On August 9, 2016 4:36:24 PM GMT+02:00, Jakub Jelinek wrote:
>On Mon, Aug 01, 2016 at 12:44:30PM +0200, Richard Biener wrote:
>> Note that reading the dwarf standard, it looks like it accepts a
>location
>> which means we could do an implicit location description using
>>
On Tue, Aug 09, 2016 at 08:33:49PM +0900, Oleg Endo wrote:
> On Mon, 2016-08-08 at 13:39 -0400, Trevor Saunders wrote:
>
>
> > First sizeof std::string is 32 on x86_64, a char *, a size_t for the
> > length, and a 16 byte union of a size_t for allocated size and a 16
> > byte buffer for short
Hi,
On Tue, Aug 09, 2016 at 05:17:31PM +0530, Prathamesh Kulkarni wrote:
> On 9 August 2016 at 16:39, Martin Jambor wrote:
>
> ...
>
> >> Instead of storing arg's precision and sign, we should store
> >> parameter's precision and sign in ipa_compute_jump_functions_for_edge ().
>
The first patch corrects a remaining issue with the GCC 6 fix for ABI
tags on template member functions. We were missing tags on them, and
I fixed that by setting tags on the template, but the proper fix is to
look at the instantiation rather than the template to find the
relevant tags, since the
On 08/09/2016 06:01 AM, Andre Vieira (lists) wrote:
[snip]
The documentation is in the ARMV8-M Security Extensions in: ARM®v8-M
Security Extensions: Requirements on Development Tools document I linked
in the email above and subsequent emails
Hi Steven:
You are right, I have build arm toolchain today, it's crash when building libgcc
since there is no valid CFG in split_all_insns_noflow.
Hi Jeff:
Maybe we need split_reg_dead_p and split_reg_unused_p like peephole2
have peep2_reg_dead_p?
On Wed, Aug 10, 2016 at 3:54 AM, Steven
Hi Jakub,
On 10/08/16 07:55, Jakub Jelinek wrote:
On Wed, Aug 10, 2016 at 07:51:08AM +1000, kugan wrote:
On 10/08/16 07:46, Jakub Jelinek wrote:
On Wed, Aug 10, 2016 at 07:42:25AM +1000, kugan wrote:
There was no new regression while testing. I also moved the testcase from
This patch to the Go frontend implements the go:linkname compiler
directive. This is a special comment that permits renaming a symbol
in Go code that imports the unsafe package. This is used for various
linker tricks. This is similar to the gccgo-specific "extern"
compiler directive.
This only
On 9 August 2016 at 22:21, Paolo Bonzini wrote:
>
>
> On 09/08/2016 20:30, Manuel López-Ibáñez wrote:
>>>
>>>
>>> + cpp_opts->warn_expansion_to_defined = cpp_warn_expansion_to_defined;
>>> + if (cpp_warn_expansion_to_defined == -1)
>>> +cpp_warn_expansion_to_defined =
On Tue, Aug 9, 2016 at 6:11 PM, Jakub Jelinek wrote:
> This test is the same as the other one. Did you mean to use
> typedef int V __attribute__((vector_size (N * sizeof (int;
> V v;
> instead or something similar (where the attribute would live on
> the type rather
This patch to the Go frontend implements the go:noinline and
go:nosplit compiler directives. Bootstrapped and ran Go testsuite on
x86_64-pc-linux-gnu. Committed to mainline.
Ian
Index: gcc/go/gofrontend/MERGE
===
---
I have followup patches that use this class in selftests in various
places, so this patch moves it to selftest.h.
Successfully bootstrapped on x86_64-pc-linux-gnu.
OK for trunk?
gcc/ChangeLog:
* input.c (class selftest::temp_source_file): Move to
selftest.h.
This patch adds some initial selftesting of selftest.c/h itself.
This may seem like overdoing it, but I have some followup patches
that add non-trivial logic to selftest.c, which we should verify, so it
makes sense to test the existing functionality and then build on that.
