The following patch avoids rewriting virtual SSA after each if-converted
loop (possibly).
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2016-08-18 Richard Biener
* ssa-iterators.h (ssa_vuse_operand): New inline.
* tree-if-conv.c (ifc_temp_va
Hi!
While looking if something doesn't need to be updated for the
i386-builtin.def addition (I think we are ok, this file doesn't need to be
shipped for plugins, as it is only included in i386.c, the first time
i386.c is built no matter what and next time we'll have it listed in
.deps/i386.Po), I'
On Thu, Aug 18, 2016 at 9:39 AM, Jakub Jelinek wrote:
> Hi!
>
> While looking if something doesn't need to be updated for the
> i386-builtin.def addition (I think we are ok, this file doesn't need to be
> shipped for plugins, as it is only included in i386.c, the first time
> i386.c is built no ma
The following patch makes it possible to add statistic counters to
update-ssa. Clone materialization ends up updating SSA form from
a context with current_pass being NULL - wrapping materialize_all_clones
into a pass fixes this.
Bootstrap / regtest running on x86_64-unknown-linux-gnu, ok for tru
On Wed, Aug 17, 2016 at 10:11 PM, H.J. Lu wrote:
> builtin_memset_gen_str returns a register used for memset, which only
> supports integer registers. But a target may use vector registers in
> memmset. This patch adds a TARGET_GEN_MEMSET_VALUE hook to duplicate
> QImode value to mode derived fr
The atteched patch fixes the S/390 test case insv-1.c with -m31.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/testsuite/ChangeLog
* gcc.target/s390/insv-1.c: Fix test when running with -m31.
>From 4021fe9c4703e29b9446a24584ebd998c591d7ed Mon Sep 17 00:00:00 2001
From: Dominik
On Wed, Aug 17, 2016 at 09:09:07PM -0600, Jeff Law wrote:
> On 08/17/2016 02:23 AM, Richard Biener wrote:
> >On Tue, Aug 16, 2016 at 6:06 PM, Jeff Law wrote:
> >>On 08/16/2016 08:01 AM, Tamar Christina wrote:
> >>>
> >>>
> >>>Hi All,
> >>>
> >>>This patch turns off the usage of section anchors for
THe attached patch improves checking of teh results of the
subtests "a" and "f". As they share the same "vone" instruction,
the duplicate scan-assembler-times was bogus. Moved "f" to a
separate function to fix this. Also double check that no extra
"vgmf" instructions are used.
Ciao
Dominik ^_^
On 16/08/16 05:16 +, Bernd Edlinger wrote:
Hi Jonathan,
I think this would be an improvement, although I still can't get the
assertion to fail:
Probably because the memory is still initialized to zero,
when it is used for the first time.
Try this:
#include
#include
#include
struct A
Hi all,
This fixes a bug in the vector load functions in which they load the
vector in the wrong order for big endian systems. This patch flips the
order conditionally in the vec_concats.
No testcase given because plenty of existing tests for vld functions.
Ran regression tests on aarch64_be-none
Hello,
Currently, the VxWorks target overrides the defaults for debug info
output options (DWARF version, strictness) in a target-specific options
hook. This patch creates macros so that these defaults can be overriden
in config headers. In particular, this will make it easier to have
different
Hello,
This enhances location description generation so that the generated
opcodes for integer literals are as space-efficient when HOST_WIDE_INT
is 64-bits large than when it's 32-bits large. In particular, this
reduces the size of the opcodes generated to produce big unsigned
literals using smal
Hello,
For -gdwarf-3 and newer, the DWARF back-end sometimes generates DWARF
procedures to factorize complex location descriptions. DWARF procedures
can be materialized as DW_TAG_dwarf_procedure DIEs, but actually any DIE
that can hold a DW_AT_location attribute is also accepted.
Unlike what I t
Hello,
A check in dwarf2out_imported_module_or_decl prevents valid strict
DWARF2 constructs such as DW_TAG_imported_declaration from being emitted
in dwarf2out_imported_module_or_decl_1.
