Hi,
ix86_expand_set_or_movmem is trying to use widest possible vector mode
available.
This does not help for ryzen, because 512 operations are performed by halves.
This patch by itself does not affect generated code because memcpy/memset
expansion
tables needs to be updated.
Bootstrapped/regtest
Hi,
at the moment we do dispatch scheduling which is trying to optimize for
Buldrozer's
decoder. This does not make much sense as both decoders are significantly
different.
Bootstrapped/regtested x86_64-linux, comitted.
Honza
* i386.c (has_dispatch): Disable for Ryzen.
Index: i386.c
=
Sorry for the delay. I lost track of this discussion.
The patch looks good to me.
I haven't checked the unit tests because I don't understand them. I
assume they all succeed when you run the tests so they are all ok.
I don't remember seeing a definite answer about the C-only syntax used in
err
Hello,
this moves (and extends a bit) one more transformation from fold-const.c
to match.pd. The single_use restriction is necessary for consistency with
the existing X+CST1 CMP CST2 transformation (if we do only one of the 2
transformations, gcc.dg/tree-ssa/vrp54.c regresses because DOM fails
On 2017/9/28 4:09, Joseph Myers wrote:
On Thu, 28 Sep 2017, Liu Hao wrote:
Colorized diagnostics used to be disabled for MinGW targets (on which
the macro `_WIN32` is defined), and this patch enables it.
I'd hope this is all to do with MinGW host, and nothing to do with the
target.
Ping? Ar
On 2017/10/8 19:55, Hannes Domani wrote:
So why not just enable it on Win10?
It is up to you, GCC maintainers. If dropping support for Windows prior
to Windows 10 TH2 is an option, I may provide another patch, which I
can't test because I primarily work on Windows 7. XD
--
Best regards,
LH_
On 2017/10/8 20:24, Hannes Domani wrote:
Am Sonntag, 8. Oktober 2017, 14:02:48 MESZ hat Liu Hao
Folgendes geschrieben:
> On 2017/10/8 19:55, Hannes Domani wrote:
>
> > So why not just enable it on Win10?
>
>
> It is up to you, GCC maintainers. If dropping support for Windows prior
> to
Hello world,
I have committed a few small fixes for -fdump-fortran-original and
friends.
Since there was interest on c.l.f from somebdoy who actually wanted to
use the dump for something useful, I have also clarified in the docs
that this is something that is not guaranteed to work between relea
Hi Thomas,
I thought that the suggestion to add the original input lines was not
bad. It would be even better if -fdump-tree-original could do that. I
often time find myself putting a line under the spotlight in a
contained subroutine.
Thanks for working on -fdump-parse-tree.
Cheers
Paul
On 8
FAIL: gcc.dg/cold-1.c (test for excess errors)
Excess errors:
/usr/local/gcc/gcc-20171008/gcc/testsuite/gcc.dg/cold-1.c:10:5: warning:
implicit declaration of function 'abort' [-Wimplicit-function-declaration]
/usr/local/gcc/gcc-20171008/gcc/testsuite/gcc.dg/cold-1.c:10:5: warning:
in
On Okt 08 2017, Eric Botcazou wrote:
> * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
> of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
This breaks gcc.dg/plugin/must-tail-call-2.c, gcc.dg/torture/pr81083.c
and gcc.dg/torture/pr82264.c:
warning: conflicti
Hi Paul,
I thought that the suggestion to add the original input lines was not
bad. It would be even better if -fdump-tree-original could do that. I
often time find myself putting a line under the spotlight in a
contained subroutine.
This should be doable.
What should be the perferred format
Hello,
Le 08/10/2017 à 15:15, Paul Richard Thomas a écrit :
I thought that the suggestion to add the original input lines was not
bad. It would be even better if -fdump-tree-original could do that.
There is -fdump-tree-original-lineno. I think it does that.
Cheers
Mikael
On 06/10/2017 20:00, Petr Ovtchenkov wrote:
On Fri, 6 Oct 2017 18:01:36 +0200
François Dumont wrote:
...
The test itself simulate "stop and go" istream usage.
stringstream is convenient for behaviuor illustration, but in "real life"
I can assume socket or tty on this place.
At the very minim
Red Hat's internal GCC testing does runs with -fstack-protector. This
exposed failures with a few of the stack clash protection tests.
