On Thu, Dec 13, 2012 at 1:02 PM, Jakub Jelinek wrote:
>> > Various TM tests ICE when built with -fgnu-tm -fsanitizer=address.
>> > The problem is that asan.c pass adds calls to builtins that weren't there
>> > before and TM is upset about it. The __asan_report* are all like
>> > abort, in correct
Hi Uros,
With your new fix that add if-then-else splitting for memory operand I
got expected performance speed-up - +6.7% for Atom and +8.4% for SNB.
We need to do all testing this weekend and I will get you our final
feedback on Monday.
Thanks ahead for all your help.
Yuri.
2012/12/13 Uros Bizj
Hi Richard,
> + "add\\t%w0, %w2, %w, xt"
>
> ^^^ %w1
Got spot. I guess that pattern hasn't fired yet then! I'll fix it.
> > This patch significantly reduces the number of redundant
> > uxtw instructions seen in a variety of programs.
> >
> > (There are further patterns th
Honza left an ICE when we cannot compute number_of_iterations_exit
in remove_redundant_iv_tests, but that can for sure happen by design
(see bugzilla).
The following patch fixes it, bootstrap and regtest pending on
x86_64-unknown-linux-gnu.
Richard.
2012-12-14 Richard Biener
PR tre
> Hi Richard,
>
> > + "add\\t%w0, %w2, %w, xt"
> >
> > ^^^ %w1
>
> Got spot. I guess that pattern hasn't fired yet then! I'll fix it.
Now fixed in v3.
> I should have said that I am indeed running with REE enabled. It has
> some impact (about 70 further UXTW removed from
Hi,
below are two variants for PR 55355, one for trunk and for 4.7, the
second one, without dumping, for 4.6. On 4.6 they fix an ICE for a
too-large integer on i686, which perhaps can also happen on 4.7 and
trunk even though it does not for their particular testcase.
I've bootstrapped and tested
This fixes PR55687 - there is a latent issue in SCEV analysis.
For (int) (short) {(unsigned short) i_2, +, 1 }_4 no_evolution_in_loop_p
returns false when asked whether that CHREC has an evolution in loop 4
(which is obviously wrong). The reason is that the machinery is not
good at filtering NOPs
On Fri, Dec 14, 2012 at 1:04 PM, Martin Jambor wrote:
> Hi,
>
> below are two variants for PR 55355, one for trunk and for 4.7, the
> second one, without dumping, for 4.6. On 4.6 they fix an ICE for a
> too-large integer on i686, which perhaps can also happen on 4.7 and
> trunk even though it doe
On Thu, 2012-12-13 at 10:02 +0100, Jakub Jelinek wrote:
> On Thu, Dec 13, 2012 at 10:38:13AM +0400, Dmitry Vyukov wrote:
> > On Wed, Dec 12, 2012 at 11:50 PM, Jakub Jelinek wrote:
> > > Various TM tests ICE when built with -fgnu-tm -fsanitizer=address.
> > > The problem is that asan.c pass adds ca
On Fri, 2012-12-14 at 13:44 +0400, Dmitry Vyukov wrote:
> On Thu, Dec 13, 2012 at 1:02 PM, Jakub Jelinek wrote:
> >> > Various TM tests ICE when built with -fgnu-tm -fsanitizer=address.
> >> > The problem is that asan.c pass adds calls to builtins that weren't there
> >> > before and TM is upset a
On Thu, 2012-12-13 at 10:40 +0400, Dmitry Vyukov wrote:
> On Thu, Dec 13, 2012 at 1:58 AM, Richard Henderson wrote:
> > On 12/12/2012 11:50 AM, Jakub Jelinek wrote:
> >> 2012-12-12 Jakub Jelinek
> >>
> >> PR sanitizer/55508
> >> * builtin-attrs.def (ATTR_TMPURE_NOTHROW_LEAF_LIST,
>
On Thu, Aug 09, 2012 at 07:31:58PM -0700, Richard Henderson wrote:
> Constant fold, and less typing than, GET_MODE_BITSIZE with
> another mode substitution.
