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:
Looking at adpcm_decoder we have the same idiom as in adpcm_coder:
if (bufferstep_79 != 0)
goto ;
else
goto ;
;;succ: 6 [50.0%] (
Hi!
On Fri, 8 Jan 2016 09:16:55 +0100, Richard Biener wrote:
> On Wed, 16 Dec 2015, Tom de Vries wrote:
> > On 10/12/15 14:14, Tom de Vries wrote:
> > > > On Thu, 3 Dec 2015, Tom de Vries wrote:
> > > > > Essentially we have two situations:
> > > > > - in the host compiler, there is no need for t
On Thu, 14 Jan 2016, Thomas Schwinge wrote:
> Hi!
>
> On Fri, 8 Jan 2016 09:16:55 +0100, Richard Biener wrote:
> > On Wed, 16 Dec 2015, Tom de Vries wrote:
> > > On 10/12/15 14:14, Tom de Vries wrote:
> > > > > On Thu, 3 Dec 2015, Tom de Vries wrote:
> > > > > > Essentially we have two situation
On Wed, Jan 13, 2016 at 9:04 PM, Tom de Vries wrote:
> Hi,
>
> At r231739, there was an ICE when checking code generated by
> oacc_xform_loop, in case the source contained an error.
>
> Due to seen_error (), gimplification during oacc_xform_loop bailed out, and
> an uninitialized var was introduce
On Thu, Jan 14, 2016 at 8:38 AM, Jeff Law wrote:
>
> As noted in the BZ, DOM does not exploit VRP information to create
> additional equivalences in the arms of conditionals.
>
> This can cause DOM to miss relatively simple optimizations that show up in
> the adpcm benchmark as well as within GCC
If/when mips-ps-3d.md is moved from reduc_* to reduc_*_scal optabs (patch here:
https://gcc.gnu.org/ml/gcc-patches/2016-01/msg00311.html ), there will be no
uses of the old reduc_* optabs remaining. This patch removes those optabs and
the migration path.
Bootstrapped + check-gcc + check-g++ on x86
(Previous message: https://gcc.gnu.org/ml/gcc-patches/2015-12/msg02159.html)
On Sat, Dec 26, 2015 at 18:58 PM, Richard Biener
wrote:
>> I'm not sure whether adding a pass_copy_prop is the right thing here, but
>> since
>> loop-header-copying can create such opportunities, it seems good to take
This PR is a request for better array bounds diagnostics. The C++ FE says
"multidimensional array must have bounds for all dimensions except the first",
and we can do something similar in the C FE.
Bootstrapped/regtested on x86_64-linux, ok for trunk?
2016-01-14 Marek Polacek
PR c/69
On 07/01/16 12:47, Alan Lawrence wrote:
Here's an updated version, also covering the min/max patterns I missed before.
I've now managed to do some testing with a stage 1 compiler, by compiling all
tests in gcc.dg/vect at -O2 -ftree-vectorize -mips3d -march=mips64r2 -mabi=n32
$x -ffast-math -ffini
Alan Lawrence writes:
> On 07/01/16 12:47, Alan Lawrence wrote:
> > Here's an updated version, also covering the min/max patterns I missed
> before.
> > I've now managed to do some testing with a stage 1 compiler, by
> > compiling all tests in gcc.dg/vect at -O2 -ftree-vectorize -mips3d
> > -march
On Mon, Jan 11, 2016 at 11:46:21AM +0100, Dominik Vogt wrote:
> The following two patches fix some test cases for S/390.
Patch 4:
Adapt the test pattern in bind_c_array_params_2.f90 to S/390
assemply output.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/testsuite/ChangeLog
The attached patch fixes a test failure because of a missing full
stop at the end of an S/390 option help text.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/ChangeLog
* config/s390/s390.opt (mtune): Add full stop.
>From 6d4de7bb9ff9643b4cf3c967ea5ccfd1591d0799 Mon Sep 17 00:0
Now bootstrapped and regtested against GCC v5.
OK for 5-branch thus?
