Hi Savin,
I've committed your patch for PR59784; the fix seems small enough to
accept directly. Thanks a lot for catching this.
Thanks,
Chung-Lin
2014-01-30 Savin Zlobec
PR target/59784
* config/nios2/nios2.c (nios2_fpu_insn_asm): Fix asm output of
SFmode to DFmode ca
Hello,
As detailed in http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58960#c6, we
fail to use the DF liveness info in the register pressure sensitive
scheduling for the new blocks as we do not properly compute it in this
case. The patch fixes this by avoiding to use the sched-pressure for the
n
We've previously tweaked a few rules to avoid creating builtin
candidates with dependent types, but things keep slipping through the
cracks. This should be a more complete solution.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit aca790abae74da31099bcfb3ceb1459b6c2b6f05
Author: Jason Mer
Hi,
This continues the series of -maltivec=be patches, this one handling
vec_splat. Since vec_splat is no longer treated as a true intrinsic,
its default behavior for little endian is changed to use right-to-left
element indexing for selecting the element to splat. With -maltivec=be
for little e
If a parameter of a covariant virtual function is volatile, we can't
just use the parameter directly in the call we build for the thunk, or
the gimple verifier will complain. We need to copy it into a temporary
first.
Tested x86_64-pc-linux-gnu. OK for trunk?
commit 0cbf28df997d109615545f113
Minor change to support some new functionality in newlib. Committed.
* config/msp430/msp430.opt (-minrt): New.
* config/msp430/msp430.h (STARTFILE_SPEC): Link alternate runtime
if -minrt given.
(ENDFILE_SPEC): Likewise.
Index: gcc/config/msp430/msp430.opt
===
On Sat, 25 Jan 2014, Marc Glisse wrote:
> Good luck, catching up always seems to take forever...
Thanks!
And thanks for your explanations. With those, I am fine. (Might it make
sense to add some of the background you shared to the documentation to
describe the limitations? "No" or "Yes, but I
Hi,
The attached patch adds some new features to gcov-tool. It aims to
rewrite LIPO profile for reuse, debug or performance tuning purpose.
-l, --modu_list Only use the modules in this file
-r, --string Replace string in path
-u, --use_imports_file
Some doc improvements. I decided it's time to regenerate the HTML
pages too.
Committed to trunk.
commit bbf5d74cf6f1399413217bc70ebfecb360efe822
Author: Jonathan Wakely
Date: Wed Jan 29 20:31:07 2014 +
PR libstdc++/57226
* doc/xml/manual/debug.xml (debug.gdb): Update docu
On Wed, 2014-01-29 at 21:38 +, Jonathan Wakely wrote:
> On 29 January 2014 21:17, Oleg Endo wrote:
> > My original intention was to eliminate code bloat when doing something
> > like throw std::logic_error ("cold coffee");
> > If the const char* overloads are inlined it will emit code to const
We counted the pack and all the elements of the pack, giving us a final
count one too high.
Tested x86_64-pc-linux-gnu, applying to trunk and 4.8.
commit b75c08746dc30ac50e908e8d520042f4157281f9
Author: Jason Merrill
Date: Wed Jan 29 16:41:23 2014 -0500
PR c++/59989
* pt.c (expand_
On Wed, Jan 29, 2014 at 09:19:35PM +, Joseph S. Myers wrote:
> The c/ and c-family/ changes are OK. (One of the other things left to fix
> would be that it looks like convert_arguments could do with locations for
> each argument, rather than just a single main location.)
Thanks. Yes, tweak
On 29 January 2014 21:17, Oleg Endo wrote:
> My original intention was to eliminate code bloat when doing something
> like throw std::logic_error ("cold coffee");
> If the const char* overloads are inlined it will emit code to construct
> an std::string from const char* in user code where the exce
Hi,
On 01/29/2014 09:49 PM, Jason Merrill wrote:
OK, thanks.
Applied.
By the way, when I said 3/4 uses (in fact, at least 4 in the file), I
meant the pattern:
tree name = TYPE_NAME (type);
if (TREE_CODE (name) == TYPE_DECL)
name = DECL_NAME (name);
which I noticed
On Wed, 29 Jan 2014, Marek Polacek wrote:
> On Wed, Jan 29, 2014 at 05:06:43PM +, Joseph S. Myers wrote:
> > I do think we need a public ubsan todo list, if there isn't already one,
> > and sanitization of calls through incompatible types should go on that
> > list.
