On 07/28/2015 09:52 AM, Nathan Sidwell wrote:
I've committed this patch to the gomp4 branch to redo the launch API.
I'll post a version for trunk once the versioning patch gets approved
committed.
This changes the API in a number of ways, allowing device-specific
knowledge to be moved
Ping?
1) updated version patch
https://gcc.gnu.org/ml/gcc-patches/2015-08/msg00010.html
2) https://gcc.gnu.org/ml/gcc-patches/2015-08/msg00204.html
An infrastructure piece from Thomas, who noticed liboffloadmic didn't have the
include paths to #include gomp-constants.h.
nathan
On 08/02/15
David Malcolm dmalc...@redhat.com writes:
On Wed, 2015-08-05 at 16:22 -0400, Trevor Saunders wrote:
On Wed, Aug 05, 2015 at 11:34:28AM -0400, David Malcolm wrote:
On Wed, 2015-08-05 at 11:28 -0400, David Malcolm wrote:
On Wed, 2015-08-05 at 13:47 +0200, Richard Biener wrote:
On Wed, Aug
In this testcase a tentative parse was failing, and after an error we
called cp_parser_skip_to_end_of_statement, which messes with generic
function scope. So we shouldn't call that function during tentative
parsing; we don't need to worry about error recovery then, anyway.
Tested
On 08/06/15 17:56, Cesar Philippidis wrote:
On 08/06/2015 01:41 AM, Nathan Sidwell wrote:
I've committed this to fix the spinlock problem Cesar fell over. While
there I added more checking on the worker dimension.
I hit a couple of more bugs with the spinlocks. First, the address space
On 08/06/15 18:33, Cesar Philippidis wrote:
Looking at set_oacc_fn_attrib, it appears that const values are also
considered dynamic. See the attached test case more more info. Is that
the expected behavior? If not, I could take a look at this after I
finished my reduction patch.
It's annoying
Hi Pinski,
On 06/08/15 10:48, pins...@gmail.com wrote:
On Aug 6, 2015, at 11:40 AM, Renlin Li renlin...@arm.com wrote:
Hi all,
This is a simple patch to add a new cmovdi_insn_uxtw rtx pattern to aarch64
backend.
For the following simple test case:
unsigned long
foo (unsigned int a,
Le 19/04/2015 16:54, Mikael Morin a écrit :
Hello,
while working on pr65792, I noticed that gfc_trans_scalar_assign's
l_is_temp and dealloc flags are used only once, and at the same place.
This patch merges them together.
The calls are changed from
gfc_trans_scalar_assign (...blah...,
Hi,
the attached patch adds documentation for three options.
Committed to mainline.
Bye,
-Andreas-
gcc/ChangeLog:
2015-08-06 Andreas Krebbel kreb...@linux.vnet.ibm.com
* config/s390/s390.opt: Clarify description for -mzvector
* doc/invoke.texi: Add documentation for
Dear all,
the attached patch fixes a regression introduced by my patches for the
F2008-style allocate(). In this case a function returning an array of BT_CLASS
objects can not be conv'ed using conv_expr_descriptor, but needs to be
conv_expr_reference()'ed, because the _data component has the
With the updated patch the test gfortran.dg/ieee/large_1.f90 compiles, but
fails at run time due to the lines
if (.not. ieee_support_underflow_control(x1)) call abort
and
if (.not. ieee_support_underflow_control(x2)) call abort
IIRC Uros said that underflow id not supported for
Can you please also introduce tests for ieee exceptions for long
double and __float128 types (as is done for real and double in
gfortran.dg/ieee/ieee_1.F90) as well as test for supported rounding
modes as done for real and double in gfortran.dg/ieee/rounding_1.f90 ?
On x86_64, these new
Hi all,
This is a simple patch to add a new cmovdi_insn_uxtw rtx pattern to
aarch64 backend.
For the following simple test case:
unsigned long
foo (unsigned int a, unsigned int b, unsigned int c)
{
return a ? b : c;
}
With this new pattern, the new code-generation will be:
cmpw0,
On Aug 6, 2015, at 11:40 AM, Renlin Li renlin...@arm.com wrote:
Hi all,
This is a simple patch to add a new cmovdi_insn_uxtw rtx pattern to aarch64
backend.