Regards,
Alexander
On Wed, Jan 13, 2016 at 03:41:26PM +0300, Kirill Yukhin wrote:
> Hello Sasha,
> On 12 Jan 14:57, Alexander Fomin wrote:
> > This patch addresses PR target/69228. Expanding non-mask builtins
> > for prefetch ga
On Thu, Jan 14, 2016 at 11:26 AM, Alan Lawrence wrote:
> If/when mips-ps-3d.md is moved from reduc_* to reduc_*_scal optabs (patch
> here:
> https://gcc.gnu.org/ml/gcc-patches/2016-01/msg00311.html ), there will be no
> uses of the old reduc_* optabs remaining. This patch removes those optabs and
On Thu, Jan 14, 2016 at 11:29 AM, Alan Lawrence wrote:
> (Previous message: https://gcc.gnu.org/ml/gcc-patches/2015-12/msg02159.html)
> On Sat, Dec 26, 2015 at 18:58 PM, Richard Biener
> wrote:
>
>>> I'm not sure whether adding a pass_copy_prop is the right thing here, but
>>> since
>>> loop-he
On Wed, Jan 13, 2016 at 06:39:26PM +0100, Martin Jambor wrote:
> Hi,
>
> this patch contains changes to the configuration mechanism and offload
> bits, so that users can build compilers with HSA support.
>
> It is a re-post of
> https://gcc.gnu.org/ml/gcc-patches/2015-12/msg00714.html, which, has
Hi Guys
Whilst checking PR 69194 I noticed that the new testcase
(gcc.target/arm/pr69194.c) was not being run with a Linux hosted
toolchain configured as --target=arm-eabi. Investigating further I
found that the check_effective_target_arm_neon_ok_nocache proc in
target_supports.exp was
On Wed, Jan 13, 2016 at 06:39:28PM +0100, Martin Jambor wrote:
> the patch below adds the HSA-specific plugin for libgomp. The plugin
> implements the interface mandated by libgomp and takes care of finding
> any available HSA devices, finalizing HSAIL code and running it on
> HSA-capable GPUs.
>
On Wed, Jan 13, 2016 at 06:39:30PM +0100, Martin Jambor wrote:
> @@ -726,14 +730,14 @@ struct GTY((tag("GSS_OMP_CONTINUE")))
>tree control_use;
> };
>
> -/* GIMPLE_OMP_SINGLE, GIMPLE_OMP_TEAMS, GIMPLE_OMP_ORDERED */
> +/* GIMPLE_OMP_SINGLE, GIMPLE_OMP_ORDERED */
I think in this patch GIMPLE
On Wed, Jan 13, 2016 at 06:39:31PM +0100, Martin Jambor wrote:
> Hi,
>
> the pass manager changes required for HSA have already been committed
> to trunk so all that remains are these additions to the pass pipeline.
>
> This bit has already been approved by Richi in
> https://gcc.gnu.org/ml/gcc-p
On Wed, Jan 13, 2016 at 06:39:32PM +0100, Martin Jambor wrote:
> + cgraph_node *clone = node->create_virtual_clone
> + (vec (), NULL, NULL, "hsa");
Nicer formatting would be
cgraph_node *clone
= node->create_virtual_clone (vec (),
On Wed, Jan 13, 2016 at 06:39:33PM +0100, Martin Jambor wrote:
> the following patch adds a BRIG (binary representation of HSAIL)
> representation description. It is within a single header file
> describing the binary structures and constants of the format.
>
> The file comes from the HSA Foundat
On 13 January 2016 at 18:17, Richard Henderson wrote:
> On 01/12/2016 08:53 AM, Richard Henderson wrote:
>>
>> The problem in this PR is that we never got around to flushing out the
>> vector
>> support for transactions for anything but x86. My goal here is to make
>> this as
>> generic as possib
Hi Christian,
* gcc.target/arm/attr-neon.c: Use dg-add-options to add the
command line options necessary to enable Neon support.
is this last one needed ? the __attribute__ ((target("fpu=neon"))) is
here to test without -mfpu=neon. So we are losing something here.
Ah, good point. We
On Wed, Jan 13, 2016 at 06:39:34PM +0100, Martin Jambor wrote:
> +struct function_linkage_pair
> +{
> + function_linkage_pair (tree decl, unsigned int off):
I would have expected a space before :, or better yet on the next line:
function_linkage_pair (tree decl, unsigned int off)
: function
On 01/13/16 17:44, Jakub Jelinek wrote:
On Wed, Jan 13, 2016 at 11:35:08PM +0100, Marek Polacek wrote:
On Wed, Jan 13, 2016 at 02:29:21PM -0800, Cesar Philippidis wrote:
--- a/gcc/gimplify.c
+++ b/gcc/gimplify.c
@@ -5994,6 +5994,11 @@ oacc_default_clause (struct gimplify_omp_ctx *ctx, tree
dec
Hi Martin,
while addressing Jakub's feedback can you also fix the typos below?
