On Thu, Feb 19, 2015 at 05:52:26AM -0800, H.J. Lu wrote:
> On Wed, Feb 18, 2015 at 11:17 PM, Mark Wielaard wrote:
> >> >> Now it becomes a monthly topic:
> >> >>
> >> >> https://sourceware.org/ml/binutils/2015-01/msg00089.html
> >> >
> >> > Thanks, I hadn't seen that before. Alan Modra makes some
As discussed last year, we've converted the GNAT main documentation
(gnat_rm.texi and gnat_ugn.texi) to reST/sphinx, so the master doc
can now be found under gcc/ada/doc.
We're keeping automatically generated .texi files for now under gcc/ada
so that people only having texinfo tools can still gene
Hi,
this patch reverses the abort logic in pr30957-1.c, such that it aborts on
failure rather than on success.
OK for stage4?
Thanks,
- Tom
2015-02-20 Tom de Vries
* gcc.dg/pr30957-1.c (main): Abort on failure.
---
gcc/testsuite/gcc.dg/pr30957-1.c | 6 +++---
1 file changed, 3 i
Hi Bernd!
On Thu, 19 Feb 2015 10:28:46 +0100, Bernd Schmidt
wrote:
> issue when trying to
> get at the libgcc for the nvptx accel compiler after it's been
> installed. The libgcc Makefile puts it in the wrong place -
> gcc/nvptx-none/accel/nvptx-none instead of gcc/host/accel/nvptx-none.
> T
Hi!
On Thu, 19 Feb 2015 14:09:29 +0100, Jakub Jelinek wrote:
> On Tue, Feb 17, 2015 at 09:55:32PM +0100, Bernd Schmidt wrote:
> > On 02/17/2015 06:10 PM, Jakub Jelinek wrote:
> > >
> > >What exact testcase are you trying to fix with this patch, and how do you
> > >think offloading of code using v
On Fri, Feb 20, 2015 at 10:33:38AM +0100, Thomas Schwinge wrote:
> On Thu, 19 Feb 2015 14:09:29 +0100, Jakub Jelinek wrote:
> > On Tue, Feb 17, 2015 at 09:55:32PM +0100, Bernd Schmidt wrote:
> > > On 02/17/2015 06:10 PM, Jakub Jelinek wrote:
> > > >
> > > >What exact testcase are you trying to fix
Hi!
On Thu, 19 Feb 2015 11:51:02 +0100, Jakub Jelinek wrote:
> On Thu, Feb 19, 2015 at 11:48:17AM +0100, Thomas Schwinge wrote:
> > Like this?
>
> Yes.
>
> > commit 56c0312469f583ba3fa9fa2777981742ab6d6c75
> > Author: Thomas Schwinge
> > Date: Thu Feb 19 11:41:23 2015 +0100
> >
> > If w
On Fri, Feb 20, 2015 at 10:25:54AM +0100, Tom de Vries wrote:
> this patch reverses the abort logic in pr30957-1.c, such that it aborts on
> failure rather than on success.
That sounds really weird. From the description it looks like it is a known bug
that we don't return -0.0.
If 0.0 is the righ
On Wed, Nov 19, 2014 at 08:52:40PM +0100, Bernd Schmidt wrote:
> Another change that's required is (something like) the following. For ptx,
> we need to know whether to output something as a .func (callable from ptx
> code) or a .kernel (callable from the host). That means we need to mark the
> ker
Prior to this change, -fno-inline was stopping all the inlining
unconditionally, even when requested with a pragma Inline_Always.
This was inconsistent with the behavior of the "always_inline" gcc attribute,
and this change fixes this.
It also removes an internal shortcircuit which was preventing
When the compiler is invoked with several --RTS= switches, with different
values designated the same runtime directory, the compiler reports an
error that RTS cannot be specified several times. This patch fixes
this problem.
The test for this is to invoke the compiler with two switches --RTS=, one
This fixes an assertion failure in gigi triggered by an order-of-elaboration
issue for the type of a parameter of a subprogram which is both the renaming
of another subprogram and marked Inline. In this case the compiler was using
expansion (aka front-end inlining) but this doesn't always play nic
Hi!
