On Tue, 15 Apr 2014, Paolo Carlini wrote:
Anyway, the real issue is indeed that implementing those bits requires a new
virtual function, and that would break the ABI.
What is the status of the ABI half-break plan (abi_tag and such),
necessary to get the remaining pieces of C++11?
--
Marc Gl
>> Also the old standard seems to require that ios_base::fixed |
>> ios_base::scientific (or any other combination) falls through to the
>> uppercase test; I was trying to use abi_tag for a solution as not only
>> would two versions of _S_format_float be necessary, but also num_get
>> due to the pr
This is a ping about a patch I had posted sometime back.
Previous post: http://gcc.gnu.org/ml/gcc-patches/2014-03/msg01332.html
It was reviewed by Cary, but he said that a C++ front end maintainer
needs to approve it:
http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00175.html
The patch is attached.
On Tue, Apr 15, 2014 at 6:12 PM, Richard Sandiford
wrote:
> cc:ing Jason, who's the C++ maintainer.
FWIW: I created http://gcc.gnu.org/bugzilla/show_bug.cgi?id=60850
>
> Daniel Gutson writes:
>> ping for maintainer.
>>
>> Could this be considered for 4.8.3 please?
>>
>> Thanks,
>>
>>Daniel
Rong, David, Dehao, Teresa
I would like to have some rought idea of what we could merge this stage1. There
is
certainly a lot of interesting stuff on the google branch including AutoFDO,
LIPO,
the multivalue profile counters that may be used by the new devirtualization
bits
and more. I also thin
Hi,
On 04/15/2014 11:19 PM, Rüdiger Sonderfeld wrote:
This patch adds support for std::time_get::get and std::time_get::do_get
[locale.time.get]. Currently do_get is not virtual because this caused
segfault errors for me.
Those should be isolated and a compiler bug report opened including a
mi
> + /* If HOST_WIDE_INT is big enough then represent the bound as
> + a constant value. Note that we need to make sure the type
> + is signed or unsigned. We cannot just add an unsigned
> + constant if the value itself is positive. Some DWARF
> + consume
* libstdc++-v3/include/std/iomanip (_Get_time): New struct.
(get_time): New manipulator [ext.manip].
(operator>>): New overloaded function.
* libstdc++-v3/testsuite/27_io/manipulators/extended/get_time/char/1.cc:
* libstdc++-v3/testsuite/27_io/manipulators/extended/get_time/char/2.cc:
* libstdc
Described in [ext.manip].
* libstdc++-v3/include/std/iomanip (_Put_time): New struct.
(put_time): New manipulator.
(operator<<): New overloaded function.
* libstdc++-v3/testsuite/27_io/manipulators/extended/put_time/char/1.cc:
* libstdc++-v3/testsuite/27_io/manipulators/extended/put_time/char/
This patch adds support for std::time_get::get and std::time_get::do_get
[locale.time.get]. Currently do_get is not virtual because this caused
segfault errors for me.
* libstdc++-v3/include/bits/locale_facets_nonio.h (time_get::get):
(time_get::do_get): New method (C++11).
* libstdc++-v3/incl
cc:ing Jason, who's the C++ maintainer.
Daniel Gutson writes:
> ping for maintainer.
>
> Could this be considered for 4.8.3 please?
>
> Thanks,
>
>Daniel.
>
>
> On Tue, Apr 1, 2014 at 2:46 PM, Daniel Gutson
> wrote:
>>
>> I just realized I posted the patch in the wrong list.
>>
>>
>> ---
Robert Suchanek writes:
>> Hmm, marking them fixed was supposed to be a temporary reload-only thing,
>> until the move to LRA. It should never be worse to spill to these GPRs
>> over spilling to the stack, if the value isn't live across a call.
>
> I would say this also affects IRA/LRA integratio
Hi,
this patch prevents ipa-devirt to devirtualize to functions that are
not exported and have no address takem from virtual table; those are
obviously not going to be targets of virtual calls.
Honza
* ipa-devirt.c (referenced_from_vtable_p): New predicate.