(noticed because I needed to reuse those hunks in gomp-nvptx)
On Wed, 13 Jan 2016, Martin Jambor wrote:
> +/* Return a target argument consisiting of DEVICE identifier, value
> identifier
> + ID, and the actual VALU
Hi!
During complex lowering, we were walking bbs in bb index # order, which
means sometimes we could visit SSA_NAME uses before visiting definitions.
Also, even if we walk in rpo order as the following patch does, sometimes
we need to first use SSA_NAMEs before their definitions are lowered at lea
Hi!
This PR is an ICE while expand_field_assignment attempts to "simplify"
some (zero_extract:DI (subreg:DI (reg:OI ...) 0) (const_int 32) (const_int 0))
or what on ia64. The problem is that ia64 and various other backends add
various very large integral modes (OI in this case, but x86_64 even XI
Hi!
I've bootstrapped/regtested on x86_64-linux and i686-linux the following fix
from Francois that has been sitting in the PR for a few months.
The problem is that the testcase assumed that the last real_kind
has maximum precision and maximum range. That is usually the case,
but is not the case
On Wed, Jan 13, 2016 at 06:39:35PM +0100, Martin Jambor wrote:
> +for (phi = hbb->m_first_phi;
> + phi;
> + phi = phi->m_next ? as_a (phi->m_next): NULL)
Space before :
Ok with that change.
Jakub
2016-01-14 Uros Bizjak
* lib/target-supports.exp (check_effective_target_issignaling):
New procedure.
* gcc.dg/pr61441.c: Require issignaling effective target.
Tested on x86_64-linux-gnu, CentOS 5.11 and Fedora 23.
OK for mainline?
Uros.
diff --git a/gcc/testsuite/gcc.dg/pr61441.
This applies some refactoring to vect_build_slp_tree to make a fix for
PR66856 easier (where I need to add some reference counting code to
stmts in SLP tree). Doing this w/o refactoring turned out to be
a "bit" unwieldly.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2
On 01/14/2016 03:31 PM, Jakub Jelinek wrote:
PR target/68269
* combine.c (expand_field_assignment): Punt if compute_mode is
unsupported scalar mode.
Ok.
Bernd
On Wed, Jan 13, 2016 at 09:05:47PM +0300, Alexander Monakov wrote:
> I'm testing 'target teams' implementation for gomp-nvptx branch, and I'm
> seeing test failures that, I think, are caused by faulty tests (unlike NVPTX,
> MIC offloading uses 1 team similar to native execution, so the issues are n
On 11/01/16 16:39, Jakub Jelinek wrote:
On Mon, Jan 11, 2016 at 05:11:21PM +0100, Christophe Lyon wrote:
I tested a similar version on my side. It just makes the test become
UNSUPPORTED for arm/aarch64 + newlib. They used to pass, though.
Is anything bad on that? The test tests functions that
Hi Christian,
That works in my local tests. Are you OK with it ?
yeah, I checked that it tests the attribute in my configuration as well.
just to make things less obfuscated, What about /* {
dg-additional-options "-mfpu=fp-armv8 } */ instead, eventually with a
selector for armv7-a.
just an id
In my fix for 67104 I knew that I wasn't handling RANGE_EXPRs, but
thought that it wasn't necessary, and left an assert to make sure.
Apparently I was wrong, so this patch implements the necessary range
splitting when the constexpr code wants to store to a single element of
a range.
Tested x8
Hello Sasha,
On 14 Jan 15:08, Alexander Fomin wrote:
> Now bootstrapped and regtested against GCC v5.
> OK for 5-branch thus?
Your patch is OK for gcc-5 branch.
--
Thanks, K
>
> Regards,
> Alexander
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2016-01-14 Richard Biener
PR tree-optimization/68060
* tree-vect-loop.c (vect_is_simple_reduction): Check the
outer loop reduction is only used in the inner loop before
detecting a double r
On Thu, 14 Jan 2016, Jakub Jelinek wrote:
> Hi!
>
> During complex lowering, we were walking bbs in bb index # order, which
> means sometimes we could visit SSA_NAME uses before visiting definitions.