On Fri, 20 Feb 2015 10:47:13 +0100, Jakub Jelinek wrote:
> On Wed, Nov 19, 2014 at 08:52:40PM +0100, Bernd Schmidt wrote:
> > Another change that's required is (something like) the following. For ptx,
> > we need to know whether to output something as a .func (callable from ptx
> > code) or a
If the type of the operand of a type conversion is defined as an
access to a class-wide interface type, and the target interface
type is defined in a package visible at the point of declaration
of the access type through a limited-with clause, then the compiler
may silently skip generating code for
If a statement appears where a declaration is expected, then an
appropriate error message is given, but the tree is malformed
leading to a later crash if -gnatQ is used.
The following should compile with -gnatQ without giving a
"compilation abandoned" message:
1. procedure AssignInD is
Make Non_Preemptive_FIFO_Within_Priorities a standard dispatching policy
name as defined in RM D.2.4(2/2).
Create the language-defined library package Ada.Dispatching.Non_Preemptive,
as defined in RM D.2.4(2.2/3). This package is marked as unimplemented
because no target environment supports it.
Within the compiler class-wide aspects use a leading-underscore internal
identifier. If the placement of the aspect is illegal, the error message
must mention the original form of the aspect.
Compiling class_a.ads must yield:
class_a.ads:5:10:
aspect "Type_Invariant_Class" only allowe
On Thu, 19 Feb 2015, Uros Bizjak wrote:
> Hello!
>
> I would like to port following patches to 4.8 branch:
Ok.
Thanks,
Richard.
> 2014-11-05 Uros Bizjak
>
> Backport from mainline
> 2015-01-22 Wei Mi
>
> PR rtl-optimization/64557
> * dse.c (record_store): Call get_addr f
On Thu, 19 Feb 2015, Tom de Vries wrote:
> On 19-02-15 14:07, Richard Biener wrote:
> > On Thu, 19 Feb 2015, Jakub Jelinek wrote:
> >
> > > On Thu, Feb 19, 2015 at 01:44:46PM +0100, Richard Biener wrote:
> > > > I'd call it a bug though, and we do have internal fns in
> > > > generic already thus
Hi all!
Here is simple patch that moves asan phase just behind sanopt for all
O0/O1/O2/O3 modes which gives +7% on x86 SPEC2006 (ref dataset). Regression
testing and sanitized GCC bootstrapping were successfully done.
Performance data attached.
--Marat
PATCH REF
On Fri, Feb 20, 2015 at 02:59:51PM +0300, Marat Zakirov wrote:
> Here is simple patch that moves asan phase just behind sanopt for all
> O0/O1/O2/O3 modes which gives +7% on x86 SPEC2006 (ref dataset).
> Regression testing and sanitized GCC bootstrapping were successfully done.
This isn't a regre
On 20-02-15 10:42, Jakub Jelinek wrote:
On Fri, Feb 20, 2015 at 10:25:54AM +0100, Tom de Vries wrote:
this patch reverses the abort logic in pr30957-1.c, such that it aborts on
failure rather than on success.
That sounds really weird. From the description it looks like it is a known bug
that
On 02/12/2015 05:57 PM, Jan Hubicka wrote:
From: mliska V
Date: Fri, 23 Jan 2015 14:58:36 +0100
Subject: [PATCH] Fix PR ipa/64693.
gcc/ChangeLog:
2015-02-12 Martin Liska
PR ipa/64693
* ipa-icf.c (sem_item_optimizer::execute): Call
identify_address_sensitive_classes.
On 02/10/2015 11:34 AM, Richard Biener wrote:
Bernds patch is ok with the unrelated gcc.c hunk removed.
Committed in the following form (adapted to some recent changes) after
bootstrap and retest.
