Hi Cesar!
On Mon, 20 Feb 2017 20:42:59 -0800, Cesar Philippidis
wrote:
> This patch introduces a new -finform-parallelism flag to report any
> detected parallelism encountered by the compiler. Initially, it's being
> used to report how oaccdevlow partitions OpenACC loops. Currently, if
> you wan
On Tue, 21 Feb 2017, Jeff Law wrote:
> On 02/21/2017 02:01 AM, Richard Biener wrote:
> > On Mon, 20 Feb 2017, Marc Glisse wrote:
> >
> > > On Mon, 20 Feb 2017, Jakub Jelinek wrote:
> > >
> > > > As mentioned by Jason in the PR, we've regressed on the following
> > > > testcase
> > > > since we s
Hi,
On Wed, Feb 22, 2017 at 08:58:06AM +0100, Thomas Schwinge wrote:
> >
> > Rename the "openmp" group of optimizations to "omp"
> >
> > gcc/
> > * dumpfile.h (OPTGROUP_OPENMP): Rename to OPTGROUP_OMP. Adjust
> > all users.
> > * dumpfile.
I am testing the following patch to fix PR79666 - we may not create
a symbolic range with expressions that may introduce (not already present)
undefined overflow.
With a new predicate like operand_not_min_p we can possibly improve
this for symbolic ops with a useful range, but I'm not sure we'd
e
Hello,
this is a fix for PR 78140 which is about LTO WPA of Firefox taking
1GB memory more than gcc 6.
It works by reusing the ipa_bits and value_range that we previously
had directly in jump functions and which are just too big to be
allocated for each actual argument in all of Firefox. Reusing
On 21/02/17 19:53, Matthew Fortune wrote:
> Eric Botcazou writes:
>>> Agreed. I don't think things like WORD_MODE_OPERATIONS should change
>>> rtl semantics, just optimisation decisions.
>
> Sorry, I did cover two topics in one email. My point was about whether
> simplifying:
>
> (subreg:OUTER
Hi,
I've implemented -Wstring-plus-int (like the one in CLANG) for GCC. I've
bootstrapped GCC with my patches and runned dejaGNU tests.
My patches are for both C and C++ (like CLANG).
Unlike CLANG, this option is not enabled by default (or GCC won't
bootstrap with -Werror).
Xi Ruoyao (5):
Mo
This patch moves prototype of char_type_p into c-common.h, so we can
use it in c-family/*.
gcc/ChangeLog:
2017-02-22 Xi Ruoyao
* c-family/c-common.h (char_type_p): New prototype
* c/c-typeck.c (char_type_p): Make the function extern
* cp/cp-tree.h (char_type_p): Remove
This patch adds warning option -Wstring-plus-int for C++.
gcc/ChangeLog:
2017-02-22 Xi Ruoyao
* c-family/c-common.h (maybe_warn_string_plus_int): new prototype
* c-family/c-warn.h (maybe_warn_string_plus_int): new function
* c-family/c-opt: new option -Wstring-plus-int
This patch adds warning option -Wstring-plus-int for C.
gcc/ChangeLog:
2017-02-22 Xi Ruoyao
* c/c-typeck.c (parser_build_binary_op, build_modify_expr): checking
for -Wstring-plus-int
---
gcc/c/c-typeck.c | 28
1 file changed, 28 insertions(+)
diff --gi
This patch adds document of -Wstring-plus-int.
gcc/ChangeLog:
2017-02-22 Xi Ruoyao
* doc/invoke.texi (Warning Options): Document -Wstring-plus-int
---
gcc/doc/invoke.texi | 8 +++-
1 file changed, 7 insertions(+), 1 deletion(-)
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.
This patch adds tests for -Wstring-plus-int.
gcc/ChangeLog:
2017-02-22 Xi Ruoyao
* testsuite/c-c++-common/Wstring-plus-int.c: New test
* testsuite/g++.dg/Wstring-plus-int-1.C: New test
* testsuite/g++.dg/Wstring-plus-int-2.C: New test
* testsuite/gcc.dg/Wstring
> I will keep working on this code post GCC 7 as the topic is bugging me now
> :-)
I'm OK to lend a hand, but what's left for GCC 7 to make MIPS happy?