>
> I've put that TODO o
On Wed, 29 Jan 2014, Marek Polacek wrote:
> Finally I managed to get this patch into regression-free shape. It
> should improve the column info for quite a lot of warnings, fixing
> PR59940 on the way. I had to add location_t to bunch of functions;
> SET_EXPR_LOCATION is of no use here - we're d
On Wed, 2014-01-29 at 15:21 +, Jonathan Wakely wrote:
> On 26/01/14 15:15 +0100, Oleg Endo wrote:
> >> Files in libstdc++-v3/src/c++98/ seem to be never compiled with C++11.
> >> Thus I can think of two options:
> >> 1) Add const char* ctors for C++98 and C++11.
> >> 2) Add #ifdef'ed declaratio
Here the problem was that when we built up A<'X'> the arguments to A
were 'X' and the empty set for the parameter pack. When we then use
that to deduce the arguments for C, we got confused by the empty set and
tried to treat it as another argument, leading to bad times. Fixed by
calling expan
OK, thanks.
By the way, when I said 3/4 uses (in fact, at least 4 in the file), I meant the
pattern:
tree name = TYPE_NAME (type);
if (TREE_CODE (name) == TYPE_DECL)
name = DECL_NAME (name);
which I noticed yesterday. Could be a new macro? (post 4.9 maybe)
Sure. M
This deprecates this GNU extension which isn't really a conforming
allocator anyway.
PR libstdc++/21609
* include/ext/array_allocator.h: Add deprecated attribute.
Tested x86_64-linux, committed to trunk.
I'll update the gcc-4.9/changes.html page to note this.
commit e6682598d145
In this testcase, tsubst_friend_function was getting confused trying to
instantiate the friend template because it didn't recognize that it was
a template. Fixed by keeping the TEMPLATE_DECL on the friends list and
dealing with that appropriately in tsubst_friend_function.
is_specialization_of
On 29 January 2014 20:06, François Dumont wrote:
> Here is the patch that simply consider 55083 as not supported except in
> normal mode. This is a temporary workaround for 4.9 release so I prefer not
> to introduce a dg-profile-mode-unsupported or something like that. Those
> tests will simpl
Here is the patch that simply consider 55083 as not supported
except in normal mode. This is a temporary workaround for 4.9 release so
I prefer not to introduce a dg-profile-mode-unsupported or something
like that. Those tests will simply appear as not supported for debug and
parallel mode
On Wed, Jan 29, 2014 at 05:06:43PM +, Joseph S. Myers wrote:
> I do think we need a public ubsan todo list, if there isn't already one,
> and sanitization of calls through incompatible types should go on that
> list.
I've put that TODO on my TODO. :)
(But firstly the wiki has to let me creat
Finally I managed to get this patch into regression-free shape. It
should improve the column info for quite a lot of warnings, fixing
PR59940 on the way. I had to add location_t to bunch of functions;
SET_EXPR_LOCATION is of no use here - we're dealing with
{VAR,PARM}_DECLs, which do not carry a
On January 29, 2014 6:03:15 PM GMT+01:00, Markus Trippelsdorf
wrote:
>This is a backport of r194470 to the gcc-4.7 branch. It fixes a case
>were n_functions in coverage_obj_finish is 0 and the resulting array
>contains 0x1 elements.
>
>(Mike Hommey pointed out that Fixefox PGO-build is br
Hi,
On 01/29/2014 06:00 PM, Jason Merrill wrote:
On 01/29/2014 11:48 AM, Paolo Carlini wrote:
Rather than duplicate this code, let's factor it out into a separate
function.
Yeah, you are totally right, we have at least 3/4 uses of that. Care to
propose a name?
is_cxx_auto?
Good. I'm finishin
On 01/29/2014 06:41 PM, Cary Coutant wrote:
in this bug we ICE in dwarf2out.c:is_base_type with -g, because it doesn't
know how to handle the TEMPLATE_TYPE_PARM coming from the C++ front-end and
representing 'auto' in this kind of C++1y code.
That it shouldn't ICE and return 0 instead I'm pretty
On 23/01/14 08:58, Kyrill Tkachov wrote:
On 16/01/14 18:10, Kyrill Tkachov wrote:
Hi all,
The Cortex-A53 and Cortex-A57 cores support the CRC32 and Crypto extensions to
the ARMv8-A architecture. This patch adds that information to their definitions
in aarch64-cores.def.