(maybe_record_node, lik
On Tue, 15 Apr 2014, Jason Merrill wrote:
It's just vectors, because they're an extension; the patch I checked in
covered the standard language.
Like this? (regtested on x86_64-linux-gnu)
2014-04-16 Marc Glisse
gcc/cp/
* decl.c (reshape_init_r): Handle a single element of vector t
I typically leave the mode off when the operand accepts a CONST_INT as
I've had problems with patterns matching CONST_INTs otherwise, as
CONST_INT rtx's do not have a mode (or have VOIDmode).
(yes, I know gcc is supposed to accomodate that, but like I said, I've
had problems...)
As Robert pointed out here:
http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00416.html
we're a bit too eager when folding stuff into an 'X' constraint.
The value at expand time is sensible, but after that asm_operand_ok
allows arbitrary rtx expressions, including any number of registers
as well
Hi,
This patch cleans up some warnings when building due to missing operand
modes.
"trampoline_init" in rl78.md still produces warnings but I'm not
entirely sure about how best to fix that insn and I didn't want to break
anything.
Regards,
Richard
2014-04-15 Richard Hulme
*
"Moore, Catherine" writes:
>> -Original Message-
>> From: Moore, Catherine
>> Sent: Tuesday, April 15, 2014 8:49 AM
>> To: Rozycki, Maciej; Richard Sandiford
>> Cc: Matthew Fortune; gcc-patches@gcc.gnu.org; Moore, Catherine
>> Subject: RE: [PATCH] [MIPS] Fix operands for microMIPS SW16, SH
On 15/04/14 21:51 +0200, Václav Zeman wrote:
On 04/15/2014 08:29 PM, Jonathan Wakely wrote:
A minor simplification that removes a longstanding TODO note.
Tested x86_64-linux, committed to trunk.
diff --git a/libstdc++-v3/include/bits/shared_ptr.h
b/libstdc++-v3/include/bits/shared_ptr.h
ind
> -Original Message-
> From: Moore, Catherine
> Sent: Tuesday, April 15, 2014 8:49 AM
> To: Rozycki, Maciej; Richard Sandiford
> Cc: Matthew Fortune; gcc-patches@gcc.gnu.org; Moore, Catherine
> Subject: RE: [PATCH] [MIPS] Fix operands for microMIPS SW16, SH16 and
> SB16
>
>
>
> > -
On Tue, Apr 15, 2014 at 09:59:10AM -0400, Jason Merrill wrote:
> On 04/14/2014 11:10 AM, Marek Polacek wrote:
> >+ else if (TREE_CODE (val) == IDENTIFIER_NODE)
> >+{
> >+ tree t = lookup_name (val);
> >+ if (t && TREE_CODE (t) == CONST_DECL)
> >+return DECL_INITIAL (t);
> >+}
Three little fixes. Tested x86_64-linux, committed to trunk.
commit 17beaac60c01164cc496da57a2a9ced7a487d17d
Author: Jonathan Wakely
Date: Tue Apr 15 19:41:42 2014 +0100
* testsuite/24_iterators/insert_iterator/requirements/container.cc: Do
not use uninitialized members in mem-
On 29/03/14 14:54 -0400, Ed Smith-Rowland wrote:
All,
In string_view I botched the noexcept specification of operations
like find and friends with CharT* arguments.
I'm a little surprised the inconsistency between string_view and
string_view.tcc didn't error. In fact, in one repo thats a li
G++ accepts these tests but it shouldn't, and clang doesn't, so this
makes them valid C++11.
