact of old versions of the DFP code. The
> condition can never be false; delete it.
>
>
> 2014-12-08 Segher Boessenkool
>
> gcc/
> * config/rs6000/rs6000.md (*add3): Remove condition.
Okay.
Than
On Mon, Dec 8, 2014 at 9:18 AM, Segher Boessenkool
wrote:
>
> 2014-12-08 Segher Boessenkool
>
> gcc/
> * config/rs6000/40x.md (ppc403-compare): Remove "compare".
> config/rs6000/440.md (ppc440-compare): Remove "compare".
> config/rs6000/476.md (ppc476-compare): Remove "c
On Mon, Dec 8, 2014 at 9:18 AM, Segher Boessenkool
wrote:
> These now are the only remaining patterns that have type "compare".
> And they shouldn't: "two" is a better type for them.
>
>
> 2014-12-08 Segher Boessenkool
>
> gcc/
> * config/rs6000/rs6000.md (*anddi3_2rld_dot, *anddi3_rld_
On Mon, Dec 8, 2014 at 9:18 AM, Segher Boessenkool
wrote:
> Now that everything is in place for letting GCC use the carry-using and
> carry-producing machine insns as separate RTL insns, switch over all
> remaining patterns that clobber CA without telling the compiler. This
> are the multiple-pre
Move acquisition/release of the JIT mutex from
jit-recording.c:gcc::jit::recording::context::compile
into
jit-playback.c: gcc::jit::playback::context::compile
and reduce the amount of code guarded by the mutex somewhat.
For acquisition, we didn't need the mutex when building the tempdir and
bu
Hello!
The compilation with gentoo glibc 2.20 emits following warning:
../../../gcc-svn/trunk/fixincludes/server.c:195:10: warning: ignoring
return value of ‘getcwd’, declared with attribute warn_unused_result
[-Wunused-result]
The manpage says:
On failure, these functions return NULL, and errn
On 12/09/14 02:49, Zhenqiang Chen wrote:
Do you need to verify SETA and SETB satisfy single_set? Or has that
already been done elsewhere?
A is NEXT_INSN (insn)
B is prev_nonnote_nondebug_insn (insn),
For I1 -> I2 -> B; I2 -> A;
LOG_LINK can make sure I1 and I2 are single_set, but not A and B.
Another simplification of playback::context::compile.
Committed to trunk as r218527.
Index: gcc/jit/ChangeLog
===
--- gcc/jit/ChangeLog (revision 218526)
+++ gcc/jit/ChangeLog (revision 218527)
@@ -1,3 +1,12 @@
+2014-12-09 David Mal
On 12/09/14 05:39, Alexander Ivchenko wrote:
Hi,
Whether the *linux* target supports ifuncs or not is defined here:
linux_has_ifunc_p (void)
{
return OPTION_BIONIC ? false : HAVE_GNU_INDIRECT_FUNCTION;
}
Bionic right now supports indirect functions, but there is no way to
notify the compile
On 12/09/14 10:38, Sebastian Pop wrote:
Richard Biener wrote:
On Mon, Dec 8, 2014 at 10:49 PM, Steve Ellcey wrote:
expected? Should this test also check flag_thread_jumps? Or should
that be getting checked somewhere else?
-fthread-jumps is an RTL optimization flag and ignored on GIMPLE.
The issue here is that TYPE_OVERFLOW_TRAPS, TYPE_OVERFLOW_UNDEFINED,
and TYPE_OVERFLOW_WRAPS macros work on integral types only, yet we
pass e.g. pointer_type/real_type to them. This patch adds proper
checking for these macros and adds missing guards to various places.
This looks pretty straightfo
Hello!
The compilation with gentoo glibc 2.20 emits following warning:
../../../gcc-svn/trunk/libcpp/directives.c:2411:28: warning: ignoring
return value of ‘int vasprintf(char**, const char*, __gnuc_va_list)’,
declared with attribute warn_unused_result [-Wunused-result]
The manpage says:
If me
Richard Biener wrote:
> On Mon, Dec 8, 2014 at 10:49 PM, Steve Ellcey wrote:
> > expected? Should this test also check flag_thread_jumps? Or should
> > that be getting checked somewhere else?