> Also, even if we walk in rpo order as the following patch does, sometimes
> we need to first us
Hi Christian,
On 14/01/16 12:37, Christian Bruel wrote:
Here is the rebased patch after the #pragma GCC target warning fixes.
I also disabled the builtins initialisations when float-abi is solft as you suggested
This is ok with a couple of nits below addressed:
+/* Set up all the NEON builti
> -Original Message-
> From: Matthew Fortune [mailto:matthew.fort...@imgtec.com]
> Sent: Thursday, January 14, 2016 5:49 AM
> To: Alan Lawrence; Moore, Catherine
> Cc: gcc-patches@gcc.gnu.org
> Subject: RE: [PATCH][MIPS] Migrate reduction optabs in mips-ps-3d.md
>
> Alan Lawrence writes
I am currently testing the following patch to fix PR69117 where
during value-numbering we end up using SSA name info (points-to info
in this case but also range-info in general) from the value leader
which may be defined under different conditions than the use is.
The only fix I can think of appr
PING.
On Tue, Jan 5, 2016 at 8:07 AM, H.J. Lu wrote:
> On Mon, Dec 14, 2015 at 2:08 PM, H.J. Lu wrote:
>> On Mon, Dec 14, 2015 at 12:43 PM, Jason Merrill wrote:
>>> On 12/14/2015 03:39 PM, H.J. Lu wrote:
On Mon, Dec 14, 2015 at 12:16 PM, Jason Merrill wrote:
>
> On 12/12/2015
On 14/01/16 15:02, Andre Vieira (lists) wrote:
Unfortunately c99_functions is a very wide net. For instance, newlib supports
the ceill, but doesn't support
wscanf_s nor any bounds checking function I think.
wscanf_s is not c99
(it is in the optional annex k of c11, which is
likely to be remov
On Thu, 14 Jan 2016, Jakub Jelinek wrote:
> That is weird, because c is also a firstprivate var in target and
> (implicitly) shared in teams, so if omp lowering/expansion is not buggy,
> you should see the exactly same problem with that.
> Wonder if we use GOMP_MAP_FIRSTPRIVATE_INT when we shouldn'
On Thu, Jan 14, 2016 at 07:08:23PM +0300, Alexander Monakov wrote:
> On Thu, 14 Jan 2016, Jakub Jelinek wrote:
> > That is weird, because c is also a firstprivate var in target and
> > (implicitly) shared in teams, so if omp lowering/expansion is not buggy,
> > you should see the exactly same probl
Hi,
This patch removes an attribute which was not required
and replaces the previous checks with an alternative
function call.
Regtested with x86_64 and committed to gomp4.
Jim
Index: gcc/c/ChangeLog.gomp
===
--- gcc/c/ChangeLog
On 14/01/16 12:22, Richard Biener wrote:
On Thu, Jan 14, 2016 at 11:26 AM, Alan Lawrence wrote:
If/when mips-ps-3d.md is moved from reduc_* to reduc_*_scal optabs (patch here:
https://gcc.gnu.org/ml/gcc-patches/2016-01/msg00311.html ), there will be no
uses of the old reduc_* optabs remaining.
On S/390 we have address style shift counts. So it is possible to
have a reg + const_int shift count. To address this it appeared to be
convenient to really handle this as address operand. For that reason
the "Y" constraint used for shift counts is an extra memory
constraint. Unfortunately a sh
Hi,
For test-case containing only the following declaration:
static struct undefined_struct object;
gcc rejects it at -O0 in assemble_variable() with error "storage size
of is unknown",
however no error is reported when compiled with -O2.
AFAIU that happens because at -O2, analyze_function() remov
So far whenever we wanted to disable an alternative we have used mode
attributes emitting constraints matching an earlier alternative
assuming that due to this the later alternative will never be chosen.
With this patch a `disabled' attribute is defined which can be used to
explicitly disable an a
This patch introduces substitution patterns to add PLUS const_int, and
AND operands to patterns and uses this to rewrite the existing rotate
pattern.
gcc/ChangeLog:
2016-01-14 Andreas Krebbel
* config/s390/predicates.md (addrreg_or_constint_operand)
(const_int_6bitset_operand)
This removes the Y constraint from the tabort pattern definition. In
this case it is easier without using substitutions.
gcc/ChangeLog:
2016-01-14 Andreas Krebbel
* config/s390/s390.md ("*tabort_1"): Change predicate to
addrreg_or_constint_operand. Add a second alternative t
This finally removes the Y constraint from the vector patterns while
folding some of them using a code iterator.
gcc/ChangeLog:
2016-01-14 Andreas Krebbel
* config/s390/subst.md (SUBST mode iterator): Add ashiftrt.