Tested x86_64-linux, committed to trunk.
commit f1517e2ae280691724472cbd0f6b31fa98f313d0
Author: Jonathan Wakely
Date: Tue Apr 15 19:39:15 2014 +0100
PR c++/60786
* testsuite/20_util/
Let me ping this. There's no hurry, but it may have got lost with 4.9
approaching.
http://gcc.gnu.org/ml/gcc-patches/2014-03/msg01205.html
On Sun, 23 Mar 2014, Marc Glisse wrote:
On Mon, 3 Mar 2014, Richard Biener wrote:
That's a bit much of ad-hoc pattern-matching ... wouldn't be
p = mallo
Ping
http://gcc.gnu.org/ml/gcc-patches/2014-03/msg00051.html
On Sun, 2 Mar 2014, Marc Glisse wrote:
Hello,
inlining operator new (with LTO or otherwise), I noticed that it has a
complicated implementation, which makes it hard to use this inlined code for
optimizations. This patch does two t
The parameters names should all be reserved names.
Also remove some trailing whitespace at the end of lines.
Tested x86_64-linux, committed to trunk.
commit b6e5e08880da7d80c1f14c132bd4bd8eed301205
Author: Jonathan Wakely
Date: Tue Apr 15 19:01:36 2014 +0100
* include/std/atomic: Ugli
A minor simplification that removes a longstanding TODO note.
Tested x86_64-linux, committed to trunk.
commit 7769b63f43899b901bba08e5b2b3a6806e2195f2
Author: Jonathan Wakely
Date: Tue Apr 15 19:00:47 2014 +0100
* include/bits/shared_ptr.h (shared_ptr::shared_ptr(nullptr_t)): Use
Add const to functions that would change meaning in C++14.
Tested x86_64-linux, committed to trunk.
commit f30d35d5a7aa3ff2e0d0e4010aecaf9f5a5fb9ed
Author: Jonathan Wakely
Date: Tue Apr 15 18:55:45 2014 +0100
* include/bits/atomic_base.h (__atomic_base<_PTp*>::_M_type_size): Add
Hi,
back in January in
http://gcc.gnu.org/ml/gcc-patches/2014-01/msg00848.html Eric pointed
out a testcase where the problem was SRA not scalarizing an aggregate
because it was involved in a throwing statement. The reason is that
SRA is likely to need to append new statements after each one where
> ISTR some more ???/FIXMEs and/or special-casings we could remove with
> that. As followup, of course.
It would be better to remove them all at once, so if you have specifics...
> The single reason why we don't have virtual operands at -O0 is compile-time
> btw - SSA rewrite doesn't come for fr
Testcase weirdness?
for (i < 0; i < N; ++i)
{
arr[i] = i;
expect[i] = __builtin_bswap64 (i);
if (y) /* Avoid vectorisation. */
abort ();
}
i < 0 :)
duplicated in all 3 testcases btw.
-eric
On Tue, Apr 15, 2014 at 4:25 AM, Kyrill Tkachov wrote:
> Hi all,
>
> This
On Tue, Apr 15, 2014 at 6:06 PM, Evgeny Stupachenko wrote:
> I've separated the patch into 3.
> The patch passes x86 bootstrap.
>
> 1st part:
>
> 2014-04-15 Evgeny Stupachenko
>
>* config/i386/i386.c (slm_cost): Fixing vec_to_scalar_cost for
>Silvermont according latency table.
On Tue, Apr 15, 2014 at 5:21 PM, Rainer Orth
wrote:
> As reported in the PR, gcc/configure currently fails to detect native
> TLS support on x86_64-*-solaris2* with a 64-bit gas since it feeds it
> 32-bit TLS code. I haden't noticed this so far since I've been using a
> 32-bit gas here (no idea w
On 15 April 2014 16:18, Ramana Radhakrishnan
wrote:
> On 04/14/14 23:16, Christophe Lyon wrote:
>>
>> Hi Ramana,
>>
>> Here is an updated version of my proposal to include tests for Neon
>> intrinsics.
>>
>> wrt to my previous post, I have made a few changes:
>> - renamed the test files, removing
On Tue, Apr 15, 2014 at 3:52 AM, Richard Biener
wrote:
> On Mon, Apr 14, 2014 at 4:52 PM, Patrick Palka wrote:
>> Hi everyone,
>>
>> Many source files currently define a global function that is not
>> previously declared within that source file because the source file did
>> not include the appro
On Tue, Apr 15, 2014 at 3:51 AM, Richard Biener
wrote:
> On Mon, Apr 14, 2014 at 4:51 PM, Patrick Palka wrote:
>> Hi everyone,
>>
>> This patch wraps a bunch of locally-used, non-debug functions in an
>> anonymous namespace. These functions can't simply be marked as "static"
>> because they are
> Added a clarifying comment to the code and reinstated the TODO for the
> double case. OK to push?