The latter already protects the emission of newer DWARF tags with
appropriate checks, so the one in the former
On Thu, Aug 18, 2016 at 11:32:45AM +0200, Pierre-Marie de Rodat wrote:
> Currently, the VxWorks target overrides the defaults for debug info
> output options (DWARF version, strictness) in a target-specific options
> hook. This patch creates macros so that these defaults can be overriden
I believ
> This patch (2nd of 4) includes various changes to Ada related C files
> required for DJGPP support
>
> ChangeLog entry:
>
> 2016-07-30 Andris Pavenis
>
> * ada/ctrl_c.c: Do not use macro SA_RESTART for DJGPP.
>
> * ada/gsocket.h: Do not support sockets for DJGPP.
>
> * ada/init.
> 2016-07-30 Andris Pavenis
>
> * ada/gcc-interface/Makefile.in (LIBGNAT_TARGET_PAIRS): Define for
> DJGPP target
> (EH_MECHANISM): Define to -gcc for DJGPP
> * ada/system-djgpp.ads: New file
>
> Andris
> +++ b/gcc/ada/system-djgpp.ads
> @@ -0,0 +1,148 @@
> +--
I don't think I meant for this test to use -std=gnu++0x when I added
it, that probably came from copying 20_util/function/cons/addressof.cc
It certainly works OK with C++98, which is how tr1::function is most
likely to be used.
* testsuite/tr1/3_function_objects/function/10.cc: Remove un
The following fixes PR77282, tested on x86_64-unknown-linux-gnu, applied
as obvious.
Richard.
2016-08-18 Richard Biener
PR tree-optimization/77282
* tree-ssa-pre.c (eliminate_dom_walker::before_dom_children):
When doing auto-parallelizing also prevent use of PHIs that
On 18/08/16 07:00, Sandra Loosemore wrote:
> On 08/11/2016 04:31 AM, Thomas Preudhomme wrote:
>
>> diff --git a/gcc/doc/fragments.texi b/gcc/doc/fragments.texi
>> index
>> b6d8541c8ca820fa732363a05221e2cd4d1251c2..abf4e128671bb4751d21f24bb69625593d3c839e
>> 100644
>> --- a/gcc/doc/fragments.texi
>
Hi Sandra,
On 18/08/16 07:00, Sandra Loosemore wrote:
On 08/11/2016 04:31 AM, Thomas Preudhomme wrote:
diff --git a/gcc/doc/fragments.texi b/gcc/doc/fragments.texi
index
b6d8541c8ca820fa732363a05221e2cd4d1251c2..abf4e128671bb4751d21f24bb69625593d3c839e
100644
--- a/gcc/doc/fragments.texi
+++ b
As discussed in the PR, for __atomic_* functions we shouldn't accept
pointer-to-function and pointer-to-VLA as arguments.
Bootstrapped/regtested on x86_64-linux, ok for trunk?
2016-08-18 Marek Polacek
PR c/71514
* c-common.c (get_atomic_generic_size): Disallow pointer-to-funct
On 08/17/16 23:15, Jeff Law wrote:
On 08/12/2016 07:31 AM, Martin Liška wrote:
On 08/09/2016 09:03 PM, Andi Kleen wrote:
It could potentially make things a lot slower. I don't think it's a good
idea to do this by default.
-Andi
Ok, alternative can be a warning in the driver that would inform
On Thu, Aug 18, 2016 at 11:33:06AM +0200, Pierre-Marie de Rodat wrote:
> Hello,
>
> This enhances location description generation so that the generated
> opcodes for integer literals are as space-efficient when HOST_WIDE_INT
> is 64-bits large than when it's 32-bits large. In particular, this
> re
On 09/08/16 02:22 +0300, Ville Voutilainen wrote:
Implement the latest proposed resolution of LWG 2756.
* include/std/optional (Optional_base(const _Tp&))
(Optional_base(_Tp&&), using _Base::_Base): Remove.
(optional(nullopt_t)): New.
(optional(_Up&&)): Invoke base directly with in
On Thu, Aug 18, 2016 at 11:33 AM, Pierre-Marie de Rodat
wrote:
> Hello,
>
> For -gdwarf-3 and newer, the DWARF back-end sometimes generates DWARF
> procedures to factorize complex location descriptions. DWARF procedures
> can be materialized as DW_TAG_dwarf_procedure DIEs, but actually any DIE
>
On 05/08/16 22:45 -0700, Tim Shen wrote:
On Fri, Aug 5, 2016 at 4:08 AM, Jonathan Wakely wrote:
I think these all need to use ::new (__ptr) with qualification, see
below.
