On Wed, Sep 24, 2014 at 6:48 AM, Jonathan Wakely wrote:
> OK for trunk - thanks.
Submitted
--
Regards,
Tim Shen
Jason,
this is my second attempt to get type_in_anonymous_namespace_p return true on
compound types built from types in anonymous namespace. I need this for LTO
TBAA where I do need the anonymous types to be closed upwards.
The original thread is here
https://gcc.gnu.org/ml/gcc-patches/2014-07/m
On 2014.09.26 at 04:50 +0200, Jan Hubicka wrote:
> What is the earlierst binutils release fixing the bug? I will add it into
> changes.html for 5.0.
binutils-2.24
--
Markus
> > Thank you! Now when I have your attention, perhaps we could discuss the
> > original
> > motivation of the change that exposed this bug.
> > I was building libreoffice with profile feedback and I run into a message
> >
> > cannot load any more object with static TLS
> >
> > that took me a whil
> Thank you! Now when I have your attention, perhaps we could discuss the
> original
> motivation of the change that exposed this bug.
> I was building libreoffice with profile feedback and I run into a message
>
> cannot load any more object with static TLS
>
> that took me a while to track as I
> From: Christophe Lyon [mailto:christophe.l...@linaro.org]
> Sent: Thursday, September 25, 2014 10:08 PM
>
> While attempting to try this, I noticed that more precisely the test
> is currently UNSUPPORTED on aarch64_be,
> which is because check_effective_target_bswap only accepts istarget
> aarc
> The plugin API doesn't have a way to mark a symbol as TLS, but it
> doesn't really matter since the linker simply overrides the
> placeholder from the claimed file with the symbol provided in the
> replacement. Unfortunately, I excluded common symbols from this logic
> in gold, so the symbol isn'
Ping?
> -Original Message-
> From: Tony Wang [mailto:tony.w...@arm.com]
> Sent: Tuesday, September 16, 2014 11:01 AM
> To: 'gcc-patches@gcc.gnu.org'
> Cc: Richard Earnshaw; Ramana Radhakrishnan
> Subject: RE: [PATCH 3/3,ARM,libgcc]Code size optimization for the fmul/fdiv
> and dmul/ddi
Ping?
> -Original Message-
> From: Tony Wang [mailto:tony.w...@arm.com]
> Sent: Tuesday, September 16, 2014 11:01 AM
> To: 'gcc-patches@gcc.gnu.org'
> Cc: Richard Earnshaw; Ramana Radhakrishnan
> Subject: RE: [PATCH 2/3,ARM,libgcc]Code size optimization for the fmul/fdiv
> and dmul/ddiv f
Ping?
> -Original Message-
> From: Tony Wang [mailto:tony.w...@arm.com]
> Sent: Tuesday, September 16, 2014 11:01 AM
> To: 'gcc-patches@gcc.gnu.org'
> Cc: Richard Earnshaw; Ramana Radhakrishnan
> Subject: RE: [PATCH 1/3,ARM,libgcc]Code size optimization for the fmul/fdiv
> and dmul/ddiv f
The plugin API doesn't have a way to mark a symbol as TLS, but it
doesn't really matter since the linker simply overrides the
placeholder from the claimed file with the symbol provided in the
replacement. Unfortunately, I excluded common symbols from this logic
in gold, so the symbol isn't getting
Hi,
ipa_intraprocedural_devirtualization was for few weeks used only to check that
new get_dynamic_type code can handle all cases old detect_type_change did.
