Hi!
Here is an attempt to adjust gimple_build_assign* documentation.
Ok for trunk?
Note, apparently the documentation has not been adjusted for
the gimple -> gassign * etc. changes, David, can you please work on
adjusting gimple.texi to match the reality after your changes?
2014-12-02 Jakub Jel
On Tue, Dec 02, 2014 at 08:37:19AM +0100, Jakub Jelinek wrote:
> Anyway, sorry for missing this, supposedly I've grepped for it in
> subdirectories of gcc/ but not in the second level subdirectories.
> And, apparently I've missed it also some time ago when removing unnecessary
> last NULL arguments
On Tue, Dec 02, 2014 at 05:13:52AM +, Gopalasubramanian, Ganesh wrote:
> The following patch implements that. Bootstrapped/regtested on x86_64-linux
> and i686-linux, ok for trunk?
>
> Our aarch64 build also breaks as mentioned in
> https://gcc.gnu.org/ml/gcc-patches/2014-12/msg00119.html
Y
Hi Manuel,
Manuel López-Ibáñez wrote:
This patch actually does not touch linemap but I will appreciate
Dodji's comments about the approach. The problem is that in case of
long lines, the column hint of 120 might be too small, thus we do not
have enough locations within one line to point to a hig
The following patch implements that. Bootstrapped/regtested on x86_64-linux
and i686-linux, ok for trunk?
Our aarch64 build also breaks as mentioned in
https://gcc.gnu.org/ml/gcc-patches/2014-12/msg00119.html
Regards
Ganesh
On Mon, Dec 1, 2014 at 5:48 PM, Michael Meissner
wrote:
> in my change on November 24th (adding the support to use scalar floating point
> values in Altivec registers) there was a regression when Spec 2000 was
> compiled
> for 32-bit big endian power7 systems. The rs6000_legitimize_reload_addres
Ping^2
Thanks,
bin
On Tue, Nov 25, 2014 at 4:19 PM, Bin.Cheng wrote:
> Ping. Anybody have a look?
>
> Thanks,
> bin
>
> On Tue, Nov 18, 2014 at 4:34 PM, Bin Cheng wrote:
>> Hi,
>> This is the patch implementing ldp/stp optimization for aarch64. It
>> consists of two parts. The first one is p
Hi,
This is an obvious patch fixing wrong documentation for
TARGET_SCHED_FUSION_PRIORITY.
Html file generated as expected now. Will commit it in 24h if no objection.
2014-12-02 Bin Cheng
* target.def (fusion_priority): Wrap code with @smallexample.
* doc/tm.texi: Regenerated.
On 23/10/14 17:00 +0100, Jonathan Wakely wrote:
This adds:
std::notify_all_at_thread_exit()
std::promise<>::set_value_at_thread_exit()
std::promise<>::set_exception_at_thread_exit()
std::packaged_task<>::make_ready_at_thread_exit()
There's a linked list of callbacks that run after TLS destructor
I think the jit patches introduced a problem when you rebuild within a
directory that contains an old build (i.e., no brand new build
directory). The gcc/Makefile creates a symlink for xgcc with the full
driver name without first removing the symlink. The right procedure
(gathered from other Mak
On Mon, 2014-12-01 14:52:05 +0100, Jakub Jelinek wrote:
> On Fri, Nov 28, 2014 at 08:02:23PM +0100, Jakub Jelinek wrote:
> > As possible follow-up, I wonder if gimple_build_assign_with_ops
> > isn't too long and too verbose either, couldn't we just
> > use overloads of gimple_build_assign instead?
This patch actually does not touch linemap but I will appreciate
Dodji's comments about the approach. The problem is that in case of
long lines, the column hint of 120 might be too small, thus we do not
have enough locations within one line to point to a higher column
(like 132 in the testcase). Gi
On 12/01/2014 05:48 PM, Gopalasubramanian, Ganesh wrote:
> +2014-12-01 Ganesh Gopalasubramanian
> +
> + * config/aarch64/aarch64.md (define_insn "prefetch"): New.