Successfully
Successfully bootstrapped on x86_64-pc-linux-gnu.
OK for trunk?
gcc/ChangeLog:
* toplev.c (finalize): Set aux_info_file, asm_out_file, and
stack_usage_file to NULL after fclose calls.
---
gcc/toplev.c | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git
On 10/08/16 08:51, kugan wrote:
Hi Jakub,
On 10/08/16 07:55, Jakub Jelinek wrote:
On Wed, Aug 10, 2016 at 07:51:08AM +1000, kugan wrote:
On 10/08/16 07:46, Jakub Jelinek wrote:
On Wed, Aug 10, 2016 at 07:42:25AM +1000, kugan wrote:
There was no new regression while testing. I also moved
This patch to the gccgo manual documents the new options and compiler
directives. It also updates the section on function names to suggest
-fgo-pkgpath rather than -fgo-prefix; -fgo-prefix still works, but
-fgo-pkgpath is preferred. Bootstrapped on x86_64-pc-linux-gnu.
Committed to mainline.
On Tue, Aug 09, 2016 at 11:53:26PM -0400, Michael Meissner wrote:
> Index: gcc/config/rs6000/rs6000.c
> ===
> --- gcc/config/rs6000/rs6000.c(revision 239301)
> +++ gcc/config/rs6000/rs6000.c(working copy)
> @@ -7739,6
As we are starting to run the testsuite through the power9 simulator using a
compiler built with the --with-cpu=power9 configuration option, we discovered
that one of the tests (20021120-1.c) failed due to the wrong instruction being
generated. It turns out that the wY constraint (which in turn
On 08/08/16 18:48, Andrew Pinski wrote:
> On Fri, Aug 5, 2016 at 12:18 AM, Andrew Pinski wrote:
>> Hi,
>> On ThunderX, load (and store) pair that does a pair of two word
>> (32bits) load/stores is slower in some cases than doing two
>> load/stores. For some internal
ping https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01867.html
Thanks,
Prathamesh
On 28 July 2016 at 19:05, Prathamesh Kulkarni
wrote:
> On 8 June 2016 at 19:53, Richard Biener wrote:
>> On Fri, 3 Jun 2016, Jim Wilson wrote:
>>
>>> On Mon, May
On 9 August 2016 at 16:39, Martin Jambor wrote:
> Hi,
>
> On Tue, Aug 09, 2016 at 01:41:21PM +0530, Prathamesh Kulkarni wrote:
>> On 8 August 2016 at 19:33, Martin Jambor wrote:
>> >> >> +class ipcp_bits_lattice
>> >> >> +{
>> >> >> +public:
>> >> >> + bool
On Tue, Aug 9, 2016 at 11:20 AM, Yuri Rumyantsev wrote:
> Yes it is impossible since all basic blocks are handled from outer
> loops to innermost so we can not have the sequence with wrong
> dependence, i.e. we cached that reference is independent (due to
> safelen) but the
On Thu, Aug 04, 2016 at 10:47:27AM -0600, Jeff Law wrote:
> On 08/04/2016 06:36 AM, Michael Matz wrote:
> > Hi,
> >
> > On Wed, 27 Jul 2016, Marek Polacek wrote:
> >
> > > And this is the rest. Either I just adjusted a falls through comment,
> > > or I added __builtin_fallthrough (). These
On Tue, Aug 9, 2016 at 4:51 AM, Kugan Vivekanandarajah
wrote:
> Hi Richard,
>
> Thanks for the review.