Bernd
Index: gcc/ChangeLog
On 02/12/2015 05:57 PM, Jan Hubicka wrote:
From: mliska V
Date: Fri, 23 Jan 2015 14:58:36 +0100
Subject: [PATCH] Fix PR ipa/64693.
gcc/ChangeLog:
2015-02-12 Martin Liska
PR ipa/64693
* ipa-icf.c (sem_item_optimizer::execute): Call
identify_address_sensitive_classes.
Hi!
On Wed, 18 Feb 2015 17:17:08 +0100, Jan Hubicka wrote:
> looking across the ODR violation messages in libreoffice and Chromium I found
> some false positives and some confused messages. This patch fixes them. In
> partiuclar
> - I introduced nasty vtable corruption when breaking out my typ
On Fri, Feb 20, 2015 at 01:36:15PM +0100, Tom de Vries wrote:
> On 20-02-15 10:42, Jakub Jelinek wrote:
> >On Fri, Feb 20, 2015 at 10:25:54AM +0100, Tom de Vries wrote:
> >>this patch reverses the abort logic in pr30957-1.c, such that it aborts on
> >>failure rather than on success.
> >
> >That sou
On 18/02/15 10:35 +0100, François Dumont wrote:
Hello
I am still studying hashtable performances and especially how to
reduce overhead compared to tr1 implementation. Most of the overhead
is coming from the additional modulo operations required with the new
data model. Having a closer look
On Fri, Feb 20, 2015 at 02:15:24PM +0100, Thomas Schwinge wrote:
> > * ipa-devirt.c (odr_subtypes_equivalent_p): Fix formating.
> > (compare_virtual_tables): Be smarter about skipping typeinfos;
> > do sane output on virtual table table mismatch.
> > (warn_odr): Be ready for forward
> case SYMBOL_SMALL_GOTTPREL:
>{
> - rtx tmp_reg = gen_reg_rtx (Pmode);
> + /* In ILP32, the mode of dest can be either SImode or DImode,
> + while the got entry is always of SImode size. The mode of
> + dest depends on how dest is used: if dest is assign
On 18 February 2015 at 11:35, Kyrill Tkachov wrote:
> Ok for trunk and 4.9?
>
> Thanks,
> Kyrill
>
> 2015-02-17 Kyrylo Tkachov
>
> * config/aarch64/aarch64.md (*aarch64_lshr_sisd_or_int_3):
> Mark operand 0 as earlyclobber in 2nd alternative.
> (1st define_split below *aarch64_lshr
On 11 February 2015 at 10:57, Alan Lawrence wrote:
> gcc/ChangeLog:
>
> PR target/64997
> * config/aarch64/aarch64.md (*xor_one_cmpl3): Use FP_REGNUM_P
> as split condition; force split via '#' in output pattern.
OK /Marcus
Hi,
On Fri, Feb 20, 2015 at 06:16:47AM +0100, Jan Hubicka wrote:
>
> ...
>
> This is followup patch that also fixes ;; issue.
> It makes the alignment propagation to do proper lattice operations instead
> of requiring perfect match and dropping everything to bottom otherwise.
> Martin, does it lo
In some error messages, the aspect name Type_Invariant'Class
appeared as Type_Invariant_Class, this is now fixed. The
following is compiled with -gnatl -gnatj60:
1. package Class_Aspect is
2.type A_T is tagged private;
3.procedure P (Arg : Integer) with
4. Pre'Clas
an indirect call through an access-to-protected-subprogram value is always
considered an external call. This patch makes that happen (previously, the
compiler was using internal calls in some cases).
The following test should run to completion, silently. (Before the bug fix, it
would hang, because
On Thu, Feb 19, 2015 at 12:37:10 +0100, Thomas Schwinge wrote:
> On Fri, 26 Sep 2014 16:36:21 +0400, Ilya Verbin wrote:
> > gcc/
> > * Makefile.in
>
> > (install-cpp, install-common, install_driver, install-gcc-ar): Do not
> > install for the offload compiler.
>
> > gcc/cp/
> > *
This updates the Python printers to work with the new std::tuple
layout as well as the old one (I changed std::tuple for GCC 5 to
remove the empty base class that every tuple derived from).