>
> -fthread-jumps is an RTL optimization flag and ignored on GIMPLE.
Does it make sense to add a -f[n
On Mon, Dec 08, 2014 at 09:01:39AM -0600, Segher Boessenkool wrote:
> > Why are you removing the alternative instead of clobbering XER[CA]?
>
> I should have mentioned, sorry.
>
> We don't want to clobber CA on *every* add, and reload can generate more
> adds out of thin air. And CA is a fixed r
> Marcus Shawcroft wrote:
> > OK for commit?
> >
> > ChangeLog:
> > 2014-11-24 Wilco Dijkstra
> >
> > * gcc/config/aarch64/aarch64-protos.h (tune-params):
> > Add reasociation tuning parameters.
> > * gcc/config/aarch64/aarch64.c (TARGET_SCHED_REASSOCIATION_WIDTH):
> >
This was for x86_64-apple-darwin14. The patch also works for
x86_64-apple-darwin10.
Dominique
> Le 6 déc. 2014 à 01:49, Dominique d'Humières a écrit :
>
> Bootstrap just finished with the patch.
>
> Thanks,
>
> Dominique
On Tue, Dec 09, 2014 at 11:48:44AM -0500, David Edelsohn wrote:
> We definitely will have to define addptrM3 and restore addM3 when
> rs6000 switches to LRA.
Can LRA deal with adding a clobber of a hard register? How would
you express that in the patterns, anyway?
Segher
On Tue, Dec 9, 2014 at 11:24 AM, Eric Botcazou wrote:
>> It's unfortunate that GCC does not have separate patterns for a safe
>> ADD used by reload.
>
> It does, the addptrM3 pattern, but it works only with LRA at the moment.
Hi, Eric
Segher and I previously discussed addptrM3, but I thought tha
I mean that there are a lot of people tracking spec2006 stability and
therefore the issue should be on track in future.
And that I can create the test case, but it would be as big as several
GCC functions.
Will work on reducing the test case.
On Tue, Dec 9, 2014 at 7:20 PM, Richard Henderson wro
On 12/09/2014 06:53 AM, Varvara Rainchik wrote:
> 2014-12-09 Varvara Rainchik
>
> * config.h.in: Regenerate.
> * configure: Regenerate.
> * configure.ac: Add GCC_CHECK_EMUTLS.
> * libgomp.h: Add check for USE_EMUTLS: this case
> is equal to HAVE_TLS.
>
> It's unfortunate that GCC does not have separate patterns for a safe
> ADD used by reload.
It does, the addptrM3 pattern, but it works only with LRA at the moment.
--
Eric Botcazou
On Tue, Dec 09, 2014 at 06:59:27PM +0300, Evgeny Stupachenko wrote:
> I've tried to get smaller reproducer, however currently it is
> complicated as several functions in GCC.
Just add a gcc_assert where you are changing the code, testing for
what you want to avoid.
Then just delta reduce it or cre
On 5 December 2014 at 14:36, Kyrill Tkachov wrote:
> Hi all,
>
> Andrew posted this patch sometime ago (before stage1 closed) and I had
> rebased it on top of the
> other macro fusion patches in that series.
> This is a respin of that patch with the comment about not calling
> get_attr_type
> repe
On 12/09/2014 07:59 AM, Evgeny Stupachenko wrote:
> However patch is fixing spec2006 benchmark. Shouldn't that be enough
> for regression testing?
>
No. Spec is not free.
r~
On Mon, 2014-12-08 at 14:13 +0100, Rainer Orth wrote:
> Jeff Law writes:
>
> > On 12/04/14 15:42, Rainer Orth wrote:
> >> David Malcolm writes:
> >>
> >>> assumed -fgnu89-inline until a recent upstream fix;
> >>> see http://lists.gnu.org/archive/html/dejagnu/2014-10/msg00011.html
> >>>
> >>> Re
On 24 November 2014 at 13:46, Wilco Dijkstra wrote:
>> Richard Earnshaw wrote:
>> If all cores seem to benefit from FP reassociation set to 4, then it
>> seems odd that 4 is not also the default for generic.
>>
>> Andrew, you may need to pick a target-specific value for ThunderX; I
>> think Wilco
I've tried to get smaller reproducer, however currently it is
complicated as several functions in GCC.