Tested aarch64-none-elf
> in this bug we ICE in dwarf2out.c:is_base_type with -g, because it doesn't
> know how to handle the TEMPLATE_TYPE_PARM coming from the C++ front-end and
> representing 'auto' in this kind of C++1y code.
>
> That it shouldn't ICE and return 0 instead I'm pretty sure, I'm less sure
> about the next
Yup, these are OK. Thanks!
On Wed, 29 Jan 2014, Richard Biener wrote:
> testing reveals diagnostic regressions
>
> FAIL: gcc.dg/call-diag-2.c abort (test for warnings, line 12)
> FAIL: gcc.dg/call-diag-2.c abort (test for warnings, line 15)
> FAIL: gcc.dg/invalid-call-1.c (test for warnings, line 16)
>
> which may be har
The handling of the ARM ctor-returns-this ABI in the new decloning patch
failed to add a RETURN_EXPR. Fixed thus.
Tested with cross-compiler to arm-eabi, applying to trunk.
commit e1f2153a7facc47b886b24a5b3507bb4bea3e447
Author: Jason Merrill
Date: Wed Jan 29 09:16:51 2014 -0500
PR c++
On 01/29/14 03:53, Nick Clifton wrote:
Hi Alex, Hi Jeff,
There is a typo in the MN10300 store_movm pattern. It calls
mn10300_store_multiple_operation to generate a bit mask of registers
to be pushed, which it then passes to mn10300_print_reg_list. But
mn10300_store_multiple_operati
This is a backport of r194470 to the gcc-4.7 branch. It fixes a case
were n_functions in coverage_obj_finish is 0 and the resulting array
contains 0x1 elements.
(Mike Hommey pointed out that Fixefox PGO-build is broken using 4.7
because of this bug)
Bootstrapped and tested on x86_64-unkno
On Wed, Jan 29, 2014 at 8:52 AM, Jakub Jelinek wrote:
> On Wed, Jan 29, 2014 at 08:39:55AM -0800, H.J. Lu wrote:
>> -m16 is used by Linux kernel, which checks if the compiler supports -m16
>> and uses it if it does. Adding gas --code16gcc check is an additional change
>> in Linux kernel. Clang al
OK.
Jason
On 01/29/2014 11:48 AM, Paolo Carlini wrote:
Rather than duplicate this code, let's factor it out into a separate
function.
Yeah, you are totally right, we have at least 3/4 uses of that. Care to
propose a name?
is_cxx_auto?
Jason
On 01/29/2014 05:37 AM, Andreas Schwab wrote:
Jason Merrill writes:
+// { dg-final { scan-assembler "a1.*(0x\[0-9a-f\]+)\[ \t\]*# DW_AT_type.*\\1.
DW_TAG_unspecified_type.*DW_AT_specification\[\n\r\]{1,2}\[^\n\r\]*(0x\[0-9a-f\]+)\[
\t\]*# DW_AT_type.*\\2. DW_TAG_base_type" } }
This regexp
OK.
Jason
On Wed, Jan 29, 2014 at 08:39:55AM -0800, H.J. Lu wrote:
> -m16 is used by Linux kernel, which checks if the compiler supports -m16
> and uses it if it does. Adding gas --code16gcc check is an additional change
> in Linux kernel. Clang already supports -m16, which generates objects
> directly. Ad
On 01/29/2014 05:42 PM, Jason Merrill wrote:
On 01/28/2014 04:35 PM, Paolo Carlini wrote:
+ if (is_cxx ())
+{
+ tree name = TYPE_NAME (type);
+ if (TREE_CODE (name) == TYPE_DECL)
+name = DECL_NAME (name);
+ if (name == get_identifier ("auto"))
+return 0;
On 01/29/14 02:47, Richard Biener wrote:
I wonder if you can't simply prune the edges for the OpenMP regions
(similar as to how we could prune some of the edges from setjmp
to calls before that setjmp call)? Eventually this asks for delaying
of abnormal edge creation (but that's what you do any
On 01/28/2014 04:35 PM, Paolo Carlini wrote:
+ if (is_cxx ())
+ {
+ tree name = TYPE_NAME (type);
+ if (TREE_CODE (name) == TYPE_DECL)
+ name = DECL_NAME (name);
+ if (name == get_identifier ("auto"))
+ return 0;
Rather than duplicate this
On Wed, Jan 29, 2014 at 6:41 AM, Alexander Monakov wrote:
>
> On Wed, 29 Jan 2014, H.J. Lu wrote:
>> Since the .code16gcc directive was added to binutils back in 1999,
>> it is older than the minimum binutils required for x86 GCC. There
>> is no need to specify a separate minimum binutils for it.
