On Fri, Dec 16, 2011 at 09:47:52AM +0400, Andrey Belevantsev wrote:
> Thus, if an insn has any note that prevents it from being cloned, we
> need to fix the above place. (If an insn should not even be moved
> from its block, then CANT_MOVE should be set to 1 in the same
> function.) Then create_c
> Well, ppc and spe support arbitrary permutation, which are the only other
> two listed with vec_perm.
>
> As for i386, we support arbitrary permutation beginning with ssse3.
>
> So, don't add vect_any_perm. Use vec_perm and figure out what's really
> wrong with slp-perm-9.
Thanks, that's right,
Hi,
I am working on user-directed and compiler-directed function
multiversioning which has been discussed in these threads:
1) http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02344.html
2) http://gcc.gnu.org/ml/gcc/2011-08/msg00298.html
The gist of the discussions for user-directed multiversioni
On 16.12.2011 0:42, Jakub Jelinek wrote:
On Thu, Dec 15, 2011 at 09:37:45AM -0800, Richard Henderson wrote:
On 12/15/2011 09:32 AM, Jakub Jelinek wrote:
PR debug/51557
* sel-sched-ir.c (create_copy_of_insn_rtx): Copy REG_ARGS_SIZE
notes.
* gcc.dg/pr51557.c: New
> This also fails on 4.6 branch. Can we move it to 4.6 branch?
Sure.
--
Eric Botcazou
As Georg-Johann Lay wrote:
> > Is there any need to still keep the -mtiny-stack option any longer at
> > all?
> In the proposed patch the option is needed to trigger the
> appropriate multilib.
OK, then it's fine with me. avr-libc will be taught to handle it,
sooner or later.
--
cheers, J"org
I cc'ed gcc-patches by mistake. This is a patch still in progress.
Please ignore.
-Sri.
On Thu, Dec 15, 2011 at 1:58 PM, wrote:
> Reviewers: davidxl,
>
> Message:
> I have uploaded a new patch. I did not split the patch into two because
> I felt the ifunc dispatch patch does nothing as a stand
On Thu, Oct 27, 2011 at 3:39 PM, Eric Botcazou wrote:
>> 2011-10-27 Chung-Lin Tang
>>
>> PR rtl-optimization/49720
>> * simplify-rtx.c (simplify_relational_operation_1): Detect
>> infinite recursion condition in "(eq/ne (plus x cst1) cst2)
>> simplifies to (eq/ne x (cst2
Reviewers: davidxl,
Message:
I have uploaded a new patch. I did not split the patch into two because
I felt the ifunc dispatch patch does nothing as a standalone. The
framework now will do aggressive cloning and it will mark the clones
specialized for core2 with "-mtune=core2". I have already sub
We were ignoring designators in cases where they aren't meaningful, but
we ought to reject them instead.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit b719497ead3a8be29d1842fa8715409c82fd1bd3
Author: Jason Merrill
Date: Wed Dec 14 19:02:00 2011 -0500
PR c++/51458
* decl.c (
On 15/12/2011 17:44, Mike Stump wrote:
> On Dec 15, 2011, at 1:43 AM, Kai Tietz wrote:
>> This patch takes care that we are using for operator new/delete
>> replacement test static version on mingw-targets. As the shared (DLL)
>> version isn't able to have operator overload within DLL itself, as
In PR51280, LTO does ICE because the object file uses TM builtin but the
TM is not enabled.
In the patch, it displays a error message if the builtin is not defined
and due to TM.
I moved is_tm_builtin() from calls.c to trans-mem.c. I splitted it into
2 functions is_tm_builtin/is_tm_builtin_code.
Committed. Fixes issues with the full epilogue not being emitted.
* config/rl78/rl78.md (return): Rename to rl78_return.
* config/rl78/rl78.c (rl78_expand_epilogue): Use new name.
(rl78_expand_eh_epilogue): Use new name.
(rl78_calculate_death_notes): Likewise.