However patch is fixing spec2006 benchmark. Shouldn't that be enough
for regression testing?
Thanks,
Evgeny
On Tue, Dec 9, 2014 at 4:29 PM, Evgeny Stupachenko wrote:
> The case comes from spec2
On 11/21/2014 01:23 PM, Richard Biener wrote:
On Fri, Nov 21, 2014 at 12:52 PM, Martin Liška wrote:
On 11/20/2014 05:41 PM, Richard Biener wrote:
On Thu, Nov 20, 2014 at 5:30 PM, Martin Liška wrote:
Hello.
Following patch fixes ICE in IPA ICF. Problem was that number of
non-debug
statemen
From: Charles Baylis
gcc/ChangeLog:
Charles Baylis
* config/aarch64/aarch64-builtins.c
(aarch64_types_storestruct_lane_qualifiers): Mark last argument with
qualifier_struct_load_store_lane_index.
gcc/testsuite/ChangeLog:
Charles Baylis
* gcc.target/aar
From: Charles Baylis
gcc/ChangeLog:
Charles Baylis
* config/aarch64/arm_neon.h (__ST2_LANE_FUNC): Add explicit lane bounds
check.
(__ST3_LANE_FUNC): Likewise.
(__ST4_LANE_FUNC): Likewise.
gcc/testsuite/ChangeLog:
Charles Baylis
* gcc.target/aar
From: Charles Baylis
gcc/ChangeLog
Charles Baylis
* config/aarch64/arm_neon.h (__LD2_LANE_FUNC): Add explicit lane
bounds check.
(__LD3_LANE_FUNC): Likewise.
(__LD4_LANE_FUNC): Likewise
gcc/testsuite/ChangeLog:
Charles Baylis
* gcc.target/aarch
From: Charles Baylis
gcc/ChangeLog:
Charles Baylis
PR target/63870
* config/aarch64/aarch64-builtins.c (enum aarch64_type_qualifiers):
Add qualifier_struct_load_store_lane_index.
(aarch64_types_loadstruct_lane_qualifiers): Use
qualifier_struct_load_st
From: Charles Baylis
This patch series moves the checking of lane indices for vld[234](q?)_lane and
vst[234](q?)_lane intrinsics so that it occurs during builtin expansion.
The q register variants are checked directly, but since the d register variants
use the same intrinsics, these are checked
On Tue, Dec 9, 2014 at 2:11 PM, Yuri Rumyantsev wrote:
> Richard,
>
> Here is updated patch2 with the following changes:
> 1. Delete functions phi_has_two_different_args and find_insertion_point.
> 2. Use only one function for extended predication -
> predicate_extended_scalar_phi.
> 3. Save gsi
On Mon, Oct 27, 2014 at 8:19 PM, Kito Cheng wrote:
>
> This patch update `Bit operations` section in libgcc.text, most bit
> operation function is take an unsigned integer instead of signed
> integer in libgcc/libgcc2.c [1], and it seem more make sense :)
>
> ChangeLog
> 2014-10-28 Kito Cheng
>
Ok, the following patch works for Android:
2014-12-09 Varvara Rainchik
* config.h.in: Regenerate.
* configure: Regenerate.
* configure.ac: Add GCC_CHECK_EMUTLS.
* libgomp.h: Add check for USE_EMUTLS: this case
is equal to HAVE_TLS.
* team.c: Like
On 18/11/14 11:28, Yangfei (Felix) wrote:
On 11/18/2014 11:48 AM, Yangfei (Felix) wrote:
+(define_expand "doloop_end"
+ [(use (match_operand 0 "" "")) ; loop pseudo
+ (use (match_operand 1 "" ""))] ; label
+ ""
+{
+ /* Currently SMS relies on the do-loop pattern to recognize loops
Hello!
The fix for PR64037 [1] introduced severe problems, including
bootstrap failure on alphaev6-linux-gnu. Non-bootstrapped testsuite
results [2] can be compared with bootstrapped gcc with a revert [3].
As discussed in the PR, the problem was in a missed SI-DI zero_extend.
[1] https://gcc.gnu.
