On Mon, 3 Mar 2014, Richard Sandiford wrote:
AIUI:
Reading back the references don't yield any dissenting
flash-backs, FWIW.
So, a (use fp) then a (clobber fp)? That was probably just too
weird for me to think of, much like a hypercorrect ending of the
previous clause. :)
Thanks for dealing
When using Cilk Plus (-fcilkplus), it makes sense to automatically link
the run-time library (-lcilkrts).
This patch mimics libgomp by adding a .spec file; I am not 100% sure
whether the .spec file is needed, but the pthread tests in libgomp imply
that it makes sense. (libgomp also checks for
From: Jakub Jelinek [mailto:ja...@redhat.com]
n-size = n1.size;
+ for (i = 0, mask = 0xff; i n-size; i++, mask = 8)
This should be mask = BITS_PER_UNIT for consistency.
Indeed.
And, as has been said earlier, the testcase should be a runtime testcase
(in
Hi,
we have merged the gcc-4_8-branch into linaro/gcc-4_8-branch up to
revision 208264 as r208471.
This will be part of our 2014.03 release.
Yvan
On Mon, 10 Mar 2014, Rainer Orth wrote:
Richard Biener rguent...@suse.de writes:
Ouch. But as lto-plugin is a host module it should link against
the host libgcc, no? During stage1, that is. So the question is
why does it use the gcc/ compiler?
For me it's using the host gcc:
On 03/10/2014 08:13 PM, Uros Bizjak wrote:
OK for mainline SVN and release branches?
Sure. You don't need approval for pa
Thanks,
Andrew.
On 03/10/2014 08:13 PM, Uros Bizjak wrote:
OK for mainline SVN and release branches?
Sure. You don't need approval for patches that are obviously
correct/trivial.
Thanks,
Andrew.
The current Nios II prologue/epilogue code has a bug where the frame
pointer points to the start of the register save area, rather than the
frame slot where FP is saved (as specified the Nios II ABI).
This was only discovered relatively recently, as dwarf-based unwinding
is used most of the time,
Hi Mikael,
here is a fix for a wrong code issue, where we pass a descriptor with
broken bounds when the actual argument is a transposed array and the
dummy an assumed shape dummy.
The bug comes from the interaction of the transpose optimization,
which creates a descriptor with transposed
On Tue, Mar 11, 2014 at 02:53:39PM +0800, Thomas Preud'homme wrote:
+2014-03-10 Thomas Preud'homme thomas.preudho...@arm.com
+
+ PR tree-optimization/60454
+ * gcc.c-torture/execute/pr60454.c (fake_swap32): Testcase to track
+ regression of PR60454.
The ChangeLog entry is
Hi John,
On Tue, Mar 11, 2014 at 02:54:18AM +, John Carr wrote:
A comment in arm/sync.md notes We should consider issuing a inner
shareability zone barrier here instead. Here is my first attempt
at a patch to emit weaker memory barriers. Three instructions seem
to be relevant for user
From: Jakub Jelinek [mailto:ja...@redhat.com]
The ChangeLog entry is wrong, the file is new, so you should just say:
* gcc.c-torture/execute/pr60454.c: New test.
Done.
But more importantly:
--- /dev/null
+++ b/gcc/testsuite/gcc.c-torture/execute/pr60454.c
@@ -0,0 +1,25 @@
On Mon, Mar 10, 2014 at 10:29 AM, bin.cheng bin.ch...@arm.com wrote:
Hi,
When I investigating PR60363 which is caused by previous patch for PR60280,
I found there is a latent bug in remove_forwarder_block_with_phi because GCC
doesn't update loop's latch information. Without this patch,
On Mon, Mar 10, 2014 at 11:14 AM, Jakub Jelinek ja...@redhat.com wrote:
On Mon, Mar 10, 2014 at 11:10:05AM +0100, Uros Bizjak wrote:
Well, better is non-obvious, while it is smaller (which is good for
initialization and thus rarely executed code), the common case is that
*jcr_list is 0 (gcj
On Tue, Mar 11, 2014 at 06:48:37PM +0800, Thomas Preud'homme wrote:
I also added a typedef unsigned uint32_t for when sizeof(unsigned) == 4. I
hope it's right.