--
Eric Botcazou
> Eric - you added these improvements so you might want to double-check
> effects on Ada code (and see if the above suggestion is worth
> the effort).
I think that all the relevant testcases are in the gnat.dg testsuite.
--
Eric Botcazou
The list of ARM CPUs in readings.html is now looking rather dated. This
patch updates the list to refer to the most commonly available parts (or
at least, the product families by which they are known).
Committed.
R.
? backends.html~
? index.html.~1.866.~
? readings.html.~1.260.~
? svn.html.~1.18
> > I will keep working on this code post GCC 7 as the topic is bugging me
> now
> > :-)
>
> I'm OK to lend a hand, but what's left for GCC 7 to make MIPS happy?
I have just successfully bootstrapped MIPS with just the pending (amended)
patch (3). i.e. with this patch (1) reverted so I did not ne
> Le 21 févr. 2017 à 23:48, Alan Modra a écrit :
>
> On Sat, Feb 18, 2017 at 12:39:08PM +0100, Dominique d'Humières wrote:
>>> I'm slightly concerned about the test and how it'll behave on targets with
>>> small address spaces. If it's a problem we can fault in adjustments.
>>
>> The test fail
On Tue, Feb 21, 2017 at 3:49 PM, Jan Hubicka wrote:
>> 2017-02-21 Bin Cheng
>>
>> PR tree-optimization/77536
>> * tree-ssa-loop-manip.c (niter_for_unrolled_loop): New function.
>> (tree_transform_and_unroll_loop): Use above function to compute the
>> estimated niter of unrolled loop and use it
Hi,
This is a followup patch fixing a left over issue in vect_do_peeling. In the
first patch,
I used scale_loop_frequencies to scale up profiling counter for epilog_loop
because
scale_loop_profile doesn't allow scaling up for the now. This patch adds
another call
to scale_loop_profile after sc
> Hi Honza,
OK,
thanks!
Honz
> There is the 3rd version patch fixing mentioned issues. Is it OK?
>
> Thanks,
> bin
> diff --git a/gcc/testsuite/gcc.dg/vect/pr79347.c
> b/gcc/testsuite/gcc.dg/vect/pr79347.c
> index 586c638..6825420 100644
> --- a/gcc/testsuite/gcc.dg/vect/pr79347.c
> +++ b/gcc/
The following fixes PR79673.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2017-02-22 Richard Biener
PR tree-optimization/79673
* tree-ssa-pre.c (compute_avail): Use wide_int_to_tree to
convert the [TARGET_]MEM_REF offset INTEGER_CST
On 02/22/2017 04:00 AM, Xi Ruoyao wrote:
Hi,
I've implemented -Wstring-plus-int (like the one in CLANG) for GCC. I've
bootstrapped GCC with my patches and runned dejaGNU tests.
My patches are for both C and C++ (like CLANG).
Unlike CLANG, this option is not enabled by default (or GCC won't
boo
Oh great thanks!
Best regards,
Thomas
On 17/02/17 22:52, JonY wrote:
On 02/17/2017 11:31 AM, Thomas Preudhomme wrote:
Here you are:
2017-01-24 Thomas Preud'homme
* configure.ac (--enable-mingw-wildcard): Add new configurable
feature.
* configure: Regenerate.
* con
On 02/22/2017 04:32 AM, Dominique d'Humières wrote:
Le 21 févr. 2017 à 23:48, Alan Modra a écrit :
On Sat, Feb 18, 2017 at 12:39:08PM +0100, Dominique d'Humières wrote:
I'm slightly concerned about the test and how it'll behave on targets with
small address spaces. If it's a problem we can
> Le 22 févr. 2017 à 17:06, Jeff Law a écrit :
>
> On 02/22/2017 04:32 AM, Dominique d'Humières wrote:
>>
>>> Le 21 févr. 2017 à 23:48, Alan Modra a écrit :
>>>
>>> On Sat, Feb 18, 2017 at 12:39:08PM +0100, Dominique d'Humières wrote:
> I'm slightly concerned about the test and how it'll
On Tue, Feb 21, 2017 at 12:37:32PM -0800, Jason Merrill wrote:
> On Sat, Feb 18, 2017 at 12:17 AM, Jakub Jelinek wrote:
> > On Fri, Feb 17, 2017 at 09:37:09PM -0500, Jason Merrill wrote:
> >> On Fri, Feb 17, 2017 at 1:52 PM, Jakub Jelinek wrote:
> >> > - && die->die_tag != DW_TAG_member)
> >
On 02/21/2017 02:23 PM, Jakub Jelinek wrote:
Hi!