>
> On 29 April 2016 at 20:47, Richard Biener wrote:
>> On Sun, Apr 17, 2016 at 1:14 AM, kugan
>>
Hi,
On Tue, Aug 09, 2016 at 01:41:21PM +0530, Prathamesh Kulkarni wrote:
> On 8 August 2016 at 19:33, Martin Jambor wrote:
> >> >> +class ipcp_bits_lattice
> >> >> +{
> >> >> +public:
> >> >> + bool bottom_p () { return lattice_val == IPA_BITS_VARYING; }
> >> >> + bool top_p
Richard,
The patch proposed by you does not work properly for
g++.dg/vect/pr70729-nest.cc test since the reference for this->S_n has
been cached as dependent for outer loop and loop is not vectorized:
g++ -Ofast -fopenmp -mavx2 pr70729-nest.cc -c
-fdump-tree-vect-details
grep 'LOOP VECTORIZED'
ping https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01869.html
Thanks,
Prathamesh
On 28 July 2016 at 19:07, Prathamesh Kulkarni
wrote:
> Hi,
> The following patch adds test-cases for divmod transform.
> I separated the SImode tests into separate file from DImode
Richard,
I checked that this move helps.
Does it mean that I've got approval to integrate it to trunk?
2016-08-09 14:33 GMT+03:00 Richard Biener :
> On Tue, Aug 9, 2016 at 1:26 PM, Yuri Rumyantsev wrote:
>> Richard,
>>
>> The patch proposed by you
On Tue, Aug 09, 2016 at 12:18:11PM +0200, Marek Polacek wrote:
> > > On Wed, 27 Jul 2016, Marek Polacek wrote:
> > >
> > > > And this is the rest. Either I just adjusted a falls through comment,
> > > > or I added __builtin_fallthrough (). These were the cases where I was
> > > > fairly sure
ping https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01969.html
Thanks,
Prathamesh
On 29 July 2016 at 22:13, Prathamesh Kulkarni
wrote:
> On 29 July 2016 at 05:40, Prathamesh Kulkarni
> wrote:
>> On 28 July 2016 at 20:14, Ramana
Hi Sandro,
As far as I can see, this is OK barring a couple of minor wrinkles and
a question:
For coarray_failed_images_err.f90 and coarray_image_status_err.f90 you
have used the option -fdump-tree-original without making use of the
tree dump.
Mikael asked you to provide an executable test with
Hi.
As mention in [1], enabling -fprofile-update=atomic when -pthread is logical
thing and is quite expected default behavior.
Ready for trunk?
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58306#c21
>From 7135fcf5f7b8f2f3c55a6e80048660c1beea5052 Mon Sep 17 00:00:00 2001
From: marxin
On Mon, 2016-08-08 at 13:39 -0400, Trevor Saunders wrote:
> First sizeof std::string is 32 on x86_64, a char *, a size_t for the
> length, and a 16 byte union of a size_t for allocated size and a 16
> byte buffer for short strings. I suppose some of this is required by
> the C++ standard,
I
On Tue, Aug 9, 2016 at 1:26 PM, Yuri Rumyantsev wrote:
> Richard,
>
> The patch proposed by you does not work properly for
> g++.dg/vect/pr70729-nest.cc test since the reference for this->S_n has
> been cached as dependent for outer loop and loop is not vectorized:
>
> g++
On 08/08/16 05:19, Sandra Loosemore wrote:
> On 07/25/2016 07:17 AM, Andre Vieira (lists) wrote:
>> [PATCH 0/7, GCC, V8M] ARMv8-M Security Extensions
>>
>> Hello,
>>
>> This is a respin of a previous patch series for ARMv8-M Security
>> Extensions. In this version I have removed one patch, rebased
On Tue, Aug 09, 2016 at 08:53:54AM +0200, Richard Biener wrote:
> Hmm, I don't think we should see -0.0 as +0.0 with -fno-signed-zeros.
> As far as I can see this is a memory load/store op and we may not
> transform, say,
>
> double x = a[i];
> b[i] = x;
>
> into a copy that changes -0.0 to
On Tue, 9 Aug 2016, Jakub Jelinek wrote:
> On Tue, Aug 09, 2016 at 08:53:54AM +0200, Richard Biener wrote:
> > Hmm, I don't think we should see -0.0 as +0.0 with -fno-signed-zeros.
