On Mon, Jan 11, 2016 at 04:41:32PM +, Kyrill Tkachov wrote:
> Hi all,
>
> This patch fixes the test gcc.target/aarch64/pr66776.c for -mcpu=cortex-a53.
> Currently we don't handle the (if_then_else (cond) (zero_extend r1)
> (zero_extend r2))
> form of CSEL, so we end up recursing into the oper
On Mon, Jan 11, 2016 at 11:56:50AM +, Jiong Wang wrote:
> There are quite a few redundant type conversions in arm_neon.h, all of
> them are intrinsics taking argument of vector float type and return result
> of vector unsigned integer type.
>
> The problem is currently we support UNOP and UNOP
From: Sebastian Pop
* graphite-sese-to-poly.c (set_scop_parameter_dim): Remove.
(cleanup_loop_iter_dom): Remove.
(build_loop_iteration_domains): Remove.
(build_scop_context): Remove.
(build_scop_iteration_domain): Remove.
(add_loop_constraints): New
On Jan 15, 2016, at 2:47 AM, David Malcolm wrote:
> FWIW, I do something similar in multiline.exp's _build_multiline_regex,
> which attempts to have a complete list of metacharacters (though I
> believe some of these are not valid for POSIX filenames);
Only ‘\’ and ‘\0’ are invalid. The rest are
On 01/12/2016 11:11 AM, Martin Sebor wrote:
On 01/11/2016 10:20 PM, Jason Merrill wrote:
On 12/22/2015 09:32 PM, Martin Sebor wrote:
+ if (is_attribute_p ("aligned", name)
+ || is_attribute_p ("vector_size", name))
+{
+ /* Attribute argument may be a dependent indentifier. */
+
On Jan 15, 2016, at 2:37 AM, Jakub Jelinek wrote:
> HSA Foundation grants express permission to any current Founder, Promoter,
> Supporter Contributor, Academic or Associate member of HSA Foundation to
> copy and redistribute UNMODIFIED versions of this specification
So, this isn’t the GNU way.
Calling redirect_edge_var_map_empty after each pass was slowing things
down because hash_table::empty () cleared all slots even if the hash
table was already empty.
Tested on x86_64-linux-gnu, where it gives a 1% compile time improvement
for fold-const.ii at -O and -O2. OK to install?
Thanks,
Ri
On Wed, 2016-01-13 at 10:00 -0500, David Malcolm wrote:
> Ping: https://gcc.gnu.org/ml/gcc-patches/2016-01/msg00250.html
>
> On Wed, 2016-01-06 at 09:50 -0500, David Malcolm wrote:
> > The attached patch adds information on various things to the
> > gcc-6/changes.html page:
> >
> > * source-range
On Fri, Jan 15, 2016 at 17:45:22 +0100, Jakub Jelinek wrote:
> On Fri, Jan 15, 2016 at 07:38:14PM +0300, Ilya Verbin wrote:
> > On Fri, Jan 15, 2016 at 17:09:54 +0100, Jakub Jelinek wrote:
> > > On Fri, Jan 15, 2016 at 05:02:34PM +0100, Martin Jambor wrote:
> > > > How do other accelerators cope wi
On Fri, Jan 15, 2016 at 06:00:10PM +, Richard Sandiford wrote:
> Calling redirect_edge_var_map_empty after each pass was slowing things
> down because hash_table::empty () cleared all slots even if the hash
> table was already empty.
>
> Tested on x86_64-linux-gnu, where it gives a 1% compile
On Thu, 2016-01-14 at 21:50 -0600, Peter Bergner wrote:
> This patch adds support for __builtin_cpu_init(), __builtin_cpu_is() and
> __builtin_cpu_supports() builtins for PowerPC. We use the same API as the
> x86* builtins of the same name. These builtins uses the new GLIBC 2.23
> feature where w
The crux of the problem in this PR is that type_dependent_expression_p
returns true for a FUNCTION_DECL that is not actually type-dependent.
This leads tsubst_decl to attempt to perform template argument
substitution on the template arguments of the FUNCTION_DECL, which
do not necessarily correspon
On 01/15/2016 07:00 PM, Richard Sandiford wrote:
Calling redirect_edge_var_map_empty after each pass was slowing things
down because hash_table::empty () cleared all slots even if the hash
table was already empty.
Tested on x86_64-linux-gnu, where it gives a 1% compile time improvement
for fold-
Marcin,
your implementation looks very good to me. Thanks!
But please be aware that we deprecated the support of g5 and g6 and intend to
remove that code from
the back-end with the next GCC version. So I would prefer if you could remove
all the
!TARGET_CPU_ZARCH stuff from the implementation a
Does this version look better? I am not sure if this the right place
to put the new helper, so let me know if there is a better spot for it.
gcc/testsuite/lib/
* prune.exp (prune_file_path): Sanitize path used in regex.