Committed to branch dmalcolm/jit:
Currently there's almost no type-checking within libgccjit - if client
code uses the API in such a way as to generate bogus code, this is
likely to violate assumptions made later on within GCC proper, leading
to errors deep inside GCC (e.g. with
internal c
Hi,
Some testcases need explicit GCC options to properly run, like
gcc.target/i386/sse2-init-v2di-2.c has
/* { dg-options "-O2 -msse4 -march=core2 -dp" } */
-march=core2 is specified explicitly. But with multlib, like
make check-gcc RUNTESTFLAGS="--target_board='unix{-march=k8}'
-march=k8 is
Diego wrote:
>> Ok to commit?
>
> OK. Thanks.
Done in r207265.
-Y
2014/1/29 Bernd Schmidt :
> I was worried the answer was going to be something like this. These are
> produced by two different compilers/linkers, aren't they? This seems really
> fragile to me and a recipe for hard-to-debug silent failures.
Yes.
> First we let the driver compile and load all the
On Wed, Jan 29, 2014 at 06:33:21PM +0300, Kirill Yukhin wrote:
> I think its time to remove `XPASS' from corresponding tests.
> On 03 Jan 22:11, Jakub Jelinek wrote:
> > Hi!
> >
> > On Fri, Jan 03, 2014 at 08:58:30PM +0100, Toon Moene wrote:
> > > I don't doubt that would work, what I'm interested
Fix a couple of white space nits, committed.
/Marcus
2014-01-29 Marcus Shawcroft
* config/aarch64/aarch64.c (aarch64_expand_mov_immediate)
(aarch64_legitimate_address_p, aarch64_class_max_nregs): Adjust
whitespace.diff --git a/gcc/config/aarch64/aarch64.c b/gcc/config
On Wed, Jan 29, 2014 at 4:24 PM, Richard Biener
wrote:
> On Wed, Jan 29, 2014 at 11:02 AM, Jakub Jelinek wrote:
>> On Tue, Jan 28, 2014 at 07:26:47AM +0100, Jakub Jelinek wrote:
>>> > I wonder if this is just some of --enable-checking tests in dwarf2out
>>> > going wild
>>> > or if it is just ex
Hi Jakub,
> -Original Message-
> From: Jakub Jelinek [mailto:ja...@redhat.com]
> Sent: Wednesday, January 29, 2014 6:31 AM
> To: Iyer, Balaji V
> Cc: 'Jason Merrill'; 'Jeff Law'; 'Aldy Hernandez'; 'gcc-patches@gcc.gnu.org';
> 'r...@redhat.com'
> Subject: Re: [PING] [PATCH] _Cilk_for for C
The definition of _FP_I_TYPE in AArch64 libgcc sfp-machine.h is wrong,
this patch ensures the definition matches that provided by glibc's
AArch64 sfp-machine.h
Committed
/Marcus
2014-01-29 Marcus Shawcroft
* config/aarch64/sfp-machine.h (_FP_I_TYPE): Define
as long long.dif
Hi,
I tested the following patch. It makes inliner to ignore empty EH clenaup
regions regadless they are internal or external in anticipation they will be
removed. It also make tree-eh to not do any cleanups pre-inline. This is
independent change, but I wanted to stress the code bit more.
Here ar
gnatmake, gnatclean, gnatname and the gnat driver now fail immediately
if the main project is an aggregate project or if there is an aggregate
library project in the project tree.
Tested on x86_64-pc-linux-gnu, committed on trunk
2014-01-29 Vincent Celier
* clean.adb (Gnatclean): Fail
Hello,
I think its time to remove `XPASS' from corresponding tests.
On 03 Jan 22:11, Jakub Jelinek wrote:
> Hi!
>
> On Fri, Jan 03, 2014 at 08:58:30PM +0100, Toon Moene wrote:
> > I don't doubt that would work, what I'm interested in, is (cat verintlin.f):
>
> Well, you need gather loads for that
If exceptions aren't propagated, the optimized path of
Exp_Ch9.Build_Protected_Subprogram_Body could be used.