I did just a minor change in order to avoid a warning in the last hunk.
Tested on s390 and s390x with different arch flags.
Please apply t
I've changed the s390_extzv_shift_ok function and have added extzv
patterns for zEC12. I also tried to implement an extzvsi expander but
ran into some issues which need to be fixed first.
Tested on s390 and s390x with z196 and zEC12.
Please apply to mainline. Thanks!
Bye,
-Andreas-
gcc/config
I only changed DSI to GPR. This is semantically the same but more
common in the backend.
Tested on s390 and s390x with z196 and zEC12.
Please apply to mainline. Thanks!
Bye,
-Andreas-
gcc/config/s390/s390.md | 61 ++!!
1 file changed, 3 insertions
I did the change suggested by Uli in the *insv_rnsbg_srl pattern.
Tested on s390 and s390x with z196 and zEC12.
Please apply to mainline. Thanks!
Bye,
-Andreas-
gcc/config/s390/s390.md | 55
1 file changed, 55 insertions(+)
Index: gcc/config
Hi all,
here is another attempt to make gfortran support user-requested backtraces.
A patch in this direction was already proposed by FX in March, but did
not make it in so far. It was last discussed in June, cf.
http://gcc.gnu.org/ml/fortran/2012-06/msg00131.html and follow-ups,
where the consen
My understanding of shifts of constants in Go has improved. This patch
more accurately reflects the language spec when it comes to reporting
errors for invalid shifts. Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Committed to mainline.
Ian
diff -r 9be532343312 go/expressions.
The attached revised patch disables two asan tests introduced in r194458
which are invalid
on darwin. The g++.dg/asan/interception-test-1.C test should be dg-skip-if'd
on darwin since
mac function interposition is used instead of interception. The
c-c++-common/asan/swapcontext-test-1.c
test
On Fri, Dec 14, 2012 at 10:13:47AM -0500, Jack Howarth wrote:
> 2012-12-14 Jack Howarth
>
> PR sanitizer/55679
> * g++.dg/asan/interception-test-1.C: Skip on darwin.
> * lib/target-supports.exp (check_effective_target_swapcontext): Use
> check_no_compiler_messages to test
Yvan noticed that the set of memory models has already been
validated by expand_builtin_atomic_store, so the test here
in expand_atomic_store is fully redundant.
Committed.
r~
* optabs.c (expand_atomic_store): Elide redundant model test.
diff --git a/gcc/optabs.c b/gcc/optabs.c
index 9
On 2012-12-13 23:08 , Lawrence Crowl wrote:
+ class iterator
+ {
+ public:
+inline iterator ();
+inline iterator (value_type **, value_type **);
+inline value_type &operator * ();
+void slide ();
+inline iterator &operator ++ ();
+inline bool operator != (const iterato
In this bug, when we see the invalid initializer we changed the type of
x to be error_mark_node. But we had already built an ARRAY_REF to it
using the incomplete type. Then the back end gets confused by this
invalid expression wrapped in a valid one.
Fixed by not messing with the type of x.
Ping^2?
On 7 December 2012 09:34, Christophe Lyon wrote:
> Ping?
> http://gcc.gnu.org/ml/gcc-patches/2012-11/msg02558.html
>
> Thanks,
>
> Christophe.
On Fri, Dec 14, 2012 at 04:59:02AM +, Iyer, Balaji V wrote:
> --- tree-vect-loop.c (revision 194483)
> +++ tree-vect-loop.c (working copy)
> @@ -234,8 +234,8 @@
> if (flag_enable_cilk && pragma_simd_assert_requested_p
> (loop->pragma_simd_index))
>
Hello Jakub,
Please see my responses below.
Thanks,
Balaji V. Iyer.