(escape_regex_chars): New.
Signed-off-by: Zachary T
On Tue, Oct 20, 2015 at 4:37 PM, Bernd Schmidt wrote:
> On 10/15/2015 12:37 PM, H.J. Lu wrote:
>>
>> On Thu, Oct 15, 2015 at 1:44 AM, Richard Biener
>> wrote:
>>>
>>> On Wed, Oct 14, 2015 at 6:21 PM, H.J. Lu wrote:
By default, there is no visibility on builtin functions. When there is
On Jan 15, 2016, at 10:40 AM, Zachary T Welch wrote:
> Does this version look better?
Ok.
> I am not sure if this the right place to put the new helper, so let me know
> if there is a better spot for it.
So, someone that wants to rehome the helper is free to do that.
On Fri, 15 Jan 2016, Jakub Jelinek wrote:
> On Fri, Jan 15, 2016 at 07:38:14PM +0300, Ilya Verbin wrote:
> > On Fri, Jan 15, 2016 at 17:09:54 +0100, Jakub Jelinek wrote:
> > > On Fri, Jan 15, 2016 at 05:02:34PM +0100, Martin Jambor wrote:
> > > > How do other accelerators cope with the situation wh
On Fri, Jan 15, 2016 at 10:19:13PM +0300, Alexander Monakov wrote:
> Sorry, can you clarify -- what do you mean by "can't offload"?
I meant stuff like setjmp/longjmp, exceptions?, alloca (I know your changes
might fix this one), computed goto, non-local goto, and the like, which I
believe nvptx do
The following patch fixes
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=69030
The patch was bootstrapped and tested on x86 and x86-64.
Committed as rev. 232445.
2016-01-15 Vladimir Makarov
PR rtl-optimization/69030
* lra-spills.c (remove_pseudos): Check nrefs and make
There can be multiple gcc_options instances, each with
a call to
init_options_struct
matched with a call to
finalize_options_struct
whereas the
opts_obstack
is a singleton. Each gcc_options instance can potentially use the
opts_obstack singleton.
r230264 (aka 25faed340686df8d7bb2242dc8d0428
It was difficult to track down the memory corruption bug fixed by the
previous patch (PR jit/68446). The following patch attempts to make
it easier to find that kind of thing by adding "poisoning" code:
(A) when memory blocks are returned to the memory_block_pool's free
list (e.g. by an obsta
On Fri, 15 Jan 2016, Jakub Jelinek wrote:
> On Fri, Jan 15, 2016 at 10:19:13PM +0300, Alexander Monakov wrote:
> > Sorry, can you clarify -- what do you mean by "can't offload"?
>
> I meant stuff like setjmp/longjmp, exceptions?, alloca (I know your changes
> might fix this one), computed goto, n
On Fri, Jan 15, 2016 at 10:46:32PM +0300, Alexander Monakov wrote:
> On Fri, 15 Jan 2016, Jakub Jelinek wrote:
>
> > On Fri, Jan 15, 2016 at 10:19:13PM +0300, Alexander Monakov wrote:
> > > Sorry, can you clarify -- what do you mean by "can't offload"?
> >
> > I meant stuff like setjmp/longjmp, e
* graphite-isl-ast-to-gimple.c: Fix comment.
* graphite-scop-detection.c (defined_in_loop_p): New.
(canonicalize_loop_closed_ssa): Do not add close phi nodes for SSA
names defined in loop.
gcc/testsuite
* gcc.dg/graphite/pr68976.c: New test.
---
gcc/graphi
On Fri, Jan 15, 2016 at 11:19 AM, Sebastian Pop wrote:
> On Fri, Jan 15, 2016 at 7:58 AM, Tom de Vries wrote:
>> During scop detection/canonicalize_loop_closed_ssa_form, an exit phi is
>> introduced in the loop for _24:
>> ...
>> :
>> # _58 = PHI <_24(22)>
>> ...
>> Note that _24 is not defin
On Tue, Jan 12, 2016 at 03:23:32PM +0100, Jakub Jelinek wrote:
> But looking at GOMP_PLUGIN_target_task_completion, I see we have a bug in
> there,
> gomp_mutex_lock (&team->task_lock);
> if (ttask->state == GOMP_TARGET_TASK_READY_TO_RUN)
> {
> ttask->state = GOMP_TARGET_TASK_FINISHED
Hi!
I've noticed
../../gcc/ada/adaint.c: In function 'char* __gnat_locate_exec_on_path(char*)':
../../gcc/ada/adaint.c:2799:34: warning: deprecated conversion from string
constant to 'char*' [-Wwrite-strings]
if (path_val == NULL) path_val = "";
^
warning, fix
Hi!