Done. I didn't add a testcase for this. Do you think that we need one/some?
No, I don't think we need a test. I doubt an
Selftests can use class selftest::temp_source_file to write out files
for testing input-handling, and the files are unlinked in the dtor.
This leads to stale entries in input.c's cache of file content, which
could lead to errors if a temporary filename gets reused during a run
of the selftests.
W
On Thu, May 12, 2016 at 04:12:21PM +0200, Martin Liška wrote:
> --- a/gcc/asan.c
> +++ b/gcc/asan.c
> @@ -243,6 +243,11 @@ static unsigned HOST_WIDE_INT asan_shadow_offset_value;
> static bool asan_shadow_offset_computed;
> static vec sanitized_sections;
>
> +/* Set of variable declarations tha
On 2016-06-22 05:25, David Malcolm wrote:
Clang has an option -fdiagnostics-parseable-fixits, which emits a
machine-readable version of the fixits, for use by IDEs. (The only
IDE I know of that supports this format is Xcode [1]; does anyone
know of other IDEs supporting this? I'd be very happy
Now that all various switch fallthrough bugfixes and adjustments were
committed, and this patch has shrunk considerably, I'm presenting the latest
version. The changes from the last version are not huge; we don't warn for a
fall through to a user-defined label anymore, and I made some tiny changes
Tested on powerpc64le-unknown-linux-gnu, aarch64-linux-gnu,
and x86_64-redhat-linux.
2016-08-18 Marek Polacek
PR c/7652
* Makefile.in (insn-attrtab.o-warn, insn-dfatab.o-warn,
insn-latencytab.o-warn, insn-output.o-warn, insn-emit.o-warn): Add
-Wno-switch-fallthr
On 08/18/2016 11:40 AM, Jakub Jelinek wrote:
I believe that is the preferred way, rather than macros.
The macros are yet another thing that would need to be undone if we ever
start supporting multiple targets in the same binary.
I don't see any advantages of introducing the macros.
Ah, I unders
On Thu, Aug 18, 2016 at 03:50:07PM +0200, Marek Polacek wrote:
> +case GIMPLE_BIND:
> + {
> + gbind *bind = as_a (stmt);
> + return last_stmt_in_scope (
> + gimple_seq_last_stmt (gimple_bind_body (bind)));
> + }
> +
> +case GIMPLE_TRY:
> + {
> + gtry
On 07/08/16 23:36 -0600, Sandra Loosemore wrote:
On 08/04/2016 06:37 PM, Jonathan Wakely wrote:
I've been working on some changes to the libstdc++ manual recently.
A lot of the changes are just using DocBook markup with more semantic
meaning (e.g. or instead of using for
everything that shou
On 08/18/2016 01:29 PM, Trevor Saunders wrote:
HOST_BITS_PER_WIDE_INT is now always 64, so you don't need that check.
Good point, thank you! I’ll adapt the patch.
--
Pierre-Marie de Rodat
On 08/18/2016 02:39 PM, Richard Biener wrote:
Ok.
Thank you for the quick review! This is commited, now (revision 239575).
--
Pierre-Marie de Rodat
On 08/18/2016 08:01 AM, Pierre-Marie de Rodat wrote:
On 08/18/2016 11:40 AM, Jakub Jelinek wrote:
I believe that is the preferred way, rather than macros.
The macros are yet another thing that would need to be undone if we ever
start supporting multiple targets in the same binary.
I don't see an
On Wed, Aug 17, 2016 at 11:21 PM, Uros Bizjak wrote:
> On Wed, Aug 17, 2016 at 10:11 PM, H.J. Lu wrote:
>> builtin_memset_gen_str returns a register used for memset, which only
>> supports integer registers. But a target may use vector registers in
>> memmset. This patch adds a TARGET_GEN_MEMSE
This patch allows the preprocessor to offer suggestions for misspelled
directives, taking us from e.g.:
test.c:5:2: error: invalid preprocessing directive #endfi
#endfi
^
to:
test.c:5:2: error: invalid preprocessing directive #endfi; did you mean #endif?