>
> * dwarf2out.c (gen_enumeration_type_die): Add DW_AT_const_value
> as unsigned or int depending on type and value used.
OK. Thanks!
-cary
Hi!
The (admittedly ugly) reassoc stmt positioning stuff requires that
we maintain uids in ascending order within each bb (equal uid for several
adjacent stmts is ok), including debug stmts.
We assign those initially, and for stmts we add we make sure to copy the
uid from some adjacent insn.
But,
Hi,
a lot of time ago I noticed that these parameters are unused: should I
prepare a ChangeLog for the below or we have stylistic, etc, reasons for
keeping the parameters?
Thanks,
Paolo.
PS: I also see many int return types in the various unify* which could
as well be bool. Opinions about t
3d part:
2014-04-15 Evgeny Stupachenko
* config/i386/i386.c (x86_add_stmt_cost): Fixing vector cost model for
Silvermont.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 0ae3cda..2522b5c 100644
--- a/gcc/config/i386/i386.c
+++ b/gcc/config/i386/i386.c
@@ -8
2d part:
2014-04-15 Evgeny Stupachenko
* config/i386/x86-tune.def (TARGET_SLOW_PHUFFB): Target for slow byte
shuffle on some x86 architectures.
* config/i386/i386.h (TARGET_SLOW_PHUFFB): Ditto.
* config/i386/i386.c (expand_vec_perm_even_odd_1): Avoid byte shuffles
Hi,
I've separated the patch into 3.
The patch passes x86 bootstrap.
1st part:
2014-04-15 Evgeny Stupachenko
* config/i386/i386.c (slm_cost): Fixing vec_to_scalar_cost for
Silvermont according latency table.
(intel_cost): Ditto.
diff --git a/gcc/config/i386/i386.c b/gc
On Tue, Apr 15, 2014 at 4:13 PM, Eric Botcazou wrote:
> Hi,
>
> I recently stumbled on the ??? comment in tree-ssa-uninit.c and wondered why
> virtual operands are not enabled at -O0. As demonstrated by the attached
> patch, this would make it possible to unXFAIL a few uninit-*.c testcases.
>
> T
The bug in the PR is that is_copy_constructible>
performs overload resolution for function(const function&), which
instantiates the SFINAE constraints on the function(F) constructor
template, which instantiates is_convertible, which is
ill-formed when bar is incomplete.
When bar is later complete
As reported in the PR, gcc/configure currently fails to detect native
TLS support on x86_64-*-solaris2* with a 64-bit gas since it feeds it
32-bit TLS code. I haden't noticed this so far since I've been using a
32-bit gas here (no idea why).
The following patch fixes this by making sure 64-bit co
> -Original Message-
> From: Jason Merrill [mailto:ja...@redhat.com]
> Sent: Monday, April 14, 2014 9:49 PM
> To: Zamyatin, Igor; Jakub Jelinek
> Cc: GCC Patches (gcc-patches@gcc.gnu.org); Iyer, Balaji V
> Subject: Re: [PATCH, PR60189, Cilk+] Fix for ICE with incorrect Cilk_sync
> usage
On 03/21/2014 12:16 PM, Paul Pluzhnikov wrote:
Ok for trunk once it opens in stage 1?
OK.
Jason
On 04/15/2014 10:13 AM, Marc Glisse wrote:
I can add || VECTOR_TYPE_P (type), yes, but I thought you might have
ideas about other cases that might have been forgotten, maybe arrays or
something (I didn't have time to test any further), and thus on what the
right test should be. If it is just vect
As I said in the PR comments, I see no useful reason to allow
std::atomic<> to support zero-sized types.
