Hi
The trunk patch r177357 fixed an arm target test failure, which also affects
google/gcc-4_6, OK to port it to google/gcc-4_6?
Tested with
make check-gcc RUNTESTFLAGS="--target_board=arm-sim/arch=armv7-a
arm.exp=vfp-1.c"
thanks
Carrot
2011-08-12 Guozhi Wei
Backport r177357 from
On Thu, Aug 11, 2011 at 09:08:41PM -0500, Sebastian Pop wrote:
> > This patch falls to compile under clang3.0svn with the error...
> >
> > ../../gcc-4.7-20110811/gcc/graphite-sese-to-poly.c:643:8: error: unknown
> > type name 'isl_id'
> > static isl_id
On 08/11/2011 07:18 PM, Lawrence Crowl wrote:
On 8/11/11, Andrew MacLeod wrote:
The __sync_mem_fetch_{add,sub,and,xor,or} routines perform the operation
atomically, and return the value that was in memory before the operation
was performed. As I was working on switching the c++ wrappers to use
> This patch falls to compile under clang3.0svn with the error...
>
> ../../gcc-4.7-20110811/gcc/graphite-sese-to-poly.c:643:8: error: unknown type
> name 'isl_id'
> static isl_id *
> ^
> ../../gcc-4.7-20110811/gcc/graphite-sese-to-poly.c:647:3: error:
Ping ^ 2
On Fri, Aug 5, 2011 at 8:56 AM, Ye Joey wrote:
> Ping
>
> On Wed, Aug 3, 2011 at 8:44 AM, Joey Ye wrote:
>>
>> This patch fixes PR49437 with a single line change in ARM backend
>> and a regression test case for ARM target
>>
>> ChangeLog:
>> 2011-08-02 Matthew Gretton-Dann
>>
t;context = isl_set_add_constraint (scop->context, c);
> + }
> }
>
>if (ub)
> @@ -1474,6 +1750,25 @@ add_param_constraints (scop_p scop, ppl_Polyhedron_t
> context, graphite_dim_t p)
>ppl_Polyhedron_add_constraint (context, cstr);
>ppl_delete_Li
A new patch with lightly modified the test case.
David
On Thu, Aug 11, 2011 at 5:02 PM, Xinliang David Li wrote:
> Compile the following program using 4.6 or trunk compiler and run it,
> the program will seg fault. The problem is that the D3_Spec's primary
> vtable has a null entry for virtual f
Compile the following program using 4.6 or trunk compiler and run it,
the program will seg fault. The problem is that the D3_Spec's primary
vtable has a null entry for virtual function id which it overrides.
The root cause is that the base class's virtual function list for the
derived class is cop
The following patch addresses
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42554#c15
by extending the logic used in...
Author: mrs
Date: Fri Mar 19 10:19:52 2010
New Revision: 157563
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=157563
Log:
PR ada/42554
* configure.ac: Only pas
On 8/11/11, Andrew MacLeod wrote:
> The __sync_mem_fetch_{add,sub,and,xor,or} routines perform the operation
> atomically, and return the value that was in memory before the operation
> was performed. As I was working on switching the c++ wrappers to use
> these new routines, I discovered we also
Signed-off-by: Sebastian Pop
---
gcc/graphite-clast-to-gimple.c | 55 ++--
gcc/graphite-cloog-util.c |1 +
gcc/graphite-poly.c|8 +
gcc/graphite-poly.h|3 +
gcc/graphite-sese-to-poly.c| 301 +++-
gcc/graphite
Signed-off-by: Sebastian Pop
---
gcc/graphite-poly.c |7 ++-
gcc/graphite-poly.h |4 +-
gcc/graphite-sese-to-poly.c | 146 ---
3 files changed, 145 insertions(+), 12 deletions(-)
diff --git a/gcc/graphite-poly.c b/gcc/graphite-pol
Signed-off-by: Sebastian Pop
---
gcc/graphite-blocking.c| 12 +++-
gcc/graphite-clast-to-gimple.c |9 +
gcc/graphite-cloog-util.c | 11 ++-
gcc/graphite-cloog-util.h |2 --
gcc/graphite-dependences.c | 11 ++-
gcc/graphite-flatt
Signed-off-by: Sebastian Pop
---
gcc/graphite-poly.c | 12 -
gcc/graphite-poly.h |6 ++
gcc/graphite-sese-to-poly.c | 122 ++
3 files changed, 104 insertions(+), 36 deletions(-)
diff --git a/gcc/graphite-poly.c b/gcc/graphite-po
2011-07-21 Tobias Grosser
* configure: Regenerated.
