Hi!
This patch adds VI_256 vec_interleave_{high,low}mode as well as
using it in the vector expander.
While it needs 3 insns for each, the first two will be actually CSEd
if both patterns are expanded (the usual case from the vectorizer, e.g.
for vect-strided-store-u32-i2.c), so we end up with 2
Hi!
mulv2di3 can be expanded the same as mulv2di3.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2011-10-14 Jakub Jelinek ja...@redhat.com
* config/i386/sse.md (mulv2di3): Macroize using VI8_AVX2
iterator.
(ashlmode3): Use VI248_AVX2 iterator
Richard Henderson schrieb:
On 10/13/2011 12:00 PM, Georg-Johann Lay wrote:
What do you propose?
o A command line option that is on per default like
-mnoreturn-tail-calls or -mjmp-noreturn
The command-line-option. I think I prefer -mjump-noreturn,
as the inverse -mno-noreturn-tail-calls is
On 10/13/2011 10:07 PM, H.J. Lu wrote:
On Thu, Oct 13, 2011 at 11:15 AM, Richard Kenner
ken...@vlsi1.ultra.nyu.edu wrote:
The answer to H.J.'s Why do we do it for MEM then? is simply
because no one ever thought about not doing it
No, that's false. The same expand_compound_operation /
Weddington, Eric schrieb:
This is yet another attempt to fix PR46278 (fake X addressing).
After the previous clean-ups it is just a small change.
caller-saves.c tries to eliminate call-clobbered hard-regs allocated to
pseudos around function calls and that leads to situations that
On Oct 13, 2011, at 4:19 PM, Paolo Carlini wrote:
Ok for mainline?
Ok.
On 13 Oct 2011, at 23:38, Iain Sandoe wrote:
On 13 Oct 2011, at 23:22, Mike Stump wrote:
+/* Add $LDBL128 suffix to long double builtins for ppc darwin. */
static void
-darwin_patch_builtin (int fncode)
+darwin_patch_builtin (enum built_in_function fncode)
This is a property of the
On Oct 13, 2011, at 5:02 PM, Nicola Pero wrote:
I actually forgot to post a tiny bit that is required to support
the additional objc/objc-map.h and objc/objc-map.c files. It's
part of the same patch. Apologies.
Hum, looks fairly obvious to me. :-)
On Oct 13, 2011, at 10:40 PM, Jan Kratochvil wrote:
On Wed, 12 Oct 2011 16:18:07 +0200, Jan Kratochvil wrote:
On Wed, 12 Oct 2011 16:07:24 +0200, Tristan Gingold wrote:
I fear that this may degrade performance of other debuggers. What about
adding a command line option ?
I can test idb,
On Thu, 13 Oct 2011, Jakub Jelinek wrote:
Hi!
Andrew mentioned on IRC he found walk_stmt_load_store_addr_ops
doesn't handle COND_EXPR weirdo first argument well, the following
patch is an attempt to handle that.
I've noticed similar spot in verify_ssa, though in that case I'm not
sure
This patch, originally from Chung-Lin Tang, fixes PR50496.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50496
Can someone please review and commit it?
Thanks.
Bootstrapped and tested on x86_64-pc-linux-gnu.
PR middle-end/50496
* cfgrtl.c (try_redirect_by_replacing_jump): Treat
we've been building the FPR routines for ages (for compatibility with
system tools)..
This implements their use and also the GPRs - the latter makes an
appreciable reduction in code size,
the former is neutral for non-fp intensive code but, for example,
makes an appreciable reduction to the
On Oct 14, 2011, at 2:05 AM, Iain Sandoe wrote:
This implements their use and also the GPRs - the latter makes an appreciable
reduction in code size,
OK for trunk?
Ok. Watch for problems with async stack walking (hitting sample in Activity
Monitor, or the walking done by CrashReporter)...
As per the PR audit trail, there is no reason to retain this special-
casing for Darwin.
(given that current GCC is not build-able using Darwin toolsets of
the vintage that required the case).
Mike has OK'd this off-list - but, since Ralf commented on the
previous version, I'd like to give
As per the PR audit trail, there is no reason to retain this in the
building of GCC.
As for its use as a general option in tool-builds;
With current darwin toolsets it has the potential to cause issues when
using convenience libs containing common.
OK for trunk?
Iain
gcc/ada:
PR
This fixes PR50723, a common error when inserting stmts after some
others and using GSI_NEW (which will make the iterator point to
the _first_ newly inserted stmt).