> +
Looks good to me.
r~
On Mon, 1 Dec 2014, Jeff Law wrote:
> > diff --git a/gcc/c-family/c.opt b/gcc/c-family/c.opt
> > index 8f5d76c..283c632 100644
> > --- a/gcc/c-family/c.opt
> > +++ b/gcc/c-family/c.opt
> > @@ -1043,6 +1043,9 @@ Instrument only functions marked with bnd_instrument
> > attribute.
> > static-libmpx
The initialization of color diagnostics occurs too late to affect, for
example, the "unrecognized command line option" error. Unfortunately,
in the compiler proper we cannot initialize colors just after
diagnostic_initialize() because the FEs replace the pretty-printer
with their own by destroying
On Mon, 2014-12-01 at 10:38 -0700, Jeff Law wrote:
> On 11/27/14 18:44, Segher Boessenkool wrote:
> > Currently reg_dead_at_p returns 0 for registers that are set but never
> > used. This patch repairs that oversight. This fixes PR59278.
> >
> >
> > 2014-11-27 Segher Boessenkool
> >
> > gcc/
>
> I also wonder if compare-elim ought to be helping here. Isn't that the
> point here, to eliminate the comparison and instead get it for free as
> part of the arithmetic? If so, is it the fact that we have memory
> references that prevents compare-elim from kicking in?
Yes, compare-elim doesn't
On Mon, 1 Dec 2014, Richard Biener wrote:
> +/* Combine two successive divisions. */
> +(for div (trunc_div ceil_div floor_div round_div exact_div)
This doesn't seem correct for all kinds of division and signedness of
arguments.
TRUNC_DIV_EXPR (C division) and EXACT_DIV_EXPR should be OK regar
On Mon, 2014-12-01 at 12:09 +0100, Richard Biener wrote:
> On Mon, Dec 1, 2014 at 8:21 AM, Oleg Endo wrote:
> > Hi,
> >
> > When running the testsuite on a sh-elf configuration, some test cases
> > fail due to multiple definitions of the function '_init'. This is
> > because on sh-elf every funct
in my change on November 24th (adding the support to use scalar floating point
values in Altivec registers) there was a regression when Spec 2000 was compiled
for 32-bit big endian power7 systems. The rs6000_legitimize_reload_address
function generated reg+offset address for scalar values. This r
> > 2014-11-19 Martin Jambor
> >
> > * ipa-prop.h (ipa_alignment): New type.
> > (ipa_jump_func): New field alignment.
> > (ipcp_transformation_summary) New type.
> > (ipcp_grow_transformations_if_necessary): Declare.
> > (ipa_node_agg_replacements): Removed.
> > (ipcp_t
Ping^2.
-cary
On Wed, Oct 29, 2014 at 12:04 PM, Cary Coutant wrote:
> Ping?
>
> On Mon, Oct 20, 2014 at 10:31 AM, Cary Coutant wrote:
>> This patch adds support for the mips target in gold.
>>
>> OK to commit?
>>
>> -cary
>>
>>
>> 2014-10-20 Cary Coutant
>>
>> * configure (--enable-g
On 08/04/14 02:24, Zhenqiang Chen wrote:
ChangeLog:
2014-05-22 Zhenqiang Chen
Part of PR rtl-optimization/61225
* config/i386/i386-protos.h (ix86_peephole2_rtx_equal_p): New
proto.
* config/i386/i386.c (ix86_peephole2_rtx_equal_p): New function.
* reg
[+cc Michael Eager]
> Rather than having to lobby to keep it unchanged because we jumped the gun,
> can we lobby to get the number assigned in the near future rather than in
> the potentially far future? That feels more cooperative to me :-)
>
> Would that make Michael happier?
I'm pretty confid
On Mon, Dec 01, 2014 at 02:50:57PM -0700, Jeff Law wrote:
> On 12/01/14 14:44, Cary Coutant wrote:
> >>Presumably we don't have any sense when the values will be assigned, right?