I looked at all the failures. In every case we were still getting
proper protection, but the addition of stack-protector code subtly
changes the code generation
reation of a new block for a dispatch label.
This also breaks gcc.c-torture/compile/951222-1.c:
during GIMPLE pass: lower
/opt/gcc/gcc-20171008/gcc/testsuite/gcc.c-torture/compile/951222-1.c:4:1:
internal compiler error: in gimple_call_arg, at gimple.h:3159
0x118ab1b gimple_call_arg
../../
On Fri, Oct 6, 2017 at 4:37 PM, Martin Sebor wrote:
> On 10/06/2017 12:06 PM, Bernd Edlinger wrote:
>>
>> On 10/06/17 17:43, Martin Sebor wrote:
>>>
>>> On 10/06/2017 07:25 AM, Bernd Edlinger wrote:
On 10/05/17 18:16, Martin Sebor wrote:
>
> In my (very quick) tests the warning a
On Fri, Oct 6, 2017 at 2:06 PM, Bernd Edlinger
wrote:
> On 10/06/17 17:43, Martin Sebor wrote:
>> On 10/06/2017 07:25 AM, Bernd Edlinger wrote:
>>> On 10/05/17 18:16, Martin Sebor wrote:
In my (very quick) tests the warning appears to trigger on all
strictly incompatible conversions, eve
In check.c, gfc_check_x and gfc_check_fn_r were indentical functions
(well other than the names). I have removed gfc_check_x and updated
its uses to those of gfc_check_fn_r. After a bootstrap and regression
testing I've committed the attached patch.
2017-10-08 Steven G. Kargl
* check
> This also breaks gcc.c-torture/compile/951222-1.c:
>
> during GIMPLE pass: lower
> /opt/gcc/gcc-20171008/gcc/testsuite/gcc.c-torture/compile/951222-1.c:4:1:
> internal compiler error: in gimple_call_arg, at gimple.h:3159 0x118ab1b
> gimple_call_arg
You're right, not cle
Hi all,
I'm going to commit this patch if no objections pop up.
We (FreeBSD folks) have added a new target tripplet for armv7*-*-freebsd*
To be able to build GCC we need the below changes.
Results will be posted to the usual place, once they completed
While here I remove an obsolete define
Hi all,
here is the patch witch adds support for armv7*-*-freebsd* on the gcc-7
branch.
The difference towards trunk is the target_cpu_cname.
I'll commit this one if no objections pop up.
TIA,
Andreas
2017-10-08 Andreas Tobler
* config.gcc: (armv7*-*-freebsd*): New target.
Hi all,
and here is the patch for gcc-6/5 branch to add support for
armv7*-*-freebsd*
The difference towards trunk is the missing of target_cpu_cname which is
solved differently.
I'm aware of the gcc-5 branch status and I will not apply before gcc-5.5
is released. I'll also will not apply
* expr.c (expand_expr_real_1) : For a SSA_NAME, do
not set SUBREG_PROMOTED_VAR_P on the sub-register if it may contain
uninitialized bits.
* loop-iv.c (iv_get_reaching_def): Disqualify all subregs.
2017-10-08 Eric Botcazou
* gcc.c-torture/execute/20171008-1.c:
Currently we have two patterns for every conditional jump: one jumping
if some condition is met, and the other jumping if it is not met. We
don't need that second form because all our conditions have an opposite
already, and the generic code can deal with that just fine.
This patch deletes that s
This cleans up formatting of many jump patterns. Mostly it is writing
output statements as brace blocks, and deleting default arguments. It
also makes indentation a little bit more consistent, and uses b%T0
(instead of bctr/blr alternatives) in two more places.
Tested as usual; committing to tru
On Sun, 8 Oct 2017, Andreas Schwab wrote:
> On Okt 08 2017, Eric Botcazou wrote:
>
> > * builtins.def (BUILT_IN_SETJMP): Declare as library builtin instead
> > of GCC builtin if DONT_USE_BUILTIN_SETJMP is defined.
>
> This breaks gcc.dg/plugin/must-tail-call-2.c, gcc.dg/torture/pr81083.
This patch allows you to configure GCC to explicitly (not) add
-lssp_nonshared -lssp link options when the -fstack-protector options
are provided, by overriding the inferred default setting for
TARGET_LIBC_PROVIDES_SSP. The specific use case this is for is a
bare-metal target where the runtime
28 matches
Mail list logo