In our gcc package build, libgfortran is built with -Werror=format-security
and errors on this file. While it is a false positive, because
cmdmsg_values[i] for any valid i don't contain % characters, IMNSHO it is
better to use "%s", msg anyway to make it clear that msg should not be
interpret
> I've noticed
> ../../gcc/ada/adaint.c: In function 'char*
> __gnat_locate_exec_on_path(char*)':
> ../../gcc/ada/adaint.c:2799:34: warning: deprecated conversion from
> string constant to 'char*' [-Wwrite-strings]
>if (path_val == NULL) path_val = "";
> ^
> wa
Hi!
As discussed in bugzilla some time ago, this patch decreases size of
cp_token on 64-bit hosts from 24 bytes to 16 bytes and on 32-bit
hosts from 16 bytes to 12. As for C++ all tokens are preparsed, it is quite
important.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2
Hi Jakub,
Of course, that's OK; obvious even - good for trunk.
Thanks
Paul
On 15 January 2016 at 21:07, Jakub Jelinek wrote:
> Hi!
>
> In our gcc package build, libgfortran is built with -Werror=format-security
> and errors on this file. While it is a false positive, because
> cmdmsg_values[i
On January 15, 2016 6:50:51 PM GMT+01:00, Mike Stump
wrote:
>On Jan 15, 2016, at 2:37 AM, Jakub Jelinek wrote:
>> HSA Foundation grants express permission to any current Founder,
>Promoter,
>> Supporter Contributor, Academic or Associate member of HSA Foundation
>to
>> copy and redistribute UNMO
Dear All,
Following an exchange with Dominique on #gfortran, I fixed PR54070
comment #23. The changes are in trans-array.c and are listed in the
ChangeLogs below.
Committed to trunk as revision 232450. I will wait some weeks before
committing to 5-branch. This patch should have made deferred char
OK.
Jason
OK.
Jason
Adding a testcase and reverting this patch:
[PATCH] remove parameter_rename_map
This map was used in the transition to the new scop detection: with the new scop
detection, we do not need this map anymore.
* graphite-isl-ast-to-gimple.c (gcc_expression_from_isl_ast_expr_id):
Remove u
On 01/14/2016 07:00 PM, Martin Sebor wrote:
Among the bugs fixed by the flexible array patch (r231665) was
c++/68490. I forgot to include a test for this bug in the commit
so I'm adding it via the attached patch.
(Please let me know if adding new passing tests is considered
trivial and I don't
On 01/14/2016 10:01 PM, Martin Sebor wrote:
In anticipation of needing to do something I put together the attached
patch that rolls this change into version 10, letting version 9 and
prior roll it back. I also mention it in the manual. What the patch
doesn't do is add a warning.
Looks good, b
On 15/01/16 19:38, Andreas Krebbel wrote:
Marcin,
your implementation looks very good to me. Thanks!
But please be aware that we deprecated the support of g5 and g6 and intend to
remove that code from
the back-end with the next GCC version. So I would prefer if you could remove
all the
!TARG
On 01/15/2016 01:04 PM, David Malcolm wrote:
It was difficult to track down the memory corruption bug fixed by the
previous patch (PR jit/68446). The following patch attempts to make
it easier to find that kind of thing by adding "poisoning" code:
(A) when memory blocks are returned to the memo
On Fri, 15 Jan 2016, Prathamesh Kulkarni wrote:
> On 15 January 2016 at 03:27, Joseph Myers wrote:
> > On Thu, 14 Jan 2016, Prathamesh Kulkarni wrote:
> >
> >> Hi,
> >> For test-case containing only the following declaration:
> >> static struct undefined_struct object;
> >> gcc rejects it at -O0
See the PR for details, but basically, the plus operations are special so you
can't just split out one of the alternatives to a different pattern.
This merges the two-instruction add case back into the main plus pattern, and
then adds peepholes and splitters to generate the same code as before.
O
Hi,
In straight-line strength reduction, it can sometimes happen that more
than one conditional candidate can be predicated upon a PHI node P.
During processing of the first conditional candidate, a new PHI node may
be introduced in the same block as P, with new computations introduced
in predeces
On 01/14/2016 11:14 AM, Jeff Law wrote:
On 01/14/2016 12:49 AM, Jakub Jelinek wrote:
On Thu, Jan 14, 2016 at 08:46:43AM +0100, Jakub Jelinek wrote:
On Thu, Jan 14, 2016 at 12:38:52AM -0700, Jeff Law wrote:
+ /* An integral type with more precision, but the object
+ only takes on values [0
On 01/13/2016 01:10 AM, Jeff Law wrote:
I'm going to focus on adpcm for the moment, in particular adpcm_coder.