In theory you could have __CHAR_BIT__ different from 8 and what you care
about is that uint32_t has exactly 32 bits, so the check
Ping^2
http://gcc.gnu.org/ml/gcc-patches/2014-02/msg01487.html
Kyrill
On 05/03/14 09:37, Kyrill Tkachov wrote:
Ping?
On 25/02/14 10:08, Kyrill Tkachov wrote:
Hi all,
The problem solved in this patch is that when gcc is configured with
--with-arch=armv8-a gcc will go into
On Mon, Mar 10, 2014 at 8:22 PM, David Malcolm dmalc...@redhat.com wrote:
Gimple function dumps contain the types of parameters, but not of the
return type.
The attached patch fixes this omission; here's an example of the
before/after diff:
$ diff -up /tmp/pr23401.c.004t.gimple.old
On 25 February 2014 10:08, Kyrill Tkachov kyrylo.tkac...@arm.com wrote:
Hi all,
The problem solved in this patch is that when gcc is configured with
--with-arch=armv8-a gcc will go into aarch64-arches.def, pick the
representative CPU (Cortex-A53 for ARMv8-A) and use that CPUs ISA flags. Now
On 03/11/14 03:48, Thomas Preud'homme wrote:
diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 748805e..b6d7d93 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,8 @@
+2014-03-07 Thomas Preud'homme thomas.preudho...@arm.com
+
+ PR tree-optimization/60454
+ *
Hi Marcus,
On 11/03/14 11:25, Marcus Shawcroft wrote:
On 25 February 2014 10:08, Kyrill Tkachov kyrylo.tkac...@arm.com wrote:
Hi all,
The problem solved in this patch is that when gcc is configured with
--with-arch=armv8-a gcc will go into aarch64-arches.def, pick the
representative CPU
On Tue, Mar 11, 2014 at 11:58:01AM +0100, Richard Biener wrote:
On Mon, Mar 10, 2014 at 11:14 AM, Jakub Jelinek ja...@redhat.com wrote:
On Mon, Mar 10, 2014 at 11:10:05AM +0100, Uros Bizjak wrote:
Well, better is non-obvious, while it is smaller (which is good for
initialization and thus
On Sat, 8 Mar 2014, Gerald Pfeifer wrote:
Thanks for the time and diligence writing this up, Richi!
On Thu, 6 Mar 2014, Richard Biener wrote:
-files; if @option{-flto} is not passed to the linker, no
-interprocedural optimizations are applied.
+files; if @option{-fno-lto} is not passed
This fixes offsetted constraint handling in the constraint builder
and the solver. The existing code didn't consider non-equal-sized
fields and thus missed including all those that are overlapping
a offsetted variable.
Fixed like the following, bootstrapped and tested on
Currently, castring away volatile from a pointer makes -Wcast-qual
claim that __attribute__((noreturn)) was cast away (see bugzilla
55383). The attached patch, originally written by Manuel López-Ibáñez
and updated to match trunk by me, correctes that. No regressions on
gcc from applying this patch
On 03/11/2014 08:03 AM, Paolo Carlini wrote:
+ if (DECL_INHERITED_CTOR_BASE (fun)
+ TREE_CODE (fun) == TEMPLATE_DECL)
+{
+ ret = false;
+ if (complain)
+ error (inherited constructors inherit %constexpr% from
+ the designated base);
+}
To correct my
This moves handling of constant folding and operand order canonicalization
to the folding worker instead of doing that only when building a stmt
as fallback. Matches what fold_unary vs. fold_build does.
Applied to branch.
Richard.
2014-03-11 Richard Biener rguent...@suse.de
*
Why not also -std=gnu++14?
This converts remaining parts of SCCVN to use the
gimple_match_and_simplify interface. Most importantly
(apart from one legacy case) we no longer build GENERIC
trees from GIMPLE to do simplification nor track
whether it's maybe worthwhile doing so.
Committed to branch (a few more optimization
Hi!
As described in the PR, the r208165 change regressed following test.