Bootstrap and regtest running on x86_64-unknown-linux-gnu.
Richard.
2011-10-14 Richard Guenther rguent...@suse.de
PR
2011/10/14 Georg-Johann Lay a...@gjlay.de:
Weddington, Eric schrieb:
This is yet another attempt to fix PR46278 (fake X addressing).
After the previous clean-ups it is just a small change.
caller-saves.c tries to eliminate call-clobbered hard-regs allocated to
pseudos around function calls
On Fri, Oct 14, 2011 at 1:12 AM, Tom de Vries tom_devr...@mentor.com wrote:
On 10/12/2011 02:19 PM, Richard Guenther wrote:
On Wed, Oct 12, 2011 at 8:35 AM, Tom de Vries vr...@codesourcery.com wrote:
Richard,
I have a patch for PR50672.
When compiling the testcase from the PR with
On Wed, Oct 12, 2011 at 7:16 PM, Michael Matz m...@suse.de wrote:
Hi,
this adds a mean to retain restrict information without relying on
restrict casts. In the patch it's emitted by the gimplifier when it sees
a norestrict-restrict cast (which from then on is useless), at which
point also
On Thu, Oct 13, 2011 at 6:57 PM, Andi Kleen a...@firstfloor.org wrote:
Or I am missing someting?
I often see the x86 vectorizer with -mtune=generic generate a lot of
complicated code just to adjust for potential misalignment.
My thought was just if the alias oracle knows what the original
Hello,
while testing my constructor draft patch with FGSL, I found two bugs
(4.6/4.7 regressions): PR target/50721 (segfault on x86-64 after
execution - one of the rare -O0 only bugs) - and while trying to debug
it, I found this bug.
The problem is that for -fcheck=pointer, gfortran expects
On Thu, 13 Oct 2011 16:12:17 +0100
Richard Earnshaw rearn...@arm.com wrote:
On 13/10/11 15:56, Joseph S. Myers wrote:
Indeed, vector initializers are part of the target-independent GNU
C language and have target-independent semantics that the elements
go in memory order, corresponding to
Hello guys,
Here is a bunch of tests which check basic vectorization abilities to
generate AVX2 instructions.
testsuite/ChangeLog entry is:
2011-10-14 Kirill Yukhin kirill.yuk...@intel.com
* gcc.target/i386/avx2-vpaddd-3.c: New test.
* gcc.target/i386/avx2-vpaddw-3.c: Ditto.
This patch, originally from Chung-Lin Tang, fixes PR50496.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50496
Can someone please review and commit it?
A proper patch submission should include a description of the problem and a
rationale for the proposed fix (unless it is trivial).
--
Eric
This is a ping. Change affects Atom only and was made because it
really gives better performance on this architecture. This fact
actually leads to the thought that old value is just a simple
misprint.
Please look.
Vladimir
2011/9/30 Vladimir Yakovlev vbyakov...@gmail.com:
This patch fixes
On Thu, Oct 13, 2011 at 3:25 PM, Kai Tietz ktiet...@googlemail.com wrote:
Hello,
this new version addresses the comments from you.
On gimplify.c's gimplify_expr we didn't handled the case that operands
for TRUTH-AND/OR/XOR expressions need to have same operand-size in
case of transformation
Denis Chertykov schrieb:
Georg-Johann Lay :
Weddington, Eric schrieb:
This is yet another attempt to fix PR46278 (fake X addressing).
After the previous clean-ups it is just a small change.
caller-saves.c tries to eliminate call-clobbered hard-regs allocated to
pseudos around function
On Fri, Oct 14, 2011 at 03:13:45PM +0400, Kirill Yukhin wrote:
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx2-vpaddb-3.c
@@ -0,0 +1,49 @@
+/* { dg-do run } */
+/* { dg-options -mavx2 -O2 -ftree-vectorize -save-temps } */
+/* { dg-require-effective-target avx2 } */
+
...
+
+/* { dg-final {
Consider this testcase:
$ cat test.cpp
class GCAlloc {
};
class BaseAlloc {
};
class String;
class Base {
public:
virtual void destroy( String *str ) const =0;
};
class String: public GCAlloc {
const Base *m_class;
public:
enum constants {
};
String( const char *data );
This cleans up the chain streaming code under a single core function
and various entry points. In the process it fixes the timeout problem
in c1limits-externalid.cc (we were just messing up the chain when
streaming it in reverse).