Since all reported issues are fixed now, I removed the code.
lto-profiledbootstrapped/regtested x86_64-linux
* ipa-prop.c (ipa_intr
On 19 September 2014 12:21, Tejas Belagod wrote:
> The reason we avoided using type-punning using unions was that reload would
> get confused with potential subreg(mem) that could be introduced because of
> memory xfer caused by unions and large int modes. As a result, we would get
> incorrect or
On 09/25/2014 04:48 PM, Sandra Loosemore wrote:
> On 09/25/2014 02:04 PM, Sandra Loosemore wrote:
>> While doing some arm-none-eabi testing, I noticed that a bunch of
>> gcc.dg/vect tests were causing the target to hang from trying to execute
>> code compiled with "-mfpu=neon -mfloat-abi=softfp", o
On 09/25/2014 02:04 PM, Sandra Loosemore wrote:
While doing some arm-none-eabi testing, I noticed that a bunch of
gcc.dg/vect tests were causing the target to hang from trying to execute
code compiled with "-mfpu=neon -mfloat-abi=softfp", on a target that
doesn't support those instructions. And,
Hi,
this patch contains two things, the first is an upgrade to OKRA
bindings in libgomp of the hsa branch provided by Saravanan Ekanathan
and the second is a brand new README.hsa file that explains what is
necessary to run the generated code and how to do it, in this brave
new HSAIL 1.0p and new O
Hi,
this big patch is work of Ganesh Gopalasubramanian with a small number
of my modifications. It upgrades the hsa branch of gcc to HSAIL
version 1.0p (so far we've been producing 0.95).
Bootstrapped to confirm there were no warnings, tested on a collection
of OMP testcases, committed to the HS
Hi,
this patch removes omp-expansion generated control flow from simple
omp parallel loops (identified by the previous patch) and makes the
functions generated from these loops with attribute "hsakernel" which
means that it also adds two parameters denoting the iteration size and
group size (which
Hi,
this patch is preparatory work for the next one in the series. It
identifies simple and single omp parallel for loops and marks them as
such. A loop is eligible if there is just one in its parallel
construct, there is no other omp construct in that construct and there
are other limitations l
Hi,
in the hsa branch, we currently compile functions with the hsa
attribute to an HSA kernel and when such a function is called, we
translate into a call to libgomp which then feeds the grid and group
sizes (basically the number of threads and how they are structured).
For GPGPU we of course wan
Hi,
this patch enhances the interface in between gcc and the current HSA
libgomp entry point to communicate the HSA grid and group size.
Bootstrapping of course only showed there were no warnings but I've
done it anyway, my small collection of OMP tests did not regress.
Committed to the hsa branc
Yeah, the comments are big, but the body is the smallest.
Send from Lenovo A788t.
Michael Eager wrote:
>On 09/25/14 10:38, Michael Eager wrote:
>> On 09/25/14 07:03, Chen Gang wrote:
>>> Need use VOID instead of SI, or when real VOIDmode comes, it does not
>>> match SImode, so cause issue. Th
On Sep 25, 2014, at 1:04 PM, Sandra Loosemore wrote:
> While doing some arm-none-eabi testing, I noticed that a bunch of gcc.dg/vect
> tests were causing the target to hang from trying to execute code compiled
> with "-mfpu=neon -mfloat-abi=softfp", on a target that doesn't support those
> inst
2014-09-25 14:07 GMT+01:00 Jiong Wang :
>
> On 25/09/14 12:25, Christophe Lyon wrote:
>>
>>>
>> I have observed regressions in the g++ testsuite: pr49847 now FAILs
>> after this patch.
>
> no.
>
> even without my patch, the regression still happen.
>
> or you could specify -fno-shrink-wrap, gcc sti
Hi,
according to my type checker, ipa-prop is only place where we produce a variant
of
a METHOD_TYPE that is FUNCTION_TYPE or viarant that has different parameters.
The code in question is producing new prototype to remove unused arguments,
but I do not think it should eclare the new type to be v
Apart from those minor adjustments I think this looks good, but I'd
like to know that it has been tested with -fsanitize=thread, even if
only lightly tested.
Hi
Dmitry, who reported the bug, confirmed the fix. Can I go ahead and
commit ?
François
Hi!
Resending this email. Seems some spam filter ate it due to the many
email addresses...
Following up on my suggestion to put all email addresses into <...>
(cf. https://gcc.gnu.org/ml/gcc/2014-09/msg00298.html) here's an
actual patch. Quite a mechanical change, along with a few clean-ups of
s
On Tue, 2014-09-23 at 23:27 +, Joseph S. Myers wrote:
[...]
