On 03/27/2017 04:27 PM, Richard Biener wrote:
> On Mon, Mar 27, 2017 at 4:14 PM, Richard Biener
> wrote:
>> On Mon, Mar 27, 2017 at 2:47 PM, Martin Liška wrote:
>>> Hello.
>>>
>>> As described in the PR, we can create a PHI node in einline that has no
>>> argument.
>>> That can cause ICE in devi
My recent refactoring of enable_shared_from_this broke support for an
obscure use case, where shared_ptr "owns" a pointer to function (and
the deleter does something with it, like unregister a callback from a
plugin).
The problem is simply that I made the new
_M_enable_shared_from_this_with overl
On Mon, Mar 27, 2017 at 7:30 PM, Marek Polacek wrote:
> The code in fold_comparison calls save_expr on an expression and then tries to
> set a location of the expression. But since save_expr calls fold, it can
> produce an integer constant, so we must be more careful when setting its
> location.
On Mon, Mar 27, 2017 at 11:48 PM, Martin Jambor wrote:
> Hi,
>
> On Mon, Mar 27, 2017 at 12:15:10PM -0600, Jeff Law wrote:
>> On 03/27/2017 05:40 AM, Martin Liška wrote:
>> > Hello.
>> >
>> > There's alternative approach suggested by Martin Jambor.
>> > Patch can bootstrap on ppc64le-redhat-linux
> PR ada/80117
> * system-linux-arm-ilp32.ads: New file.
> * gcc-interface/Makefile.in (LIBGNAT_TARGET_PAIRS_COMMON): Use it
> for aarch64-linux with -mabi=ilp32.
OK, thanks.
Arno
On Tue, Mar 28, 2017 at 9:45 AM, Martin Liška wrote:
> On 03/27/2017 04:27 PM, Richard Biener wrote:
>> On Mon, Mar 27, 2017 at 4:14 PM, Richard Biener
>> wrote:
>>> On Mon, Mar 27, 2017 at 2:47 PM, Martin Liška wrote:
Hello.
As described in the PR, we can create a PHI node in ein
Hi!
On Thu, 23 Mar 2017 16:46:19 +0100, Tom de Vries wrote:
> I've run the gcc testsuite for target nvptx-none and ran into "test for
> excess errors" FAILs due to:
> ...
> sorry, unimplemented: target cannot support alloca.
> ...
>
> This patch marks those testcases as requiring alloca.
I obs
On 28/03/17 10:23, Thomas Schwinge wrote:
Hi!
On Thu, 23 Mar 2017 16:46:19 +0100, Tom de Vries wrote:
I've run the gcc testsuite for target nvptx-none and ran into "test for
excess errors" FAILs due to:
...
sorry, unimplemented: target cannot support alloca.
...
This patch marks those testcas
Thank you Andrew. The patch is committed,
Claudiu
> -Original Message-
> From: Andrew Burgess [mailto:andrew.burg...@embecosm.com]
> Sent: Monday, March 27, 2017 9:58 PM
> To: Claudiu Zissulescu
> Cc: gcc-patches@gcc.gnu.org; francois.bed...@synopsys.com
> Subject: Re: [PATCH] [ARC] Updat
Thank you Andrew. The patch is committed,
Claudiu
P.S. Thomas can you mark fixed the Bugzilla entry for this patch. Thanks!
Hello.
The fix of the PR, where intermediate format is currently dumped in a bit
different manner.
I believe it can share majority of file creation (and destruction) with normal
format.
Apart from that I refined usage string from:
Usage: gcov [OPTION]... SOURCE|OBJ...
to:
Usage: gcov [OPTION..
Hi!