For the following simple test case:
unsigned long
foo (unsigned int a, unsigned int b, unsigned int c)
{
return a ? b
Le 25/07/2015 20:33, Mikael Morin a écrit :
Le 21/07/2015 23:10, Paul Richard Thomas a écrit :
On 21 July 2015 at 14:53, Mikael Morin mikael.mo...@sfr.fr wrote:
Hello,
The fix for PR61831 committed recently [1] introduced/uncovered a NULLL
pointer dereference with iso_varying_string, because
Le 21/07/2015 13:52, Mikael Morin a écrit :
Hello,
this is a followup to the bug report/patch at:
https://gcc.gnu.org/ml/gcc-patches/2015-07/msg01062.html
It's a fortran quite unimportant bug, so I bet nobody cares.
However, the bug could have been detected when the co_reduce support was
Dear Mikael,
Well spotted! This is fine for trunk, since it is 'obvious' once seen.
Cheers
Paul
On 6 August 2015 at 12:07, Mikael Morin mikael.mo...@sfr.fr wrote:
Le 19/04/2015 16:54, Mikael Morin a écrit :
Hello,
while working on pr65792, I noticed that gfc_trans_scalar_assign's
Hi Mikael,
This is fine for backporting.
Thanks for doing this.
Paul
On 6 August 2015 at 12:09, Mikael Morin mikael.mo...@sfr.fr wrote:
Le 25/07/2015 20:33, Mikael Morin a écrit :
Le 21/07/2015 23:10, Paul Richard Thomas a écrit :
On 21 July 2015 at 14:53, Mikael Morin mikael.mo...@sfr.fr
Hi,
this fixes an inconsistency with the dwarf reg size table when using
-m31 -mzarch. In this mode we use 64 bit register while still
adhering to the 31 bit ABI. This changed some of the GPR sizes
returned by __builtin_init_dwarf_reg_size_table. It is not a big
problem since it only changed
Hi Mikael,
This is OK for trunk.
Thanks for resurrecting the patch.
Cheers
Paul
On 6 August 2015 at 12:11, Mikael Morin mikael.mo...@sfr.fr wrote:
Le 29/07/2015 20:35, Mikael Morin a écrit :
Hello,
I'm unburrying the patch from the thread starting at:
Hi,
a transaction rollback does not restore the contents of vector
registers. So we have to add clobbers for them as we are doing
already for the FPRs.
Bootstrap and regression testing is fine on z13, zEC12, and z196 with
31 and 64 bit.
Committed to mainline.
Bye,
-Andreas-
gcc/ChangeLog:
Hi,
Sorry to bother people again. Is this OK to go now?
Thanks!
David.
On Mon, 29 Jun 2015, David Sherwood wrote:
Hi,
I have added new STRICT_MAX_EXPR and STRICT_MIN_EXPR expressions to
support the
IEEE versions of fmin and fmax. This is done by recognising the math
Le 29/07/2015 20:35, Mikael Morin a écrit :
Hello,
I'm unburrying the patch from the thread starting at:
https://gcc.gnu.org/ml/gcc-patches/2014-03/msg00439.html
I provide the patch in two flavors read-only (without whitespace
changes) and write-only (with them).
This has been tested on
Hello,
this avoids an error found with bootstrap-ubsan.
Regression tested on x86_64-unknown-linux-gnu. OK for trunk?
Mikael
2015-08-05 Mikael Morin mik...@gcc.gnu.org
* hwint.h (sext_hwi): Rewrite without undefined behaviour on
negative SRC.
diff --git a/gcc/hwint.h
Committed to trunk as r226697.
Backported to gcc-5-branch as r226698.
gcc/testsuite/ChangeLog:
* jit.dg/test-benchmark.c (main): Record all elapsed times at each
optimization level, and print a summary at the end.
---
gcc/testsuite/jit.dg/test-benchmark.c | 18 +++---
On Thu, Aug 6, 2015 at 11:26 AM, FX fxcoud...@gmail.com wrote:
Can you please also introduce tests for ieee exceptions for long
double and __float128 types (as is done for real and double in
gfortran.dg/ieee/ieee_1.F90) as well as test for supported rounding
modes as done for real and double
On Thu, Aug 06, 2015 at 08:36:36PM +0100, Richard Sandiford wrote:
David Malcolm dmalc...@redhat.com writes:
On Wed, 2015-08-05 at 16:22 -0400, Trevor Saunders wrote:
On Wed, Aug 05, 2015 at 11:34:28AM -0400, David Malcolm wrote:
On Wed, 2015-08-05 at 11:28 -0400, David Malcolm wrote:
I have applied and committed these patches, both in gcc and binutils-gdb.