* config/cloog.m4: Do not define CLOOG_ORG
and in gcc/
2011-07-21 Tobias Grosser
* Makefile.in (graphite-clast-to-gimple.o, graphite-cloog-util.o):
Remove graphite-cloog-util.h.
* graphite-clast-to
* graphite-clast-to-gimple.c (new_clast_name_index): Store a copy
of the string, no just a reference.
(clast_name_index): Add a new field, that specifies if we need to free the
name.
(free_clast_name_index): If necessary, free the name string.
(clast_name_index_elt_info): Calculate the
Signed-off-by: Sebastian Pop
---
gcc/graphite-sese-to-poly.c |1 +
1 files changed, 1 insertions(+), 0 deletions(-)
diff --git a/gcc/graphite-sese-to-poly.c b/gcc/graphite-sese-to-poly.c
index 7e23c9d..05280e7 100644
--- a/gcc/graphite-sese-to-poly.c
+++ b/gcc/graphite-sese-to-poly.c
@@ -19
2011-07-21 Tobias Grosser
* configure: Regenerated.
* config/cloog.m4: Remove support for CLooG-ppl and CLooG-parma,
both cloog.org and legacy versions. The only supported version will
be CLooG with the isl backend.
Signed-off-by: Sebastian Pop
---
ChangeLog
Signed-off-by: Sebastian Pop
---
gcc/graphite-cloog-util.c | 10 +-
1 files changed, 5 insertions(+), 5 deletions(-)
diff --git a/gcc/graphite-cloog-util.c b/gcc/graphite-cloog-util.c
index 9bc24a0..83cfb54 100644
--- a/gcc/graphite-cloog-util.c
+++ b/gcc/graphite-cloog-util.c
@@ -236
2011-07-26 Sebastian Pop
* doc/invoke.texi: Document CLooG-ISL requirement for Graphite.
Signed-off-by: Sebastian Pop
---
gcc/doc/install.texi | 24
1 files changed, 8 insertions(+), 16 deletions(-)
diff --git a/gcc/doc/install.texi b/gcc/doc/install.texi
2011-07-21 Tobias Grosser
* configure: Regenerated.
* configure.ac: Require cloog isl 0.16.3
Signed-off-by: Sebastian Pop
---
ChangeLog|5 +
configure|6 +++---
configure.ac |2 +-
3 files changed, 9 insertions(+), 4 deletions(-)
diff --git a/ChangeLo
Hi,
here are the updated patches for the conversion of Graphite to
ISL following the comments from Tobias and Sven. The patches
are passing regression testing on amd64-linux.