In
On Thu, Dec 15, 2011 at 4:17 AM, Richard Guenther wrote:
> + fprintf (mstream, "\t@-touch -r %s %s.tem && mv %s.tem %s\n",
> + input_name, input_name, input_name, input_name);
Is touch -r portable?
Thanks,
Andrew Pinski
On Thu, Dec 15, 2011 at 09:37:45AM -0800, Richard Henderson wrote:
> On 12/15/2011 09:32 AM, Jakub Jelinek wrote:
> > PR debug/51557
> > * sel-sched-ir.c (create_copy_of_insn_rtx): Copy REG_ARGS_SIZE
> > notes.
> >
> > * gcc.dg/pr51557.c: New test.
>
> There are plenty of other no
On 12/14/2011 04:25 AM, Jakub Jelinek wrote:
> * config/i386/sse.md (vcond,
> vcond, vcondv2di):
> Use general_operand instead of nonimmediate_operand for
> operand 5 and no predicate for operands 1 and 2.
> * config/i386/i386.c (ix86_expand_int_vcond): Optimize
>
OK.
Jason
This removes a new fail that happens on bleeding edge linux with
the testcase in the subject line.
The behavior of the es_ES locale with respect to grouping has changed
in some version of glibc around 2.14.90-x. It used to have no grouping,
and now it has the usual "/003/003" grouping that one
On Thu, Dec 15, 2011 at 4:42 AM, Richard Guenther wrote:
>
> This fixes PR51564 - we were streaming the TREE_ASM_WRITTEN flag
> for TYPE_DECLs as-is, confusing the ltrans stage dwarf2out.c which
> then only partly emit debug information, eventually ICEing
> at dwarf2out.c:19288.
>
> The following
> -Original Message-
> From: Georg-Johann Lay [mailto:a...@gjlay.de]
> Sent: Thursday, December 15, 2011 11:30 AM
> To: gcc-patches@gcc.gnu.org
> Cc: Denis Chertykov; Weddington, Eric; Anatoly Sokolov
> Subject: [Patch,AVR]: Hookize PRINT_OPERAND*
>
> Implements target hooks
>
> TARGET_
On 12/15/2011 10:22 AM, Michael Zolotukhin wrote:
>> This is vect_perm. Why are you inventing a new one?
> As far as I understand, vect_perm is true if target supports at least
> some vector-permutation, while vect_any_perm is intended to be true if
> arbitrary permutation is supported (like in av
Implements target hooks
TARGET_PRINT_OPERAND
TARGET_PRINT_OPERAND_ADDRESS
TARGET_PRINT_OPERAND_PUNCT_VALID_P
Ok?
Johann
* config/avr/avr-protos.h (print_operand): Remove.
(print_operand_address): Remove.
* config/avr/avr.h (PRINT_OPERAND): Remove.
(PRINT_OPERAND_
> This is vect_perm. Why are you inventing a new one?
As far as I understand, vect_perm is true if target supports at least
some vector-permutation, while vect_any_perm is intended to be true if
arbitrary permutation is supported (like in avx). It was introduced
because vectorization began to occu
Looks good in general.
1) The profile information should be omitted when profile information
is not available
2) have a test case showing the call chain.
Ok for google branches with the above changes.
thanks,
David
On Thu, Dec 15, 2011 at 12:30 AM, Dehao Chen wrote:
> I change to use the bfd_
On Dec 15, 2011, at 1:43 AM, Kai Tietz wrote:
> This patch takes care that we are using for operator new/delete
> replacement test static version on
> mingw-targets. As the shared (DLL) version isn't able to have
> operator overload within DLL itself, as a DLL is finally-linked for
> PE-COFF.
> O
Joerg Wunsch wrote:
> As Georg-Johann Lay wrote:
>
>> With the patch it looks like so:
>>
>> .
>> ./tiny-stack
>> ./avr25
>> ./avr25/tiny-stack
>> ./avr3
>> ./avr31
>> ./avr35
>> ./avr4
>> ./avr5
>> ./avr51
>> ./avr6
>>
>> -mtiny-stack is a partial multilib option now:
>
> Is there any need to st
Hi!