Tested on x86_64. Committed to branch.
*
* pph-streamer-out.c (pph_out_tree_1): Rename from pph_out_any_tree.
Update all users.
diff --git a/gcc/cp/pph-streamer-out.c b/gcc/cp/pph-streamer-out.c
index ffbe710..e337a31 100644
--- a/gcc/cp/pph-streamer-out.c
+++ b/gcc/cp/pph-streamer-out.c
@@ -619,16 +619,15 @@
* pph-streamer-out.c (pph_out_tree_1): Do not call pph_trace_tree
unless flag_pph_tracer is set.
diff --git a/gcc/cp/pph-streamer-out.c b/gcc/cp/pph-streamer-out.c
index e337a31..d534b42 100644
--- a/gcc/cp/pph-streamer-out.c
+++ b/gcc/cp/pph-streamer-out.c
@@ -1916,7 +1916,8 @@
On 11-10-13 17:55 , Gabriel Charette wrote:
I'm not sure exactly how you skip headers already parsed now (we
didn't used to when I wrote this code and that was the only problem
remaining in the line_table (i.e. duplicate entries for guarded
headers in the non-pph compile)), but couldn't you
Hello!
This is a ping. Change affects Atom only and was made because it
really gives better performance on this architecture. This fact
actually leads to the thought that old value is just a simple
misprint.
This patch fixes performance regression with -mtune=atom. Changing
atom cost
On Fri, Oct 14, 2011 at 8:18 AM, Jakub Jelinek ja...@redhat.com wrote:
mulv2di3 can be expanded the same as mulv2di3.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2011-10-14 Jakub Jelinek ja...@redhat.com
* config/i386/sse.md (mulv2di3): Macroize using
This follows up Michas testcase where we fail to handle the
conservatively propagated restrict tags properly. The following
patch simplifies handling of restrict in the oracle and thus
only excludes NONLOCAL (as designed), but not ESCAPED from
conflict checking.
Bootstrapped on
On 07/13/11 16:03, Richard Earnshaw wrote:
* config/arm/arm.c (store_multiple_sequence): Avoid cases where
the base reg is stored iff compiling for Thumb1.
* gcc.target/arm/pr49641.c: New test.
Ping. Richard, you replied to the mail but didn't comment on the patch.
Bernd
On Thu, Oct 13, 2011 at 10:40 AM, Artem Shinkarov
artyom.shinkar...@gmail.com wrote:
On Thu, Oct 13, 2011 at 10:23 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Oct 13, 2011 at 10:59 AM, Mike Stump mikest...@comcast.net wrote:
On Oct 12, 2011, at 2:37 PM, Artem Shinkarov
Thanks, done.
Anything else?
K
On Fri, Oct 14, 2011 at 3:53 PM, Jakub Jelinek ja...@redhat.com wrote:
On Fri, Oct 14, 2011 at 03:13:45PM +0400, Kirill Yukhin wrote:
--- /dev/null
+++ b/gcc/testsuite/gcc.target/i386/avx2-vpaddb-3.c
@@ -0,0 +1,49 @@
+/* { dg-do run } */
+/* { dg-options
On 10/13/2011 10:37 PM, Paolo Carlini wrote:
+ if ((TYPE_PTR_P (type1) TYPE_PTR_P (type2))
+ || (TYPE_PTRMEM_P (type1) TYPE_PTRMEM_P (type2))
+ || TYPE_PTRMEMFUNC_P (type1))
You don't need to check TYPE_PTR_P or TYPE_PTRMEM_P for type2 here (or
in the condition above)
On 10/13/2011 07:19 PM, Paolo Carlini wrote:
-@item -Wno-format-zero-length @r{(C and Objective-C only)}
+@item -Wno-format-zero-length @r{(C, C++, Objective-C and Objective-C++ only)}
I would just remove the {...only} to match the other -Wformat items.
OK.
Jason
On Fri, Oct 14, 2011 at 3:42 PM, Artem Shinkarov
artyom.shinkar...@gmail.com wrote:
On Thu, Oct 13, 2011 at 10:40 AM, Artem Shinkarov
artyom.shinkar...@gmail.com wrote:
On Thu, Oct 13, 2011 at 10:23 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Oct 13, 2011 at 10:59 AM, Mike
Tristan == Tristan Gingold ging...@adacore.com writes:
Tristan I am not against this patch, my only concern is that there are many
Tristan many dwarf consumers and I have no idea how they will react to this
Tristan change.
