On Sat, Jul 09, 2011 at 07:22:18PM -0700, H.J. Lu wrote:
> On Thu, Jul 07, 2011 at 10:29:53AM -0700, H.J. Lu wrote:
> > Hi,
> >
> > On Linux/x86-64, when we pass
> >
> > RUNTESTFLAGS="--target_board='unix{-mx32}'"
> >
> > to GCC tests, we can't check lp64/ilp32 for availability of 64bit x86
> >
On Thu, Jul 07, 2011 at 10:29:53AM -0700, H.J. Lu wrote:
> Hi,
>
> On Linux/x86-64, when we pass
>
> RUNTESTFLAGS="--target_board='unix{-mx32}'"
>
> to GCC tests, we can't check lp64/ilp32 for availability of 64bit x86
> instructions. This patch adds ia32 and x32 effetive targets. OK for
> tru
On Sat, Jul 9, 2011 at 6:43 PM, Richard Henderson wrote:
> On 07/09/2011 06:42 PM, H.J. Lu wrote:
>> On Sat, Jul 9, 2011 at 6:34 PM, Richard Henderson wrote:
>>> I developed this patch while working on the dwarf2 pass series.
>>> This was before I bypassed the entire problem by removing the
>>> !
On 07/09/2011 06:42 PM, H.J. Lu wrote:
> On Sat, Jul 9, 2011 at 6:34 PM, Richard Henderson wrote:
>> I developed this patch while working on the dwarf2 pass series.
>> This was before I bypassed the entire problem by removing the
>> !deep branch prediction paths.
>>
>> Ideally, we'd do this generi
On Sat, Jul 9, 2011 at 6:34 PM, Richard Henderson wrote:
> I developed this patch while working on the dwarf2 pass series.
> This was before I bypassed the entire problem by removing the
> !deep branch prediction paths.
>
> Ideally, we'd do this generically from gimple. Less ideally,
> but still
I developed this patch while working on the dwarf2 pass series.
This was before I bypassed the entire problem by removing the
!deep branch prediction paths.
Ideally, we'd do this generically from gimple. Less ideally,
but still better, is to always emit rtl, and support that in
the middle end wit
On Sat, Jul 9, 2011 at 4:12 PM, Richard Henderson wrote:
> On 07/09/2011 04:05 PM, H.J. Lu wrote:
>>> Is this change actually necessary? I would think that the
>>> predicate has already been adjusted...
>>>
>>
>> Since we always use short version for x32, there is no need to call.
>> x86_64_zext_
The $(LIBGCC2_CFLAGS) removed in the patch is prehistoric
(before first commit in the FSF repo) but was once needed or at
least seemed like a good idea; I can't recall which. These
days, its major effect is to add "-g", which with the recent
dwarf2 changes is lethal, if you're unlucky. With the d
On 07/09/2011 04:05 PM, H.J. Lu wrote:
>> Is this change actually necessary? I would think that the
>> predicate has already been adjusted...
>>
>
> Since we always use short version for x32, there is no need to call.
> x86_64_zext_immediate_operand.
Yes, but using the shorter condition, i.e. al
On Sat, Jul 9, 2011 at 3:46 PM, Richard Henderson wrote:
> On 07/09/2011 02:41 PM, H.J. Lu wrote:
>> Hi,
>>
>> X32 uses movl instead of movabs for trampoline. OK for trunk?
>>
>> Thanks.
>>
>> H.J.
>> ---
>> 2011-07-09 H.J. Lu
>>
>> * config/i386/i386.c (ix86_trampoline_init): Use movl i
The attached patch fixes the boehm-gc testsuite on hppa2.0w-hp-hpux11.11.
Without it, libtool always generates an informational warning when linking
causing the entire boehm-gc testsuite to fail.
Ok? Ralf would you please install in libtool tree if ok.
2011-07-09 John David Anglin
PR
On Sat, Jul 9, 2011 at 3:46 PM, Richard Henderson wrote:
> On 07/09/2011 02:41 PM, H.J. Lu wrote:
>> Hi,
>>
>> X32 uses movl instead of movabs for trampoline. OK for trunk?
>>
>> Thanks.
>>
>> H.J.