> The code for compiling a .s file should:
[...]
> * use the $(target_noncanonical)-gcc-$(version) name for the driver rather
> than plain "gcc", to maximise the chance that it is actually the same
> compiler the JIT library was buil
Hi!
On Tue, 8 Jul 2014 07:28:24 -0700, Cesar Philippidis
wrote:
> On 07/07/2014 02:55 AM, Thomas Schwinge wrote:
> > On Sun, 6 Jul 2014 16:10:56 -0700, Cesar Philippidis
> > wrote:
> >> This patch is the first step to enabling parallel reductions in openacc.
> --- /dev/null
> +++ b/gcc/testsu
While doing some arm-none-eabi testing, I noticed that a bunch of
gcc.dg/vect tests were causing the target to hang from trying to execute
code compiled with "-mfpu=neon -mfloat-abi=softfp", on a target that
doesn't support those instructions. And, this was caused by tests
specifying "dg-do ru
Hi,
this patch moves all code dealing with polymorphic_call_context to
ipa-polymorphic-call.c in hope that things are more manageable this way.
This code contains most of the magic of analyzing GIMPLE and determining
contextes.
ipa-devirt.c still contains several essentially independent things - t
On 08/04/14 05:04, Alan Modra wrote:
Two small changes to gcc code to support newer obstacks:
1) gcc currently calls _obstack_begin, which requires some ugly casts
on alloc and free functions, and the casts will change when/if 64-bit
obstacks are available. It's cleaner to use the convenience f
On 09/01/14 21:24, Andi Kleen wrote:
Having a nop area at the beginning of each function can be also
also useful for other things. For example it can be used to patch
functions at runtime to point to different functions, to do
binary updates without restarting the program (like ksplice or
simila
On Thu, Sep 25, 2014 at 10:33:17AM -0700, Michael Collison wrote:
> The problem is the "CONST_INT 0", not a large constant. This constant is
> not accepted by the predicate, but is accepted by the constraint.
Yes, bad choice of words, sorry. Just read "big" as "not matching the
predicate". The
On Thu, Sep 25, 2014 at 05:12:24PM +0100, Alan Lawrence wrote:
> So: original patch with updated commenting attached...Segher, is there any
> chance you could test this on powerpc too? (in combination with patch 2/14,
> which will need to be applied first; you can skip patch 1, and >=4.)
2+3/14,
On 09/01/14 04:29, Ilya Tocar wrote:
AVX512 added new 16 xmm registers (xmm16-xmm31).
Those registers require evex encoding.
Only 512-bit wide versions of instructions have evex encoding with
avx512f, but all versions have it with avx512vl.
Most instructions have same macroized pattern for 128/2
On 09/23/14 01:14, Maxim Ostapenko wrote:
2014-09-04 Jakub Jelinek
Max Ostapenko
* common.opt: New option.
* doc/invoke.texi: Describe new option.
* gcc.c (execute): Don't free first string early, but at the end
of the function. Call retry_ice if c
On 09/24/14 06:07, Felix Yang wrote:
Hi Jeff,
Thanks for the comments. I updated the patch adding some enhancements.
Bootstrapped on x86_64-suse-linux. Please apply this patch if OK for trunk.
Three points:
1. For multiple-set register, it is not qualified to have a equiv
no
On Sep 25, 2014, at 1:02 AM, Uros Bizjak wrote:
>>> IMO, these should be converted to dg-additional-options.
>>
>> That would be something like following, except that compat framework doesn't
>> support dg-additional-options:
>> WARNING: compat.exp does not support dg-additional-options
>
> Rain
On Thu, Sep 25, 2014 at 4:12 PM, Kirill Yukhin wrote:
> Hello,
> Patch in the bottom extends plus/minus/abs/andnot patterns
> to support AVX-512.
> I've used questionable hack in the patterns.