#endfi
^
endif
I can sel
On 2016/8/16 7:06 PM, Thomas Schwinge wrote:
> Hi!
>
> On Mon, 15 Aug 2016 19:25:48 +0800, Chung-Lin Tang
> wrote:
>> per the discussion on the bugzilla PR page, reductions on OpenACC loop
>> directives will automatically get a copy clause mapping on an enclosing
>> parallel construct (unless bo
On Thu, Aug 18, 2016 at 1:18 AM, Richard Biener
wrote:
> On Wed, Aug 17, 2016 at 10:11 PM, H.J. Lu wrote:
>> builtin_memset_gen_str returns a register used for memset, which only
>> supports integer registers. But a target may use vector registers in
>> memmset. This patch adds a TARGET_GEN_MEM
On Thu, Aug 18, 2016 at 11:28:02AM -0400, David Malcolm wrote:
> This patch allows the preprocessor to offer suggestions for misspelled
> directives, taking us from e.g.:
>
> test.c:5:2: error: invalid preprocessing directive #endfi
> #endfi
> ^
>
> to:
>
> test.c:5:2: error: invalid prep
On Thu, Aug 18, 2016 at 4:53 PM, H.J. Lu wrote:
> On Wed, Aug 17, 2016 at 11:21 PM, Uros Bizjak wrote:
>> On Wed, Aug 17, 2016 at 10:11 PM, H.J. Lu wrote:
>>> builtin_memset_gen_str returns a register used for memset, which only
>>> supports integer registers. But a target may use vector regist
> I'd prefer to make updates atomic in multi-threaded applications.
> The best proxy we have for that is -pthread.
>
> Is it slower, most definitely, but odds are we're giving folks
> garbage data otherwise, which in many ways is even worse.
It will likely be catastrophically slower in some cases
On 08/18/2016 09:51 AM, Andi Kleen wrote:
I'd prefer to make updates atomic in multi-threaded applications.
The best proxy we have for that is -pthread.
Is it slower, most definitely, but odds are we're giving folks
garbage data otherwise, which in many ways is even worse.
It will likely be ca
On Thu, Aug 18, 2016 at 08:51:31AM -0700, Andi Kleen wrote:
> > I'd prefer to make updates atomic in multi-threaded applications.
> > The best proxy we have for that is -pthread.
> >
> > Is it slower, most definitely, but odds are we're giving folks
> > garbage data otherwise, which in many ways i
On August 18, 2016 5:51:31 PM GMT+02:00, Andi Kleen wrote:
>> I'd prefer to make updates atomic in multi-threaded applications.
>> The best proxy we have for that is -pthread.
>>
>> Is it slower, most definitely, but odds are we're giving folks
>> garbage data otherwise, which in many ways is eve
On August 18, 2016 5:54:49 PM GMT+02:00, Jakub Jelinek wrote:
>On Thu, Aug 18, 2016 at 08:51:31AM -0700, Andi Kleen wrote:
>> > I'd prefer to make updates atomic in multi-threaded applications.
>> > The best proxy we have for that is -pthread.
>> >
>> > Is it slower, most definitely, but odds are
On Thu, 18 Aug 2016, Marek Polacek wrote:
> As discussed in the PR, for __atomic_* functions we shouldn't accept
> pointer-to-function and pointer-to-VLA as arguments.
>
> Bootstrapped/regtested on x86_64-linux, ok for trunk?
OK.
--
Joseph S. Myers
jos...@codesourcery.com
On 07/26/2016 09:53 AM, Dominik Vogt wrote:
Finally a patch that works and is simple. Bootstrapped and
regression tested on s390, s390x biarch and x86_64. The new patch
exploits the known alignment of (stack pointer +
STACK_DYNAMIC_OFFSET) as described earlier (see below). I think
that is the
input.c has a fixture class for running each selftest with a fresh
line_table, and logic for looping over various interesting line_table
test cases.
