Hi Gerald, Joseph
On 9 Dec 2011, at 23:45, Gerald Pfeifer wrote:
On Wed, 30 Nov 2011, Iain Sandoe wrote:
There is really no point in having a flag to control the Objective
C (orObjective C++) exceptions [@throw, as opposed to throw]
scheme, since weactually only support one scheme for each
I've committed the attached patch to resolve 51449. Even though the function is
entirely inlined, we need to emit its coverage counters.
tested in i686-pc-linux-gnu with profiled bootstrap.
nathan
2011-12-10 Nathan Sidwell
PR gcov-profile/51449
* coverage.c (coverage_end_fu
On 9 Dec 2011, at 20:56, Iain Sandoe wrote:
[[** I'm checking out whether it's feasible to switch off PIC for
m64 Darwin .. so one could just do the test -fno-PIC ..
... works for m32 - but PIC is jammed on for x86/m64 ... ]]
I guess the problem is that, whilst we *could* arrange to all
Hi,
this patch addresses ABI change about class-member-functions for
32-bit IA Windows native target. For
it the calling-convention is now __thiscall, as described by
manufactor's ABI description.
ChangeLog
2011-12-10 Kai Tietz
PR libgcj/50053
* java/lang/natClass.cc (java::
Kai Tietz writes:
> + /* Class constructor/destructor have for 32-bit native Windows ABI
> + __thiscall calling convention. */
/* Class constructors/destructors have __thiscall calling
convention for 32-bit native Windows ABI. */
Andreas.
--
Andreas Schwab, sch...@linux-m6
On Fri, Dec 9, 2011 at 4:56 PM, Richard Sandiford
wrote:
> Carrot Wei writes:
>> Since it also affects 4.6 branch, can this and r176270 also be ported
>> to gcc4.6?
>
> Always worth asking, but in this case, I'm not sure it's appropriate.
> The patch is pretty invasive, and I don't think the bug
Attached patch applied.
2011-12-12 François Dumont
* include/profile/unordered_set: Minor formatting changes.
(unordered_set<>::_M_profile_destruct,
unordered_multiset<>::_M_profile_destruct): Fix implementation
to not
rely on normal implementation details any
On 10 December 2011 15:35, Benjamin Kosnik wrote:
>
> This patch just removes/restructures some of the doxygen markup to
> avoid warnings when generating the documentation. Most of the libstdc++
> headers are pretty doxygen clean now.
By the way, I recently made this feature request for Doxygen:
On 12/09/2011 03:26 PM, Dodji Seketeli wrote:
Two overall comments:
1) Yeah, let's hold this for 4.8.
2) TEMPLATE_ID_TYPE in this patch is too closely related to
BOUND_TEMPLATE_TEMPLATE_PARM; I'd like to see all the
BOUND_TEMPLATE_TEMPLATE_PARM_P checks go away so that things know how to
hand
This looks good.
Jason
Hi,
this removes all the occurrences of int64_t in the host code, as well as some
gratuitous occurrences of int32_t (there are real ones in DFP and LTO code).
Tested on i586-suse-linux and x86_64-suse-linux. Any objections?
Are the LTO files present in the gcc directory compiled when LTO is dis
On 12/02/11 19:26, Patrick Marlier wrote:
Hi,
PR51347 shows up a problem due to the TM IPA rework.
tree_function_versioning segfault because the cfg of old_decl (alias) is
NULL.
Indeed, an alias can get called but tm cg data are in the parent of the
alias.
[Richard, this one's for you :-)].
In reviewing Patrick Marlier's last patch I noticed that the tree dump
files for C++ objects were all convoluted when it came to displaying
certain clones.
It turns out C++'s dump_function_name() needs the DECL_LANG_SPECIFIC
bits to properly display the demangled names. However,
ipa_tm_creat
Hi all,
I have committed (Rev. 182190) the attached patch as obvious.
a) In one case, gfortran was using "%lld" with "long" instead of "%ld"
(or instead of long long).
The proper data type would be something like ptrdiff_t, long long or
similar. However, different systems require different p
Without this patch, transaction commits of different transactions get
potentially merged, which breaks how we handle transactional regions.
Patch provided by Richard Henderson.
OK for trunk?
commit 48d236d4c40fdb7111308fe88844068603e235eb
Author: Torvald Riegel
Date: Thu Dec 8 11:57:05 2011 +01
Here is the most recent version, which is now based on both Richard
Henderson's fix for voidify_wrapper_expr (patch2) and Jason Merrill's
conditional MUST_NOT_THROW_EXPR (patch3).
patch2: voidify_wrapper_expr wasn't handling MUST_NOT_THROW_EXPR. The
patch adds a reasonable default handling that a
Previously, GET_MODE_2XWIDER_MODE had no specific definition for
vector modes; yes you'd get _a_ type that contained twice as many
bits, but usually nothing interesting.