No functional change, so no testcase.
Tested on x86_64-pc-linux-gnu, committed on trunk
2014-01-29 Tristan Gingold
* exp_ch9.adb (Is_Exception_Safe): Return true if no except
On Wed, Jan 29, 2014 at 11:02 AM, Jakub Jelinek wrote:
> On Tue, Jan 28, 2014 at 07:26:47AM +0100, Jakub Jelinek wrote:
>> > I wonder if this is just some of --enable-checking tests in dwarf2out
>> > going wild
>> > or if it is just expensive sanity checking code?
>> > I used to have chroot envir
A generic instance should be analyzed with the value of SPARK_Mode
defined at the point of instantiation. Now, the following code
compiles without errors:
$ gcc -c -gnatec=test.adc pinst.adb
-- test.adc
1. pragma SPARK_Mode (On);
-- pinst.ads
1. with Ada.Containers.Formal_Doubly_Link
On 26/01/14 15:15 +0100, Oleg Endo wrote:
Files in libstdc++-v3/src/c++98/ seem to be never compiled with C++11.
Thus I can think of two options:
1) Add const char* ctors for C++98 and C++11.
2) Add #ifdef'ed declarations to libstdc++-v3/include/std/stdexcept and
add a new file libstdc++-v3/src/c
This patch prevents the copy of too big names to fixed-size buffers from
overflowing. Instead, when too big strings are provided, treat them as
invalid and return an empty string.
The execution of the following example must print "OK":
$ gnatmake foo && ./foo
with Ada.Text_IO; use Ada.Text_IO;
w
This procedure was empty, so no need to insert a call to it.
No functional change.
Tested on x86_64-pc-linux-gnu, committed on trunk
2014-01-29 Tristan Gingold
* exp_ch9.adb (Build_Protected_Entry): Do not call
Complete_Entry_Body anymore.
* rtsfind.ads (RE_Complete_Si
The rules for pragma SPARK_Mode have changed. In particular, the mode
is not inherited anymore from a the spec to the body, and the body
should not have a mode if the spec did not have one. These new rules
are checked now. A minor change is that SPARK_Mode is set even on
subprograms for which Comes
This patch removes a spurious initialization on the string that builds the
name of task that is a record component, when Initialize_Scalars is enabled.
This is both efficient, and prevents anomalies in the handling of dynamic
objects on the secondary stack, that would result in a mangled name for s
Hi,
a very minor ICE on invalid regression, but since we explicitly allow
for redeclarations (also see comments in declare_global_var) we may as
well avoid crashing, at least in mainline. Tested x86_64-linux.
Thanks,
Paolo.
/
/cp
2014-01-29 Paolo Carlini
PR c++/58
On 27/01/14 23:37 +, Jonathan Wakely wrote:
On 27 January 2014 20:35, Jonathan Wakely wrote:
On 27 January 2014 20:12, Marc Glisse wrote:
On Mon, 27 Jan 2014, Jonathan Wakely wrote:
This is the best I've come up with to avoid dereferencing an invalid
pointer when calling vector::data() on
std::allocator_traits::allocate(a, n, hint) could fail to compile
if the const_void_pointer passed as the hint was a non-trivial type,
because it was passed to a varargs function.
std::vector was not using std::allocator_traits for calls to
A::allocate and A::deallocate.
Also remove some redunda
Hi,
On Wed, 29 Jan 2014, Jakub Jelinek wrote:
> > > It is also problem of inliner quality decisions and memory
> > > use/compile time. The in-memory representation of unnecesary EH is
> > > quite big.
> > >
> > > I am quite ignorant in this area, but for -O0 can't we simply
> > > disable all
On Wed, 29 Jan 2014, H.J. Lu wrote:
> Since the .code16gcc directive was added to binutils back in 1999,
> it is older than the minimum binutils required for x86 GCC. There
> is no need to specify a separate minimum binutils for it.
It's not clear to me why the compiler should be involved in pro
On 01/28/2014 01:52 PM, Ilya Verbin wrote:
The table is used in libgomp (see my patch [1]), as proposed by Jakub
(see [2]). The idea is that the order of entries in the host and
target tables must be identical. This allows to set up one-to-one
correspondence between host and target addresses.
This fixes the rest, adding p + (q - p) pattern matching and
(p + o1) + o2 associating (similar to the fold-const code).
And it adds testcases for the whole series.
Bootstrapped and tested on x86_64-unknown-linux-gnu and applied.