> > As far as I can see this is a memory load/store op and we may not
> > transform, say,
> >
> > double x =
On 8 August 2016 at 19:33, Martin Jambor wrote:
> Hi,
>
> thanks for following through. You'll need an approval from Honza, but
> I think the code looks good (I have looked at the places that I
> believe have changed since the last week). However, I have discovered
> one new
On 08/06/2016 05:34 AM, ayush goel wrote:
> On 5 August 2016 at 4:09:00 AM, Pedro Alves (pal...@redhat.com) wrote:
>> On 08/02/2016 12:38 AM, Manuel López-Ibáñez wrote:
>>>
>>> If there is something wrong or missing, ideally we would like to know
>>> so that Ayush can work on fixing it before the
Richard Biener writes:
> The following patch avoids overaligned types created from IPA parameter
> replacement. It is said to help mipsel which still suffers from the
> backend-looks-at-type-alignment-for-parameter-passing-ABI bug.
>
> Bootstrap and regtest pending on
On Mon, Aug 8, 2016 at 7:04 PM, Jeff Law wrote:
> On 08/08/2016 05:58 AM, Marek Polacek wrote:
>>
>> On Mon, Aug 08, 2016 at 12:56:20PM +0100, Roger Sayle wrote:
>>>
>>>
>>> The following patch is an attempt to finally fully close PR
>>> middle-end/21137.
>>> As explained in the
Hello!
syscall test fails on CentOS 5.11 with:
exec_linux_test.go:167:25: error: reference to undefined identifier
'syscall.CLONE_NEWNET'
Unshareflags: syscall.CLONE_NEWNET,
^
FAIL: syscall
Attached patch fixes the failure by providing CLONE_NEWNET definition.
The fallback part of HARD_REGNO_CALLER_SAVE_MODE, choose_hard_reg_mode,
returns DFmode for SImode when TARGET_E500_DOUBLE. This confuses
lra when attempting to save ctr around a call.
Arseny, the bug reporter, has regression tested this patch.
OK to apply?
PR target/71680
*
On Tue, 9 Aug 2016, Prathamesh Kulkarni wrote:
> On 8 August 2016 at 19:33, Martin Jambor wrote:
> > Hi,
> >
> > thanks for following through. You'll need an approval from Honza, but
> > I think the code looks good (I have looked at the places that I
> > believe have changed
On Mon, 8 Aug 2016, Bernd Edlinger wrote:
> Hi!
>
>
> As described in https://gcc.gnu.org/bugzilla/show_bug.cgi?id=71083
> we generate unaligned accesses because tree-predcom rewrites
> bitfield and packed accesses in a way that looses the proper
> alignment information.
>
> The attached patch
On Wed, 3 Aug 2016, Martin Liška wrote:
> I've installed (r239066) the script which is used by maintainers to
> update PRs in a batch mode.
I think it would be good to add a comment at the top that
describes what the scripts does and how to invoke the script?
Gerald
On Fri, 5 Aug 2016, Jeff Law wrote:
> On 08/05/2016 07:36 AM, Richard Biener wrote:
> > @@ -560,6 +562,14 @@ fsm_find_control_statement_thread_paths
> > edge taken_edge = profitable_jump_thread_path (path, bbi, name,
> > arg);
> > if (taken_edge)
> > {
> > +
On Mon, Aug 8, 2016 at 11:58 AM, Andre Vehreschild wrote:
> Hi Paul,
>
> thanks for the review. Committed to trunk as r239237.
>
> I will commit to gcc-6 in one week.
If you want to catch the 6.2 release then please make sure to have it on the
branch this Friday.
Richard.