I tend to think that this is the wrong standard to apply. In this case
On 10/14/2011 03:57 PM, Jason Merrill wrote:
On 10/13/2011 07:19 PM, Paolo Carlini wrote:
-@item -Wno-format-zero-length @r{(C and Objective-C only)}
+@item -Wno-format-zero-length @r{(C, C++, Objective-C and
Objective-C++ only)}
I would just remove the {...only} to match the other -Wformat
Hi,
On Fri, 14 Oct 2011, Richard Guenther wrote:
But I'm not sure. I think we should delete this check from
verify_ssa and instead have a corresponding check in
verify_stmts (which already properly walks trees) that
for an SSA name we encounter we do have a properly linked use
(see
On Fri, Oct 14, 2011 at 05:53:28PM +0400, Kirill Yukhin wrote:
Thanks, done.
Anything else?
First of all, most of the testcases look very similar, the only changes
in between many of them are (unimportant) function names and different type.
So, I think it would be much better to just write
Could anyone checkin that?
Thanks,
Vladimir
2011/10/14 Uros Bizjak ubiz...@gmail.com:
Hello!
This is a ping. Change affects Atom only and was made because it
really gives better performance on this architecture. This fact
actually leads to the thought that old value is just a simple
Hi!
This patch improves generated code for SSE4.1 and even more
for AVX2 on the attached testcases. SSE4.1 has pmuldq
(where SSE2 only had pmuludq), so can handle signed widening shifts fine
too.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2011-10-14 Jakub Jelinek
Hi!
This patch allows to vectorize negations using 32-byte vectors.
2011-10-14 Jakub Jelinek ja...@redhat.com
* config/i386/sse.md (negmode2): Use VI_AVX2 iterator instead
of VI_128.
--- gcc/config/i386/sse.md.jj 2011-10-14 13:05:58.0 +0200
+++
Hi,
On 10/13/2011 10:37 PM, Paolo Carlini wrote:
+ if ((TYPE_PTR_P (type1) TYPE_PTR_P (type2))
+ || (TYPE_PTRMEM_P (type1) TYPE_PTRMEM_P (type2))
+ || TYPE_PTRMEMFUNC_P (type1))
You don't need to check TYPE_PTR_P or TYPE_PTRMEM_P for type2 here (or
in the condition above)
On 09/28/2011 06:38 PM, Eric Botcazou wrote:
[Vlad, if you have a few minutes, would you mind having a look at the couple of
questions at the end of the message? Thanks in advance].
No problem.
Here are the results of the investigation. Pseudo 116 needs to be assigned a
hard register. It
When the pragma Short_Circuit_And_Or is used, no part of the right operand
of an and or or operator should be executed if the left operand would
short-circuit the evaluation of the corresponding and then or or else.
However, run-time checks associated with such operands were being evaluated
On 14/10/11 11:42, Julian Brown wrote:
On Thu, 13 Oct 2011 16:12:17 +0100
Richard Earnshaw rearn...@arm.com wrote:
On 13/10/11 15:56, Joseph S. Myers wrote:
Indeed, vector initializers are part of the target-independent GNU
C language and have target-independent semantics that the elements
AI05-0053 forbids the use of the aliased keyword in the object declaration of an
extended return statement. This avoids semantic complications with return
objects that are not of an immutably limited type, and which therefore are not
necesarily built in place.
Compiling the following in -gnat12
On Fri, 14 Oct 2011, Julian Brown wrote:
* The vectorizer can use whatever layout it wants for vectors in
either endianness. Vectorizer vectors never interact with either
GCC generic (source-level) vectors, nor the NEON intrinsics. Also
they never cross ABI boundaries.
I don't
On 10/13/2011 11:31 PM, Georg-Johann Lay wrote:
Richard Henderson schrieb:
On 10/13/2011 12:00 PM, Georg-Johann Lay wrote:
What do you propose?
o A command line option that is on per default like
-mnoreturn-tail-calls or -mjmp-noreturn
The command-line-option. I think I prefer
On 10/14/2011 12:19 PM, Tobias Burnus wrote:
while testing my constructor draft patch with FGSL, I found two bugs
(4.6/4.7 regressions): PR target/50721 (segfault on x86-64 after
execution - one of the rare -O0 only bugs)
That bug turned out to be a out-of-bounds problem between Fortran and
On 10/14/2011 07:19 AM, Jakub Jelinek wrote:
2011-10-14 Jakub Jelinek ja...@redhat.com
* config/i386/sse.md (negmode2): Use VI_AVX2 iterator instead
of VI_128.