> >>Any chance we could speed that along a bit?
> >
> >As Jason said, the value in the current draft is unlikely to chan
On 12/01/14 14:44, Cary Coutant wrote:
Presumably we don't have any sense when the values will be assigned, right?
Any chance we could speed that along a bit?
As Jason said, the value in the current draft is unlikely to change,
and I think he and I can probably lobby to keep it unchanged if the
> Presumably we don't have any sense when the values will be assigned, right?
> Any chance we could speed that along a bit?
As Jason said, the value in the current draft is unlikely to change,
and I think he and I can probably lobby to keep it unchanged if there
any danger that the numbering will
> On Fri, Nov 21, 2014 at 08:18:12PM +0100, Martin Jambor wrote:
> > Hi,
> >
> > when debugging PR 63814 I noticed that when cgraph_node::create_clone
> > was using redirect_edge_duplicating_thunks to redirect two edges to a
> > thunk of a clone, two thunks were created, one for each edge. The
>
Hi
Here is another proposal that consider all your remarks except one.
I finally prefer to go with std::vector of pointers. Dynamically
allocating Catalog_info allow to avoid numerous copies of locale when we
find this pointer from the catalog info.
On 28/11/2014 11:49, Jonathan Wakely
> > Hi,
> >
> > PR 63814 is caused by cgraph_edge_brings_value_p misidentifying an
> > edge to an expanded artificial thunk as an edge to the original node,
> > which then leads to crazy double-cloning and doubling the thunks along
> > the call.
> >
> > This patch fixes the bug by strengthening t
On 11/25/14 15:57, Mark Wielaard wrote:
This implements the DWARFv5 noreturn proposal:
http://dwarfstd.org/ShowIssue.php?issue=140331.1
TREE_THIS_VOLATILE on a FUNCTION_DECL node means the function does not
return normally. This catches the traditional noreturn GNU attribute,
the C11 _Noreturn k
On 11/25/14 18:39, David Malcolm wrote:
Running various JIT testcases under valgrid showed this one-time leak:
8,464 (336 direct, 8,128 indirect) bytes in 1 blocks are definitely lost in
loss record 144 of 147
at 0x4A081D4: calloc (in
/usr/lib64/valgrind/vgpreload_memcheck-amd64-linux.so)
On 11/25/14 18:39, David Malcolm wrote:
Various functions in tree-pretty-print.c take a param
pretty_printer *buffer
and there's also a static pretty_printer *buffer;
Additionally, pretty_printer instances are not buffers; they
*contain* buffers (inasmuch as they have a field "buffer").
This
On Mon, 1 Dec 2014, Richard Biener wrote:
The following fixes the oldest bug in my list of assigned PRs, namely
combining A / CST / CST' to A / (CST * CST'). extract_muldiv does
this in fold-const.c (but it fails to optimize to zero when CST * CST'
overflows).
Bootstrapped and tested on x86_64
On Mon, 1 Dec 2014, Richard Biener wrote:
The following fixes PR64126.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2014-12-01 Richard Biener
PR middle-end/64126
* match.pd: Allow conversions in ~A + 1 -> -A, add -A - 1 -> ~A
and -1 - A ->
On 11/14/14 10:31, Ilya Enkovich wrote:
Hi,
Here are few tests for MPX wrappers. Will add more in case patches #1,2 are OK.
Thanks,
Ilya
--
2014-11-14 Ilya Enkovich
* lib/mpx-dg.exp (mpx_link_flags): Set path to wrappers library.
* gcc.target/i386/mpx/calloc-1-lbv.c: New.
On 11/24/14 07:09, Ilya Enkovich wrote:
--
Here is a version with updated documentation. Added -static-libmpx and
-static-libmpxwrappers options. Added description on runtime library
dependencies.