>
On Mon, 8 Aug 2016, Prathamesh Kulkarni wrote:
> Hi Richard,
> In the attached version, I tried to address your suggestions from:
> https://gcc.gnu.org/ml/gcc-patches/2016-08/msg00279.html
>
> In ccp_finalize we do:
> wide_int nonzero_bits = wide_int::from (val->mask, precision,
>
On Sat, 6 Aug 2016, Prathamesh Kulkarni wrote:
> Hi Richard,
> The patch for PR71078 broke the test-cases for arm and aarch64 bare
> metal targets :/
> In the attached patch, restricting the tests to c99_runtime.
> Sorry for the breakage.
> Ok for trunk ?
Ok.
Richard.
> Thanks,
> Prathamesh
>
On Mon, 8 Aug 2016, Jakub Jelinek wrote:
> Hi!
>
> Only +0.0 stores can be optimized into memset, -0.0 can't, so if we are
> honoring signed zeros, we should make sure the constant is positive.
>
> Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
Hmm, I don't think we
The following fixes a missed CFG cleanup opportunity (requiring its
iteration) that ultimatively causes us to create a dead if-converted
loop variant (discovered as dead by CFG cleanup run by ifcvt).
The issue is that when removing unreachable code CFG cleanup does not
schedule predecessors of
On Fri, 5 Aug 2016, Kyrill Tkachov wrote:
>
> On 01/08/16 15:31, Kyrill Tkachov wrote:
> >
> > On 11/07/16 18:55, Bernd Schmidt wrote:
> > > On 07/11/2016 04:52 PM, Kyrill Tkachov wrote:
> > > > Based on that, I think that code block is a useful optimisation, we just
> > > > need
> > > > to
Another small follow-up changes instrumentation of MOD exprs
to not instrument divisors different from SSA_NAME.
Patch survives:
make check -k -j10 RUNTESTFLAGS="tree-prof.exp"
Ready for trunk?
Thanks,
Martin
>From 00aecc0dd74c4546a1882bdbbb0f66fbf39a5408 Mon Sep 17 00:00:00 2001
From: marxin
On Tue, Aug 02, 2016 at 11:02:56PM +0930, Alan Modra wrote:
> This is a patch for a problem in lra, triggered by the rs6000
> backend not allowing SImode in floating point registers.
Ping? https://gcc.gnu.org/ml/gcc-patches/2016-08/msg00113.html
Note that to recreate the problem with the
On Tue, Aug 9, 2016 at 12:58 AM, kugan
wrote:
> Hi Jakub,
>
> Thanks for the review.
>
> On 08/08/16 16:40, Jakub Jelinek wrote:
>>
>> On Mon, Aug 08, 2016 at 01:36:51PM +1000, kugan wrote:
>>>
>>> diff --git a/gcc/tree-ssanames.h b/gcc/tree-ssanames.h
>>> index
On 08/09/16 07:24, Martin Liška wrote:
Hi.
As mention in [1], enabling -fprofile-update=atomic when -pthread is logical
thing and is quite expected default behavior.
Ready for trunk?
[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=58306#c21
This certainly requires changes to invoke.texi,
On 08/05/2016 01:55 PM, Richard Biener wrote:
Hi Richard.
Please don't use std::string. For string building you can use obstacks.
Alright let's talk details then so I can write things up in a way you
approve of.
Take for instance simple uses like all the tree_*check_failed routines,
On 08/09/2016 03:17 PM, Aldy Hernandez wrote:
On 08/05/2016 01:55 PM, Richard Biener wrote:
Hi Richard.
Please don't use std::string. For string building you can use obstacks.
Alright let's talk details then so I can write things up in a way you
approve of.
Take for instance simple uses
Hi,
The test-case in PR72835 is failing with -O2 and passing with
-fno-tree-reassoc. It also passes with -O2 -fno-tree-vrp.
diff of .115t.dse2 and .116t.reassoc1 for the c++ testcase is as
follows, which looks OK.
+ unsigned int _16;
+ unsigned int _17;
+ unsigned int _18;
:
_1 =
On 08/08/2016 05:26 PM, Senthil Kumar Selvaraj wrote:
I picked out the commit where you'd added SYMBOL_REF handling (rev
#190252), and patched that with the below code. Bootstrapped and
regtested on x86_64-pc-linux - the results were identical with and
without the patch. Is this good enough for
On 08/08/16 12:56, Martin Liška wrote:
Hello.