Ok.
r~
On 10/13/2011 05:01 PM, Bernd Schmidt wrote:
On 09/30/11 17:29, Bernd Schmidt wrote:
This patch allows a backend to set a new scheduler flag, DO_PREDICATION,
which will make the haifa scheduler try to move insns across jumps by
predicating them. On C6X, the primary benefit is to fill jump delay
On Thu, Oct 13, 2011 at 11:51 PM, Paolo Bonzini bonz...@gnu.org wrote:
On 10/13/2011 10:07 PM, H.J. Lu wrote:
On Thu, Oct 13, 2011 at 11:15 AM, Richard Kenner
ken...@vlsi1.ultra.nyu.edu wrote:
The answer to H.J.'s Why do we do it for MEM then? is simply
because no one ever thought about
So, I had not fixed c1limits-externalid.cc. I had simply messed up
merging decls into chains.
The problem here is that we insert decls into a chain *before* we
finish reading them. So, when pph_merge_into_chain inserts the
partially read DECL into the chain (or returns an existing DECL), the
On 09/07/2011 05:38 AM, Alexander Monakov wrote:
Hello,
The patch repairs a problem when we attempt to substitute an insn like
(... (cmp (mem (reg:DI ax)) (reg:SI ax))) (note different modes) through
(set (reg:DI ax) (reg:DI dx)), which leaves the (reg:SI ax) part of the
comparison intact,
On 09/13/2011 12:42 PM, Alexander Monakov wrote:
Fixed as follows, bootstrapped and regtested on x86_64-linux and ia64-linux
(without java, with one recent SRA patch reverted to unbreak bootstrap) with
sel-sched enabled at -O2. OK for trunk?
Ok with a small code format change below.
(a small
On 10/13/2011 01:35 PM, Michael Spertus wrote:
+int main() {
+ assert(typeid(bF::type)
+ == typeid(typesA,C,D,A,C,D,B,C,C,D,B,E));
+ assert(typeid(dbF::type) == typeid(typesA,B,E));
+ assert(typeid(dbint::type) == typeid(types));
+ return 0;
+}
Let's make this a compile-time test
On 10/14/11 17:35, Vladimir Makarov wrote:
On 10/13/2011 05:01 PM, Bernd Schmidt wrote:
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg02053.html
It is hard to read the patch without function names.
Oh, you mean without -p? Not sure how that happened, svn diff seems to
add them when run on my
Index: tree-asan.c
===
--- tree-asan.c (revision 0)
+++ tree-asan.c (revision 0)
@@ -0,0 +1,512 @@
+/* AddressSanitizer, a fast memory error detector.
+ Copyright (C) 2011 Free Software Foundation, Inc.
+ Contributed by Kostya
On 10/14/2011 07:18 AM, Jakub Jelinek wrote:
+ /* This would be 2 insns shorter if
+ rperm[i] = GEN_INT (((~i 1) 2) + i / 2);
+ has been used instead (both vpslrq insns wouldn't be needed),
+ but vec_widen_*mult_hi_* is usually used together with
+ vec_widen_*mult_lo_* and
On Fri, Oct 14, 2011 at 7:15 AM, Vladimir Yakovlev vbyakov...@gmail.com wrote:
Could anyone checkin that?
Please provide a suitable patch which can be applied.
H.J.
Thanks,
Vladimir
2011/10/14 Uros Bizjak ubiz...@gmail.com:
Hello!
This is a ping. Change affects Atom only and was
On 14/10/11 16:21, Joseph S. Myers wrote:
On Fri, 14 Oct 2011, Richard Earnshaw wrote:
One way we might address this is to redefine our 128-bit vector types as
structs of low/high Dwords. Each Dword remains a vector (apart from
64-bit lane types), but the Dword order then matches the ABI
On 10/14/2011 05:36 PM, H.J. Lu wrote:
There is a testcase at
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50696
It passes with my patch.
Cool, so let's wait for the results of testing.