cd .../build/gcc/
make -k check RUNTESTFLAGS=graphite.exp
cd .../build/x86_64-unknown-linux-gnu/libgomp/testsuite
make -k
On Thu, Aug 11, 2011 at 04:59:43PM -0500, Sebastian Pop wrote:
> >>> + {
> >>> + isl_dim *dc = isl_set_get_dim (scop->context);
> >>> + int nb_in = isl_dim_size (dc, isl_dim_set);
> >>> + int nb_out = 1 + DR_NUM_DIMENSIONS (dr);
> >>> + int nbp = scop_nb_params (scop);
> >>> + isl_d
>>> + {
>>> + isl_dim *dc = isl_set_get_dim (scop->context);
>>> + int nb_in = isl_dim_size (dc, isl_dim_set);
>>> + int nb_out = 1 + DR_NUM_DIMENSIONS (dr);
>>> + int nbp = scop_nb_params (scop);
>>> + isl_dim *dim = isl_dim_alloc (scop->ctx, nbp, nb_in, nb_out);
>>> + int i;
>>
On Thu, Aug 11, 2011 at 16:00, Sven Verdoolaege wrote:
> On Thu, Aug 11, 2011 at 03:10:30PM -0500, Sebastian Pop wrote:
>> On Thu, Aug 11, 2011 at 14:28, Sven Verdoolaege wrote:
>> > On Thu, Aug 11, 2011 at 01:16:45PM -0500, Sebastian Pop wrote:
>> >> +
>> >> + /* FIXME: This function will be re
On Thu, Aug 11, 2011 at 03:23:13PM -0500, Sebastian Pop wrote:
> As we are using this function only on parameters, get_name should
> return a unique name. I guess that the name in isl_id is only used
> for debugging purposes, as the ISL manual states that "Identifiers
> with the same name but diff
On Thu, Aug 11, 2011 at 03:10:30PM -0500, Sebastian Pop wrote:
> On Thu, Aug 11, 2011 at 14:28, Sven Verdoolaege wrote:
> > On Thu, Aug 11, 2011 at 01:16:45PM -0500, Sebastian Pop wrote:
> >> +
> >> + /* FIXME: This function will be renamed isl_map_insert_dims and
> >> + documented in a later
On Thu, Aug 11, 2011 at 08:14:05PM +0100, Tobias Grosser wrote:
> I think the best would be to provide our ctx to cloog when allocating
> the CloogState.
Yes.
skimo
> I also removed LINEMAP_POSITION_FOR_COLUMN, it did the EXACT same thing as
> linemap_position_for_column, so maintaining both in parallel seems like
> overkill to me. The only thing I can think of is that it's more optimal as
> it's inlined (but if that's really needed we can always make
> li
Removed LINEMAP_POSITION_FOR_COLUMN, it did the EXACT same thing as
linemap_position_for_column, so maintaining both in parallel seems like
overkill to me. The only thing I can think of is that it's more optimal as it's
inlined (but if that's really needed we can always make
linemap_position_fo
On 08/11/2011 10:49 AM, Rainer Orth wrote:
There might be an alternative implementation that is less invasive to
the C++ frontend, though: add
&& TARGET_DECL_NAMESPACE_STD_P (decl)
in write_unscoped_name, defaulting to true, override it in sol2.h (which
gets included via tm.h) and have
On Thu, Aug 11, 2011 at 2:50 PM, Richard Guenther wrote:
>> Currently, conversion from floating point to integer on 64bit targets
>> goes through DImode temporary, due to missing BUILT_IN_ICEIL (and
>> other) builtins that can convert directly from FP to integer.
> I'm not sure about the naming
On Thu, Aug 11, 2011 at 14:14, Tobias Grosser wrote:
> This needs to be adapted to my cloog.org interface patch.
I will adapt my patch set to be on top of your patch.
>> +/* Return an ISL identifier from the name of the ssa_name E. */
>> +
>> +static isl_id *
>> +isl_id_for_ssa_name (scop_p s,
On Thu, Aug 11, 2011 at 14:28, Sven Verdoolaege wrote:
> On Thu, Aug 11, 2011 at 01:16:45PM -0500, Sebastian Pop wrote:
>> + if (1)
>> + {
>> + /* For now remove the isl_id's from the context before
>> + translating to CLooG: this code will be removed when the
>> + domain will a
The following patch will be applied to google branches.
2011-08-11 David Li
* l-ipo.c (process_module_scope_static_func): Promote
aliases of static symbol in LIPO mode.
Index: l-ipo.c
===
--- l-ipo.c (revision
On Thu, Aug 11, 2011 at 10:45 AM, Dodji Seketeli wrote:
>> As I mentioned, we have the same problem in pph where we need to force
>> a location (i.e. the lexer is assigning new locations, but we don't
>> want it to when we are replaying pre-processor tokens), so just a
>> "builtin_location" field
On Thu, Aug 11, 2011 at 20:19, H.J. Lu wrote:
> On Thu, Aug 11, 2011 at 8:23 AM, Rainer Orth
> wrote:
>> Paolo Bonzini writes:
>>
>>> On 08/10/2011 06:05 PM, Rainer Orth wrote:
>> >> True: it is called once per multilib.