It appears the key blocks are:
;; basic block 14, loop depth 1, count 0, freq 9100, maybe hot
;;prev block 13, next block 15, flags: (NEW, REACHABLE)
;;pred: 12 [100.0
This patch changes the function cilk_gimplify_call_params_in_spawned_fn to use
gimplify_arg instead of gimplify_expr. It fixes an ICE when calling a function
with a constructed empty class as the argument.
Bootstrapped and regression tested on x86_64-linux.
2016-01-15 Ryan Burn
PR c
On Thu, 2016-01-14 at 17:58 +, Jonathan Wakely wrote:
> On 07/01/16 17:47 +0100, Torvald Riegel wrote:
> >The attached patch makes some exceptions transaction-safe, as require by
> >the Transactional Memory TS. I believe I addressed all feedback for the
> >previous version of this patch (in pa
On 13/01/16 16:26 +, Jonathan Wakely wrote:
On 08/01/16 13:59 +, Jonathan Wakely wrote:
I'm only checking for those functions for *-*-*gnu* targets, as I
don't know of any other targets where it's an issue. Solaris and
the BSDs don't define those functions. If it affects other targets we
On 01/14/2016 01:55 AM, Jeff Law wrote:
[ Replying to myself again, mostly to make sure we've got these thoughts
in the archives. ]
Anyway, going back to adpcm_decode, we do end up splitting this path:
# vpdiff_12 = PHI
if (sign_41 != 0)
goto ;
else
goto ;
;;succ:
The PR is actually due to a defect in the standard, which I reported
today. The reporter said we're missing a check for is_constructible that would ensure we go to bullet (9.4)
and make the example in the PR ill-formed. I didn't add that check
because it's redundant, we don't need to check is_cons
On Fri, 8 Jan 2016, Andrew Pinski wrote:
> On Fri, Jan 8, 2016 at 4:05 PM, Joseph Myers wrote:
> > The Octeon 3 processor does not support the MIPS paired-single
> > instructions. This results in illegal instruction errors in the
> > testsuite when vectorization tests try to use those instructio
Hi,
bootstrapping on i686-linux revealed the need for the following simple
patch. I've run into two types of compilation errors on
powerpc-ibm-aix (no htolenn functions and ASM_GENERATE_INTERNAL_LABEL
somehow expanding to undeclared rs6000_xcoff_strip_dollar). I plan to
workaround them quickly b
This removes stray closing braces in the docs for dg-error, dg-warning
etc.
OK for trunk?
commit 1cb064263cfcfa14da81585886750f01a5611c7e
Author: Jonathan Wakely
Date: Sat Jan 16 00:11:27 2016 +
* doc/sourcebuild.texi (Directives): Remove extra closing braces.
diff --git a/gcc/doc/
Hi!
The following testcase is miscompiled on i686-linux at -O3.
The bug is in DSE record_store, which for group_id < 0 uses mem_addr
set to result of get_addr (base->val_rtx) (plus optional offset),
which is fine for canon_true_dependence with other MEMs in that function,
but we also store that ad
On Fri, Jan 15, 2016 at 03:32:33PM -0700, Jeff Law wrote:
> +bool
> +ssa_name_has_boolean_range (tree op)
> +{
> + gcc_assert (TREE_CODE (op) == SSA_NAME);
> +
> + /* Boolean types always have a range [0..1]. */
> + if (TREE_CODE (TREE_TYPE (op)) == BOOLEAN_TYPE)
> +return true;
> +
> + /*
> -Original Message-
> From: Myers, Joseph
> Sent: Friday, January 15, 2016 6:28 PM
> To: Andrew Pinski
> Cc: GCC Patches; Moore, Catherine; matthew.fort...@imgtec.com
> Subject: Ping Re: Handle Octeon 3 not supporting MIPS paired-single
> instructions
>
> On Fri, 8 Jan 2016, Andrew Pins
Our code for printing source code can apply an x-offset when printing very
wide source lines, which attempts to ensure that the caret will be printed
before line-wrapping occurs (it doesn't attempt to prevent line-wrapping,
but the old implementation didn't either).
The current implementation has
On 01/04/2016 07:32 AM, Ajit Kumar Agarwal wrote:
-Original Message- From: Jeff Law [mailto:l...@redhat.com]
Sent: Wednesday, December 23, 2015 12:06 PM To: Ajit Kumar Agarwal;
Richard Biener Cc: GCC Patches; Vinod Kathail; Shail Aditya Gupta;
Vidhumouli Hunsigida; Nagaraju Mekala Subje
On January 15, 2016 11:06:40 PM GMT+01:00, Bill Schmidt
wrote:
>Hi,
>
>In straight-line strength reduction, it can sometimes happen that more
>than one conditional candidate can be predicated upon a PHI node P.
>During processing of the first conditional candidate, a new PHI node
>may
>be introdu
101 - 160 of 160 matches
Mail list logo