I noticed that in C++, for a directive like:
#pragma acc parallel default (hello)
... we currently print two error messages for the same location:
[...]:35:30: error: expected 'none' before 'hello'
[...]:35:30: error: expected ')' before 'hello'
I think it should be enough to j
Not actually solving those testcases but currently scatters/gathers
are costed as a single vector load/store. Micro-architectural Intel
seems to have gotten to the point (with skylake) to hide all but
the scalar loads/stores which seem to take up as many load/store
ports as expected. I'm not exp
On 17/03/17 16:51 +0200, Ville Voutilainen wrote:
diff --git a/libstdc++-v3/include/std/optional
b/libstdc++-v3/include/std/optional
index 3f540ec..e67ba89 100644
--- a/libstdc++-v3/include/std/optional
+++ b/libstdc++-v3/include/std/optional
@@ -95,125 +95,127 @@ _GLIBCXX_BEGIN_NAMESPACE_VERSIO
In the context of PR79671 it was noticed that forcing a ref-all
pointer access via a simple static_cast (obj) and C
with may_alias on it didn't work.
But it should.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2017-03-28 Richard Biener
PR middle-
Hi,
The below trivial patch fixes some more testsuite failures for the avr
target. They fail for avr because they assume ints are 32 bits or
wider. The patch uses explicit __{U}INT32_TYPE__ for targets with
smaller int size.
Committed as obvious.
Regards
Senthil
2017-03-28 Senthil Ku
Hi,
This patch fixes a latent bug in vectorizer, specifically, vect_do_peeling
doesn't mark newly created edge
with irreducible flag if the vectorized loop is in irreducible region.
Function checking_verify_loop_structure
needs to be explicitly called in vectorizer otherwise the issue is covered
Hi,
This patch is to fix PR80153. As analyzed in the PR, root cause is tree_affine
lacks
ability differentiating (unsigned)(ptr + offset) and (unsigned)ptr +
(unsigned)offset,
even worse, it always returns the former expression in aff_combination_tree,
which
is wrong if the original expression
Hi,
This patch is to fix a latent vectorizer bug uncovered by previous one.
Function optimize_mask_stores
creates if-then structure in CFG in order to revert if-conversion
transformation. It adds newly created
then_bb to outer loop when vectorizing two levels loop nest and masked_store is
in t
On Tue, Mar 28, 2017 at 2:01 PM, Bin Cheng wrote:
> Hi,
> This patch fixes a latent bug in vectorizer, specifically, vect_do_peeling
> doesn't mark newly created edge
> with irreducible flag if the vectorized loop is in irreducible region.
> Function checking_verify_loop_structure
> needs to be
On Tue, Mar 28, 2017 at 2:01 PM, Bin Cheng wrote:
> Hi,
> This patch is to fix a latent vectorizer bug uncovered by previous one.
> Function optimize_mask_stores
> creates if-then structure in CFG in order to revert if-conversion
> transformation. It adds newly created
> then_bb to outer loop
On Tue, Mar 28, 2017 at 2:01 PM, Bin Cheng wrote:
> Hi,
> This patch is to fix PR80153. As analyzed in the PR, root cause is
> tree_affine lacks
> ability differentiating (unsigned)(ptr + offset) and (unsigned)ptr +
> (unsigned)offset,
> even worse, it always returns the former expression in af
This documents the new _GLIBCXX_RELEASE macro in the main manual, not
just the appendix on ABI version history.
* doc/xml/manual/abi.xml: Add xml:id anchor.
* doc/xml/manual/using.xml (manual.intro.using.macros): Document
_GLIBCXX_RELEASE. Link to new anchor for __GLIBCXX_
tree-call-cdce.c was updating the edge probabilities and counts but
it wasn't updating the corresponding block information. Among other
things, this tricked the register allocator into thinking that the
libm call was relatively hot and that it wasn't worth assigning
call-clobbered registers to val
> From: Matthew Fortune
>
> Given the acceptance that MIPS PIC behaviour is semi-unique then checking
> MIPS
> complies with the rules around pre-processor behaviour doesn't add much
> value.
> I'm happy to skip these tests on the basis that software can't make the same
> assumptions about MIPS an
On Mar 24, 2017, at 3:19 AM, Richard Biener wrote:
>
> On Thu, Mar 23, 2017 at 7:20 PM, Bill Schmidt
> wrote:
>> Hi,
>>
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80158 reports an ICE in
>> SLSR while building 416.gamess on x86_64. This is a latent (but
>> previously harmless) bug that wa
I am testing the following patch that avoids putting non-final copies
into the CCP lattice which eventually results in released SSA names in
the IL.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
Richard.