The problem is that VRP inserts a useless ASSERT_EXPR right before
__builtin_unreachable () (obviously, the uses of the ASSERT_EXPR
lhs aren't and can't be used by anything), which then prevents
Hi!
This patch fixes the ssa-ifcombine-10.c regression.
The thing is that the uselessly added ASSERT_EXPR makes vrp1 change
the cfg slightly like this:
bb 2:
_4 = x_3(D) 1;
if (_4 == 0)
goto bb 5;
else
goto bb 3;
bb 3:
_5 = x_3(D) 4;
if (_5 != 0)
-goto bb 5;
Hi,
On 03/11/2014 02:10 PM, Jason Merrill wrote:
On 03/11/2014 08:03 AM, Paolo Carlini wrote:
+ if (DECL_INHERITED_CTOR_BASE (fun)
+ TREE_CODE (fun) == TEMPLATE_DECL)
+{
+ ret = false;
+ if (complain)
+error (inherited constructors inherit %constexpr% from
+
On Tue, 11 Mar 2014, Magnus Reftel wrote:
Currently, castring away volatile from a pointer makes -Wcast-qual
claim that __attribute__((noreturn)) was cast away (see bugzilla
55383). The attached patch, originally written by Manuel López-Ibáñez
and updated to match trunk by me, correctes that.
On Tue, Mar 11, 2014 at 05:10:45PM +0100, Gerald Pfeifer wrote:
On Tue, 11 Mar 2014, Magnus Reftel wrote:
Currently, castring away volatile from a pointer makes -Wcast-qual
claim that __attribute__((noreturn)) was cast away (see bugzilla
55383). The attached patch, originally written by
OK.
Jason
This is a 4.9 regression due to a non-uglified name being used in
debug mode containers.
Tested x86_64-linux, normal and debug mode, committed to trunk.
commit 39cc9a9e195aa90027fcc40083b913016b83f0e5
Author: Jonathan Wakely jwak...@redhat.com
Date: Tue Mar 11 10:47:52 2014 +
PR
Committed to branch dmalcolm/jit:
I tracked down the crash in dwarf2out.c mentioned in the commit message
for 6e624036e3f47c6ca4ae2e99974b74aa29f11eca:
base_types wasn't being flushed between invocations, leading to it
containing garbage that might be collected (it's not GTY-labelled), and
On Tue, 11 Mar 2014, Tobias Burnus wrote:
When using Cilk Plus (-fcilkplus), it makes sense to automatically link the
run-time library (-lcilkrts).
This patch mimics libgomp by adding a .spec file; I am not 100% sure whether
the .spec file is needed, but the pthread tests in libgomp imply
On Tue, 11 Mar 2014, Magnus Reftel wrote:
Currently, castring away volatile from a pointer makes -Wcast-qual
claim that __attribute__((noreturn)) was cast away (see bugzilla
55383). The attached patch, originally written by Manuel L?pez-Ib??ez
and updated to match trunk by me, correctes that.
Committed to branch dmalcolm/jit:
Turning optimizations up from 0 to 3 showed a segfault on the 2nd
iteration of test-linked-list.c in get_data_from_adhoc_loc whilst
garbage-collecting.
Investigation revealed the issue to be a CFG from the previous compile
being kept alive by this GC root in
This patch backports the important parts of three commits from trunk
to the 4.8 branch.
Tested x86_64-linux, committed to the 4.8 branch.
commit 48a2f3af4fcff86695e6f12c1490833f5679ad44
Author: Jonathan Wakely jwak...@redhat.com
Date: Tue Mar 11 18:20:47 2014 +
Backport from
On 27/01/14 17:58 +, Jonathan Wakely wrote:
This fixes a tsan warning in shared_ptr by replacing the non-atomic
load with a call to _M_get_use_count() which does a relaxed atomic
load.
Tested x86_64-linux, committed to trunk. Will commit to 4.8 soon too.