Currently, we utilize pow2 profile histogram to track gimple STMTs like this:
ssa_name_x % value.
void
__gcov_pow2_profiler (gcov_type *counters, gcov_type value)
{
if (value & (value - 1))
counters[0]++;
else
counters[1]++;
}
Hi!
Due to my recent cleanup of special_function_p the darwin bootstrap is
currently broken because the longjmp function misses the noreturn
attribute. Therefore I revived an old solaris_longjmp_noreturn
fixinclude rule.
See the attached patch.
Boot-strap on i686-apple-darwin11 and
On 08/09/16 04:41, Martin Liška wrote:
Another small follow-up changes instrumentation of MOD exprs
to not instrument divisors different from SSA_NAME.
Patch survives:
make check -k -j10 RUNTESTFLAGS="tree-prof.exp"
ok, thanks
On 08/08/16 13:03, Martin Liška wrote:
v3: fixed wrong defines in libgcc/Makefine.in
ok, thanks
On Fri, 5 Aug 2016, Jeff Law wrote:
> On 08/05/2016 07:36 AM, Richard Biener wrote:
> > @@ -560,6 +562,14 @@ fsm_find_control_statement_thread_paths
> > edge taken_edge = profitable_jump_thread_path (path, bbi, name,
> > arg);
> > if (taken_edge)
> > {
> > +
On Tue, Aug 09, 2016 at 12:24:03PM +0200, Jakub Jelinek wrote:
> On Tue, Aug 09, 2016 at 12:18:11PM +0200, Marek Polacek wrote:
> > > > On Wed, 27 Jul 2016, Marek Polacek wrote:
> > > >
> > > > > And this is the rest. Either I just adjusted a falls through comment,
> > > > > or I added
On Tue, Aug 9, 2016 at 2:05 PM, Yuri Rumyantsev wrote:
> Richard,
>
> I checked that this move helps.
> Does it mean that I've got approval to integrate it to trunk?
Yes, if it survives bootstrap & regtest.
Richard.
> 2016-08-09 14:33 GMT+03:00 Richard Biener
I have applied the MIPS specific fix for PR65345. It is a mechanical
change. This fixes the MIPS failures in atomic/pr65345-4.c.
gcc/
PR c/65345
* config/mips/mips.c (mips_atomic_assign_expand_fenv):
Use create_tmp_var_raw instead of create_tmp_var.
Thanks,
Matthew
ayush goel writes:
> This patch https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01302.html
> is still pending review.
The build portions are OK, assuming a global maintainer has approved of
adding gnulib in general.
Tested on Linux-x64.
2016-08-09 Ville Voutilainen
Implement LWG 2744 and LWG 2754.
* include/std/any (any(ValueType&&)): Constrain with __is_in_place.
(any(in_place_type_t<_ValueType>, _Args&&...)): Use _Decay.
(any(in_place_type_t<_ValueType>,
On Fri, Aug 5, 2016 at 11:48 AM, Richard Biener wrote:
>
> This fixes PR72772 by avoing placing a degenerate PHI in each
> forwarder block loop init creates when creating simple preheaders.
> The solution is to simply split the single loop entry edge which
> is also way cheaper
On Wed, Aug 03, 2016 at 10:56:08AM -0600, Jeff Law wrote:
> On 07/27/2016 10:53 AM, Marek Polacek wrote:
> > These are the cases where I wasn't sure if the falls through were
> > intentional
> > or not.