Thanks,
Ilya
--
2014-11-24 Ilya Enkovich
* c-family/c.opt (fchkp-use-wrappers):
On 11/24/14 07:06, Ilya Enkovich wrote:
Normally GCC target libraries assigned to the FSF would use GPL+exception
rather than LGPL (especially if the library might be linked in
statically), to keep predictable what requirements are imposed by linking
your program with GCC. libquadmath is an exce
On 11/25/14 14:16, Sebastian Pop wrote:
Sebastian Pop wrote:
>I will bootstrap and regression test this patch on x86_64-linux and
>powerpc64-linux. I will also run it on our internal benchmarks, coremark, and
>the llvm test-suite.
>
>I will also include a longer testcase that makes sure we do n
On 12/01/2014 07:01 AM, Fabien Chêne wrote:
2014-11-03 21:18 GMT+01:00 Fabien Chêne :
2014-10-09 15:34 GMT+02:00 Jason Merrill :
[...]
If the USING_DECL is returned, the code below will be rejected as
expected, but the error message will not mention the line where the
USING_DECL appears as the
Latest results for 4.9.x
-tgc
Testresults for 4.9.2:
hppa-unknown-linux-gnu
i386-pc-solaris2.9
i386-sun-solaris2.11
i486-sun-solaris2.11
i686-unknown-linux-gnu
mips-unknown-linux-gnu
mipsel-unknown-linux-gnu
sparc-sun-solaris2.9
sparc64-sun-solaris2.9
sparc-unknown-linux-gnu
Latest results for gcc 4.8.x.
-tgc
Testresults for 4.8.3
hppa-unknown-linux-gnu
x86_64-apple-darwin14.0.0
Index: buildstat.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.8/buildstat.html,v
retrieving revision 1.12
diff -u -r1.
On 12/1/2014 11:57 AM, Jeff Law wrote:
Prior to reload (ie, in DSE1) there's a bit of magic in that we do not
set frame_read on call insns. That may in fact be wrong and possibly
the source of the problem.
/* This field is only used for the processing of const functions.
These functions
On Mon, Dec 1, 2014 at 10:26 AM, Jeff Law wrote:
> On 11/26/14 15:00, Andrew Pinski wrote:
>>
>> Hi,
>>The problem here is lra_substitute_pseudo calls gen_rtx_SUBREG with
>> a VOIDmode (const_int) argument but really it should not be calling
>> gen_rtx_SUBREG directly instead it should be usin
On Wed, Nov 26, 2014 at 2:01 PM, Andrew Pinski wrote:
> On Wed, Nov 26, 2014 at 2:00 PM, Andrew Pinski wrote:
>> Hi,
>> The problem here is lra_substitute_pseudo calls gen_rtx_SUBREG with
>> a VOIDmode (const_int) argument but really it should not be calling
>> gen_rtx_SUBREG directly instead i
On 11/26/14 12:43, Jakub Jelinek wrote:
On Fri, Nov 21, 2014 at 04:20:44PM +0400, Dmitry Vyukov wrote:
Yes, I think it's the way to go.
I've just committed the following revision to clang that removes -pie
when compiling with tsan:
http://llvm.org/viewvc/llvm-project?view=revision&revision=22252
On 11/26/14 15:00, Andrew Pinski wrote:
Hi,
The problem here is lra_substitute_pseudo calls gen_rtx_SUBREG with
a VOIDmode (const_int) argument but really it should not be calling
gen_rtx_SUBREG directly instead it should be using
gen_lowpart_if_possible. This patch fixes that and adds a test
On 11/27/14 18:44, Segher Boessenkool wrote:
[What a great subject line].
This simplifies the code right before the new code of the previous patch
to use one of the new helper functions that introduces.
2014-11-27 Segher Boessenkool
gcc/
* combine.c (try_combine): Use is_parallel_o
On 11/26/14 14:59, Segher Boessenkool wrote:
On Wed, Nov 26, 2014 at 11:14:36AM -0700, Jeff Law wrote:
Are you talking about create_log_links? There can be no duplicates there
(anymore), that would be multiple defs of the same reg in the same insn,
indeed.
Yes, I was referring to the code in c
OK.