2017-03-28 Richard Biener
PR tree-optimization/78644
*
On 03/28/2017 02:09 AM, Richard Biener wrote:
On Mon, Mar 27, 2017 at 11:48 PM, Martin Jambor wrote:
Hi,
On Mon, Mar 27, 2017 at 12:15:10PM -0600, Jeff Law wrote:
On 03/27/2017 05:40 AM, Martin Liška wrote:
Hello.
There's alternative approach suggested by Martin Jambor.
Patch can bootstrap
On 03/28/2017 06:45 AM, Richard Sandiford wrote:
tree-call-cdce.c was updating the edge probabilities and counts but
it wasn't updating the corresponding block information. Among other
things, this tricked the register allocator into thinking that the
libm call was relatively hot and that it was
For gcc-6 I fixed a bug in generate_canonical that allowed it to
return a value outside the permitted range (due to rounding), but that
was done by looping until we got a value that was in range, which
violates the complexity requirements.
John Salmon suggested a better fix in bugzilla, which is
Hi,
The gcc.dg/ipa/iinline-attr.c test has been failing for MIPS when using the n32
and n64 ABIs.
These two ABIs imply a 64-bit target, which means that the align_loops,
align_jumps, and align_functions flags are set to 8 by default.
Because the MIPS backend doesn't use the TARGET_OVERRIDE_OPTIO
> That needs to use ptr_mode, not Pmode.
I don't think so, the whole computation is in Pmode. Could you try something
similar to what is done in the 'else' arm of the big surrounding conditional?
--
Eric Botcazou
Hello!
As argued in the PR [1], limiting -mpreferred-stack-boundary=3 only
for non-SSE 64bit targets represent artificial limitation, restricting
compiler functionality and user freedom.
Attached patch allows -mpreferred-stack-boundary=3 in all cases. The
compiler will align stack when needed, im
Committed as r246546:
https://gcc.gnu.org/viewcvs/gcc?view=revision&revision=246546
Cheers,
Janus
2017-03-27 23:03 GMT+02:00 Janus Weil :
> Forgot to mention: The patch was pre-approved by Jerry, but of course
> I'm happy to make corrections (if necessary).
>
> Cheers,
> Janus
>
>
> 2017-03-27
In GCC PR 80226 Than reports a Go frontend crash when compiling a
function that returns multiple zero-sized values. This patch that he
wrote fixes the problem. Bootstrapped and ran Go testsuite on
x86_64-pc-linux-gnu. Committed to mainline.
Ian
2017-03-28 Than McIntosh
PR go/80226
* go-gc
The following patch fixes
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=80193
The patch was successfully bootstrapped on x86-64.
Committed as rev. 246554.
Index: ChangeLog
===
--- ChangeLog (revision 246553)
+++ ChangeLog (working
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 Spanish team of translators. The file is available at:
http://translationproject.org/latest/gcc/es.po
(This file, 'gcc-7.1-b20170226.es.po',
On Tue, Mar 28, 2017 at 11:35:35AM +0200, Thomas Schwinge wrote:
> I noticed that in C++, for a directive like:
>
> #pragma acc parallel default (hello)
>
> ... we currently print two error messages for the same location:
>
> [...]:35:30: error: expected 'none' before 'hello'
> [...]
On Tue, Mar 28, 2017 at 08:27:54AM +0200, Tom de Vries wrote:
> this patch fixes testsuite PR80220 - "relative line numbers don't work when
> put between braces".
What is the advantage of putting the line numbers between braces?
Isn't it easier to just drop those?
Jakub
With PIC a @ha relocation isn't generated, so skip that test then.
Tested with {-m32,-m64}{,-fPIC,-fpic}; committing to trunk.
Segher
2017-03-28 Segher Boessenkool
gcc/testsuite/
PR testsuite/43496
* gcc.target/powerpc/gcse-1.c: Skip scan-assembler-times "@ha" if
gen
Hi all,
The attached patch resolves this problem by moving the code that invokes the
child I/O procedure into nml_read_obj where it belongs. This allows the normal
flow of code that parses the namelist decorations before attempting to read the
object data.
One new test case is provided. Tes
41 matches
Mail list logo