Paolo
Richard Henderson schrieb:
On 10/13/2011 11:31 PM, Georg-Johann Lay wrote:
Richard Henderson schrieb:
On 10/13/2011 12:00 PM, Georg-Johann Lay wrote:
What do you propose?
o A command line option that is on per default like
-mnoreturn-tail-calls or -mjmp-noreturn
The command-line-option.
On Fri, Oct 14, 2011 at 9:23 AM, Paolo Bonzini bonz...@gnu.org wrote:
On 10/14/2011 05:36 PM, H.J. Lu wrote:
There is a testcase at
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=50696
It passes with my patch.
Cool, so let's wait for the results of testing.
Paolo
Here is the complete
On 10/13/2011 11:16 PM, Jakub Jelinek wrote:
2011-10-14 Jakub Jelinek ja...@redhat.com
* config/i386/sse.md (vec_interleave_highmode,
vec_interleave_lowmode): Add AVX2 expanders for VI_256
modes.
* config/i386/i386.c (expand_vec_perm_interleave3): New function.
On 10/14/2011 06:23 PM, Georg-Johann Lay wrote:
+@item -mjump-to-noreturn
+@opindex mjump-to-noreturn
+Use a jump instruction instead of a call instruction when calling a
+no-return functions. This option is active if optimization is turned
+on and just affects the way a call instruction is
Denis Chertykov schrieb:
2011/10/10 Georg-Johann Lay a...@gjlay.de:
toplev.c complains about unwind tables currently require a frame pointer for
correctness.
This patchlet supplies a fix to avoid build warnings/test fails in that it
sets
flag_omit_frame_pointer to 0 if unwind needs FP.
I applied the following patch to the 4.6 branch to backport the fix for
libobjc/50002.
It makes sense to backport it to 4.6.x so that it appears in 4.6.2. It really
is quite a bug,
and the fix is simple/safe (and, the ObjFW guys were particularly keen on it).
Thanks
Index: class.c
I applied the following patch to backport the fix for libobjc/49883 to GCC 4.6
so that
it appears in 4.6.2. This is the clang-related problem that was recently
discussed.
Again, it's an important fix, but safe, with users (the ObjFW guys) asking for
it in 4.6.2,
which made total sense, so I
On Fri, Oct 14, 2011 at 10:13 AM, Vladimir Yakovlev
vbyakov...@gmail.com wrote:
Patch is attached.
Thanks,
Vladimir
2011-10-14 Yakovlev Vladimir vladimir.b.yakov...@intel.com
^ Should be in .
* gcc/config/i386/i386.c (atom_cost): Changed cost for
David pointed out that I had a typo in the AIX code in my builtin changes on
October 11th. I've checked this patch in as obvious.
2011-10-14 Michael Meissner meiss...@linux.vnet.ibm.com
* config/rs6000/rs6000.c (rs6000_init_builtins): Fix typo in my
change on October 11th,
Can I apply this fix to the 4.6 branch as well ?
Some users are asking for it to be backported to the 4.6 branch.
It's not a regression, because the property implementation is new in GCC 4.6,
but it is still a serious
bug if you're trying to use properties with ObjC++, as you can't use
We should check for an initializer before complaining about lack
thereof. :)
Tested x86_64-pc-linux-gnu, applying to trunk.
commit fc150101bb79b3b328263e272a44eecd1083e6cd
Author: Jason Merrill ja...@redhat.com
Date: Thu Oct 13 21:57:21 2011 -0400
PR c++/50507
* method.c
Hi,
submitter complains that, at variance with C++11, __is_base_of doesn't
handle an incomplete base type (the first parameter). The reason seems
simple: in finish_trait_expr we try to complete *both* types instead of
doing it where/when necessary.
Tested x86_64-linux. Ok?
Thanks,
Paolo.
If you configure a biarch Linux/Sparc compiler defaulting to
32-bit, but give --with-cpu= for a v9 cpu it erroneously
turns on 64-bit in TARGET_DEFAULT.
PR target/50354 reports the breakage of the opposite case after the change:
configuring for sparc64-linux --with-cpu=v8 used to build a
On 10/14/2011 08:23 PM, Paolo Carlini wrote:
Hi,
submitter complains that, at variance with C++11, __is_base_of doesn't
handle an incomplete base type (the first parameter). The reason seems
simple: in finish_trait_expr we try to complete *both* types instead
of doing it where/when
On Fri, Oct 14, 2011 at 09:21:15AM -0700, Richard Henderson wrote:
So what you're doing here is the low-part permutation:
0 4 1 5 2 6 3 7
followed by a shift to get
4 . 5 . 6 . 7 .