> Instead of writing dozen similar patterns with masking
> I've simply substed them, prohibiting non-mask
push_cfun() fails when there's no cfun stack. With this patch, we use
set_cfun if not stack is available.
This fixes the 16 Fortran guality regressions.
Now guality tests all pass, for all languages.
Committed to branch.
commit a6f19a625bc6f662db6f23679503458f22721de9
Author: Aldy Hernandez
On Thu, Sep 25, 2014 at 11:42 AM, Kirill Yukhin wrote:
> Hello,
> As suggested, this is splitted out part of [52/n] patch,
> which introduces new vec2mask and mask2vec insn patterns.
>
> As suggested, I've got rid off use of UNSPEC_CVTINT2MASK
> unspec. Unfortunatelly, only partially.
> I suppose,
On Thu, Sep 25, 2014 at 7:44 PM, Uros Bizjak wrote:
>>> The failure was caused by barrier detection code, which failed to
>>> detect barrier after call insn was to be split when
>>> NOTE_CALL_ARG_LOCATION was present. This problem caused
>>> -fcompare-debug failure.
>>>
>>> Digging a bit deeped,
On 09/24/14 13:39, Uros Bizjak wrote:
Hello!
The failure was caused by barrier detection code, which failed to
detect barrier after call insn was to be split when
NOTE_CALL_ARG_LOCATION was present. This problem caused
-fcompare-debug failure.
Digging a bit deeped, and as hinted in the PR, the
On Thu, Sep 25, 2014 at 7:26 PM, Jeff Law wrote:
> On 09/24/14 13:39, Uros Bizjak wrote:
>>
>> Hello!
>>
>> The failure was caused by barrier detection code, which failed to
>> detect barrier after call insn was to be split when
>> NOTE_CALL_ARG_LOCATION was present. This problem caused
>> -fcompa
On 09/01/2014 09:41 PM, Ed Smith-Rowland wrote:
+ cpp_define (pfile, "__cpp_attribute_deprecated=201309");
Don't we support attribute deprecated in C++11?
Jason
On 09/25/14 10:38, Michael Eager wrote:
On 09/25/14 07:03, Chen Gang wrote:
Need use VOID instead of SI, or when real VOIDmode comes, it does not
match SImode, so cause issue. This patch can fix this issue and pass
testsuite.
Did you forget to attach the patch?
Never mind. My eyes were play
On 09/25/14 07:03, Chen Gang wrote:
Need use VOID instead of SI, or when real VOIDmode comes, it does not
match SImode, so cause issue. This patch can fix this issue and pass
testsuite.
Did you forget to attach the patch?
--
Michael Eagerea...@eagercon.com
1960 Park Blvd., Palo Alto, CA 94
On 09/01/2014 09:41 PM, Ed Smith-Rowland wrote:
+ /* Return type deduction was added as an extension to C++11
+and was standardized for C+14. */
+ cpp_define (pfile, "__cpp_return_type_deduction=201304");
When I try to use it with -std=c++11 I get
wa.C:1:8: error:
On Tue, 2014-09-23 at 23:27 +, Joseph S. Myers wrote:
[...]
> The code for compiling a .s file should:
[...]
> * use libiberty's pexecute to run subprocesses, not "system" (building up
> a string to pass to the shell always looks like a security hole, though in
> this case it may in fact be s
Segher,
The problem is the "CONST_INT 0", not a large constant. This constant is
not accepted by the predicate, but is accepted by the constraint.
On 09/25/2014 03:12 AM, Segher Boessenkool wrote:
On Wed, Sep 24, 2014 at 09:17:23PM -0700, Andrew Pinski wrote:
On Wed, Sep 24, 2014 at 9:13 PM,
Hi Jonathan,
>>it does: {i386-pc, sparc-sun}-solaris2.1[01] bootstraps completed
>>successfully with it.
>
> Great. I hope the slightly-modified version I eventually checked in
> still works too :)
it does indeed, as just verified by a i386-pc-solaris2.11 bootstrap :-)
Rainer
--
--
On 09/24/14 13:39, Uros Bizjak wrote:
Hello!