This patch exposes the above in selftest.h so that such
location-handling tests can be written in other files, renaming the
class from temp_line_tabl
On 08/11/2016 09:28 PM, Andi Kleen wrote:
If sym1 results in a return value that is some useful tree and inv1
is true and cst1 is true via this call:
The only way for get_single_symbol to return a non NULL tree
is to hit the return at the end -- and that always initializes
inv and neg.
Right.
On 08/10/2016 09:51 AM, Thomas Preudhomme wrote:
*** gcc/ChangeLog ***
2016-08-01 Thomas Preud'homme
* doc/fragments.texi (MULTILIB_REUSE): Mention that only options in
MULTILIB_OPTIONS should be used. Small wording fixes.
* genmultilib: Memorize set of all option
On Thu, 2016-08-18 at 15:50 +0200, Marek Polacek wrote:
> Now that all various switch fallthrough bugfixes and adjustments were
> committed, and this patch has shrunk considerably, I'm presenting the
> latest
> version. The changes from the last version are not huge; we don't
> warn for a
> fall t
On Thu, Aug 18, 2016 at 03:59:32PM +0200, Marek Polacek wrote:
> * eload1.c (elimination_effects): Likewise.
So we are deleting reload one char at a time now? :-)
Segher
So to recap, the problem with this BZ is we have a maybe-undefined
SSA_NAME in the IL. That maybe-undefined name is used as a condition
inside a loop and we unswitch that condition.
tree-ssa-loop-unswitch.c already tries to avoid doing that, but uses the
optimistic ssa_undefined_value_p fun
On Thu, Aug 18, 2016 at 12:28:18PM -0500, Segher Boessenkool wrote:
> On Thu, Aug 18, 2016 at 03:59:32PM +0200, Marek Polacek wrote:
> > * eload1.c (elimination_effects): Likewise.
>
> So we are deleting reload one char at a time now? :-)
If only! ;)
I fixed the typo in my local copy, than
Version 5 of the patch with the splitter for -O0/-O1 removed, as
discussed internally. Regression tested on s390x biarch and s390.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/ChangeLog
* config/s390/s390.md ("*andc_split"): New splitter for and with
complement.
gcc/
I intend to commit the following patch on Saturday.
It improves the reporting on where an error occurs.
Without the patch, one gets
% gfc7 -c gcc/testsuite/gfortran.dg/pr61318.f90
gcc/testsuite/gfortran.dg/pr61318.f90:19:6:
use gbl_message
1
Error: Type mismatch in argument 'message' at
On 08/11/2016 08:13 PM, Martin Sebor wrote:
Attached is an updated patch with changes addressing most of
the suggestions and comments I received.
The main changes/improvements are the following:
* New option, -fprintf-return-value, enables the sprintf return
value optimization. It's disa
In comment #5 Yuri reports that r235653 introduces a runtime failure for
176.gcc which I guess is caused by the combining step in
simplify_control_stmt_condition_1() not behaving properly on operands of
type VECTOR_TYPE. I'm a bit stumped as to why it mishandles
VECTOR_TYPEs because the logic shou
On 18/08/16 18:07, David Malcolm wrote:
This isn't quite the way that fix-its are meant to be used, in my mind,
at least: the insertion text is supposed to be something that could be
literally inserted into the code (e.g. by an IDE) i.e. it should be a
code fragment, rather than a message to the
Much of diagnostic-show-locus.c currently expects a diagnostic_info *,
but it only uses the rich_location and the diagnostic_t.
Change the signature of diagnostic_show_locus from:
void
diagnostic_show_locus (diagnostic_context *,
const diagnostic_info *);
to:
void
On August 18, 2016 8:25:18 PM GMT+02:00, Patrick Palka
wrote:
>In comment #5 Yuri reports that r235653 introduces a runtime failure
>for
>176.gcc which I guess is caused by the combining step in
>simplify_control_stmt_condition_1() not behaving properly on operands
>of
>type VECTOR_TYPE. I'm a b
Hello Mikael,
This doesn't look correct, what about substrings following component or
array references?
You're right; corrected in the attached patch. I have also added a test
case for this.
PS: What about the original fix, wasn't it a dependency
problem/inaccuracy after all?