Richard.
2014-01-29 Richard Biener
PR tree-optimizat
Hi,
fix another simple ICE on invalid regression (the ICE is in
get_innermost_template_args). Tested x86_64-linux.
Thanks,
Paolo.
PS: happens in 4_8-branch too but I don't think it's worth backporting.
/
/cp
2014-01-29 Paolo Carlini
PR c++/58674
* pt.c (in
> But that will badly regress the stack usage, won't it? If so, that's a
> blocker for Ada where we have big temporaries.
For example this will presumably cause gnat.dg/stack_usage1.adb to fail.
--
Eric Botcazou
> That sounds reasonable, yeah.
But that will badly regress the stack usage, won't it? If so, that's a
blocker for Ada where we have big temporaries.
--
Eric Botcazou
This bumps (very late ...) the LTO bytecode revision from 2, 2
to 3, 0 (leaves us bumps of the minor version for the 4.8 branch
if necessary). I plan to do the next bump at the point we branch
for stage1.
Committed.
Richard.
2014-01-29 Richard Biener
* lto-streamer.h (LTO_major_ver
On Wed, Jan 29, 2014 at 03:06:45PM +0100, Michael Matz wrote:
> Hi,
>
> On Tue, 28 Jan 2014, Jan Hubicka wrote:
>
> > It is also problem of inliner quality decisions and memory use/compile
> > time. The in-memory representation of unnecesary EH is quite big.
> >
> > I am quite ignorant in this
Hi,
On Tue, 28 Jan 2014, Jan Hubicka wrote:
> It is also problem of inliner quality decisions and memory use/compile
> time. The in-memory representation of unnecesary EH is quite big.
>
> I am quite ignorant in this area, but for -O0 can't we simply disable
> all clobbers?
That sounds reason
On Wed, 29 Jan 2014, Richard Biener wrote:
>
> This removes the code that tries to avoid ICEs in the middle-end when
> facing calls through an incompatible type. The middle-end can now
> happily deal with those mismatches (and we rely on that). Also the
> code only detects the most primitive ca
On Wed, Jan 29, 2014 at 1:52 AM, Richard Biener wrote:
> On Tue, 28 Jan 2014, Uros Bizjak wrote:
>
>> On Mon, Jan 27, 2014 at 8:44 PM, H.J. Lu wrote:
>>
>> > The .code16gcc directive was added to binutils back in 1999:
>> >
>> > ---
>> >'.code16gcc' provides experimental support for generatin
Okay for trunk?
This is Ok since this was submitted quite sometime back.
Hi all,
I've committed this as r207237 with slightly fixed ChangeLog entries:
2014-01-29 Renlin Li
* config/arm/arm-arches.def (ARM_ARCH): Add armv7ve arch.
* config/arm/arm.c (FL_FOR_ARCH7VE): New.
(arm_f
On Wed, Jan 22, 2014 at 9:36 AM, Yury Gribov wrote:
> Ok to commit?
OK. Thanks.
Diego.
On Wed, Jan 29, 2014 at 01:29:53PM +0100, Paolo Carlini wrote:
> a very simple ICE on invalid regression present only in mainline.
> Tested x86_64-linux.
>
> Thanks,
> Paolo.
>
> ///
> /cp
> 2014-01-29 Paolo Carlini
>
> PR c++/58072
> * semantics.c (finish_omp_red
Hi,
a very simple ICE on invalid regression present only in mainline. Tested
x86_64-linux.
Thanks,
Paolo.
///
/cp
2014-01-29 Paolo Carlini
PR c++/58072
* semantics.c (finish_omp_reduction_clause): Check type for
error_mark_node.
/testsuite
2014-01-
On Wed, Jan 29, 2014 at 11:02:05AM +0100, Jakub Jelinek wrote:
> So, most of the time seems to be spent in cselib.c remove_useless_values
> (both from Ctrl-C in gdb profiling, and callgrind). For callgrind I've
> actually built 64-bit cc1plus with --enable-checking=release, and still
> compiled
>
On Tue, Jan 28, 2014 at 04:55:38PM +, Iyer, Balaji V wrote:
> I thought about it a bit more, and the main issue here is that we
> need access to the _Cilk_for loop's components both inside the child
> function and the parent function.
I guess for the C++ iterators, if in the _Cilk_for mo
"H.J. Lu" writes:
> The new tests failed on Linux/x86:
Woops.