The failure was caused by barrier detection code, which failed to
detect barrier after call insn was to be split when
NOTE_CALL_ARG_LOCATION was present. This problem caused
-fcompare-debug failure.
Digging a bit deeped, and as hinted in the PR, the
On Thu, Sep 25, 2014 at 08:22:29AM -0400, Andrew MacLeod wrote:
> So to be fair, I could use test_summary, but I think the concern is
> warranted because if this inconsistent ordering can happen to PASS, I
> would expect the same non-deterministic behaviour if those tests happen
> to FAIL. we j
On 09/01/2014 09:34 PM, Ed Smith-Rowland wrote:
(open_file_failed()): Not an error to not find a header file for
__has_include__.
Hmm, looks like this means that __has_include__ will silently return
false if a header exists but is unreadable; I would think that we want
it to b
Hi, I'm implementing the co_broadcast on libcafmpi right now.
2014-09-25 8:08 GMT+02:00 Tobias Burnus :
> Hi Paul,
>
> Paul Richard Thomas wrote:
>>
>> In the check.c error messages, you use 'A argument'. Should you not
>> use 'SOURCE argument', following CO BROADCAST (SOURCE, SOURCE IMAGE [,
>>
On 25/09/14 17:24, Jeff Law wrote:
On 09/25/14 09:04, Jiong Wang wrote:
new patch updated.
pass bootstrap and no regression, both check-gcc and check-g++, on the x86.
OK for trunk?
thanks.
gcc/
* shrink-wrap.c (move_insn_for_shrink_wrap): Initialize the live-in of
new created BB a
On 09/25/14 08:04, Felix Yang wrote:
Please apply this patch if OK for trunk. Thanks.
[ ... ]
Index: gcc/ChangeLog
===
--- gcc/ChangeLog(revision 215598)
+++ gcc/ChangeLog(working copy)
@@ -1,3 +1,8 @@
+2014-09-25 Felix Ya
On 19/09/14 15:35, Wilco Dijkstra wrote:
Jiong Wang wrote:
when generating instructions to access local variable, for example a local
array,
if the array size very big, then we need a temp reg to keep the intermediate
index,
then use that temp reg as base reg, so that ldr is capable of index
> On Thu, Sep 25, 2014 at 8:37 AM, H.J. Lu wrote:
> > On Thu, Sep 25, 2014 at 8:24 AM, Ian Lance Taylor wrote:
> >> On Wed, Sep 24, 2014 at 6:58 PM, Jan Hubicka wrote:
> >>>
> >>>b: 00 00
> >>> 9: R_X86_64_TPOFF32
> >>> __gcov_indirect_call_counters_ltopriv
> >
On Thu, Sep 25, 2014 at 8:37 AM, H.J. Lu wrote:
> On Thu, Sep 25, 2014 at 8:24 AM, Ian Lance Taylor wrote:
>> On Wed, Sep 24, 2014 at 6:58 PM, Jan Hubicka wrote:
>>>
>>>b: 00 00
>>> 9: R_X86_64_TPOFF32
>>> __gcov_indirect_call_counters_ltopriv
>>
>> Look at the
On 09/25/14 09:04, Jiong Wang wrote:
On 25/09/14 09:52, Zhenqiang Chen wrote:
-Original Message-
From: Jiong Wang [mailto:jiong.w...@arm.com]
Sent: Thursday, September 25, 2014 2:13 AM
To: Jeff Law; Zhenqiang Chen
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH, 2/2] shrink wrap a fun
Well, even that C source, you'd need to be careful and ensure that the
vectorized loop never went round more than once, or else the additions within
the loop would be performed in 8 bits, different from the final reduction...