> -Original Message-
> From: Jakub Jelinek [mailto:ja...@redhat.com]
> Sent: Friday, December 14, 2012 11:59 AM
> To: Iyer, Balaji V
> Cc: gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH][Cilkplus] Fix pragma simd
On Fri, Dec 14, 2012 at 05:01:48PM +, Iyer, Balaji V wrote:
> > Why do you think fatal_error is the right thing here? Why doesn't normal
> > error
> > work? Generally, if one function contains 10 #pragma simd loops that
> > require
> > vectorization and 5 out of them aren't vectorized, it i
Hi,
This patch adds the basic recognition for cpu cortex-a53 and cortex-a57.
OK to commit?
Thanks,
Yufeng
gcc/
2012-12-14 Yufeng Zhang
* config/aarch64/aarch64-cores.def: Add entries for
"cortex-a53" and
"cortex-a57".
* config/aarch64/aarch64-tune.md: Re-gener
On Thu, Dec 13, 2012 at 10:56 PM, Joey Ye wrote:
>> -Original Message-
>> From: H.J. Lu [mailto:hjl.to...@gmail.com]
>> Sent: Friday, December 14, 2012 11:55
>> To: Joey Ye
>> Cc: gcc-patches@gcc.gnu.org; Joseph Prostko
>> Subject: Re: [PATCH, libgcc] Make possible to disable JCR in crtstu
Here are four more C++ tests that fail for me when run under the GNU
simulator. I would like to shrink them to use less memory in the same
way as the other tests that I modified earlier.
OK to checkin?
Steve Ellcey
sell...@mips.com
2012-12-14 Steve Ellcey
* testsuite/21_strings/bas
Btw: Forgot to mention that it regtests cleanly and the docu parts
were tested with make pdf.
Cheers,
Janus
2012/12/14 Janus Weil :
> Hi all,
>
> here is another attempt to make gfortran support user-requested backtraces.
>
> A patch in this direction was already proposed by FX in March, but di
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
> ow...@gcc.gnu.org] On Behalf Of Jakub Jelinek
> Sent: Friday, December 14, 2012 12:09 PM
> To: Iyer, Balaji V
> Cc: gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH][Cilkplus] Fix pragma simd info being lost
>
Hello Everyone,
This patch is for the Cilk Plus branch mainly affecting the C compiler.
It will reject certain type of comparisons (e.g. ==) and complain about
volatile or const induction variables all in _Cilk_for.
Thanks,
Balaji V. Iyer.
Index: gcc/c/c-parser.c
===
On 14 December 2012 18:10, Steve Ellcey wrote:
> Here are four more C++ tests that fail for me when run under the GNU
> simulator. I would like to shrink them to use less memory in the same
> way as the other tests that I modified earlier.
>
> OK to checkin?
OK, thanks.
I am going to check in this patch as the obvious fix for PR 55688. I
tested it on mips-mti-elf.
Steve Ellcey
sell...@mips.com
2012-12-14 Steve Ellcey
PR regression/55688
* g++.dg/other/pr55650.C: Add dg-require-profiling.
diff --git a/gcc/testsuite/g++.dg/other/pr55650.C
Hi Rong,
Please review this code. This code allows the function reordering
plugin to separate hot and cold code into different ELF segments.
This would allow optimizations like mapping the hot code alone to huge
pages.
With this patch, by default, the plugin maps .text.unlikely
sections i
This patch enables the gcov-dump tool to optionally compute and dump
the working set information from the counter histogram, via a new -w option.
This is useful to help understand and tune how the compiler will use
the counter histogram, since it first computes the working set and selects
threshold
Teresa,
Yes, I didn't take enhancements in google branches into account while
porting this patch. In light of these comments, I withdraw this patch
and will wait for your patch. Once your patch is in, I will update
this patch for regular dumps.
To answer your other question, yes, the new dump inf
Here the problem was that tsubst_copy_and_build wasn't guarding the
SIZEOF_EXPR_TYPE_P code with processing_template_decl like we do in the
parser. It's still necessary here because sometimes tsubsting happens
when we're still in a template. Fixed by updating the code to match.
Tested x86_64
38 matches
Mail list logo