Both the C and C++ FEs weren't marking num_threads/schedule clause
expressions as read, so we could end up with false positive
-Wunused-but-set-{variable,parameter} warnings.
Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux,
committed to trunk. Will backport to 4.6 soon.
On 12/15/2011 09:32 AM, Jakub Jelinek wrote:
> PR debug/51557
> * sel-sched-ir.c (create_copy_of_insn_rtx): Copy REG_ARGS_SIZE
> notes.
>
> * gcc.dg/pr51557.c: New test.
There are plenty of other notes that could as well need duplication.
E.g. REG_INC, REG_FRAME_RELATED_EX
Hi!
As discussed on IRC, we don't currently match the expected pattern
in dom1, but only in dom2, yet we at least generate the same assembly.
Not worth keeping it at P1 failure.
2011-12-15 Jakub Jelinek
PR middle-end/49806
* gcc.dg/tree-ssa/vrp47.c: Add -fdump-tree-dom2 to dg-
Hi!
I've committed a testcase that I wrote in the PR51117 clobber
sinking thread, after regtesting on x86_64-linux and i686-linux.
2011-12-15 Jakub Jelinek
PR tree-optimization/51117
* g++.dg/opt/pr51117.C: New test.
--- gcc/testsuite/g++.dg/opt/pr51117.C.jj 2011-12-15
Hi!
On the following testcase we ICE on x86_64-linux because sel-sched
decides to duplicate a push insn, the original one remains in one bb,
the copy in a different bb which have both a common successor.
Unfortunately the REG_ARGS_SIZE note isn't copied over, so during the
dwarf2 pass we don't not
Hi!
The problem with this invalid testcase is that we first parse the
initializer taking into account the static storage class (i.e.
pass it as a normal static data member initializer), but later on
in grokdeclarator with error on static volatile being used together
and clear static because of it,
On 12/15/2011 04:48 AM, Andreas Tobler wrote:
> * config/i386/freebsd.h (TARGET_ASM_FILE_END): Define.
Ok everywhere.
r~
As Georg-Johann Lay wrote:
> With the patch it looks like so:
>
> .
> ./tiny-stack
> ./avr25
> ./avr25/tiny-stack
> ./avr3
> ./avr31
> ./avr35
> ./avr4
> ./avr5
> ./avr51
> ./avr6
>
> -mtiny-stack is a partial multilib option now:
Is there any need to still keep the -mtiny-stack option any long
On 15/12/2011 10:33, Kai Tietz wrote:
> -// { dg-options "-mno-align-double" { target i?86-*-cygwin* i?86-*-mingw* } }
> +// As for mingw target the the ms-bitfield switch is activated by default,
> +// make sure for this test that it is disabled.
> +// { dg-options "-mno-align-double -mno-ms-bitf
Jakub Jelinek wrote on 15/12/2011 03:51:25 PM:
> On Thu, Dec 15, 2011 at 03:35:34PM +0200, Ira Rosen wrote:
> > > This patch also fixes
> > > a problem where vect_determine_vectorization_factor would iterate the
> > same
> > > stmt twice - for some reason both the original stmt and pattern stmt
Hi,
Linux/x86 target is configured as i686-android-linux. Changing it to
i686-linuxandroid requires changes in many packages. This patch adds
DEFAULT_LIBC=LIBC_BIONIC and ANDROID_DEFAULT=1 to *-android-*. OK for
trunk in stage 1?
Thanks.
H.J.
---
2011-12-15 H.J. Lu
* config.gcc: Ad
This fixes PR51567 - we are incorrectly merging the two 'iterator'
types in
class _Deque_base {
public:
typedef _Deque_iterator iterator;
iterator _M_impl;
};
class deque : public _Deque_base {
public:
typedef _Deque_base::iterator iterator;
};
which results in dwarf2out.c being conf
Hi all,
I'd like to commit the below patch to trunk and gcc-4.6.x.