So: original patch with updated commenting attached...Segher, is ther
On Thu, 2014-09-25 at 17:23 +0200, Oleg Endo wrote:
> On Thu, 2014-09-25 at 16:04 +0100, Nick Clifton wrote:
> > Hi Alex, Hi Kaz, Hi Oleg,
> >
> > I am applying the patch below as an obvious fix for a typo in the
> > instruction sequence of the atomic_fetch_nand_soft_imask
> > pattern. I ho
Many thanks indeed! :)
--Alan
Segher Boessenkool wrote:
On Wed, Sep 24, 2014 at 04:02:11PM +0100, Alan Lawrence wrote:
However my CompileFarm account is still pending, so to that end, if you
were able to test patch 2/14 (attached inc. Richie's
s/VIEW_CONVERT_EXPR/NOP_EXPR/) on the CompileFarm
On Tue, 2014-09-23 at 23:27 +, Joseph S. Myers wrote:
[...]
> The code for compiling a .s file should:
>
> * use choose_tmpdir from libiberty rather than hardcoding /tmp (or,
> better, create the files directly with make_temp_file, and delete them
> individual afterwards);
[...]
I believe t
> On Wed, Sep 24, 2014 at 6:58 PM, Jan Hubicka wrote:
> >
> >b: 00 00
> > 9: R_X86_64_TPOFF32
> > __gcov_indirect_call_counters_ltopriv
>
> Look at the .o file where __gcov_indirect_call_counters_ltopriv is
> defined. That .o file must have the symbol marked as
On 25 September 2014 16:18, Richard Henderson wrote:
> On 09/25/2014 08:05 AM, James Greenhalgh wrote:
>>
>> On Fri, Sep 19, 2014 at 05:57:06PM +0100, Richard Henderson wrote:
>>> On 09/11/2014 01:29 AM, James Greenhalgh wrote:
+;; Predicates used by the various SIMD shift operations. These
On 09/25/2014 08:05 AM, James Greenhalgh wrote:
>
> On Fri, Sep 19, 2014 at 05:57:06PM +0100, Richard Henderson wrote:
>> On 09/11/2014 01:29 AM, James Greenhalgh wrote:
>>> +;; Predicates used by the various SIMD shift operations. These
>>> +;; fall in to 3 categories.
>>> +;; Shifts with a ra
http://cplusplus.github.io/LWG/lwg-defects.html#1339
Also make the same change to our __uninitialized_xxx_n_a extensions
and change std::vector to make use of the returned iterator instead of
recalculating it.
Tested x86_64-linux, committed to trunk.
commit 81456b32a7991351d40ca45c276892fdab5d91
Hi,
I removed my change to statements.cc and made a change as you suggested
in libgo/runtime/go-defer.c in function __go_set_defering_fn:
+#if defined(__powerpc64__) && _CALL_ELF != 2
+g->defer->__defering_fn = *(void **)defering_fn;
+#else
+g->defer->__defering_fn = defering_fn;
+#end
On Thu, Sep 25, 2014 at 8:24 AM, Ian Lance Taylor wrote:
> On Wed, Sep 24, 2014 at 6:58 PM, Jan Hubicka wrote:
>>
>>b: 00 00
>> 9: R_X86_64_TPOFF32
>> __gcov_indirect_call_counters_ltopriv
>
> Look at the .o file where __gcov_indirect_call_counters_ltopriv is
>
*PING*
I would like review of my SD-6 implementation.
Part 1 - __has_include__ built-in in libcpp.
https://gcc.gnu.org/ml/gcc-patches/2014-09/msg00083.html
Part 2 - __has_include macro and C++ language feature macros.