> >
> > This patch has been tested on powerpc64le-unknown-linux-gnu,
> > aarch64-linux-gnu,
On Thu, Jul 28, 2016 at 10:43:25PM +, Joseph Myers wrote:
> On Tue, 19 Jul 2016, James Greenhalgh wrote:
>
> > These slightly complicate the description you give above as we now want
> > two behaviours. Where the 16-bit floating point extensions are available,
> > we want to use the native
Testing with -Wimplicit-fallthrough turned up a few spots where
fall through is either a bug, or just too confusing. This patch
should be desirable even though the warning is still not in.
Eric, can you look at the c-ada-spec.c part?
Martin, does the hsa-gen.c part look correct?
What about the
This patch to the Go frontend rewrites the compiler directive support
to recognize all the compiler directives handled by the current gc
compiler. They are now turned into flags attached to a function or
function declaration, or, for the case of go:linkname, into a map
attached to the Lex object.
On Thu, Mar 28, 2013 at 2:14 PM, Jason Merrill wrote:
> When 'this' appears in an expression, it should be an rvalue rather than a
> const lvalue; in C++11, the distinction matters.
And likewise for &*this.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit
The first patch fixes a constexpr issue whereby we were complaining
about passing a non-constant empty class to a constexpr function; this
is OK because there's no actual copying involved.
The second patch implements the C++17 constexpr lambda proposal. This
was mostly pretty straightforward. A
Hi,
when emitting a BRIG function definition, we need to fixup the leading
directive after all of the instructions and everything has been added to
the section (well, we could pre-compute the value but it is not worth
it). At the moment the function emitting code calculates the pointer in
an
Hi,
since the HSA branch switched to a simple obstack from alloc-pools, it
is not necessary to have a separate new operator for each instruction
kind and it is enough if the common ancestor has one. Therefore, this
patch removes them. Similarly, I learned that it is enough to make the
delete
Hi,
the HSA/BRIG back-end does not have any support for CONST_DECSs so far.
This patch adds it, emitting readonly segment symbols with initializers
for them. One complication is that addresses pertaining to HSA readonly
segment cannot be converted to flat addresses. So when we need an
address
Hello.
Following enhancement for gcov solves issues when we cannot create a file due
to a filesystem
path length limit. Selected approach utilizes existing md5sum functions.
Patch survives make check -k RUNTESTFLAGS="gcov.exp" on x86_64-linux-gnu.
Ready for trunk?
Thanks,
Martin
>From
MIPS creates more invariants than gcc.dg/loop-8.c expects. There is
no way to represent what the original test was trying to verify given
the additional invariants being moved out of the loop.
Thanks,
Matthew
gcc/
PR rtl-optimization/9
* gcc.dg/loop-8.c: Skip for MIPS due to
Hi, this patch makes match.pd to shorten comparisons that involve a
sign-changing cast from a shorter unsigned type to a wider signed type.
This should be safe to do because a wider signed type can hold all the
possible values of a shorter unsigned type.
This change causes vrp23.c and vrp24.c to
On Mon, Aug 08, 2016 at 12:10:00PM +0100, Bin.Cheng wrote:
> On Mon, Aug 8, 2016 at 11:40 AM, James Greenhalgh
> wrote:
> > On Mon, Aug 01, 2016 at 01:18:53PM +, Bin Cheng wrote:
> >> Hi,
> >> This is the 3rd version patch implementing vcond_mask and vec_cmp patterns
Hi,
in the first pre-release HSAIL/BRIG specifications, registers had a type
which they now don't, they only have size. We have thought about
removing it from our internal representation but it is actually useful
to know what the type of the stored value is when it is actually
necessary to
Hi,
the type is built twice and the first one is not used. This patch
removes it.
Thanks,
Martin
2016-08-04 Martin Jambor
* hsa-brig.c (hsa_output_kernels): Remove unnecessary building of
kernel_dependencies_vector_type.
---
gcc/hsa-brig.c | 6 +-
1
Hi,
the rest of hsa code uses direct pointers, which I find preferable (and
since we have moved to explicit pointers also for example with the
gimple type, I believe I am not alone, so I removed the two _p typedefs
we had in hsa.h. They were not in widespread use anyway.