As reported by Uros, the
fld a
fld b
fxchg %st(1)
optimization to
fld b
fld a
misses several important cases, one is FLOAT_EXTEND memory loads where
the memory is SFmode or DFmode but we extend it to a wider mode, and
the other is when we load a k
On 02/16/2017 12:41 PM, Bernd Schmidt wrote:
We have two registers being assigned to each other:
(set (reg 213) (reg 209))
(set (reg 209) (reg 213))
These being the only definitions, we are happy to forward propagate reg
209 for reg 213 into a third insn, making a new use for reg 209. We are
Hi,
As discussed in https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68644,
the test case gcc.dg/tree-ssa/ivopts-lt-2.c does not apply to POWER
targets, as the cost model properly shows the pre-increment approach
to be preferable and results in ideal code generation. Thus, adding
powerpc*-*-* to the
ENOPATCH. Trying again...
Hi,
As discussed in https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68644,
the test case gcc.dg/tree-ssa/ivopts-lt-2.c does not apply to POWER
targets, as the cost model properly shows the pre-increment approach
to be preferable and results in ideal code generation. Thus,
On 02/22/2017 10:54 AM, Bill Schmidt wrote:
Hi,
As discussed in https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68644,
the test case gcc.dg/tree-ssa/ivopts-lt-2.c does not apply to POWER
targets, as the cost model properly shows the pre-increment approach
to be preferable and results in ideal code
Thanks, Jeff -- I caught it as soon as I hit the stupid button... :)
> On Feb 22, 2017, at 11:56 AM, Jeff Law wrote:
>
> On 02/22/2017 10:54 AM, Bill Schmidt wrote:
>> Hi,
>>
>> As discussed in https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68644,
>> the test case gcc.dg/tree-ssa/ivopts-lt-2.c do
On Tue, Feb 21, 2017 at 4:27 PM, Martin Sebor wrote:
>> Ah, I see, your patch changes attribute unused handling for local
>> variables from tracking TREE_USED to lookup_attribute. I'm not
>> opposed to this change, but I'd like to understand why the TREE_USED
>> handling wasn't working.
>
>
> In
On 02/21/2017 03:44 AM, Martin Liška wrote:
Hello.
To make documentation and output of --help consistent, I suggest to replace
inequality
signs with square brackets.
Ready to be installed?
OK.
jeff
The problem here is that we aren't properly checking what's been passed to
__underlying_type. It needs an enum, which the code checks, but that enum
mustn't be incomplete, otherwise we crash because ENUM_UNDERLYING_TYPE will
be null until the definition of the enum is completed by finish_enum.
Bo
We ICEd on the following test while warning for -Wtraditional-conversion
because at that point VAL had been turned into error_mark_node because we
require a complete type when converting arguments. Fixed by checking
for error_mark_node first so that we don't try to access it later with
TYPE_PRECIS
This patch oughta fix a couple of ice-on-invalids.
The parser.c part: I think it makes no sense for cp_parser_std_attribute_spec
to create an attribute if something went wrong, e.g. in make_pack_expansion, so
return error_mark_node to signal an issue.
The pt.c part: tsubst_pack_expansion can retu
OK.
On Wed, Feb 22, 2017 at 10:53 AM, Marek Polacek wrote:
> This patch oughta fix a couple of ice-on-invalids.
>
> The parser.c part: I think it makes no sense for cp_parser_std_attribute_spec
> to create an attribute if something went wrong, e.g. in make_pack_expansion,
> so
> return error_mar
Hmm, I suppose we don't need to allow this in SFINAE context. OK.
On Wed, Feb 22, 2017 at 10:24 AM, Marek Polacek wrote:
> The problem here is that we aren't properly checking what's been passed to
> __underlying_type. It needs an enum, which the code checks, but that enum
> mustn't be incomple
OK.