(DSI_VI): New mode iterator.
("addr_style_op_subst"):
After Y is never used anymore with SImode operands we can finally
disallow SImode (if != Pmode) in s390_decompose_address. In fact that
was the whole point of the patch series.
gcc/ChangeLog:
2016-01-14 Andreas Krebbel
* config/s390/s390.c (s390_decompose_address): Don't accept SImod
When processing substitutions the operands are renumbered. To find a
free operand number the array used_operands_numbers is used to record
the operand numbers already in use. Currently this array is used to
assign new numbers *before* all the RTXes in the vector have been
processed.
I did run in
The arithmetic shift patterns set also the condition code. This adds
more substitution potential. Depending on whether the actual result
or the CC output will be used 3 different variants of each of these
patterns are needed. This multiplied with the PLUS and the AND
operands from the earlier su
Hi all,
This the 4.9 backport of
https://gcc.gnu.org/ml/gcc-patches/2015-12/msg00708.html.
The gcc change is the same, but the testcase is put into gcc.dg/torture rather
than gcc.c-torture/execute.
This is because the testcase actually requires a -std=gnu99, and adding a
dg-options in execute.
On 01/14/2016 07:45 AM, Uros Bizjak wrote:
2016-01-14 Uros Bizjak
* lib/target-supports.exp (check_effective_target_issignaling):
New procedure.
* gcc.dg/pr61441.c: Require issignaling effective target.
Tested on x86_64-linux-gnu, CentOS 5.11 and Fedora 23.
OK for mainline?
With my fix for PR c++/21802 I changed build_min_non_dep_call_vec() to
additionally retain the KOENIG_LOOKUP_P flag of the non-dependent
expression that's been built. This change was a little too general
though, since retaining the KOENIG_LOOKUP_P flag is only necessary for
build_min_non_dep_op_ov
On 01/14/2016 07:35 AM, Jakub Jelinek wrote:
Hi!
I've bootstrapped/regtested on x86_64-linux and i686-linux the following fix
from Francois that has been sitting in the PR for a few months.
The problem is that the testcase assumed that the last real_kind
has maximum precision and maximum range.
On 01/14/2016 03:29 AM, Marek Polacek wrote:
This PR is a request for better array bounds diagnostics. The C++ FE says
"multidimensional array must have bounds for all dimensions except the first",
and we can do something similar in the C FE.
Bootstrapped/regtested on x86_64-linux, ok for trunk
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 particular, there are now more safety
checks for preconditions for th
OK.
Jason
Dear Jakub,
Thanks for that. OK by me too.
Paul
On 14 January 2016 at 18:36, Jeff Law wrote:
> On 01/14/2016 07:35 AM, Jakub Jelinek wrote:
>>
>> Hi!
>>
>> I've bootstrapped/regtested on x86_64-linux and i686-linux the following
>> fix
>> from Francois that has been sitting in the PR for a few
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..1] as determined by VRP
+ analysis.
In r230167 I made this testcase be tested on 64-bit as well, since it
now works. That was a tad optimistic for powerpc64le though. For now,
XFAIL it there.
Is this okay for trunk? Tested the testcase manually on powerpc64-linux
and powerpc64le-linux (that is, I ran "make check-gcc-c -k
RUNTESTF
On 01/14/2016 02:47 AM, Richard Biener wrote:
On Thu, Jan 14, 2016 at 8:38 AM, Jeff Law wrote:
As noted in the BZ, DOM does not exploit VRP information to create
additional equivalences in the arms of conditionals.
This can cause DOM to miss relatively simple optimizations that show up in
the
On 01/14/2016 11:23 AM, Segher Boessenkool wrote:
In r230167 I made this testcase be tested on 64-bit as well, since it
now works. That was a tad optimistic for powerpc64le though. For now,
XFAIL it there.
Is this okay for trunk? Tested the testcase manually on powerpc64-linux
and powerpc64le
On 01/13/2016 01:35 AM, Jakub Jelinek wrote:
Hi!