>> ---
>> 2011-07-09 H.J. Lu
>>
>> * config/i386/i386.c (ix86_trampoline_init): Use movl i
On Sat, Jul 9, 2011 at 3:43 PM, Richard Henderson wrote:
> On 07/09/2011 02:36 PM, H.J. Lu wrote:
>>
>> Hi,
>>
>> Thunk is in ptr_mode, not Pmode. OK for trunk?
>>
>> Thanks.
>>
>> H.J.
>> ---
>> 2011-07-09 H.J. Lu
>>
>> * config/i386/i386.c (x86_output_mi_thunk): Use ptr_mode instead
>>
The recent dwarf2 changes caused notes to be emitted between the
CRIS (casesi expanded to sort-of) tablejump and the jump-table.
I know RTH solved that particular problem another way (thanks!),
but the documented adjacency promise is about of instructions,
not notes, so it seems appropriate to also
On 07/09/2011 02:41 PM, H.J. Lu wrote:
> Hi,
>
> X32 uses movl instead of movabs for trampoline. OK for trunk?
>
> Thanks.
>
> H.J.
> ---
> 2011-07-09 H.J. Lu
>
> * config/i386/i386.c (ix86_trampoline_init): Use movl instead
> of movabs for x32.
>
> diff --git a/gcc/config/i386
On 07/09/2011 02:36 PM, H.J. Lu wrote:
>
> Hi,
>
> Thunk is in ptr_mode, not Pmode. OK for trunk?
>
> Thanks.
>
> H.J.
> ---
> 2011-07-09 H.J. Lu
>
> * config/i386/i386.c (x86_output_mi_thunk): Use ptr_mode instead
> of Pmode for vtable adjustment.
>
> diff --git a/gcc/config/
Hi,
TARGET_MEM_REF only works on ptr_mode. That means base and index parts
of x86 address operand in x32 mode may be in ptr_mode. This patch
supports 32bit base and index parts in x32 mode. OK for trunk?
Thanks.
H.J.
---
2011-07-09 H.J. Lu
* config/i386/i386.c (ix86_simplify_base
Hi,
X32 uses movl instead of movabs for trampoline. OK for trunk?
Thanks.
H.J.
---
2011-07-09 H.J. Lu
* config/i386/i386.c (ix86_trampoline_init): Use movl instead
of movabs for x32.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 04cb07d..c852719 100644
Hi,
Thunk is in ptr_mode, not Pmode. OK for trunk?
Thanks.
H.J.
---
2011-07-09 H.J. Lu
* config/i386/i386.c (x86_output_mi_thunk): Use ptr_mode instead
of Pmode for vtable adjustment.
diff --git a/gcc/config/i386/i386.c b/gcc/config/i386/i386.c
index 04cb07d..c852719 10064
On Sat, Jul 9, 2011 at 2:18 PM, Paolo Bonzini wrote:
> On 07/05/2011 04:27 PM, H.J. Lu wrote:
>>>
>>> diff --git a/gcc/explow.c b/gcc/explow.c
>>> index 7387dad..b343bf8 100644
>>> --- a/gcc/explow.c
>>> +++ b/gcc/explow.c
>>> @@ -383,18 +383,13 @@ convert_memory_address_addr_space (enum
>>>
Hi,
X32 psABI requires promoting pointers to Pmode when passing/returning
in registers. OK for trunk?
Thanks.
H.J.
--
2011-07-09 H.J. Lu
* config/i386/i386.c (ix86_promote_function_mode): New.
(TARGET_PROMOTE_FUNCTION_MODE): Likewise.
diff --git a/gcc/config/i386/i386.c b/g
When looking at the build log, I saw a -Wold-style-definition warning in
libgfortran.
I have committed fix as obvious (Rev. 176109).
Tobias
Index: libgfortran/runtime/error.c
===
--- libgfortran/runtime/error.c (Revision 176108)
++
Hi,
This patch turns on 64bit and check models for x32. OK for trunk?
Thanks.
H.J.