The main err
On 08/11/2016 08:02 AM, Jan Hubicka wrote:
Hi,
this patch adds early jump threading pass. Jump threading is one of most
common cases where estimated profile becomes corrupted, because the branches
are predicted independently beforehand. This patch simply adds early mode to
jump threader that doe
On 08/18/2016 01:22 PM, Richard Biener wrote:
On August 18, 2016 8:25:18 PM GMT+02:00, Patrick Palka
wrote:
In comment #5 Yuri reports that r235653 introduces a runtime failure
for
176.gcc which I guess is caused by the combining step in
simplify_control_stmt_condition_1() not behaving properl
On 08/18/2016 04:05 AM, Richard Earnshaw (lists) wrote:
I think it's probably best to just drop the entire parenthetical
subcluase. This is documentation of how to use MULTILIB_EXCEPTIONS not
precise documentation on what needs to be done on ARM.
In fact, it might be better to just rewrite the
On 08/16/2016 11:55 PM, Senthil Kumar Selvaraj wrote:
Jeff Law writes:
On 08/11/2016 01:40 AM, Senthil Kumar Selvaraj wrote:
Hi,
The below patch adds the AVR target to the list of targets that don't
have natural_alignment_32. It also skips ipa/propalign-*.c
tests (which expect 4 byte a
On 08/18/2016 03:33 AM, Pierre-Marie de Rodat wrote:
Hello,
A check in dwarf2out_imported_module_or_decl prevents valid strict
DWARF2 constructs such as DW_TAG_imported_declaration from being emitted
in dwarf2out_imported_module_or_decl_1.
The latter already protects the emission of newer DWARF
On 08/12/2016 01:13 PM, David Malcolm wrote:
On Thu, 2016-08-11 at 20:00 -0600, Sandra Loosemore wrote:
On 08/11/2016 02:34 PM, David Malcolm wrote:
I sometimes find myself scouring assembler output from the compiler
and trying to figure out which instructions correspond to which
lines of sourc
Tested on x86_64-linux-gnu and checked in as r239590.
Thanks!
--
Regards,
Tim Shen
If no one objects, I would like to commit the
following patch on Saturday.
Note, Bud Davis found the location of the problem for
free-form code and proposed a fixed. I've modified his
proposed patch and applied a similar change for fixed-form.
Previously, gfortran would issue a warning and rejec
On 08/16/2016 06:05 PM, Thomas Schwinge wrote:
> commit bffb0ee6c0a83b8c85cd919e1172086b51fdc452
> Author: tschwinge
> Date: Wed Aug 17 00:55:02 2016 +
>
> Repeated use of the C/C++ OpenACC routine directive
>
> gcc/
> * omp-low.c (verify_oacc_routine_clauses): Change
On 08/08/2016 07:58, Jiong Wang wrote:
>
> Adhemerval Zanella writes:
>
>>> diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config/aarch64/aarch64.c
>>> index e56398a..2cf239f 100644
>>> --- a/gcc/config/aarch64/aarch64.c
>>> +++ b/gcc/config/aarch64/aarch64.c
>>> @@ -3227,6 +3227,34 @@ aarch64
On Thu, 18 Aug 2016, Richard Biener wrote:
> On August 18, 2016 8:25:18 PM GMT+02:00, Patrick Palka
> wrote:
> >In comment #5 Yuri reports that r235653 introduces a runtime failure
> >for
> >176.gcc which I guess is caused by the combining step in
> >simplify_control_stmt_condition_1() not behav
The testcase pr33738.C for warning fails with early-vrp patch. The
reason is, with early-vrp ccp2 is folding the comparison that used to
be folded in simplify_stmt_for_jump_threading. Since early-vrp does
not perform jump-threading is not optimized there.
Attached patch adds this warning to tree-s
On 19 August 2016 at 12:09, Kugan Vivekanandarajah
wrote:
> The testcase pr33738.C for warning fails with early-vrp patch. The
> reason is, with early-vrp ccp2 is folding the comparison that used to
> be folded in simplify_stmt_for_jump_threading. Since early-vrp does
> not perform jump-threading
This fixes (ISTR we've been here before)
slpeel_duplicate_current_defs_from_edges being dependent on PHI node
ordering which can be disrupted by SSA update inserting virtual PHIs.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2016-08-19 Richard Biener
PR tr
80 matches
Mail list logo