> >
> > Just to doublecheck, are we sure that unwind.h is alw
On Thu, Aug 11, 2011 at 01:16:45PM -0500, Sebastian Pop wrote:
> + if (1)
> +{
> + /* For now remove the isl_id's from the context before
> + translating to CLooG: this code will be removed when the
> + domain will also contain isl_id's. */
> + isl_set *context = isl_set_p
On 08/11/2011 07:16 PM, Sebastian Pop wrote:
On Thu, Aug 11, 2011 at 13:03, Sebastian Pop wrote:
> On Thu, Aug 11, 2011 at 12:52, Tobias Grosser wrote:
>> I will commit this patch after the configure changes are in (and meanwhile
>> no further improvements were suggested for this patch).
On Thu, Aug 11, 2011 at 11:59 AM, Ludovic Brenta
wrote:
> OK to apply as trivial?
>
>
> 2011-08-11 Ludovic Brenta
>
> PR ada/50048
> * gcc-interface/Makefile.in (INCLUDES): replace -I- with
> -iquote.
>
> --- a/src/gcc/ada/gcc-interface/Makefile.in
> +++ b/src/gcc/ada/gcc-i
OK to apply as trivial?
2011-08-11 Ludovic Brenta
PR ada/50048
* gcc-interface/Makefile.in (INCLUDES): replace -I- with
-iquote.
--- a/src/gcc/ada/gcc-interface/Makefile.in
+++ b/src/gcc/ada/gcc-interface/Makefile.in
@@ -247,7 +247,7 @@
# Both . and srcdir are used,
2011/8/11 Georg-Johann Lay :
> This is an optimization in machine dependent reorg to
> remove redundant comparisons like in
>
> cc0 = compare (Reg, Num)
> if (cc0 == 0)
> goto L1
>
> cc0 = compare (Reg, Num)
> if (cc0 > 0)
> goto L2
>
> The second comparison is redundant an can be r
On Thu, Aug 11, 2011 at 11:19 AM, H.J. Lu wrote:
> On Thu, Aug 11, 2011 at 8:23 AM, Rainer Orth
> wrote:
>> Paolo Bonzini writes:
>>
>>> On 08/10/2011 06:05 PM, Rainer Orth wrote:
>> >> True: it is called once per multilib.
> >
> > Just to doublecheck, are we sure that unwind.h is
On Thu, Aug 11, 2011 at 8:23 AM, Rainer Orth
wrote:
> Paolo Bonzini writes:
>
>> On 08/10/2011 06:05 PM, Rainer Orth wrote:
> >> True: it is called once per multilib.
>
> Just to doublecheck, are we sure that unwind.h is always the same?
>>> Yep: it's unwind-generic.h for almost a
No tests depended on this, but since we can now stream locations
properly, I added them.
Tested on x86_64. Committed.
Diego.
* pph-streamer-in.c (pph_read_tree_body): Read
STATIC_ASSERT_SOURCE_LOCATION and tree_lambda_expr::locus.
* pph-streamer-out.c (pph_write_tree_
On 08/11/2011 07:03 PM, Sebastian Pop wrote:
On Thu, Aug 11, 2011 at 12:52, Tobias Grosser wrote:
I will commit this patch after the configure changes are in (and meanwhile
no further improvements were suggested for this patch).
Ok, thanks. Let's hope we will have a configure maintainer that
On Thu, Aug 11, 2011 at 12:52, Tobias Grosser wrote:
> I will commit this patch after the configure changes are in (and meanwhile
> no further improvements were suggested for this patch).
Ok, thanks. Let's hope we will have a configure maintainer that has some
spare cycles to go over your first
On Thu, Aug 11, 2011 at 10:14 AM, Richard Guenther
wrote:
> On Thu, Aug 11, 2011 at 6:54 PM, Gabriel Charette wrote:
>> On Thu, Aug 11, 2011 at 12:27 AM, Richard Guenther
>> wrote:
>>> On Wed, Aug 10, 2011 at 8:22 PM, Gabriel Charette wrote:
There was a bug where c_finish_options would cre
Hi,
straightforward (7.1.5/2). Tested x86_64-linux, committed to mainline.
Paolo.