This allows me to remove a fixme from the i386 backend that has
been there for years. I want to do this before introducing 3 mo
Targets like mips -mpaired-single which have support for V2SF vectors,
but no support for V2SI vectors, will expand the V2SI constant parameter
to VEC_PERM_EXPR in DImode. Which is absolutely not what we expect in
expand_vec_perm, and is somewhat less than useful.
Tested on x86_64-linux (where it
On 12/10/2011 12:28 PM, Torvald Riegel wrote:
> patch2: voidify_wrapper_expr wasn't handling MUST_NOT_THROW_EXPR. The
> patch adds a reasonable default handling that applies to this case (and
> this way, we don't need to teach language-independent code about
> MUST_NOT_THROW_EXPR).
>
> patch3: As
On 12/10/2011 12:20 PM, Torvald Riegel wrote:
> Prevent sharing of commit calls among transactions.
>
> gcc/
> * tree-ssa-tail-merge.c (gimple_equal_p): Don't treat transaction
> commits as equal.
>
> gcc/testsuite/
> * c-c++-common/tm/20111206.c: New tes
On 12/10/2011 02:16 PM, Aldy Hernandez wrote:
Using the parent node for aliases (as in your patch) makes sense, but I
don't see tree_function_versioning() segfaulting as you claim. What I
see is estimate_function_body_sizes() failing because it is trying to
analyze an edge that doesn't exist her
-12-10 Eric Botcazou
* gcc.c-torture/execute/20111210-1.c! New test.
--
Eric Botcazou
/* PR tree-optimization/50569 */
/* Reported by Paul Koning */
/* Reduced testcase by Mikael Pettersson */
struct event {
struct {
unsigned int sec;
} sent __attribute__((packed));
};
void __
On 12/09/2011 08:44 AM, Richard Sandiford wrote:
> PR middle-end/50873
> * optabs.c (maybe_legitimize_operand_same_code): Use copy_to_mode_reg
> instead of force_reg. Do nothing if the address is already a
> non-virtual pseudo register.
Ok.
r~
On 12/10/2011 02:16 PM, Aldy Hernandez wrote:
Using the parent node for aliases (as in your patch) makes sense, but I
don't see tree_function_versioning() segfaulting as you claim. What I
see is estimate_function_body_sizes()
Humm you are seeing this I guess:
http://gcc.gnu.org/ml/gcc-patches/
Make sure its set for all CMP, CMN, TST instructions, which do work
inside IT blocks. Split the TEQ pattern so that it can be predicated
in ARM mode for better compare-and-swap generation.
---
gcc/config/arm/arm.md | 42 +++---
1 files changed, 31 insertions(
Changes v2->v3:
* Drop SWP patch
* Merge the ccmode patch with the base atomics patch
* Fix the tst predication error that Ramana spotted.
Ok?
r~
Richard Henderson (3):
arm: Convert to atomic optabs.
arm: Set predicable on more instructions.
arm-linux: Add libitm support.
gcc/co
At the same time, perform post-reload splitting.
---
gcc/config/arm/arm-protos.h |7 +-
gcc/config/arm/arm.c | 817 +++--
gcc/config/arm/arm.h | 18 -
gcc/config/arm/arm.md | 26 +-
gcc/config/arm/constraints.md |5 +
gcc
---
libitm/Makefile.am |3 +
libitm/Makefile.in | 20 +++--
libitm/config/arm/hwcap.cc | 67 +
libitm/config/arm/hwcap.h| 41 ++
libitm/config/arm/sjlj.S | 135 +
On Sun, Dec 4, 2011 at 3:23 PM, Iain Sandoe
wrote:
gcc:
* config/rs6000/rs6000.c (first_altivec_reg_to_save): Amend comment.
(compute_vrsave_mask): Likewise.
(rs6000_emit_prologue): Move update of VRSave mask to save_world()
when that is in use.
libgcc:
* con
On 9 December 2011 19:08, Jakub Jelinek wrote:
> Hi!
>
> As mentioned in the PR, we ICE on the following testcase, because
> there are DRs in a GIMPLE_CALL stmt and when there is just one, we
> compute vectype for the call as if it were a load or store, but during
> computation of vectorization fa
30 matches
Mail list logo