Tested x86_64-linux, committed to trunk.
commit 6835d5ad1694f54d16c4a0d63273b12cbed78852
Author: Jonathan Wakely
Date: Tue Apr 15 13:19:32 2014 +0100
PR libstdc++/60695
*
Hi,
I recently stumbled on the ??? comment in tree-ssa-uninit.c and wondered why
virtual operands are not enabled at -O0. As demonstrated by the attached
patch, this would make it possible to unXFAIL a few uninit-*.c testcases.
Tested on x86_64-suse-linux.
* tree-ssa-operands.c (crea
On Tue, 15 Apr 2014, Jason Merrill wrote:
On 04/14/2014 06:02 PM, Marc Glisse wrote:
shouldn't the same also apply if VECTOR_TYPE_P (type), not just for
CLASS_TYPE_P (type)?
Sure. Do you want to make that change?
I can add || VECTOR_TYPE_P (type), yes, but I thought you might have ideas
a
On Tue, 15 Apr 2014, Eric Botcazou wrote:
> > I didn't yet relax some of its restrictions (we can safely
> > skip real/imag-part and array[-range?]-refs - we don't have
> > to stop at them. There is also the question whether
> > aliasing_component_refs_p still does sth useful after this
> > (if t
I'd rather not use TYPE_LINKAGE_IDENTIFIER in places that aren't dealing
with linkage. The other changes are OK.
Jason
> I didn't yet relax some of its restrictions (we can safely
> skip real/imag-part and array[-range?]-refs - we don't have
> to stop at them. There is also the question whether
> aliasing_component_refs_p still does sth useful after this
> (if those restrictions are removed).
>
> Bootstrap / regt
On 15/04/14 14:55, Richard Earnshaw wrote:
On 15/04/14 11:56, Kyrill Tkachov wrote:
Hi all,
This patch relates to PR60663 where cse got confused due to asm statements
being given a cost of zero in the arm backend. Jakub already put in a fix to
cse for 4.9.0 (http://gcc.gnu.org/ml/gcc-patches/
On 04/14/2014 06:02 PM, Marc Glisse wrote:
shouldn't the same also apply if VECTOR_TYPE_P (type), not just for
CLASS_TYPE_P (type)?
Sure. Do you want to make that change?
Jason
On 04/14/2014 11:10 AM, Marek Polacek wrote:
+ else if (TREE_CODE (val) == IDENTIFIER_NODE)
+{
+ tree t = lookup_name (val);
+ if (t && TREE_CODE (t) == CONST_DECL)
+ return DECL_INITIAL (t);
+}
I'm uncomfortable with this; we should have looked up any attributes in
th
On 15/04/14 11:56, Kyrill Tkachov wrote:
> Hi all,
>
> This patch relates to PR60663 where cse got confused due to asm statements
> being given a cost of zero in the arm backend. Jakub already put in a fix to
> cse for 4.9.0 (http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00512.html) but
> we sho
On 04/14/2014 04:06 PM, Jan Hubicka wrote:
1) write backend pass that walks vtables and for every non-hidden noreturn
function
it can produce hidden COMDAT wrapper with a fixed (GCC local) mangling.
Perhaps with a command line option to do so for all virtual functions;
while this
wo
Hi Vladimir:
> Although this patch is safe. I guess it could wait for stage 1 as right now
> we don't need this functionality.
>
> The patch is ok for the stage1 which is probably about a month away.
ping
is this patch ok now?
thanks:)
> Thanks for the patch.
This bumps LTO_major_version on trunk to avoid ICEing when
accidentially mixing objects with 4.9 ones. We'll soon start
to diverge (if we didn't already).
Committed.
Richard.
2014-04-15 Richard Biener
* lto-streamer.h (LTO_major_version): Bump to 4.