https://gcc.gnu.org/ml/gcc-patches/2014-09/msg00084.html
Part 3 - libstdc++
On Thu, Sep 25, 2014 at 4:32 PM, Alan Lawrence wrote:
> Ok, so, I've tried making reduc_plus optab take two modes: that of the
> vector to reduce, and the result; thus allowing platforms to provide a
> widening reduction. However, I'm keeping reduc_[us](min|max)_optab with only
> a single mode, as
On Wed, Sep 24, 2014 at 6:58 PM, Jan Hubicka wrote:
>
>b: 00 00
> 9: R_X86_64_TPOFF32
> __gcov_indirect_call_counters_ltopriv
Look at the .o file where __gcov_indirect_call_counters_ltopriv is
defined. That .o file must have the symbol marked as STT_TLS and it
On Wed, 2014-09-24 at 17:09 -0400, Jason Merrill wrote:
> On 09/24/2014 02:42 PM, Mark Wielaard wrote:
> > As my example shows even with older GCCs depending on whether a const
> > type or volatile type was encountered first a const volatile type comes
> > out either as a DIE chain const_type DIE -
On Thu, 2014-09-25 at 16:04 +0100, Nick Clifton wrote:
> Hi Alex, Hi Kaz, Hi Oleg,
>
> I am applying the patch below as an obvious fix for a typo in the
> instruction sequence of the atomic_fetch_nand_soft_imask
> pattern. I hope that this is OK with you.
Yes, ouch. Thanks!
I'll backport
On Thu, Sep 25, 2014 at 4:57 PM, James Greenhalgh wrote:
> * doc/tm.texi.in (MOVE_BY_PIECES_P): Reduce documentation to a stub
> describing that this macro is deprecated.
Remove it entirely and poison it in system.h?
It takes changes to only a few targets: mips, arc, s390, and sh.
On Fri, Sep 19, 2014 at 05:57:06PM +0100, Richard Henderson wrote:
> On 09/11/2014 01:29 AM, James Greenhalgh wrote:
> > +;; Predicates used by the various SIMD shift operations. These
> > +;; fall in to 3 categories.
> > +;; Shifts with a range 0-(bit_size - 1) (aarch64_simd_shift_imm)
> > +;;
On 25/09/14 09:52, Zhenqiang Chen wrote:
-Original Message-
From: Jiong Wang [mailto:jiong.w...@arm.com]
Sent: Thursday, September 25, 2014 2:13 AM
To: Jeff Law; Zhenqiang Chen
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH, 2/2] shrink wrap a function with a single loop: split
live_e
Hi Alex, Hi Kaz, Hi Oleg,
I am applying the patch below as an obvious fix for a typo in the
instruction sequence of the atomic_fetch_nand_soft_imask
pattern. I hope that this is OK with you.
Cheers
Nick
gcc/ChangeLog
2014-09-25 Nick Clifton
PR target/62218
* config/s
Hi,
After hookizing MOVE_BY_PIECES_P and migrating tree-inline.c, we are
left with only one user of MOVE_RATIO - deciding the maximum size of
aggregate for SRA.
Past discussions have made it clear [1] that keeping this use of
MOVE_RATIO is undesirable. Clearly it is now also misnamed.
The previ
Hi,
This patch wires up our new target hooks for AArch64. This also means
we can bring back the two failing SRA tests.
Bootstrapped on AArch64 with no issues.
OK for trunk?
Thanks,
James
---
gcc/
2014-09-25 James Greenhalgh
* config/aarch64/aarch64.c
(aarch64_estimate_mov
On Wed, Aug 20, 2014 at 10:21:41AM +0100, Richard Biener wrote:
> I think this is overly complicated and instead SRA should only
> use the parameters. Targets can adjust their default (like they
> do for other parameters).
>
> The default should be MOVE_RATIO which should be applied
> where the c
Hi,
This patch hookizes the use of MOVE_RATIO in
tree-inline.c:estimate_move_cost as TARGET_ESTIMATE_BLOCK_COPY_NINSNS.
This hook should return an estimate for the number of instructions
which will be emitted to copy a block of memory.
tree-inline.c uses this in inlining heuristics to estimate t
Hi,
This patch started off by Hookizing MOVE_RATIO, but pulling on that
thread made it clear that most users of MOVE_RATIO really want to know
whether move_by_pieces is going to be used or not. For that we have
MOVE_BY_PIECES_P.
We can hookize this, and clean up most other callers of MOVE_RATIO.
Ok, so, I've tried making reduc_plus optab take two modes: that of the vector to
reduce, and the result; thus allowing platforms to provide a widening reduction.