Tested x86_64-linux, committed to trunk.
commit 5b522ed478febde4220e0e75bb84a0189bea5941
Author: Jonathan W
Hi!
On Fri, 20 Feb 2015 17:36:12 +0300, Ilya Verbin wrote:
> On Thu, Feb 19, 2015 at 12:37:10 +0100, Thomas Schwinge wrote:
> > On Fri, 26 Sep 2014 16:36:21 +0400, Ilya Verbin wrote:
> > > gcc/
> > > * Makefile.in
> >
> > > (install-cpp, install-common, install_driver, install-gcc-ar): Do n
Hi!
As written in the PR, this testcase is supposed to be blacklisted on
(lp64) powerpc*-*-linux* unless powerpc_elfv2 and also on nvptx,
but apparently neither of those blacklistings really work.
tcl or dejagnu or the target expression parser are unhappy about the
missing {}s, it doesn't like con
Hi!
As mentioned in PR63892, with the addition of ICF the sibcall-{3,4}.c tests
no longer test what they meant to test. This patch duplicates the tests
and adds a volatile var bump in just one of the functions so that ICF can't
be performed and the tests really test sibcalls rather than say tail
Hi!
As written in the PR, the sibcall-3.c testcase fails on Darwin, because
!sem_item::target_supports_symbol_aliases_p () and we don't really try
redirect_callers, even when that is the best way to perform ICF (both
original and alias are local).
Bootstrapped/regtested on x86_64-linux and i686-l
On Fri, Feb 20, 2015 at 03:41:40PM +0100, Thomas Schwinge wrote:
> Well, but users (like Jakub, for example) ;-) may decide to build the
> offloading compilers without specifying --enable-languages, and that'll
> then default to include Java, and you'll end up with:
Yeah. We can perhaps tweak wha
On Thu, Feb 19, 2015 at 13:17:37 +0100, Thomas Schwinge wrote:
> My asumption is that it is always safe to install non-native (that is
> cross) GCC installations into the same prefix. (Which would resolve this
> problem of clashing file names for target and offloading compilers for
> good.)
>
> S
On Fri, Feb 20, 2015 at 01:03:26AM +0100, Jakub Jelinek wrote:
> Hi!
>
> As reported, !!x == y is quite common in the Linux kernel
> and unlike the !x == y case it usually doesn't mean mistyped
> !(x == y) or x != y. clang++ apparently doesn't warn about that
> either, and it doesn't warn even ab
On Fri, Feb 20, 2015 at 04:35:05PM +0100, Marek Polacek wrote:
> Note that first version of -Wlogical-not-parentheses didn't warn
> when LHS had a boolean type, this has been changed later on. I have
> no strong preference either way.
>
> > As the argument is already folded, it isn't easy to dete
On 02/20/15 07:45, Jakub Jelinek wrote:
Hi!
As mentioned in PR63892, with the addition of ICF the sibcall-{3,4}.c tests
no longer test what they meant to test. This patch duplicates the tests
and adds a volatile var bump in just one of the functions so that ICF can't
be performed and the tests
On 02/20/15 07:43, Jakub Jelinek wrote:
Hi!
As written in the PR, this testcase is supposed to be blacklisted on
(lp64) powerpc*-*-linux* unless powerpc_elfv2 and also on nvptx,
but apparently neither of those blacklistings really work.
tcl or dejagnu or the target expression parser are unhappy
Hi,
the following patch, which I am about to commit to the hsa branch, is
a reimplementation of how we translate gimple memory references to
HSAIL. Unlike what we have had so far, this implementation should
handle all non-bit fields loads and stores well, any bug in that
regard is an unexpected b
On 02/20/15 07:48, Jakub Jelinek wrote:
Hi!
As written in the PR, the sibcall-3.c testcase fails on Darwin, because
!sem_item::target_supports_symbol_aliases_p () and we don't really try
redirect_callers, even when that is the best way to perform ICF (both
original and alias are local).