Jason
On Mon, Dec 1, 2014 at 6:47 PM, Jeff Law wrote:
> On 11/27/14 04:19, Uros Bizjak wrote:
>> Attached patch helps RA to choose the most appropriate PIC register by
>> changing the register preference for set_got patterns. Using this
>> patch, there should really be a reason for RA to avoid ABI mand
On Mon, Dec 01, 2014 at 10:47:40AM -0700, Jeff Law wrote:
> > * config/i386/i386.md (set_got): Use "=b,?r" constraint for operand 0.
> > (set_got_labelled): Ditto.
> > (set_got_rex64): Ditto.
> > (set_rip_rex64): Ditto.
> > (set_got_offset_rex64): Ditto.
> >
> >Bootstrapped and
On 11/27/14 04:19, Uros Bizjak wrote:
Hello!
Attached patch helps RA to choose the most appropriate PIC register by
changing the register preference for set_got patterns. Using this
patch, there should really be a reason for RA to avoid ABI mandated
hard PIC reg.
Agreed.
This patch avoids ma
On Thu, Nov 13, 2014 at 5:32 PM, Martin Jambor wrote:
> Hi,
>
> this patch is that main part of the series, implementing propagation
> of ipa_polymorphic_call_contexts. Although I tried to keep its size
> checked, it is big because switching from binfos to contexts in ipa-cp
> also requires chang
On 11/27/14 18:44, Segher Boessenkool wrote:
If I2 is a PARALLEL of two SETs, split it into two instructions, I1
and I2. If there already was an I1, rename it to I0. If there
already was an I0, don't do anything.
This surprisingly simple patch is enough to let combine handle such
PARALLELs pro
Changes 'tree_type_ptr' to 'ttype' everywhere. The second attachment
is the updated patch which converts the branch to a segregated tree-type
for compilation testing. (remembering that this patch is maintained as
'tree-type.patch' in the source directory).
Andrew
* fold-const.c (fold_con
On 11/27/14 18:44, Segher Boessenkool wrote:
Now that LOG_LINKS are per regno, we can distribute them on PARALLELs
just fine. Do so. This makes PARALLELs not lose their LOG_LINKS early
when e.g. a trivial reg-reg move is combined, so that they can be used
in more useful combinations as well.
On 11/27/14 18:44, Segher Boessenkool wrote:
With this new field in place, we can have LOG_LINKS for insns that set
more than one register and distribute them properly in distribute_links.
This then allows many more PARALLELs to be combined.
Also split off new functions can_combine_{def,use}_p f
On 11/27/14 18:44, Segher Boessenkool wrote:
If one of the resulting insns is a noop set it is safe to allow the
combination, for it will never lead to a loop: any following combination
using that noop will delete it. In fact, in most cases combine deletes
the noop immediately; for cc0 targets i
On 11/27/14 18:44, Segher Boessenkool wrote:
Currently reg_dead_at_p returns 0 for registers that are set but never
used. This patch repairs that oversight. This fixes PR59278.
2014-11-27 Segher Boessenkool
gcc/
PR rtl-optimization/59278
combine (reg_dead_at_p): Consider R
> > > My patch hacks in a removal of them into ipa_reference_c_finalize, but
> > > I suspect there's a cleaner place to put this.
> > >
> > > gcc/ChangeLog:
> > > PR jit/63854
> > > * ipa-prop.c (ipa_register_cgraph_hooks): Guard insertion of
> > > ipa_add_new_function on function_insertion_h
Ping (this is a new feature submitted in stage1 so I'd really
appreciate a review before it is going to be deemed "too late").
Thanks,
Martin
On Wed, Nov 19, 2014 at 12:32:53AM +0100, Martin Jambor wrote:
> Hi,
>
...
>
> 2014-11-19 Martin Jambor
>
> * ipa-prop.h (ipa_alignment): N
Changes a few files where ttype is used as a variable. The next patch
changes the branch to use ttype as the name instead of tree_type_ptr.
bootstraps, run tests, etc.
checked into branch.