Index: gcc/lto-streamer.h
=
On Tue, 15 Apr 2014, Richard Biener wrote:
>
> This removes RTL loop unswitching (see last years discussion about
> compile-time issues of that pass). RTL loop unswitching is
> enabled together with GIMPLE loop unswitching at -O3 and by
> -floop-unswitch. It's clearly the wrong place to do high
> -Original Message-
> From: Maciej W. Rozycki [mailto:ma...@codesourcery.com]
> Sent: Tuesday, April 15, 2014 7:28 AM
> To: Richard Sandiford
> Cc: Matthew Fortune; Moore, Catherine; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] [MIPS] Fix operands for microMIPS SW16, SH16 and
> SB16
>
On Tue, Mar 4, 2014 at 11:37 PM, Nathaniel Smith wrote:
> On Tue, Feb 18, 2014 at 8:58 PM, Richard Henderson wrote:
>> On 02/16/2014 03:59 PM, Nathaniel Smith wrote:
>>> Yes, but the problem is that depending on what the user intends to do
>>> after forking, our pthread_atfork handler might help
This moves nonoverlapping_component_refs_p (now that I committed the
patch to make it O (n log n) instead of O (n^3)) to the tree
alias oracle (which is dispatched to from the RTL oracle). This
fixes the wrong-code part of the bug by moving the call to
nonoverlapping_component_refs_p at a place w
On 04/15/2014 01:13 PM, Joern Rennecke wrote:
> On 15 April 2014 10:20, Christian Bruel wrote:
>> Hello,
>>
>> I guess it's for RTL maintainers. Also interested by mode-switching.c
>> last contributors (from past ChangeLog entries) comments,
>>
>> http://gcc.gnu.org/ml/gcc-patches/2014-04/msg0019
Hi Tobias,
On Fri, 11 Apr 2014 16:04:51, Tobias Burnus wrote:
>
> Hi Tobias,
>
> On Fri, Apr 11, 2014 at 02:39:57PM +0200, Bernd Edlinger wrote:
>> On Fri, 11 Apr 2014 13:37:46, Tobias Burnus wrote:
>> Hmm,
>>
>> I was hoping somehow that only that test case is broken,
>> and needs to be fixed. Th
On Tue, 15 Apr 2014, Richard Sandiford wrote:
> > I believe you need to adjust constraints to ensure constant 0 is known to
> > produce a 16-bit instruction encoding where possible. Otherwise you'll
> > end up with suboptimal code when the instruction is in a branch delay
> > slot.
>
> Yeah,
Hi all,
This patch enables aarch64 to vectorise bswap[16,32,64] operations by using the
AdvancedSIMD forms of the rev[16,32,64] instructions.
The TARGET_VECTORIZE_BUILTIN_VECTORIZED_FUNCTION hook is extended to return the
vectorised forms of __builtin_bswap* where possible and vector bswap pa
Tested x86_64-linux, committed to trunk.
I intend to fix this for 4.9.1 and 4.8.3 too.
commit 1e13d4d7791a665bb254e0d53e96d3a5ab925023
Author: Jonathan Wakely
Date: Tue Apr 15 00:36:35 2014 +0100
PR libstdc++/60734
* include/bits/stl_tree.h (_Rb_tree::_M_end): Fix invalid ca
On Wed, Apr 2, 2014 at 12:04 PM, Jiong Wang wrote:
>
> On 25/03/14 15:44, Richard Earnshaw wrote:
>>
>> On 24/03/14 11:26, Jiong Wang wrote:
>>>
>>> This patch enables tail call optimization for long call on arm.
>>>
>>> Previously we have too strict check on arm_function_ok_for_sibcall and
>>> be
Hi all (and wide-int maintainers in particular),
I tried bootstrapping the wide-int branch on arm-none-linux-gnueabihf and
encountered some syntax errors while building wide-int.h and wide-int.cc in
expressions that tried to cast to HOST_WIDE_INT. This patch fixes those errors.
Also, in c-ada-
Hi all,
This patch relates to PR60663 where cse got confused due to asm statements
being given a cost of zero in the arm backend. Jakub already put in a fix to
cse for 4.9.0 (http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00512.html) but we
should still fix the costs in arm.