This patch fixes two issues (that could be committed separately, but the
testcase depends on both of them fixed).
The first one is in LRA, where my recent fix to move_plus_up fixed ICE on
the testcase, but actually turned it into a wrong-debug -
On 01/08/2016 02:40 PM, David Malcolm wrote:
Jakub had some concern about the use of sorry, so here's a revised
version of the patch, with the following changes:
- fixed the comment issues noted above.
- changed from a "sorry" to an "inform" (as per Jakub), passing
in the pertinent l
How do people feel about this approach to solving PR 69240?
The bug is that we don't define operator!= for RND::param_type, where
RND is any of random number distributions in or .
Rather than tediously defining it for every param_type I've added
this:
namespace __random_not_eq
{
// Derive
On Thu, 14 Jan 2016, Jonathan Wakely wrote:
How do people feel about this approach to solving PR 69240?
The bug is that we don't define operator!= for RND::param_type, where
RND is any of random number distributions in or .
Rather than tediously defining it for every param_type I've added
this
On 14/01/16 20:13 +0100, Marc Glisse wrote:
I didn't think about it much, but I am worried that __random_not_eq
will accidentally become an associated namespace for more classes than
we would expect.
Yes, it would be an associated namespace for types that derive from
the distributions, or clas
With this patch the substitution patterns added earlier are used for
the logical right shift and all the left shift patterns.
2016-01-14 Andreas Krebbel
* config/s390/s390.md ("*di3_31", "*3")
("*di3_31_and", "*3_and"): Merge into single
pattern definition ...
(
While trying to get rid of the Y constraint in the setmem patterns I
noticed that for these patterns it isn't even a problem since these
always only use the constraint with a Pmode match_operand. But while
being at it I've tried to fold some of the patterns a bit.
gcc/ChangeLog:
2016-01-14 Andr
On Thu, 14 Jan 2016, Jonathan Wakely wrote:
On 14/01/16 20:13 +0100, Marc Glisse wrote:
I didn't think about it much, but I am worried that __random_not_eq will
accidentally become an associated namespace for more classes than we would
expect.
Yes, it would be an associated namespace for typ
2016-01-14 20:21 GMT+01:00 Jonathan Wakely :
> We could constrain the generic operator== and operator!= to only match
> types that we want it to match, e.g. by having a type trait that is
> true for all our distributions and their parameter types. That would
> mean adding a specialization of it for
On 14/01/16 20:57 +0100, Marc Glisse wrote:
Once you constrain, you could even use 'std' as the associated
namespace ;-)
True! And std is already an associated namespace of the standard
distributions, so we wouldn't be making ADL look anywhere additional.
On 14/01/16 21:22 +0100, Daniel Krügler wrote:
If there were an __is_direct_base_of intrinsic (or is there?), it
seems to me that there would be no need for all these specializations
(each one nearly taking as much space as the explicit inline
definition of operator!) and there could be a single
While trying to get rid of the Y constraint in the setmem patterns I
noticed that for these patterns it isn't even a problem since these
always only use the constraint with a Pmode match_operand. But while
being at it I've tried to fold some of the patterns a bit.
gcc/ChangeLog:
2016-01-14 Andr
On 14/01/16 20:43 +, Jonathan Wakely wrote:
On 14/01/16 21:22 +0100, Daniel Krügler wrote:
If there were an __is_direct_base_of intrinsic (or is there?), it
seems to me that there would be no need for all these specializations
(each one nearly taking as much space as the explicit inline
defi
On 13.01.16 18:39, Bernd Schmidt wrote:
On 01/11/2016 07:37 PM, Andreas Tobler wrote:
+# Do a configure time check for -ldl
+AC_CHECK_LIB(dl, dlsym,
+ [link_sanitizer_common="-lrt $link_sanitizer_common"])
+
I'll give it a test run.
If that works (with -ldl instead of -lrt) it's ok.
It do
On 14/01/16 20:46 +, Jonathan Wakely wrote:
On 14/01/16 20:43 +, Jonathan Wakely wrote:
On 14/01/16 21:22 +0100, Daniel Krügler wrote:
If there were an __is_direct_base_of intrinsic (or is there?), it
seems to me that there would be no need for all these specializations
(each one nearly
On 01/10/2016 06:24 AM, Nicklas Bo Jensen wrote:
Hi,
In the GCC Internals documentation the loop depth is documented
incorrectly. From git commit 9e3536f4f it is accessed through a
function, not a field.