PS: one could also imagine adding a macro like
# ifdef __GXX_EXPERIMENTAL_CXX0X__
# define _GLIBCXX_INLINE_OR_CONSTEXPR constexpr
# else
# define _GLIBCXX_INLINE_OR_CONSTEXPR inline
# endif
to be used where no
On 08/10/2011 11:41 PM, Sebastian Pop wrote:
Hi Tobi,
The patch looks good modulo some formatting changes:
Thanks for your review. I attached the updated patch.
I will commit this patch after the configure changes are in (and
meanwhile no further improvements were suggested for this patch).
OK.
Jason
Gabriel Charette a écrit:
> That could work given _cpp_lex_direct does receive a cpp_reader as a
> parameter. We would need to add logic in _cpp_lex_direct to support
> this new field.
Correct.
>
> As I mentioned, we have the same problem in pph where we need to force
> a location (i.e. the lex
Gabriel Charette a écrit:
> On Thu, Aug 11, 2011 at 12:27 AM, Richard Guenther
> wrote:
>> On Wed, Aug 10, 2011 at 8:22 PM, Gabriel Charette wrote:
>>> There was a bug where c_finish_options would create some builtins and
>>> assign them source_locations in the linemap other than BUILTINS_LOCA
On Thu, Aug 11, 2011 at 6:54 PM, Gabriel Charette wrote:
> On Thu, Aug 11, 2011 at 12:27 AM, Richard Guenther
> wrote:
>> On Wed, Aug 10, 2011 at 8:22 PM, Gabriel Charette wrote:
>>> There was a bug where c_finish_options would create some builtins and
>>> assign them source_locations in the li
Small cleanup patch to remove two functions that just dispatched to
pph cache functions. It was useless and confusing.
Tested on x86_64. Committed.
Diego.
* pph-streamer-in.c (pph_in_shared_data): Remove. Update all users.
(pph_register_shared_data): Remove. Update all user
On Thu, Aug 11, 2011 at 12:27 AM, Richard Guenther
wrote:
> On Wed, Aug 10, 2011 at 8:22 PM, Gabriel Charette wrote:
>> There was a bug where c_finish_options would create some builtins and assign
>> them source_locations in the linemap other than BUILTINS_LOCATION == 1.
>>
>> Thus, when calling
On Thu, Aug 11, 2011 at 7:22 AM, Dodji Seketeli wrote:
> Hello,
>
> gch...@google.com (Gabriel Charette) a écrit:
>
>> diff --git a/gcc/c-family/c-opts.c b/gcc/c-family/c-opts.c
>> index 3227f7b..1af8e7b 100644
>> --- a/gcc/c-family/c-opts.c
>> +++ b/gcc/c-family/c-opts.c
>> @@ -1306,13 +1306,15 @
On Thu, Aug 11, 2011 at 08:23, Michael Matz wrote:
> The obvious question: have you measured performance (with an optimized
> cc1) on a medium sized exectutable, say cc1 itself? (I'm worrying about
> inlining that doesn't happen anymore).
I compared three things: profiled bootstrap time for C,
Paolo Bonzini writes:
> Writing that down in a wiki page (even as unformatted text) would be
> useful.
While I'm not too fond of WiKis, I've added some text to
http://gcc.gnu.org/wiki/Top-Level_Libgcc_Migration
It's ugly right now, but better than nothing.
Rainer
--
---
On Thu, Aug 11, 2011 at 08:59, Romain Geissler wrote:
> Find attached the updated patch. Retested for regression on x86_64.
Thanks. Committed at rev 177674.
Diego.
This is an optimization in machine dependent reorg to
remove redundant comparisons like in
cc0 = compare (Reg, Num)
if (cc0 == 0)
goto L1
cc0 = compare (Reg, Num)
if (cc0 > 0)
goto L2
The second comparison is redundant an can be removed.
Code like this can be seen in binary
On 08/11/2011 05:26 PM, Rainer Orth wrote:
> Actually I think the installation of all the installed target headers
> should move to libgcc's Makefiles (and the headers themselves should move
> under the libgcc/ directory).