Bootstr
VRP was constant-propagating to a loop latch edge PHI arg from
sth derived from an assert-expr. That's usually not a good idea
because it prevents coalescing (or the fixup code has to insert
a copy from a constant). If we derive the constant from an
assert then we don't even save computation of
gcc/ChangeLog:
* config/rs6000/htm.md (tcheck): Fix assembly encoding.
gcc/testsuite/ChangeLog
* gcc.target/powerpc/htm-builtin-1.c: Fix tcheck expect value.
---
diff --git a/gcc/config/rs6000/htm.md b/gcc/config/rs6000/htm.md
index 2c4689f..79fb740 100644
--- a/gcc/config/rs60
On Feb 20, 2015, at 1:42 AM, Jakub Jelinek wrote:
> On Fri, Feb 20, 2015 at 10:25:54AM +0100, Tom de Vries wrote:
>> this patch reverses the abort logic in pr30957-1.c, such that it aborts on
>> failure rather than on success.
>
> That sounds really weird. From the description it looks like it i
> > +/* Decrease alignment info DEST to be at most CUR. */
> > +
> > +static bool
> > +decrease_alignment (ipa_alignment *dest, ipa_alignment cur)
> > +{
> > + bool changed = false;
> > +
> > + if (!cur.known)
> > +return false;
>
> I really think this should be return set_alignment_to_bott
On Fri, Feb 20, 2015 at 5:35 AM, Marcus Shawcroft
wrote:
>> case SYMBOL_SMALL_GOTTPREL:
>>{
>> - rtx tmp_reg = gen_reg_rtx (Pmode);
>> + /* In ILP32, the mode of dest can be either SImode or DImode,
>> + while the got entry is always of SImode size. The mode of
>
> Hello.
>
> There's updated version that reflects how should we handle congruence classes
> that have any
> address reference. Patch can bootstrap x86_64-linux-pc and no new regression
> is introduced?
>
> Ready for trunk?
> Thanks,
> Martin
>
>
> >From d7472e55b345214d55ed49f5f10deafa9a24a
>
> Hi.
>
> This is second part which introduces better variable handling. Since readonly
> variable flag
> identification can identify new candidates, ICF should filter out
> non-readonly variables in
> execute phase.
>
> Ready for trunk?
> Thanks,
> Martin
> >From a18a4840d14b1c0d35a9e4387d
> On Fri, Feb 20, 2015 at 02:15:24PM +0100, Thomas Schwinge wrote:
> > > * ipa-devirt.c (odr_subtypes_equivalent_p): Fix formating.
> > > (compare_virtual_tables): Be smarter about skipping typeinfos;
> > > do sane output on virtual table table mismatch.
> > > (warn_odr): Be ready for forwa
On 02/19/15 10:27, Andreas Krebbel wrote:
Hi,
the attached patch makes the hoist-register-pressure testcases to work
with -m31 on 64 bit.
Ok to apply?
Bye,
-Andreas-
2015-02-19 Andreas Krebbel
* gcc.dg/hoist-register-pressure-1.c: Make S/390 target check work
with -m31 o
This documents most of the implementation-defined behaviour in the
C++11 standard, although some pieces (such as time facets and random
number distributions) need completing.
Committed to trunk.
commit 329e9a788a3c2854d8f91ecd49271826458b87af
Author: Jonathan Wakely
Date: Fri Feb 20 18:58:25 2
On 02/19/15 10:31, Aldy Hernandez wrote:
[Ughh...I'm apparently incapable of CCing gcc-patches on the first try.
My apologies again.]
Please see the long explanation in the PR:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58123
But the short explanation for this patch is that, at the very le
The following patch fixes PR64172. The details of the problem can be
found on
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64172
The patch was bootstrapped and tested on x86-64. I also checked
SPEC2000 for the patch. There is only one code change on sixtrack.
And the patch improves s
On Feb 20, 2015, at 4:36 AM, Tom de Vries wrote:
> On 20-02-15 10:42, Jakub Jelinek wrote:
>> On Fri, Feb 20, 2015 at 10:25:54AM +0100, Tom de Vries wrote:
>>> this patch reverses the abort logic in pr30957-1.c, such that it aborts on
>>> failure rather than on success.