Since the 3.0.3 release of gperf (made in May 2007), the generated func
has had the gnu_inline attribute applied to it. The gcc source however
has not been updated to include that which has lead to a mismatch.
In practice, this hasn't been an issue for two reasons:
(1) Before gcc-5, the default
I've been banging on the concepts branch for the past month after Andrew
told me it was about ready to merge, fixing bugs and streamlining things
there to get familiar with the code while I could still look at it as a
whole rather than mixed in with the rest of the compiler. But I think
I've
On 08/04/2015 04:50 AM, Nathan Sidwell wrote:
+/* Worker reduction address expander. */
+static rtx
+nvptx_expand_work_red_addr (tree exp, rtx target,
+ machine_mode ARG_UNUSED (mode),
+ int ignore)
{
- return nvptx_expand_lock_unlock
HI All,
Here is updated patch which implements Richard proposal to use vector
comparison with boolean result instead of target hook. Support for it
was added to ix86_expand_branch.
Any comments will be appreciated.
Bootstrap and regression testing did not show any new failures.
ChangeLog:
Le 06/08/2015 12:53, Andre Vehreschild a écrit :
Dear all,
the attached patch fixes a regression introduced by my patches for the
F2008-style allocate(). In this case a function returning an array of BT_CLASS
objects can not be conv'ed using conv_expr_descriptor, but needs to be
On Thu, Aug 6, 2015 at 9:27 AM, Uros Bizjak ubiz...@gmail.com wrote:
Is it ok to backport the patch to gcc-5-branch?
A minor attribute fix is needed, please update type attribute of
*vec_concatv2df for added alternatives, also for mainline.
Fixed in mainline with the following patch:
Hi Richard,
-Original Message-
From: Richard Biener [mailto:richard.guent...@gmail.com]
Sent: Wednesday, August 05, 2015 5:11 PM
To: Kumar, Venkataramanan
Cc: Jeff Law; Jakub Jelinek; gcc-patches@gcc.gnu.org
Subject: Re: [RFC] [Patch]: Try and vectorize with shift for mult expr
On Thu, 6 Aug 2015 14:00:56 +0200
Mikael Morin mikael.mo...@sfr.fr wrote:
Le 06/08/2015 12:53, Andre Vehreschild a écrit :
Dear all,
the attached patch fixes a regression introduced by my patches for the
F2008-style allocate(). In this case a function returning an array of
BT_CLASS
Hi,
the attached patch is an attempt to fix the problem reported here:
https://gcc.gnu.org/ml/gcc/2015-05/msg00305.html
The problem is that we have address style operands as shift count
operands. In order to benefit from reloading e.g. for invalid base
registers (r0) we mark the respective
On Tue, Jun 02, 2015 at 05:32:02PM +0200, Andreas Krebbel wrote:
...
+(define_insn *shiftmode3_imm
+ [(set (match_operand:GPR 0 register_operand =d)
+(SHIFT:GPR (match_operand:GPR 1 register_operand d0)
+ (match_operand 2 immediate_operand J)))]
+
+
On Wed, 2015-08-05 at 16:22 -0400, Trevor Saunders wrote:
On Wed, Aug 05, 2015 at 11:34:28AM -0400, David Malcolm wrote:
On Wed, 2015-08-05 at 11:28 -0400, David Malcolm wrote:
On Wed, 2015-08-05 at 13:47 +0200, Richard Biener wrote:
On Wed, Aug 5, 2015 at 12:57 PM, Trevor Saunders
Hi!
On Wed, 5 Aug 2015 15:10:40 +0100, David Sherwood david.sherw...@arm.com
wrote:
In lto_input_mode_table there is the following line of code:
machine_mode inner = (machine_mode) table[bp_unpack_value (bp, 8)];
Is this right? In lto_write_mode_table this inner mode is written out
In next few weeks I'm gonna sending patches for refactoring regex.