Agreed, added to my ever-growing todo list for the libgcc move. Thi
When adjusting the C++ FE code generation for base type copies
I noticed the XFAILs on g++.dg/tree-ssa/pr41186.C and investigated
things a bit. We can fix the testcase easily iff the frontend
emits the MEM_REF variants, thus the following patch.
Bootstrapped and tested on x86_64-unknown-linux-gn
On Thu, Aug 11, 2011 at 08:43, Michael Matz wrote:
> Hi,
>
> gcc.dg/graphite/run-id-pr47593.c (included for PR47653) hardcodes -m32 in
> dg-options. That's not supported everywhere and the testcase is just as
> valid without -m32. If it was the case that it needed -m32 for breaking
> that would
"Joseph S. Myers" writes:
> Actually I think the installation of all the installed target headers
> should move to libgcc's Makefiles (and the headers themselves should move
> under the libgcc/ directory).
Agreed, added to my ever-growing todo list for the libgcc move. This
would be
* gcc/gi
Paolo Bonzini writes:
> On 08/10/2011 06:05 PM, Rainer Orth wrote:
>> True: it is called once per multilib.
>>> >
>>> > Just to doublecheck, are we sure that unwind.h is always the same?
>> Yep: it's unwind-generic.h for almost all targets, just a few arm
>> targets use config/arm/unwind-a
The reduced test case for m68k showed a TLS symbol
given to emit_single_push_insn, which resulted in
a call insn inside a sequence that I didn't expect.
We don't see this on i686 because we pretend that
the TLS call isn't a call. Or something.
Bootstrapped on i586 for sanity.
Committed.
r~
Marc,
> On Thu, 11 Aug 2011, Rainer Orth wrote:
>
>> I could, but am a bit reluctant to do so since such a fix feels quite
>> fragile, and `fixes' the Solaris headers in many places where they are
>> completely correct. I'll also have to touch , ,
>> , and , that all have using std::tm etc. claus
On Thu, 11 Aug 2011, Rainer Orth wrote:
I could, but am a bit reluctant to do so since such a fix feels quite
fragile, and `fixes' the Solaris headers in many places where they are
completely correct. I'll also have to touch , ,
, and , that all have using std::tm etc. clauses.
Actually, you
Hi Paolo,
> All in all it seems to me that we are pretty close to be able to fix this
> old issue now and we are even in Stage 1, thus we can afford to take a bit
> of risk and handle possible fallout, thus I would recommend we do for now
> the above preprocessor dance (we are talking only about 4
On Thu, 11 Aug 2011, Paolo Bonzini wrote:
> On 08/11/2011 04:25 PM, Joseph S. Myers wrote:
> > > The actual problem are not the runtime libraries, which already know to
> > > search $builddir/.../libgcc for unwind.h and related files. The
> > > copyback is only for the benefit of the testsuite
"Joseph S. Myers" writes:
> On Thu, 11 Aug 2011, Rainer Orth wrote:
>
>> The actual problem are not the runtime libraries, which already know to
>> search $builddir/.../libgcc for unwind.h and related files. The
>> copyback is only for the benefit of the testsuite (gcc.target, g++.dg,
>> gnat.dg
On 08/11/2011 04:25 PM, Joseph S. Myers wrote:
> The actual problem are not the runtime libraries, which already know to
> search $builddir/.../libgcc for unwind.h and related files. The
> copyback is only for the benefit of the testsuite (gcc.target, g++.dg,
> gnat.dg, and gcc.dg) where I w
This fixes three ones that are clearly wrong.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2011-08-11 Richard Guenther
* lto-cgraph.c (input_node): Use DECL_BUILT_IN.
* tree-vrp.c (stmt_interesting_for_vrp): Likewise.
(vrp_visit_stmt): Likew
On Thu, 11 Aug 2011, Rainer Orth wrote:
> The actual problem are not the runtime libraries, which already know to
> search $builddir/.../libgcc for unwind.h and related files. The
> copyback is only for the benefit of the testsuite (gcc.target, g++.dg,
> gnat.dg, and gcc.dg) where I was too lazy
Hello,
gch...@google.com (Gabriel Charette) a écrit:
> diff --git a/gcc/c-family/c-opts.c b/gcc/c-family/c-opts.c
> index 3227f7b..1af8e7b 100644
> --- a/gcc/c-family/c-opts.c
> +++ b/gcc/c-family/c-opts.c
> @@ -1306,13 +1306,15 @@ c_finish_options (void)
> {
>size_t i;
>
> -
On 11/08/11 12:56, Richard Sandiford wrote:
> Trying to compile a certain embedded benchmark with -falign-loops=8
> or above fails due to out-of-range references to the literal pool.