However, I'm keeping reduc_[us](min|max)_optab with only a single mode, as
widening makes no sense there.
I've not gone as far as ma
Of course I missed c-exprs... The following patch fixes that.
Now bootstrapped (building stage3 target libs now).
Applied.
Richard.
2014-09-25 Richard Biener
* genmatch.c (struct c_expr): Also record capture identifier
to index map.
(c_expr::gen_transform): Use it.
Hello,
Patch in the bottom extends plus/minus/abs/andnot patterns
to support AVX-512.
I've used questionable hack in the patterns.
Instead of writing dozen similar patterns with masking
I've simply substed them, prohibiting non-mask variant in
the pattern condition. E.g.:
(define_expand "3"
[(set
On 25 September 2014 08:39, Thomas Preud'homme
wrote:
>> From: Christophe Lyon [mailto:christophe.l...@linaro.org]
>> Sent: Thursday, September 25, 2014 4:28 AM
>
>>
>> Hi Thomas,
>
> Hi Christophe,
>
>>
>> Although I could notice the improvement:
>> Pass disappears [PASS => ]:
>>
On 23 Sep 09:55, Jeff Law wrote:
> On 09/22/14 00:40, Ilya Enkovich wrote:
> >
> >Bounds don't have to vary for different pointers. E.g. p and p + 1
> >always have equal bounds. In this particular case we have function
> >pointers and all of them have default bounds.
> OK. It looked a bit odd an
Please apply this patch if OK for trunk. Thanks.
Index: gcc/lra.c
===
--- gcc/lra.c(revision 215598)
+++ gcc/lra.c(working copy)
@@ -933,7 +933,7 @@ lra_set_insn_recog_data (rtx_insn *insn)
nalt = 1;
if (nop < 0)
Need use VOID instead of SI, or when real VOIDmode comes, it does not
match SImode, so cause issue. This patch can fix this issue and pass
testsuite.
The related test code ('void' will cause CALL instead of SET):
typedef void (*T)(void);
f1 ()
{
((T) 0)();
}
The related error:
[ro
On Wed, Sep 24, 2014 at 6:51 PM, Bill Schmidt
wrote:
> Hi,
>
> PR63335 reports that the builtins vec_any_nge, vec_all_nge, vec_any_nle,
> and vec_all_nle produce incorrect results for vector double. There is
> some special handling for these intrinsics for the various vector
> integral types. Th
Hi,
This patch fixes an annoying gotcha when adding new cores or piepline
models in builds for AArch64. The "generic_sched" attribute also needs
updating in addition to aarch64-tune.md.
I see no good reason for this, we can generate that attribute in
gentune.sh quite easily.
For testing, I buil
On 25/09/14 12:25, Christophe Lyon wrote:
On 24 September 2014 20:32, Jiong Wang wrote:
On 22/09/14 19:01, Jeff Law wrote:
On 09/22/14 04:29, Jiong Wang wrote:
On 19/09/14 21:43, Jeff Law wrote:
patch attached.
please review, thanks.
gcc/
* shrink-wrap.c (move_insn_for_shrink_wrap):
The following makes us assign capture indexes dynamically which
easily allows handling of identifiers. I've used std::map
for this.
Applied.
Note that this may make debugging a little bit harder as
@0 now no longer necessarily corresponds to captures[0]
(I don't think it necessarily did before
OK.
On 09/24/2014 01:58 PM, Andrew MacLeod wrote:
On 09/24/2014 12:29 PM, Andrew MacLeod wrote:
AH. interesting.
The third run has a gcc.sum that is exactly the same as the first run.
so only the second run differs, and it seems to be from an
alphabetical sort. So run 3 and 1 match.
the gfort
With C++11 allocator semantics the swap() member might also replace
the allocator, which is only allowed in specific circumstances.
Even though the worst that could happen is we replace the allocator
with an equal one, we should avoid using swap and use the internal
_M_swap_data function instead.
1 - 100 of 128 matches
Mail list logo