The functionality has been in the FreeBSD base compiler since nearly 12
months now.
Is this ok for trunk, I know we're in stage three?
And for gcc-4.6 branch?
Tested with bootstrap/test (results follow) and 'readelf -l gcc/xg
> +# Return 1 if the target supports instructions for arbitrary permutations.
> +#
> +# This won't change for different subtargets so cache the result.
> +
> +proc check_effective_target_vect_any_perm { } {
This is vect_perm. Why are you inventing a new one?
r~
Mudflap checks a few legacy memory functions like bzero, and then passes
through the result to the C library's bzero. However, these functions
are no longer required in the latest POSIX spec (and in previous ones,
they were marked as LEGACY), so some libraries will omit this which leads
to build f
Hi David, Mike,
On 10 Dec 2011, at 23:57, David Edelsohn wrote:
On Sun, Dec 4, 2011 at 3:23 PM, Iain Sandoe
wrote:
gcc:
* config/rs6000/rs6000.c (first_altivec_reg_to_save): Amend
comment.
(compute_vrsave_mask): Likewise.
(rs6000_emit_prologue): Move update of VRSave mask
On 12/15/2011 09:22 AM, Aldy Hernandez wrote:
Richard, Jason, are you ok with just unsetting DECL_WEAK?
For now, yes.
I will come up with a suitable testcase for Patrick's case.
Once there's a testcase, we can figure out why the other patch didn't work.
Jason
Dear all,
while deferred-length character variables work in principle (though
there are issues), deferred-length components have not been implemented
at all. Currently, one gets an ICE if one tries to use them.
Dominique proposed to print an error (sorry not yet implemented), which
makes sen
http://gcc.gnu.org/ml/gcc-patches/2011-12/msg00471.html
Georg-Johann Lay wrote:
> Georg-Johann Lay wrote:
>> Denis Chertykov wrote:
>>
>> The only question that remains is what the -m64 option should be like?
>>
>> [ ] Omit it altogether
>> [ ] Leave it as is (off per default)
Dear Tobias,
>
> OK for the trunk?
OK - thanks!
Paul
>
> Tobias
>
> On 12/13/2011 06:30 PM, Tobias Burnus wrote:
>>
>> Two small fixes:
>>
>> a) There was an ICE when simplifying "THIS_IMAGE(caf)" (for
>> -fcoarray=single); solution: Simply use internally lcobound(), which is
>> identically (fo
This fixes PR51345: lib1funcs.S accesses SPH register even on devices that
don't have it.
There are 2 issues with SPH here:
* Reading non-existent SFR
* Assuming SPH = 0 lead to smaller code for devices without SPH
Devices with 8-bit SP are the ones that need size optimizations most
This patch
Small patch update: The patch now includes also primary.c changes, which
fix the DEALLOCATE issue with polymorphic scalar coarrays.
OK for the trunk?
Tobias
On 12/13/2011 06:30 PM, Tobias Burnus wrote:
Two small fixes:
a) There was an ICE when simplifying "THIS_IMAGE(caf)" (for
-fcoarray=
Richard, Jason, are you ok with just unsetting DECL_WEAK?
I will come up with a suitable testcase for Patrick's case.
Perhaps we can try unsetting (hee hee hee) DECL_WEAK as originally
proposed. Does this fix your problem?
Yep it does! ;)
Patrick.
Index: trans-mem.c
==
OK.
Jason
On Thu, Dec 15, 2011 at 03:35:34PM +0200, Ira Rosen wrote:
> > This patch also fixes
> > a problem where vect_determine_vectorization_factor would iterate the
> same
> > stmt twice - for some reason both the original stmt and pattern stmt (and
> > def stmt) are marked as relevant,
>
> Do you have
Jakub Jelinek wrote on 15/12/2011 12:54:29 PM:
> Perhaps it would be even cleaner to get rid of the pattern stmt and def
stmt
> seq distinction and just have pattern as whole be represented as
gimple_seq,
> but perhaps that cleanup can be deferred for later.