> This is because arm_reorg doesn't take label alignment into account.
>
> Fixed with the patch below. Tested on a
Hi,
gcc.dg/graphite/run-id-pr47593.c (included for PR47653) hardcodes -m32 in
dg-options. That's not supported everywhere and the testcase is just as
valid without -m32. If it was the case that it needed -m32 for breaking
that would be catched by normal multilib testing.
Committed as obvious
On 08/11/2011 08:57 AM, Richard Guenther wrote:
Any idea why the C++ FE prints that funny setbuf.a::mode instead
of simply setbuf.mode for g++.dg/warn/unit-1.C!?
Because by the time we're printing the FIELD_DECL, we've forgotten that
we're in a COMPONENT_REF context. We could do better, but i
On Thu, Aug 11, 2011 at 09:32, Richard Guenther
wrote:
> Well, they both make sense - you should just not confuse them.
> DECL_IS_BUILTIN should probably renamed to
> DECL_HAS_BUILTIN_SOURCE_LOCATION or similar.
Exactly. The problem is that now they both *seem* to have the same semantics.
> I'
On Thu, Aug 11, 2011 at 3:01 PM, Diego Novillo wrote:
> On Thu, Aug 11, 2011 at 03:27, Richard Guenther
> wrote:
>> On Wed, Aug 10, 2011 at 8:22 PM, Gabriel Charette wrote:
>>> There was a bug where c_finish_options would create some builtins and
>>> assign them source_locations in the linemap
Hi,
This is the last patch needed for x32 support.
convert_memory_address_addr_space
is called to convert a memory address without overflow/underflow. It
should be safe
to transform
(zero_extend:DI (plus:SI (FOO:SI) (const_int Y)))
to
(plus:DI (zero_extend:DI (FOO:SI)) (const_int Y))
GCC only
On Thu, Aug 11, 2011 at 03:27, Richard Guenther
wrote:
> On Wed, Aug 10, 2011 at 8:22 PM, Gabriel Charette wrote:
>> There was a bug where c_finish_options would create some builtins and assign
>> them source_locations in the linemap other than BUILTINS_LOCATION == 1.
>>
>> Thus, when calling DE
Hello,
In order to run regression test on simulator, the testglue.c will be
compiled and linked to return the return status of cases executed on
simulator. In my recent work, I built a cross compiler for arm-none-eabi.
When run regression test for libstdc++-v3 using command "make check-target
"RUN
On 08/10/2011 05:20 PM, Diego Novillo wrote:
On Mon, Aug 8, 2011 at 07:39, Romain Geissler wrote:
2011/7/20 Diego Novillo:
On Mon, Jul 18, 2011 at 03:06, Romain Geissler
I will commit this patch shortly.
Diego.
Ping !
Romain, please send me a current patch against today's trunk.
Thank
This restores some of the uninitialized warnings for memory and
applies some TLC to the machinery.
Any idea why the C++ FE prints that funny setbuf.a::mode instead
of simply setbuf.mode for g++.dg/warn/unit-1.C!? The latter is
what debug_generic_expr dumps.
Bootstrapped and tested on x86_64-unk
On Thu, 11 Aug 2011, Uros Bizjak wrote:
> Hello!
>
> Currently, conversion from floating point to integer on 64bit targets
> goes through DImode temporary, due to missing BUILT_IN_ICEIL (and
> other) builtins that can convert directly from FP to integer.
>
> Attached patch introduces these built
On Thu, Aug 11, 2011 at 04:45, Richard Guenther wrote:
>> 5- Handle CALL_EXPR in lto_materialize_tree. Despite the name this is
>> a generic tree streaming function (the renaming patch will come
>> in the next few days). Handling CALL_EXPR does not
>> affect LTO (since those nodes are
Hello!