This patch does tha
> -Original Message-
> From: Richard Biener [mailto:richard.guent...@gmail.com]
> Sent: Tuesday, April 15, 2014 4:05 PM
> To: Joey Ye
> Cc: GCC Patches
> Subject: Re: [patch] Disable if_conversion2 for Og
>
> On Tue, Apr 15, 2014 at 3:59 AM, Joey Ye wrote:
> > If-converstion is harmful
Richard Sandiford writes:
> "Maciej W. Rozycki" writes:
> > On Sat, 12 Apr 2014, Richard Sandiford wrote:
> >
> >> I went ahead and applied the adjusted version of the patch to trunk
> >> as below (because I wanted to add a testcase too).
> >
> > I believe you need to adjust constraints to ensur
On 27 March 2014 23:56, Luke Allardyce wrote:
> It looks like the new standard also requires the precision to be
> ignored for hexfloat
>
>>For conversion from a floating-point type, if floatfield != (ios_base::fixed
>>| ios_base:: scientific), str.precision() is specified as precision in the
>>c
Hi!
This patch adds two new options (compatible with clang) which allow
users to choose the behavior of undefined behavior sanitization.
By default as before, all undefined behaviors (except for
__builtin_unreachable and missing return in C++) continue after reporting
which means that you can get
"Maciej W. Rozycki" writes:
> On Sat, 12 Apr 2014, Richard Sandiford wrote:
>
>> I went ahead and applied the adjusted version of the patch to trunk
>> as below (because I wanted to add a testcase too).
>
> I believe you need to adjust constraints to ensure constant 0 is known to
> produce a 16-
Hello,
I guess it's for RTL maintainers. Also interested by mode-switching.c
last contributors (from past ChangeLog entries) comments,
http://gcc.gnu.org/ml/gcc-patches/2014-04/msg00196.html
Many thanks
Christian,
This removes RTL loop unswitching (see last years discussion about
compile-time issues of that pass). RTL loop unswitching is
enabled together with GIMPLE loop unswitching at -O3 and by
-floop-unswitch. It's clearly the wrong place to do high-level
loop transforms these days, and the cost of mai
This makes the renaming code in gigi more robust in preparation for further
changes related to renaming.
Tested on x86_64-suse-linux, applied on the mainline.
2014-04-15 Eric Botcazou
Pierre-Marie de Rodat
* gcc-interface/decl.c (gnat_to_gnu_entity) : Create a mere
As discussed last year.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
Richard.
2014-04-15 Richard Biener
* alias.c (record_component_aliases): Do not walk BINFOs.
Index: gcc/alias.c
===
--- gcc/alias.c
On Fri, Apr 4, 2014 at 7:49 AM, Thomas Preud'homme
wrote:
>> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
>> ow...@gcc.gnu.org] On Behalf Of Rainer Orth
>>
>> Just omit the { target *-*-* } completely, also a few more times.
>
> Please find attached an updated patch.
@@ -1733,6 +1743,
Hi,
now that TYPE_IDENTIFIER includes an explicit check that TYPE_NAME is
nonnull, we can remove a few redundant uses of the latter. Plus a few
additional tweaks. Tested x86_64-linux.
Thanks,
Paolo.
///
2014-04-15 Paolo Carlini
* decl.c (duplicate_decls): Remove
Gigi generates a VIEW_CONVERT_EXPR in the middle of a hot loop, which kills
the performance at run time.
Tested on x86_64-suse-linux, applied on the mainline.
2014-04-15 Eric Botcazou
* gcc-interface/utils.c (type_for_vector_element_p): New predicate.
(build_vector_type_for_
Really a minor detail, but still.
Tested on x86_64-suse-linux, applied on the mainline.
2014-04-15 Eric Botcazou
* gcc-interface/decl.c (prepend_one_attribute_pragma): Call
Is_OK_Static_Expression in lieu of Is_Static_Expression to
detect valid arguments.