On Wed, Feb 22, 2017 at 8:57 AM, Jakub Jelinek wrote:
> On Tue, Feb 21, 2017 at 12:37:32PM -0800, Jason Merrill wrote:
>> On Sat, Feb 18, 2017 at 12:17 AM, Jakub Jelinek wrote:
>> > On Fri, Feb 17, 2017 at 09:37:09PM -0500, Jason Merrill wrote:
>> >> On Fri, Feb 17, 2017 at 1:52 PM, Jakub Je
Eric Botcazou writes:
> > The condition would look like this, What do you think?
> >
> > if (!(GET_MODE_PRECISION (mode) != GET_MODE_PRECISION (innermode)
> > && GET_MODE_SIZE (mode) <= UNITS_PER_WORD
> > && GET_MODE_SIZE (innermode) <= UNITS_PER_WORD
> >
This patch allows lra to reload a lo_sum address. Given a mem with a
lo_sum address as used by ppc32, decompose_mem_address returns an
address_info with *base the lo_sum and *base_term the reg within the
lo_sum. When a lo_sum address isn't valid, we want to first try
reloading the entire lo_sum t
On Wed, 22 Feb 2017, Marek Polacek wrote:
> We ICEd on the following test while warning for -Wtraditional-conversion
> because at that point VAL had been turned into error_mark_node because we
> require a complete type when converting arguments. Fixed by checking
> for error_mark_node first so th
Hi!
If both arguments of integer division or modulo are known to be non-negative
in corresponding signed type, then signed as well as unsigned division/modulo
shall have the exact same result and therefore we can choose between those
two depending on which one is faster (or shorter for -Os), which
Hi!
This patch does a few things:
1) replaces ICEs with diagnostics + return false in
potential_constant_expression_1 for OpenMP/OpenACC/Cilk+ constructs,
I believe unless the upcoming versions of those standards say otherwise,
we should treat the constructs as invalid in constexpr
When w
On Wed, Feb 22, 2017 at 1:50 PM, Jakub Jelinek wrote:
> This patch does a few things:
> 2) all but one error in potential_constant_expression_1 to error_at with
> location_of (t)
I'd prefer to calculate the location once at the top of the function
rather than at every call to error_at. Given tha
On Wed, Feb 22, 2017 at 01:58:28PM -0800, Jason Merrill wrote:
> On Wed, Feb 22, 2017 at 1:50 PM, Jakub Jelinek wrote:
> > This patch does a few things:
> > 2) all but one error in potential_constant_expression_1 to error_at with
> > location_of (t)
>
> I'd prefer to calculate the location once a
Someone on the C++ committee observed that
std::less l{};
didn't work in G++; this fixes that.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit d04e2de711e69266b24c118eb330100859b2671c
Author: Jason Merrill
Date: Tue Feb 21 16:50:29 2017 -0800
* pt.c (do_class_deduction): Ha
Alan Modra writes:
> This patch allows lra to reload a lo_sum address. Given a mem with a
> lo_sum address as used by ppc32, decompose_mem_address returns an
> address_info with *base the lo_sum and *base_term the reg within the
> lo_sum. When a lo_sum address isn't valid, we want to first try
>
Moore, Catherine writes:
> > As this is an ABI fix, just wait a day or so in case Catherine has
> > any comment otherwise OK to commit.
> >
> I have not further comments on this patch. Please commit.
Hi Sameera,
I've been considering this patch further and have realised that my review
was preo
On 02/22/2017 11:02 AM, Jason Merrill wrote:
On Tue, Feb 21, 2017 at 4:27 PM, Martin Sebor wrote:
Ah, I see, your patch changes attribute unused handling for local
variables from tracking TREE_USED to lookup_attribute. I'm not
opposed to this change, but I'd like to understand why the TREE_USE
On Fri, Jan 27, 2017 at 03:11:46PM +, Segher Boessenkool wrote:
> 2017-01-27 Segher Boessenkool
>
> * config/rs6000/rs6000.md (*fsel4): Use
> gpc_reg_operand instead of fpr_reg_operand.
I committed the patch.