Currently, conversion from floating point to integer on 64bit targets
goes through DImode temporary, due to missing BUILT_IN_ICEIL (and
other) builtins that can convert directly from FP to integer.
Attached patch introduces these builtins and improves i.e.:
int
test (double a)
{
return
On Thu, Aug 11, 2011 at 08:23, Michael Matz wrote:
> The obvious question: have you measured performance (with an optimized
> cc1) on a medium sized exectutable, say cc1 itself? (I'm worrying about
> inlining that doesn't happen anymore).
Ah, good point. No, I haven't. I'll measure it.
Dieg
Hi,
On Wed, 10 Aug 2011, Diego Novillo wrote:
>The API in the tree streamer offers functions to read/write the
>tree header and the body, a streamer cache and everything that
>used to be inside the LTO streamer. This makes no difference to
>LTO, since it simply means that some fo
Hi Guys,
I am applying the patch below on behalf of Renesas. It adds support
to the RX backend for conditional register moves.
Tested without any regressions on an rx-elf toolchain.
Cheers
Nick
gcc/ChangeLog
2011-08-11 Kazuhiro Inaoka
* config/rx/rx.md (movsicc): Allow reg
Paolo Bonzini writes:
> On 08/10/2011 06:50 PM, Pedro Alves wrote:
>> On Wednesday 10 August 2011 17:05:08, Rainer Orth wrote:
>>> > Paolo Bonzini writes:
>>> >
> > >> True: it is called once per multilib.
> >
> > Just to doublecheck, are we sure that unwind.h is always the s
Trying to compile a certain embedded benchmark with -falign-loops=8
or above fails due to out-of-range references to the literal pool.
This is because arm_reorg doesn't take label alignment into account.
Fixed with the patch below. Tested on arm-linux-gnueabi. I also
tried compiling libav with a
On 08/10/2011 06:50 PM, Pedro Alves wrote:
On Wednesday 10 August 2011 17:05:08, Rainer Orth wrote:
> Paolo Bonzini writes:
>
> >> True: it is called once per multilib.
> >
> > Just to doublecheck, are we sure that unwind.h is always the same?
>
> Yep: it's unwind-generic.h for almost
On 08/10/2011 01:14 PM, Rainer Orth wrote:
* At the moment, SHLIB_LINK is used in gcc/Makefile.in and the various
Make-lang.in fragments to check if the target supports a shared
libgcc_s. I've introduced gcc/config/t-slibgcc (from t-slibgcc-dummy)
for this, which sets SHLIB = true, addi
Ramana,
I only see the following three combinations are meaningful,
* cmp l, lPy // keep cmp, and length is 2, on thumb
* cmp r, rI // keep cmp, and length is 4
* cmp r, L // convert to cmn, and length is 4
According to ARM ARM, for negative immediate, all encodings for cmp/cmn are
4-byte lo
Hi Uros,
Thanks for patience reviewing my English :) and for finding a bug in souces.
Updated patch is attached. It was bootstrapped successfully.
updated ChangeLog entry:
2011-08-11 Kirill Yukhin
* common/config/i386/i386-common.c (OPTION_MASK_ISA_AVX2_SET): New.
(OPTION_MAS
On Wed, 10 Aug 2011, Richard Guenther wrote:
> On Wed, 10 Aug 2011, Richard Guenther wrote:
>
> > On Wed, 10 Aug 2011, Jason Merrill wrote:
> >
> > > On 08/10/2011 08:35 AM, Richard Guenther wrote:
> > > > * call.c (build_over_call): Call memcpy unconditionally.
> > >
> > > OK. Have yo
Bernd Schmidt writes:
> On 07/21/11 13:28, Richard Sandiford wrote:
>> +static void
>> +process_bypass_2 (decl_t model, decl_t out_insn_reserv, void *data)
>> +{
>> + struct bypass_decl *bypass;
>> + decl_t in_insn_reserv;
>> +
>> + in_insn_reserv = (decl_t) data;
>> + bypass = XCNEW (struct b
1 - 100 of 103 matches
Mail list logo