Here's the commit to the 4.8
This patch fixes the latest 58678 testcase by avoiding speculative
devirtualization to the destructor of an abstract class, which can
never succeed: you can't create an object of an abstract class, so
the pointer must point to an object of a derived class, and the
derived class necessarily
Will Deacon will.dea...@arm.com wrote:
Hi John,
On Tue, Mar 11, 2014 at 02:54:18AM +, John Carr wrote:
A comment in arm/sync.md notes We should consider issuing a inner
shareability zone barrier here instead. Here is my first attempt
at a patch to emit weaker memory barriers.
Hi,
I think that NULL pointer checks should be added for all pointers
must_raw, may_raw etc, not only for the *_no_source ones.
This will make the function more robust and easier to maintain.
Indeed in the current code only the *_no_source pointers may be NULL, but
this may change in the
Hans-Peter Nilsson h...@bitrange.com writes:
On Mon, 3 Mar 2014, Richard Sandiford wrote:
AIUI:
Reading back the references don't yield any dissenting
flash-backs, FWIW.
So, a (use fp) then a (clobber fp)? That was probably just too
weird for me to think of, much like a hypercorrect
On 9 January 2014 23:39, Jonathan Wakely wrote:
PR libstdc++/59680
* src/c++11/thread.cc (__sleep_for): Fix call to ::sleep.
Tested x86_64-linux, and tested again with a hacked c++config.h to use
::sleep(), committed to trunk.
Also backported to the 4.8 branch now.
commit
Hi!
build_low_bits_mask doesn't work for vector types (even TYPE_PRECISION
alone on it is meaningless), but what we actually want is a constant with
all bits set.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2014-03-11 Jakub Jelinek ja...@redhat.com
Marc
Hi current google/main fails to build for arm because of duplicated
head file entries in gtyp-input.list.
Fixed by removing duplication in macro tm_file. This only affects arm
platform. Tested by successfully build for arm.
Patch below
--- config.gcc.orig 2014-03-11 15:10:26.849602409
Looks good to me.
Dehao
On Tue, Mar 11, 2014 at 3:22 PM, Hán Shěn (沈涵) shen...@google.com wrote:
Hi current google/main fails to build for arm because of duplicated
head file entries in gtyp-input.list.
Fixed by removing duplication in macro tm_file. This only affects arm
platform. Tested
This patch is fixing PR60505 in which the vectorizer may produce
unnecessary epilogues.
Bootstrapped and tested on a x86_64 machine.
OK for trunk?
thanks,
Cong
diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index e1d8666..f98e628 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,9 @@
During AutoFDO annotation, we want to record the annotation stats into
an elf section, so that we can calculate how much percentage of the
profile is annotated, which can be used as an indicator whether code
has changed significantly comparing with the profiled source.
Bootstrapped and
Ping ?
gcc/
2014-03-03 Kugan Vivekanandarajah kug...@linaro.org
PR target/60034
* aarch64/aarch64.c (aarch64_classify_address): Fix alignment for
section anchor.
gcc/testsuite/
2014-03-03 Kugan Vivekanandarajah kug...@linaro.org
PR
On 03/09/2014 05:39 PM, Jerry DeLisle wrote:
Hi all,
This final patch does two things.
First: In read.c it implements a simple space skipping scheme in read_decimal
where I found a lot of repeated next_char calls happening. This gives a pretty
good boost in performance and is applicable
4.8 also has this bug. OK to backport?
On Tue, Mar 11, 2014 at 6:59 PM, Jakub Jelinek ja...@redhat.com wrote:
On Tue, Mar 11, 2014 at 06:48:37PM +0800, Thomas Preud'homme wrote:
I also added a typedef unsigned uint32_t for when sizeof(unsigned) == 4. I
hope it's right.
In theory you could
From: Jakub Jelinek [mailto:ja...@redhat.com]
In theory you could have __CHAR_BIT__ different from 8 and what you care
about is that uint32_t has exactly 32 bits, so the check would need to be
if (sizeof (uint32_t) * __CHAR_BIT__ != 32)
return 0;
I could go with:
In = (0x12
The bug report http://golang.org/issue/7074 shows that math.Log2(1)
produces the wrong result on Aarch64, because the Go math package is
compiled to use a fused multiply-add instruction. This patch to the
libgo configure script will use -ffp-contract=off when compiling the
math package on
56 matches
Mail list logo