>>
>> That sounds really w
On Feb 20, 2015, at 10:12 AM, Mike Stump wrote:
> On Feb 20, 2015, at 1:42 AM, Jakub Jelinek wrote:
>> On Fri, Feb 20, 2015 at 10:25:54AM +0100, Tom de Vries wrote:
>>> this patch reverses the abort logic in pr30957-1.c, such that it aborts on
>>> failure rather than on success.
>>
>> That sound
On Fri, Feb 20, 2015 at 10:27:26 +0100, Thomas Schwinge wrote:
> On Thu, 19 Feb 2015 10:28:46 +0100, Bernd Schmidt
> wrote:
> > issue when trying to
> > get at the libgcc for the nvptx accel compiler after it's been
> > installed. The libgcc Makefile puts it in the wrong place -
> > gcc/nvptx-
On Feb 20, 2015, at 10:53 AM, Jeff Law wrote:
> On 02/19/15 10:27, Andreas Krebbel wrote:
>>
>> the attached patch makes the hoist-register-pressure testcases to work
>> with -m31 on 64 bit.
>>
>> Ok to apply?
>>
>> Bye,
>>
>> -Andreas-
>>
>>
>> 2015-02-19 Andreas Krebbel
>>
>> * gc
On Feb 20, 2015, at 6:36 AM, Ilya Verbin wrote:
> I assumed that nobody would build an offloading compiler with
> --enable-languages
> other than c,c++,fortran[,lto].
:-) You should try objc and obj-c++… With some luck, they might just work out
of the box.
Hi!
On Fri, 20 Feb 2015 11:35:18 -0800, Mike Stump wrote:
> On Feb 20, 2015, at 6:36 AM, Ilya Verbin wrote:
> > I assumed that nobody would build an offloading compiler with
> > --enable-languages
> > other than c,c++,fortran[,lto].
>
> :-) You should try objc and obj-c++… With some luck, th
On Fri, Feb 20, 2015 at 18:05:01 +0300, Ilya Verbin wrote:
> On Thu, Feb 19, 2015 at 13:17:37 +0100, Thomas Schwinge wrote:
> > My asumption is that it is always safe to install non-native (that is
> > cross) GCC installations into the same prefix. (Which would resolve this
> > problem of clashing
Hello all,
As noted in the bug, and a couple of times on StackOverflow, generic
functions and 'auto' return types do not play nice together. This
happens because the parser does not expect a level of template
parameters to be added to a function after it has already parsed 'auto'
in a return
On Fri, 2015-02-20 at 15:44 -0200, Adhemerval Zanella wrote:
> gcc/ChangeLog:
>
> * config/rs6000/htm.md (tcheck): Fix assembly encoding.
>
> gcc/testsuite/ChangeLog
>
> * gcc.target/powerpc/htm-builtin-1.c: Fix tcheck expect value.
You are correct that the tcheck insn does not have
On Fri, Feb 20, 2015 at 12:44 PM, Adhemerval Zanella
wrote:
> gcc/ChangeLog:
>
> * config/rs6000/htm.md (tcheck): Fix assembly encoding.
>
> gcc/testsuite/ChangeLog
>
> * gcc.target/powerpc/htm-builtin-1.c: Fix tcheck expect value.
Okay.
Thanks, David
Hello, gentle maintainer.
This is a message from the Translation Project robot.
A revised PO file for textual domain 'cpplib' has been submitted
by the Swedish team of translators. The file is available at:
http://translationproject.org/latest/cpplib/sv.po
(This file, 'cpplib-5.1-b20150208
cpplib-5.1-b20150208.sv.po.gz
Description: Binary data
The Translation Project robot, in the
name of your translation coordinator.
Le 16/02/2015 21:18, Bernd Edlinger a écrit :
>
> again, with attachments,
> sorry.
>
>
>>
>> Hi,
>>
>>
>> this patch fixes PR64980 and PR61960 at the same time.
>>
>> The unreduced test case for PR64230 is also included, because a previous
>> version
>> of this patch caused this test to fail b
74 matches
Mail list logo