Sounds good.
> This patch also fix
On 11-12-14 13:43 , Jing Yu wrote:
Index: config/locale/generic/c_locale.cc
===
--- config/locale/generic/c_locale.cc (revision 182019)
+++ config/locale/generic/c_locale.cc (working copy)
@@ -52,8 +52,8 @@
{
// Assu
This fixes PR51564 - we were streaming the TREE_ASM_WRITTEN flag
for TYPE_DECLs as-is, confusing the ltrans stage dwarf2out.c which
then only partly emit debug information, eventually ICEing
at dwarf2out.c:19288.
The following patch does the same as we do for TYPE_Ps, stream
that flag as false.
The following patch removes ltrans inputs as soon as they have been
processed by the ltrans phase. In parallel mode, which is controlled
by make, this is done by truncating the input file, preserving its
timestamp. Thus the patch should at most reduce the peak disk-space
needs by a factor of two
Hello,
It seems to me that [dcl.spec.auto]/1
The auto type-specifier signifies that the type of a variable
being declared shall be deduced from its initializer or that a
function declarator shall include a trailing-return-type.
and the subsequent paragraphs of that section up to [dcl
On Thu, Dec 15, 2011 at 10:02:15AM +0100, Richard Guenther wrote:
> > But it's really ugly to insert part of pattern sequence, don't you think?
>
> It indeed is. The issue in the past was ICEing with -fno-tree-dce
> when the pattern stmts did not have regular RTL expansion support
> and the vecto
Jason Merrill writes:
> On 12/14/2011 03:41 PM, Dodji Seketeli wrote:
>> @@ -8041,6 +8041,7 @@ joust (struct z_candidate *cand1, struct z_candidate
>> *cand2, bool warn)
>> {
>> conversion *t1 = cand1->convs[i + off1];
>> conversion *t2 = cand2->convs[i + off2];
>> + c
2011/12/15 Georg-Johann Lay :
> This is fix for the RP. Is uses (reg:QI 28) instead of building SUBREG of
> frame
> pointer.
>
> Johann
>
> PR target/51050
> * config/avr/avr.c (expand_epilogue): Don't build SUBREG of fp.
> (avr_prologue_setup_frame): Ditto.
Approved.
Denis.
2011/12/15 Rainer Orth :
> Kai Tietz writes:
>
>> Index: g++.dg/abi/bitfield3.C
>> ===
>> --- g++.dg/abi/bitfield3.C (revision 182092)
>> +++ g++.dg/abi/bitfield3.C (working copy)
>> @@ -3,7 +3,7 @@
>> // { dg-options "-O2" }
>
Kai Tietz writes:
> Index: g++.dg/abi/bitfield3.C
> ===
> --- g++.dg/abi/bitfield3.C(revision 182092)
> +++ g++.dg/abi/bitfield3.C(working copy)
> @@ -3,7 +3,7 @@
> // { dg-options "-O2" }
> // Cygwin and mingw32 default to
Hi,
In gcc 4.7 the default for IA mingw targets about ms-bitfields was
changed. So patch
adjusts testcase to make sure we disable it for mingw targets
ChangeLog
2011-12-15 Kai Tietz
* g++.dg/abi/bitfield3.C: Make sure test runs
for windows targets without activated ms-bitfiel
On 13 December 2011 22:39, Gabriel Dos Reis wrote:
>
> Leaving __is_final undocumented is probably a good compromise.
Checked in, without the doc update. Thanks, all.
Index: c-family/c-common.h
===
--- c-family/c-common.h (revision 1
PR libstdc++/51540
* include/bits/stl_numeric.h (partial_sum): Adjust doxygen comments.
Tested x86_64-linux, committed to trunk.