Andrew
* builtins.c (fold_builtin_atomic_always_lock_free): Don't use ttype as
a variable name.
* c
On 11/28/14 07:10, Wilco Dijkstra wrote:
Jeff Law wrote:
Do you have a testcase that shows the expected improvements from this
change? It's OK if it's specific to a target.
Have you bootstrapped and regression tested this change?
With a test for the testsuite and assuming it passes bootstrap a
Hi all,
I noticed that this test depends on adrp+add sequences being used for
constants, which only happens when the memory model used is small. If
tiny or large is used then the required adrp+add sequences are not
generated, making the test pointless for them. This patch guards the
test on t
Ping.
Thx,
Martin
On Fri, Nov 21, 2014 at 08:18:12PM +0100, Martin Jambor wrote:
> Hi,
>
> when debugging PR 63814 I noticed that when cgraph_node::create_clone
> was using redirect_edge_duplicating_thunks to redirect two edges to a
> thunk of a clone, two thunks were created, one for each edg
Ping.
Thx,
Martin
On Fri, Nov 21, 2014 at 07:59:11PM +0100, Martin Jambor wrote:
> Hi,
>
> PR 63814 is caused by cgraph_edge_brings_value_p misidentifying an
> edge to an expanded artificial thunk as an edge to the original node,
> which then leads to crazy double-cloning and doubling the thunk
On 01/12/14 15:58, H.J. Lu wrote:
On Thu, Nov 27, 2014 at 8:38 AM, Renlin Li wrote:
On 27/11/14 15:37, H.J. Lu wrote:
On Thu, Nov 27, 2014 at 7:32 AM, Renlin Li wrote:
On 26/11/14 18:12, H.J. Lu wrote:
On Wed, Nov 26, 2014 at 10:09 AM, Renlin Li wrote:
On 26/11/14 12:16, H.J. Lu wrote:
O
Hi,
I'm pretty sure that we should accept this. Tested x86_64-linux.
Thanks,
Paolo.
/cp
2014-12-01 Paolo Carlini
PR c++/60859
* decl.c (reshape_init_r): Do not reject value-initialization of
scalar array element.
/testsuite
2014-12-01 Paolo Car
On 11/28/14 18:05, John David Anglin wrote:
The attached simple change fixes a long standing regression since 4.2.
When we have stack arguments and a sibling
call, the dse pass deletes stores to the parent's stack frame when we
have a sibling call because they are are off frame
for the current fu
On Wed, 2014-11-19 at 12:58 -0700, Jeff Law wrote:
> On 11/19/14 03:46, David Malcolm wrote:
> > This fixes three leaks in IPA seen in jit testcases with valgrind:
> >
> > This one:
> > 96 bytes in 4 blocks are definitely lost in loss record 102 of 205
> > at 0x4A0645D: malloc (in
> > /usr/lib
On 12/01/2014 04:37 PM, Jakub Jelinek wrote:
for an unexpected type?
Yes: placeholder expressions are supposed only to refer to the type allowed
by the context. I updated the message to talk about the context and I added
another one for the (invalid) case when want_address == 0.
I meant that
The following patch adds linemap_assert_fails(), which is an assert
meant to be used in conditions such as:
if (linemap_assert_fails(EXPR))
handle_error();
This is useful for ICEs that we would like to detect during the
development phase but that could be handled gracefully. In the case of
line
On Mon, Dec 1, 2014 at 8:35 AM, Ilya Tocar wrote:
>> > I saw
>> >
>> > FAIL: gcc.dg/vect/costmodel/i386/costmodel-fast-math-vect-pr29925.c
>> > scan-tree-dump-times vect "vectorized 1 loops" 1
>> > FAIL: gcc.dg/vect/costmodel/x86_64/costmodel-fast-math-vect-pr29925.c
>> > scan-tree-dump-times vect
> > I saw
> >
> > FAIL: gcc.dg/vect/costmodel/i386/costmodel-fast-math-vect-pr29925.c
> > scan-tree-dump-times vect "vectorized 1 loops" 1
> > FAIL: gcc.dg/vect/costmodel/x86_64/costmodel-fast-math-vect-pr29925.c
> > scan-tree-dump-times vect "vectorized 1 loops" 1
> > FAIL: gcc.target/i386/avx256-
On Thu, Nov 27, 2014 at 8:38 AM, Renlin Li wrote:
> On 27/11/14 15:37, H.J. Lu wrote:
>>
>> On Thu, Nov 27, 2014 at 7:32 AM, Renlin Li wrote:
>>>
>>> On 26/11/14 18:12, H.J. Lu wrote:
On Wed, Nov 26, 2014 at 10:09 AM, Renlin Li wrote:
>
> On 26/11/14 12:16, H.J. Lu wrote:
>
Hi Richard,
I resend you patch1 and patch2 with minor changes:
1. I renamed flag_force_vectorize to aggressive_if_conv.
2. Use static cast for the first argument of gimple_phi_arg_edge.
I also very sorry that I sent you bad patch.
Now let me answer on your questions related to second patch.
1. Wh
On Mon, Dec 01, 2014 at 04:10:35PM +0100, Richard Biener wrote:
> On Mon, 1 Dec 2014, Ulrich Weigand wrote:
>
> > Richard Biener wrote:
> >
> > > * Makefile.in (gimple-match.o-warn): Remove -Wno-unused-parameter,
> > > add -Wno-unused-but-set-variable.
> >
> > This seems to cause
> >
> > cc
On 01/12/14 15:33, Christophe Lyon wrote:
On 27 November 2014 at 11:35, Ramana Radhakrishnan
wrote:
On Wed, Oct 29, 2014 at 10:20 AM, Jiong Wang wrote:
On 26/08/14 13:36, Richard Earnshaw wrote:
On 29/07/14 15:49, Jiong Wang wrote:
test done
===
no regression on the full toolchain test on
On Mon, 1 Dec 2014, FX wrote:
> Your change is OK (we don’t want to use the type of the result, but the type
> of the argument indeed).
>
>
> > Index: gcc/fortran/trans-intrinsic.c
> > ===
> > --- gcc/fortran/trans-intrinsic.c (r
Committed after entering stage3 in a bootstrap.
Richard.
2014-12-01 Richard Biener
* Makefile.in (gimple-match.o-warn): Use -Wno-unused instead of
-Wno-unused-variable and -Wno-unused-but-set-variable to restore
bootstrap with old GCC.
(generic-match.o-warn):
This moves more constant folding from fold_binary to the const_binop
overload with type and handles POINTER_PLUS_EXPR in the main routine
as well as fixing a minor glitch I introduced.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2014-12-01 Richard Biener
Your change is OK (we don’t want to use the type of the result, but the type of
the argument indeed).
> Index: gcc/fortran/trans-intrinsic.c
> ===
> --- gcc/fortran/trans-intrinsic.c (revision 218211)
> +++ gcc/fortran/trans-int
On Mon, Dec 01, 2014 at 04:29:05PM +0100, Pierre-Marie de Rodat wrote:
> >>+ expansion_failed (loc, NULL_RTX,
> >>+ "PLACEHOLDER_EXPR for a unexpected type");
> >
> >for an unexpected type?
>
> Yes: placeholder expressions are supposed only to refer to the type allowed
> by t
On 27 November 2014 at 11:35, Ramana Radhakrishnan
wrote:
> On Wed, Oct 29, 2014 at 10:20 AM, Jiong Wang wrote:
>> On 26/08/14 13:36, Richard Earnshaw wrote:
>>
>>> On 29/07/14 15:49, Jiong Wang wrote:
test done
===
no regression on the full toolchain test on arm-none-eabi.
>>
On 11/26/2014 06:42 PM, Jakub Jelinek wrote:
Replace implicitely with implicitly in the whole patch.
Oops! Done.
+ If CONTEXT is NULL, the behavior is the same as if the context_type field
+ was NULL_TREE. */
as if both context_type and base_decl were NULL_TREE?