--
Eric B
On Tue, 2014-04-15 at 08:21 +0200, Jakub Jelinek wrote:
> On Mon, Apr 14, 2014 at 03:48:06PM -0700, Cary Coutant wrote:
> > >> Also note that size_of_die and value_format will still choose
> > >> DW_FORM_data[1248] for dw_val_class_unsigned_const in most cases.
> > >> Don't you really want to use
> Well, there are passes that can end up duplicating loops and thus you lose
> no_vectorize on the copy for example. Clearly that's undesired, no? For
> safelen and simduid not copying them is erroring on the safe side at least,
> likewise for force_vectorize. But we do have the copy_loop_info a
Hi Jakub:
Thanks your review and approve, however I don't have commit right yet,
can you help me to commit it :)
And here is updated patch in attachment, thanks!
On Mon, Apr 14, 2014 at 8:12 PM, Jakub Jelinek wrote:
> On Wed, Apr 09, 2014 at 10:00:38PM +0800, Kito Cheng wrote:
>> `q` will out o
This avoids the push/pop_cfun calls, they are not necessary.
Bootstrapped / tested on x86_64-unknown-linux-gnu, applied.
Richard.
2014-04-15 Richard Biener
* tree-ssa-structalias.c (find_func_aliases_for_builtin_call):
Add struct function argument and adjust.
(find_f
On Tue, Apr 15, 2014 at 10:01 AM, Eric Botcazou wrote:
>> The loop flags copying should go into copy_loop_info instead of only to
>> copy_loops. Jakub - I see you remap simduid on copy - you have to do
>> sth in copy_loop_info instead I suppose. See the other callers.
>
> That also occurred to m
On Tue, Apr 15, 2014 at 3:59 AM, Joey Ye wrote:
> If-converstion is harmful to optimized debugging as it generates conditional
> execution instructions with line number information, which resulted in a
> dillusion to developers that both then-else branches are executed.
>
> For example:
> test.c:
> The loop flags copying should go into copy_loop_info instead of only to
> copy_loops. Jakub - I see you remap simduid on copy - you have to do
> sth in copy_loop_info instead I suppose. See the other callers.
That also occurred to me, but IMO it's not crystal clear; for example, ivdep
(aka sa
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2014-04-15 Richard Biener
* tree.c (iterative_hash_expr): Use enum tree_code_class
to store TREE_CODE_CLASS.
(tree_block): Likewise.
(tree_set_block): Likewise.
* tree.h (fold_build
On Mon, Apr 14, 2014 at 4:48 PM, Patrick Palka wrote:
> Hi everyone,
>
> This patch marks "static" a bunch of locally-used, non-debug functions
> within the GCC sources. Doing so addresses a subset of the warnings emitted
> when compiling the GCC sources with -Wmissing-declarations.
>
> I bootstr
On Mon, Apr 14, 2014 at 6:40 PM, Marc Glisse wrote:
> On Mon, 14 Apr 2014, Richard Biener wrote:
>
>>> + /* If the special case has a high probability, keep it. */
>>> + if (EDGE_PRED (middle_bb, 0)->probability < PROB_EVEN)
>>
>>
>> I suppose Honza has a comment on how to test this properly
>>
On Mon, Apr 14, 2014 at 4:52 PM, Patrick Palka wrote:
> Hi everyone,
>
> Many source files currently define a global function that is not
> previously declared within that source file because the source file did
> not include the appropriate header file that declares said function.
> This patch fi
On Mon, Apr 14, 2014 at 4:51 PM, Patrick Palka wrote:
> Hi everyone,
>
> This patch wraps a bunch of locally-used, non-debug functions in an
> anonymous namespace. These functions can't simply be marked as "static"
> because they are used as template arguments to hash_table::traverse, and
> the C
On Mon, Apr 14, 2014 at 4:45 PM, Eric Botcazou wrote:
> Hi,
>
> this adds support for 2 optimization hints pertaining to loops in Ada, namely
> Loop_Optimize (No_Vector) and Loop_Optimize (Vector), by reusing the Ivdep
> approach in the middle-end (ANNOTATE_EXPR node) and directly setting the
> do
1 - 100 of 106 matches
Mail list logo