Thanks,
Martin
Hi,
HSA backend now uses ggc_strdup to allocate name strings for stuff that
doesn't have any and therefore we rely on the garbage collector not
running. That works but is not nice and since the HSA branch now has an
obstack, not necessary. This patch changes the allocation to be from
the
On Mon, Aug 01, 2016 at 12:44:30PM +0200, Richard Biener wrote:
> Note that reading the dwarf standard, it looks like it accepts a location
> which means we could do an implicit location description using
> DW_OP_stack_value which gives us access to arbitrary dwarf
> expressions (and thus the
On Fri, Jul 29, 2016 at 11:47:18PM +0800, Chung-Lin Tang wrote:
> On 2016/7/21 07:13 PM, Jakub Jelinek wrote:
> > Better put every && on a separate line if the whole if (...) doesn't fit
> > on a single line.
> >
> >> > + && !n->sym->attr.cray_pointer
> >> > +
Hello!
Insn predicates have to allow all constraints during LRA, not only
after reload_completed is true.
2016-08-09 Uros Bizjak
PR target/72843
* config/i386/i386.md (*movtf_internal): Use
lra_in_progress || reload_completed instead of !can_create_pseudo_p
libgo test results in gcc-testresults for ppc64le & ppc64 are back to
what they were before the change.
Thanks.
On 08/08/2016 03:34 PM, Ian Lance Taylor wrote:
On Mon, Aug 8, 2016 at 11:14 AM, Lynn A. Boger
wrote:
The libgo tests on ppc64le and ppc64 have all
On 08/01/2016 08:29 AM, Thomas Schwinge wrote:
> On Mon, 01 Aug 2016 17:21:37 +0200, I wrote:
>> Some checking of OpenACC clauses currently done in the front ends should
>> be moved later, and be unified. (Also, I suppose, for supporting of the
>> device_type clause, such checking actually
clang recently added a new warning -Wexpansion-to-defined, which
warns when `defined' is used outside a #if expression (including the
case of a macro that is then used in a #if expression).
While I disagree with their inclusion of the warning to -Wall, I think
it is a good addition overall.
On Tue, Aug 02, 2016 at 09:16:00PM -0700, Cesar Philippidis wrote:
> 2016-08-02 Cesar Philippidis
>
> gcc/fortran/
> * openmp.c (resolve_oacc_data_clauses): Emit a warning about a
> potentially non-contiguous array pointer.
>
On Tue, Aug 09, 2016 at 05:21:26PM +0200, Marek Polacek wrote:
> Bootstrapped/regtested on x86_64-linux and ppc64le-redhat-linux.
>
> 2016-08-09 Marek Polacek
>
> PR c/7652
> gcc/c-family/
> * c-ada-spec.c (dump_generic_ada_node): Add return.
> gcc/
> *
On Sat, Aug 6, 2016 at 9:20 PM, Andreas Schwab wrote:
> On Mi, Jul 13 2016, "Bin.Cheng" wrote:
>
>> Patch re-tested/applied on trunk as r238301.
>
> This breaks gcc.dg/vect/vect-117.c on powerpc.
Hi Andreas,
Sorry for the inconvenience, I will have a
Hi,
On Tue, Aug 09, 2016 at 05:21:26PM +0200, Marek Polacek wrote:
> Testing with -Wimplicit-fallthrough turned up a few spots where
> fall through is either a bug, or just too confusing. This patch
> should be desirable even though the warning is still not in.
>
> Eric, can you look at the
On Wed, Jul 27, 2016 at 06:53:39PM +0200, Marek Polacek wrote:
> --- gcc/gcc/var-tracking.c
> +++ gcc/gcc/var-tracking.c
> @@ -1056,6 +1056,8 @@ adjust_mems (rtx loc, const_rtx old_rtx, void *data)
>? GET_MODE_SIZE (amd->mem_mode)
>
1 - 100 of 125 matches
Mail list logo