The goal is to make the gigantic _Executor class into several smaller
ones and hopefully more readable. After that, there are several
correctness/performance issues to be fixed. We may also need more
documentation, but
I've committed this to fix the spinlock problem Cesar fell over. While there I
added more checking on the worker dimension.
nathan
2015-08-06 Nathan Sidwell nat...@codesourcery.com
* config/nvptx/nvptx.c (nvptx_expand_lock_unlock): Create label.
(nvptx_validate_dims): Check worker
Applied the idea with python script alternative. Review, please.
2015-07-28 Maxim Blumenthal maxim.blument...@intel.com
* configure.ac: Add a check for xxd or python presence when the target
is intelmic or intelmicemul.
* configure: Regenerate.
*
The attached two patches are an updated version of this patch
sent in June:
https://gcc.gnu.org/ml/gcc-patches/2015-06/msg00123.html
[PATCH 02/16] gcc: Embed the driver in-process within libgccjit
(an updated version of [patch 01/16] from that kit is in trunk as of
r226530;
These were introduced by my patch yesterday that keeps TEMPLATE_ID_EXPRs
around longer for variable templates, so more places need to deal with them.
Tested x86_64-pc-linux-gnu, applying to trunk and 5.
commit 8814d2ca8d702ddca8c3e315f08275dbe3271cd6
Author: Jason Merrill ja...@redhat.com
Date:
This patch implements enough state cleanup with the driver to allow
it to be linked within libgccjit.so and repeatedly run in-process.
The state cleanup is optional and is only performed by libgccjit.
When run within the driver executables, the code does the same as
before.
This corresponds to
This patch requires the previous one, and would be committed with it;
I've split them up for ease of review.
This is an updated version of the second part of:
https://gcc.gnu.org/ml/gcc-patches/2015-06/msg00123.html
with various cleanups; in particular, providing a boolean
option to support
The attached patch fixes the only remaining IEEE bug (to my knowledge) in
gfortran.
A Fortran 2008 interpretation request was submitted two years ago about which
of the IEEE functions were supposed to be used in constant and specification
expressions. The interp, voted on by J3 in Nov 2014,
James Greenhalgh writes:
On Tue, Jul 28, 2015 at 02:12:36PM +0100, Jiong Wang wrote:
The instruction sequences for preparing argument for TLS descriptor
runtime resolver and the later function call to resolver can actually be
hoisted out of the loop.
Currently we can't because we have
James Greenhalgh writes:
On Thu, Jul 16, 2015 at 11:21:25AM +0100, Jiong Wang wrote:
Jeff Law writes:
On 06/23/2015 02:29 AM, Ramana Radhakrishnan wrote:
If you try disabling the REG_EQUAL note generation [*], you'll probably
find a
performance regression on arm32 (and probably
James Greenhalgh writes:
On Tue, Jul 21, 2015 at 01:42:35PM +0100, Jiong Wang wrote:
Jiong Wang writes:
Alexander Monakov writes:
Attachment is the patch which repair -fno-plt support for AArch64.
aarch64_is_noplt_call_p will only be true if:
* gcc is generating position
Alan Lawrence wrote:
James Greenhalgh wrote:
Hi Alan,
The arm_neon.h portion of this patch does not apply after Charles' recent
changes. Could you please rebase and resubmit the patch for review?
Thanks,
James
Ah, indeed, thanks. Here's a rebased version, using Charles' new versions of
On Wed, Aug 5, 2015 at 10:07 AM, Kirill Yukhin kirill.yuk...@gmail.com wrote:
Hello,
Is it ok to backport the patch to gcc-5-branch?
A minor attribute fix is needed, please update type attribute of
*vec_concatv2df for added alternatives, also for mainline.
Otherwise, the patch looks safe, so
I realized other targets might need to do different things with the openacc fork
and join buitins. I created a new target hook and default implementation. The
default deletes the internal functions if there is no RTL expander for them.
nathan
2015-08-06 Nathan Sidwell
On 08/06/2015 01:41 AM, Nathan Sidwell wrote:
I've committed this to fix the spinlock problem Cesar fell over. While
there I added more checking on the worker dimension.
I hit a couple of more bugs with the spinlocks. First, the address space
argument to membar wasn't being handled properly.
55 matches
Mail list logo