But you need to load a 256-bit constant from memory to get it.
Right.
I wonder if it
On 10/14/2011 08:30 PM, Paolo Carlini wrote:
On 10/14/2011 08:23 PM, Paolo Carlini wrote:
Hi,
submitter complains that, at variance with C++11, __is_base_of
doesn't handle an incomplete base type (the first parameter). The
reason seems simple: in finish_trait_expr we try to complete *both*
On Fri, Oct 14, 2011 at 19:19, Georg-Johann Lay a...@gjlay.de wrote:
Paolo Bonzini schrieb:
On 10/14/2011 06:23 PM, Georg-Johann Lay wrote:
+@item -mjump-to-noreturn
+@opindex mjump-to-noreturn
+Use a jump instruction instead of a call instruction when calling a
+no-return functions. This
On Friday 14 October 2011 18:19:00, Georg-Johann Lay wrote:
Paolo Bonzini schrieb:
On 10/14/2011 06:23 PM, Georg-Johann Lay wrote:
+@item -mjump-to-noreturn
+@opindex mjump-to-noreturn
+Use a jump instruction instead of a call instruction when calling a
+no-return functions. This
On 10/13/2011 05:22 PM, Jason Merrill wrote:
+ DECL_INITIAL (r) = error_mark_node;
While working on 50507 I noticed that we check for error_mark_node in
walk_field_subobs, so I'm changing the placeholder to void_zero_node.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit
How about using complete_type_or_else?
Jason
On 10/14/2011 11:34 AM, Jakub Jelinek wrote:
2011-10-14 Jakub Jelinek ja...@redhat.com
* config/i386/sse.md (vec_widen_smult_hi_v8hi,
vec_widen_smult_lo_v8hi, vec_widen_umult_hi_v8hi,
vec_widen_umult_lo_v8hi): Macroize using VI2_AVX2
mode iterator and any_extend
shift_right_vect,.-.L.shift_right_vect
.comm c,4096,32
.comm b,4096,32
.comm a,4096,32
.ident GCC: (GNU) 4.7.0 20111014 (experimental)
.file foo.c
.section.toc,aw
.section.text
.section.toc,aw
.LC1
From: Eric Botcazou ebotca...@adacore.com
Date: Fri, 14 Oct 2011 20:27:03 +0200
If you configure a biarch Linux/Sparc compiler defaulting to
32-bit, but give --with-cpu= for a v9 cpu it erroneously
turns on 64-bit in TARGET_DEFAULT.
PR target/50354 reports the breakage of the opposite case
Redo test to run at compile-time per Jason's suggestion
Index: libstdc++-v3/include/tr2/type_traits
===
--- libstdc++-v3/include/tr2/type_traits(revision 0)
+++ libstdc++-v3/include/tr2/type_traits(revision 0)
@@ -0,0
Looks good, thanks. I'll let Benjamin check this in.
Jason
:)
-Original Message-
From: Jason Merrill [mailto:ja...@redhat.com]
Sent: Friday, October 14, 2011 2:41 PM
To: Michael Spertus
Cc: Benjamin Kosnik; Jonathan Wakely; gcc-patches@gcc.gnu.org;
libstd...@gcc.gnu.org
Subject: Re: Intrinsics for N2965: Type traits and base classes
On Wed, Oct 12, 2011 at 03:42:12PM -0700, Richard Henderson wrote:
The generic support for vector permutation will allow for automatic
lowering to V*QImode, so all we need to add to support for these targets
is the single V16QI pattern that represents the base permutation insn.
I'm not
This patch breaks bootstrap on FreeBSD 8 and FreeBSD 9 and everyone
else using makeinfo 4.8 (when the minimum we require right now is
version 4.7).
Hence I went ahead and applied to simple patch below. If you'd
like to address this differently, happy to see a different approach.
Testing a
+@item -mjump-to-noreturn
+@opindex mjump-to-noreturn
+Use a jump instruction instead of a call instruction when calling a
+no-return functions. This option is active if optimization is turned
+on and just affects the way a call instruction is printed out.
Would emit be better here than
If one wants a 32-bit default compiler, they should build for the
sparc-linux target. And this is absolutely trivial to make happen
in the environments where this is supposedly a problem.
I have criticized so many times this combination in the past, while Jakub and
also you IIRC were
1 - 100 of 116 matches
Mail list logo