Absolutely: done.
+
Hi Jakub,
Do you think this patch is ok for upstream:
2014-12-01 Varvara Rainchik
* libgomp/libgomp.h: Eliminate case when HAVE_TLS is not defined:
always use tls emulation.
* libgomp/team.c: Likewise.
--
diff --git a/libgomp/libgomp.h b/libgomp/libgomp.h
index a1482c
On Mon, 1 Dec 2014, Ulrich Weigand wrote:
> Richard Biener wrote:
>
> > * Makefile.in (gimple-match.o-warn): Remove -Wno-unused-parameter,
> > add -Wno-unused-but-set-variable.
>
> This seems to cause
>
> cc1plus: error: unrecognized command line option
> "-Wno-unused-but-set-variable"
On Tue, Nov 11, 2014 at 9:55 AM, Richard Biener
wrote:
> On Mon, Nov 10, 2014 at 2:39 PM, Prathamesh Kulkarni
> wrote:
>> Hi,
>> This patch adds support for operator-lists to be used in expression.
>>
>> I reuse operator-list as the iterator. This is not really valid since
>> user-defined opera
Richard Biener wrote:
> * Makefile.in (gimple-match.o-warn): Remove -Wno-unused-parameter,
> add -Wno-unused-but-set-variable.
This seems to cause
cc1plus: error: unrecognized command line option "-Wno-unused-but-set-variable"
On Mon, Dec 1, 2014 at 6:37 AM, H.J. Lu wrote:
> On Fri, Nov 28, 2014 at 7:15 AM, Uros Bizjak wrote:
>> Hello!
>>
>>> As discussed for AVX512F submission, using -D in dg-options,
>>> should be replaced with parts of the testcase itself. Patch below
>>> does that.
>>> Ok for trunk?
>>>
>>> P.
OK.
Jason
On 11/28/2014 09:41 AM, Jakub Jelinek wrote:
Why do you look through ARRAY_REF here? An element of an array is its own
complete object.
That had to do with only instrumenting dereferences surrounded by handled
components, but not accesses to decls (so p->x gets instrumented but
q.x for VAR_DEC
OK, thanks.
Jason
Hi,
Manuel noticed that in this error message it would be nice to have an
'aka' for typedefs. It seems to me that simply using the standard for
types %qT is most of it, but I propose also to add a '#' (otherwise we
regress on g++.dg/inherit/error5.C). Tested x86_64-linux.
Thanks,
Paolo.
///
On Fri, Nov 28, 2014 at 7:15 AM, Uros Bizjak wrote:
> Hello!
>
>> As discussed for AVX512F submission, using -D in dg-options,
>> should be replaced with parts of the testcase itself. Patch below
>> does that.
>> Ok for trunk?
>>
>> P. S.
>> As patch itself is hard to review, this was done by
OK.
Jason
On Mon, Dec 1, 2014 at 3:12 PM, Marek Polacek wrote:
> This fixes PR64121, where we ran into a SSA corruption failure
> because we couldn't coalesce two SSA_NAME_OCCURS_IN_ABNORMAL_PHI
> SSA_NAMEs. So just skip such SSA names when looking for a base.
Not really "skip" but "stop at".
> Bootstrap
This fixes PR64121, where we ran into a SSA corruption failure
because we couldn't coalesce two SSA_NAME_OCCURS_IN_ABNORMAL_PHI
SSA_NAMEs. So just skip such SSA names when looking for a base.
Bootstrapped/regtested on ppc64-linux, ok for trunk?
2014-12-01 Marek Polacek
Jakub Jelin
On Mon, 1 Dec 2014, Jakub Jelinek wrote:
> On Fri, Nov 28, 2014 at 08:02:23PM +0100, Jakub Jelinek wrote:
> > As possible follow-up, I wonder if gimple_build_assign_with_ops
> > isn't too long and too verbose either, couldn't we just
> > use overloads of gimple_build_assign instead?
> > Either wit
1 - 100 of 120 matches
Mail list logo