I have committed the patch below under the obvious rule for this. Sorry
for the inconvenience.
gcc/testsuite/ChangeLog:
* c-c++-common/cpp/warning-zero-location-2.c: Fix error message
specifier.
diff --git a/gcc/
This removes the code that tries to avoid ICEs in the middle-end when
facing calls through an incompatible type. The middle-end can now
happily deal with those mismatches (and we rely on that). Also the
code only detects the most primitive cases like
((bar_get_x_func*) foo_get_x) (x)
but doe
This adds GIMPLE combining for /[ex] X * X as it appears for
pointer subtraction / addition combo.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2014-01-29 Richard Biener
PR tree-optimization/58742
* tree-ssa-forwprop.c (associate_plusminus): Return
Hi Alex, Hi Jeff,
There is a typo in the MN10300 store_movm pattern. It calls
mn10300_store_multiple_operation to generate a bit mask of registers
to be pushed, which it then passes to mn10300_print_reg_list. But
mn10300_store_multiple_operation is actually a predicate function
(define
Hi DJ,
I noticed a couple of RL78 patches in our local tree that ought to be
upstream, so here they are. The first is for the size of the frame
pointer register and the second is for the UMUL instruction when both
input operands are in the same register.
OK to apply ?
Cheers
Nick
g
en.
Here's what I get on m68k:
.uleb128 0x3| (DIE (0x2b) DW_TAG_subprogram)
| DW_AT_external
.ascii "a1\0" | DW_AT_name
.byte 0x1 | DW_AT_decl_file
(/daten/aranym/gcc/gcc-20140129/gcc/testsuite/g++.dg/debug
Hi,
the attached patch adds a prototype for __udiv_w_sdiv to longlong.h if
needed. This fixes tons of build warnings on s390 32 bit in glibc.
Ok?
Bye,
-Andreas-
2014-01-29 Andreas Krebbel
* longlong.h: Add __udiv_w_sdiv prototype.
diff --git a/include/longlong.h b/include/longlo
On Tue, Jan 28, 2014 at 07:26:47AM +0100, Jakub Jelinek wrote:
> > I wonder if this is just some of --enable-checking tests in dwarf2out going
> > wild
> > or if it is just expensive sanity checking code?
> > I used to have chroot environment for 32bit builds, I will need to
> > re-install it now
On Tue, 28 Jan 2014, Uros Bizjak wrote:
> On Mon, Jan 27, 2014 at 8:44 PM, H.J. Lu wrote:
>
> > The .code16gcc directive was added to binutils back in 1999:
> >
> > ---
> >'.code16gcc' provides experimental support for generating 16-bit code
> > from gcc, and differs from '.code16' in that '
On Tue, Jan 28, 2014 at 11:48 PM, Jakub Jelinek wrote:
> Hi!
>
> I've started an --enable-checking=valgrind bootstrap during the weekend,
> but only on Sunday afternoon and thus killed it on Monday morning, so it
> didn't get very far, but still reported a problem where the build tools
> had unini
On Tue, 28 Jan 2014, Jakub Jelinek wrote:
> Hi!
>
> As discussed in the PR and similarly to what has been done previously
> for __builtin_setjmp only, this patch attempts to decrease number of
> EDGE_ABNORMAL edges in functions with non-local gotos and/or setjmp or
> other functions that return t
On Tue, Jan 28, 2014 at 10:24 PM, Ian Lance Taylor wrote:
> `On Tue, Jan 28, 2014 at 1:10 PM, Thomas Schwinge
> wrote:
>>
>> OK, I agree to all of that, but I'd assume that if the compiler doesn't
>> do such value tracking to see whether all cases have been covered, it
>> also souldn't emit such
On Tue, Jan 28, 2014 at 4:17 PM, Bingfeng Mei wrote:
> I checked vectorization code, it seems that only relevant place
> vec_widen_mult_even/odd & vec_widen_mult_lo/hi are generated is in
> supportable_widening_operation. One of these pairs is selected, with priority
> given to vec_widen_mult_e
On Jan 28, 2014, at 6:38 PM, David Edelsohn wrote:
> It is not appropriate for a GCC or GLIBC maintainer to impose behavior
So, let's ask the direct question, do you oppose making -mieee conforming to
ieee? If not, they the fixes in question seem like they should be turned on
with -mieee, and
1 - 100 of 101 matches
Mail list logo