Segher
On Wed, Feb 22, 2017 at 3:44 PM, Martin Sebor wrote:
> On 02/22/2017 11:02 AM, Jason Merrill wrote:
>>
>> On Tue, Feb 21, 2017 at 4:27 PM, Martin Sebor wrote:
Ah, I see, your patch changes attribute unused handling for local
variables from tracking TREE_USED to lookup_attribute. I
On Wed, Feb 22, 2017 at 11:01:15PM +, Richard Sandiford wrote:
> Alan Modra writes:
> > This patch allows lra to reload a lo_sum address. Given a mem with a
> > lo_sum address as used by ppc32, decompose_mem_address returns an
> > address_info with *base the lo_sum and *base_term the reg with
We need to mask off tf_cleanup when we're dealing with argument
conversions; it should only apply to the top level call.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit 833d39a7919caa021848861e2035c8a1ac3e3b51
Author: Jason Merrill
Date: Wed Feb 22 16:55:24 2017 -0800
PR c++/
The C++ FE isn't setting the expr_location of the split acc loop in
combined acc parallel/kernels loop directives. This only happens for
with combined directives, otherwise cp_parser_omp_construct would be
responsible for setting the location. After fixing this bug, I was able
to resolve a couple o
On Thu, Feb 23, 2017 at 11:41:09AM +1030, Alan Modra wrote:
> lo_sum is indeed not valid for mem:SD. simplify_operand_subreg is
> where the subreg disappears.
Richard, doesn't the following say that lra is expecting to reload
exactly the lo_sum address you seem to think it should not handle in
pr
Like its c++ counterpart, the fortran FE incorrectly records the line
locations of combined acc loop directives when it lowers the construct
to gimple. Usually this isn't a problem because the fortran FE is able
to report problems with acc loops itself. However, there will be
inaccuracies if the ME
I've applied this patch to gomp-4-0-branch to remove an unused variable
inside trans-openmp.c. I'm not sure why bootstrapping does catch this
sort of error anymore. Maybe my build script of overriding the build
flags some how.
Cesar
2017-02-22 Cesar Philippidis
gcc/fortran/
* trans-openmp.c
I merged trunk revision 245668 to the gccgo branch. I also committed
the patches that have stacked up in gofrontend but have not yet been
committed to trunk, to avoid destabilizing stage 4.
Ian
tree-ssa-dse.c needs to verify when two writes have the same base
address. Right now it uses pointer equality. The testcase in BZ79578
shows that we should have been using operand_equal_p.
This one-liner fixes that oversight. Bootstrapped and regression tested
on x86_64-linux-gnu. Instal
On 02/22/2017 02:40 PM, Jakub Jelinek wrote:
Hi!
If both arguments of integer division or modulo are known to be non-negative
in corresponding signed type, then signed as well as unsigned division/modulo
shall have the exact same result and therefore we can choose between those
two depending on
On 02/21/2017 01:09 AM, Martin Liška wrote:
On 02/21/2017 12:03 AM, Alexandre Oliva wrote:
It's convenient to debug -fcompare-debug errors. You set the param to a
large number, and then non-debug insns will get the same uid in both
debug and non-debug compilations, so it's easier to compare RTL
On 02/15/2017 03:49 AM, Aldy Hernandez wrote:
On 02/13/2017 07:15 PM, Jeff Law wrote:
So it seems in your updated patch there is only one call where we ask
for LOOP_EXIT_COMPLEX, specifically the call from get_loop_location.
But I don't see how asking for LOOP_EXIT_COMPLEX from that location
w
On 21 February 2017 17:54:23 CET, charles.bay...@linaro.org wrote:
>From: Charles Baylis
>
>+/* Convert fron bytes to ints. */
s/fron/from/
>+#define ARM_NUM_INTS(X) (((X) + UNITS_PER_WORD - 1) / UNITS_PER_WORD)
>+
>+
> /* RTX costs. Make an estimate of the cost of executing the operation
>
On 02/22/2017 09:32 AM, Dominique d'Humières wrote:
Let me stand up an i686 linux instance and see if I can twiddle things without
compromising the test.
Also darwin is a -fpic platform.
Which is the key here :-) The test works fine without PIC, but once PIC
is introduced, it blows up.
It'
66 matches
Mail list logo