Please install this patch to reflect this.
Best,
Nicklas
2017-01-10 Nicklas Bo Jensen
On 01/10/2016 08:55 PM, Patrick Palka wrote:
On Mon, Jan 4, 2016 at 1:35 PM, Jeff Law wrote:
On 01/02/2016 04:26 PM, Patrick Palka wrote:
On Sat, Jan 2, 2016 at 3:21 AM, Jakub Jelinek wrote:
On Fri, Jan 01, 2016 at 10:06:34PM -0700, Jeff Law wrote:
gcc/cp/ChangeLog:
* cp-array-not
With this patch the substitution patterns added earlier are used for
the logical right shift and all the left shift patterns.
2016-01-14 Andreas Krebbel
* config/s390/s390.md ("*di3_31", "*3")
("*di3_31_and", "*3_and"): Merge into single
pattern definition ...
(
This patch adds a missing cleanup point to cilk_spawn expressions to
prevent an ICE when calling functions that return types with
non-trivial destructors.
Bootstrapped and regression tested on x86_64-linux.
2015-01-14 Ryan Burn
PR c++/69048
* cilk.c (create_cilk_wrapper_body): Call
On 14/01/16 20:41 +, Jonathan Wakely wrote:
On 14/01/16 20:57 +0100, Marc Glisse wrote:
Once you constrain, you could even use 'std' as the associated
namespace ;-)
True! And std is already an associated namespace of the standard
distributions, so we wouldn't be making ADL look anywhere ad
On 01/10/2016 08:20 PM, Patrick Palka wrote:
On Thu, Dec 31, 2015 at 10:40 AM, Patrick Palka wrote:
This patch makes it so that a gimplification failure is considered to be
an internal error under normal circumstances, so that we otherwise avoid
silently generating wrong code if e.g. a buggy fr
Hi!
Here is my attempt to fix https://gcc.gnu.org/bugzilla/show_bug.cgi?id=68463
This patch does 2 things:
I) lto-plugin doesn't claim files which contain offload sections, but don't
contain LTO sections. Instead, it writes names of files with offloading to the
temporary file and passes it to l
Predictably, I must have tweaked the patch after testing on x86.
Re-tested and committed.
r~
PR c/69272
PR tree-opt/68964
* trans-mem.c (tm_log_emit_stmt): Fix unit size to bit size.
* tree.c (build_tm_vector_builtins): Use builtin_decl_explicit_p
instead
On 01/14/2016 02:19 PM, Ryan Burn wrote:
This patch adds a missing cleanup point to cilk_spawn expressions to
prevent an ICE when calling functions that return types with
non-trivial destructors.
Bootstrapped and regression tested on x86_64-linux.
2015-01-14 Ryan Burn
PR c++/69048
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 in assemble_variable() with error "storage size
> of is unknown",
> however no error is reported when compiled with -O2
On Tue, Jan 12, 2016 at 06:48:58PM -0500, David Edelsohn wrote:
> On Tue, Jan 12, 2016 at 6:47 PM, Joseph Myers wrote:
> > On Tue, 12 Jan 2016, Michael Meissner wrote:
> >
> >> On Tue, Jan 12, 2016 at 12:18:55AM +, Joseph Myers wrote:
> >> > On Mon, 11 Jan 2016, Michael Meissner wrote:
> >> >
On 01/14/16 16:57, Joseph Myers wrote:
On Thu, 14 Jan 2016, Prathamesh Kulkarni wrote:
g++ rejects it during parsing. I tried similarly in C FE by adding a
check for decl with incomplete struct/union type in finish_decl(),
however that fails to compile the following case:
typedef struct foo fo
Could you push please? I don't have write access. Thanks - Ryan.
On Thu, Jan 14, 2016 at 4:57 PM, Jeff Law wrote:
> On 01/14/2016 02:19 PM, Ryan Burn wrote:
>>
>> This patch adds a missing cleanup point to cilk_spawn expressions to
>> prevent an ICE when calling functions that return types with
>
On 01/14/2016 03:14 PM, Ryan Burn wrote:
Could you push please? I don't have write access. Thanks - Ryan.
Will do.
Do you *want* write access? It's pretty easy to set up.
jeff
1 - 100 of 125 matches
Mail list logo