Index: include/bits/stl_numeric.h
===
--- include/bits/stl_numeric.h (revision 182353)
+
On Wed, 14 Dec 2011, Richard Guenther wrote:
>
> The following patch avoids streaming extern var and function decls
> into the global decls and types section when streaming a tree
> chain list (effectively only when processing BLOCK_VARS). This
> avoids the last LTO debug-info ICE I know of (kno
Hi,
This patch takes care that we are using for operator new/delete
replacement test static version on
mingw-targets. As the shared (DLL) version isn't able to have
operator overload within DLL itself, as a DLL is finally-linked for
PE-COFF.
ChangeLog
2012-12-15 Kai Tietz
* g++.old-
On 13/12/11 14:13, Jakub Jelinek wrote:
> On Tue, Dec 13, 2011 at 01:26:42PM +0100, Tom de Vries wrote:
>> 2011-12-13 Tom de Vries
>>
>> PR tree-optimization/51491
>> * tree-ssa-ccp.c (insert_clobber_before_stack_restore): New function.
>> (ccp_fold_stmt): Use insert_clobber_befor
On Thu, Dec 15, 2011 at 9:22 AM, Tobias Burnus wrote:
> Hi Jakub,
>
>
> On 12/14/2011 06:49 PM, Jakub Jelinek wrote:
>>
>> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk (and
>> after a while for 4.6 too)?
>
>
> The Fortran part of the patch is OK. Thanks for the patch!
And t
On Thu, 15 Dec 2011, Ira Rosen wrote:
>
>
> Jakub Jelinek wrote on 15/12/2011 09:02:57 AM:
>
> > On Thu, Dec 15, 2011 at 08:32:26AM +0200, Ira Rosen wrote:
> > > > + cond = build2 (LT_EXPR, boolean_type_node, oprnd0, build_int_cst
> > > > (itype, 0));
> > > > + gsi = gsi_for_stmt (last_stmt)
I change to use the bfd_name, which is much shorter for C++ symbols.
Thanks,
Dehao
On Thu, Dec 15, 2011 at 2:07 AM, Xinliang David Li wrote:
> Another usability related issue for C++. The long demangled function
> names will make the info messages very hard to swallow. Since there
> will be sour
Hi Jakub,
On 12/14/2011 06:49 PM, Jakub Jelinek wrote:
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk (and
after a while for 4.6 too)?
The Fortran part of the patch is OK. Thanks for the patch!
Tobias
2011-12-14 Jakub Jelinek
PR debug/51517
* trans-dec
Le 8 déc. 2011 à 19:09, Jakub Jelinek a écrit :
> On Mon, Nov 28, 2011 at 12:16:52PM +0100, Romain Geissler wrote:
>> 2011-11-28 Romain Geissler
>>
>> * builtins.def (BUILT_IN_STPNCPY_CHK): New definition.
>> * builtins.c (expand_builtin, fold_builtin_4, maybe_emit_chk_warning):
>>
Uros Bizjak wrote on 15/12/2011 09:56:12 AM:
> On Thu, Dec 15, 2011 at 5:16 AM, Michael Zolotukhin
> wrote:
> > Thanks!
> > Fixed patch is attached.
> >
> > Any other comments?
> >
> > Changelog:
> > 2011-12-14 Michael Zolotukhin
> >
> > * gcc.dg/vect/no-section-anchors-vect-31.c: Ad
Jakub Jelinek wrote on 15/12/2011 09:02:57 AM:
> On Thu, Dec 15, 2011 at 08:32:26AM +0200, Ira Rosen wrote:
> > > + cond = build2 (LT_EXPR, boolean_type_node, oprnd0, build_int_cst
> > > (itype, 0));
> > > + gsi = gsi_for_stmt (last_stmt);
> > > + if (rhs_code == TRUNC_DIV_EXPR)
> > > +{
Ping, PR middle-end/51200
Tailored from Bernd's, and added target independent test case. Now it
is a pure middle-end fix.
OK for trunk and 4.6?
Bernd Schmidt
gcc/
* expr.c (store_field): Avoid a direct store if the mode is larger
than the size of the bit field.
75 matches
Mail list logo