On Mon, Dec 8, 2014 at 5:52 PM, Martin Liška wrote:
> On 11/28/2014 10:32 AM, Richard Biener wrote:
>>
>> On Thu, Nov 27, 2014 at 6:08 PM, Martin Liška wrote:
>>>
>>> On 11/21/2014 04:21 PM, Martin Liška wrote:
On 11/21/2014 04:02 PM, Richard Biener wrote:
>
>
> On Fri,
On Mon, 8 Dec 2014, Ilya Verbin wrote:
> Hi,
>
> On 28 Nov 09:36, Richard Biener wrote:
> > On Fri, 28 Nov 2014, Ilya Verbin wrote:
> > > I found a bug here, have_{lto,offload} must be set if at least one file
> > > contains
> > > lto/offload sections, but currently they are overwritten by the l
The following finally fixes PR42108 (well, hopefully...) by using
range-information on SSA names to allow the integer divisions introduced
by Fortran array lowering being hoisted out of loops, thus detecting
them as not trapping.
I chose to enhance tree_single_nonzero_warnv_p for this and adjuste
The following patch restores proper CSE by FRE/PRE when doing
optimistic value-numbering of stores/loads. In that case we
may end up value-numbering a virtual operand to the only executable
edge value of a PHI - but as the alias walker doesn't know this
it fails to handle the PHI which results in
On 9 December 2014 at 13:52, Jiangjiji wrote:
> Hi,
> This patch converts more intrinsics to use builtin functions instead of
> the
> previous inline assembly syntax.
> Passed the glorious testsuite of Christophe Lyon.
>
> Three testcases are added for the testing of intriniscs whic
The case comes from spec2006 403.gcc (or old GCC itself).
for (i = 0; i < FIRST_PSEUDO_REGISTER; ++i)
{
vd->e[i].mode = VOIDmode;
vd->e[i].oldest_regno = i;
vd->e[i].next_regno = INVALID_REGNUM;
}
It is vectorized and only then completely peeled.
Only after peeling all
On Tue, Dec 9, 2014 at 6:36 AM, Chung-Ju Wu wrote:
> Hi, all,
>
> In the discussion thread last year:
> https://gcc.gnu.org/ml/gcc-patches/2013-05/msg01334.html
>
> I extended the script ./contrib/download_prerequisites so that it can
> download isl and cloog packages for graphite loop optimizat
Dear Andre,
The patch causes an ICE for the test gfortran.dg/unlimited_polymorphic_1.f03:
f951: internal compiler error: in gfc_add_component_ref, at fortran/class.c:236
f951: internal compiler error: Abort trap: 6
gfc: internal compiler error: Abort trap: 6 (program f951)
Abort
Reduced test fo
On Mon, Dec 8, 2014 at 10:49 PM, Steve Ellcey wrote:
> On Sat, 2014-12-06 at 19:21 +, Sebastian Pop wrote:
>
>> > I think it does not make sense to duplicate paths at -Os: I disabled the
>> > FSM
>> > jump-threading when optimizing for size like this.
>> >
>> > diff --git a/gcc/tree-ssa-threa
On 9 December 2014 at 03:26, Yangfei (Felix) wrote:
>> On 5 December 2014 at 18:44, Tejas Belagod wrote:
>> >
>> >>>
>> >>> +__extension__ static __inline float32x2_t __attribute__
>> >>> +((__always_inline__))
>> >>> +vfms_f32 (float32x2_t __a, float32x2_t __b, float32x2_t __c) {
>> >>> + retur
Richard,
Here is updated patch2 with the following changes:
1. Delete functions phi_has_two_different_args and find_insertion_point.
2. Use only one function for extended predication -
predicate_extended_scalar_phi.
3. Save gsi before insertion of predicate computations for basic
blocks if it has
On Mon, Dec 8, 2014 at 3:45 PM, Marc Glisse wrote:
> On Tue, 18 Nov 2014, Richard Biener wrote:
>
>>> I'll try to replace some more TYPE_MODE during stage3...
>>
>>
>> Btw, a convenience would be to be able to write
>>
>> HONOR_NANS (type)
>>
>> thus effectively make HONOR_* inline functions with
1 - 100 of 126 matches
Mail list logo