---
2011-07-09 H.J. Lu
* config/i386/i386.c (ix86_option_override_internal): Turn on
OPTION_MASK_ISA_64BIT for TARGET_X32. Only allow small and
small PIC models for TARGET_X32.
diff
On 07/05/2011 04:27 PM, H.J. Lu wrote:
diff --git a/gcc/explow.c b/gcc/explow.c
index 7387dad..b343bf8 100644
--- a/gcc/explow.c
+++ b/gcc/explow.c
@@ -383,18 +383,13 @@ convert_memory_address_addr_space (enum machine_mode
to_mode ATTRIBUTE_UNUSED,
case PLUS:
case MULT:
-
Hi Jeff,
Can you take a look at this?
Thanks.
H.J.
On Tue, Jul 5, 2011 at 7:27 AM, H.J. Lu wrote:
> Ping.
>
> On Sat, Jun 11, 2011 at 8:58 AM, H.J. Lu wrote:
>> Hi,
>>
>> convert_memory_address_addr_space has a special PLUS/MULT case for
>> POINTERS_EXTEND_UNSIGNED < 0. It turns out that it i
Hi,
I checked in this patch to add comments to *movsi_internal.
H.J.
---
commit 3c92ec7ad8591d3b2e3c3bc3c6b25e2bf38335cf
Author: H.J. Lu
Date: Sat Jul 9 08:03:46 2011 -0700
Add comments to *movsi_internal.
diff --git a/gcc/ChangeLog.x32 b/gcc/ChangeLog.x32
index fdff5af..57ecbe4 100644
-
Hi,
I checked in this patch to add x32 LEA zero-extend split.
H.J.
---
commit a57ad5f219b8aa7cc67c18fb3a217517c2a461f4
Author: H.J. Lu
Date: Sat Jul 9 08:00:56 2011 -0700
Support x32 LEA zero-extend split.
diff --git a/gcc/ChangeLog.x32 b/gcc/ChangeLog.x32
index a610f23..fdff5af 100644
-
---
gcc/defaults.h | 22 ++
gcc/dwarf2cfi.c | 567
gcc/dwarf2out.c | 638 +--
gcc/dwarf2out.h |7 +
4 files changed, 604 insertions(+), 630 deletions(-)
diff --git a/gcc/defaults.h b/gcc/
---
gcc/dwarf2cfi.c | 82 ---
gcc/dwarf2out.c |8 +++---
2 files changed, 40 insertions(+), 50 deletions(-)
diff --git a/gcc/dwarf2cfi.c b/gcc/dwarf2cfi.c
index d96aa70..ca9b503 100644
--- a/gcc/dwarf2cfi.c
+++ b/gcc/dwarf2cfi.c
@@ -146,6
When we record a save of the return column in the CIE, remember
that while processing the FDE. This requires propagating the
handling of PC_RTX as a representative of the return column to
more locations.
MIPS had been handling this case by hand, and is no longer required.
---
gcc/config/mips/mip
Move some definitions to defaults.h; some simply aren't needed
anymore in the respective dwarf2{cfi,out}.c files.
---
gcc/defaults.h | 33 ++
gcc/dwarf2cfi.c | 80 ---
gcc/dwarf2out.c | 37 -
3 f
Note the extreme silliness in dwarf2out.c with 3 (three) copies.
---
gcc/defaults.h |5 +
gcc/dwarf2asm.c |6 --
gcc/dwarf2out.c | 15 ---
gcc/final.c |5 -
gcc/toplev.c|4
gcc/varasm.c|4
gcc/vmsdbgout.c |5 -
7 files ch
There's no point calling lookup_cfa to re-create the CFA
we initialized just a few lines above.
---
gcc/dwarf2cfi.c | 45 +
1 files changed, 13 insertions(+), 32 deletions(-)
diff --git a/gcc/dwarf2cfi.c b/gcc/dwarf2cfi.c
index ca9b503..b4b035d 100644
Tidies up the split between the two files.
Tested on x86_64-linux.
r~
Richard Henderson (6):
Move ASM_COMMENT_START to defaults.h.
dwarf2cfi: Unify add_fde_cfi and add_cie_cfi.
dwarf2cfi: Simplify re-initialization for each function.
dwarf2cfi: Handle return column save from CIE.
dwar
We were dropping the CFI notes inside the sequence
instead of in front of it.
Tested with sh-sim.
r~
PR debug/49686
* dwarf2cfi.c (dwarf2out_frame_debug): Don't set cfi_insn here...
(create_cfi_notes): ... do it here instead.
diff --git a/gcc/dwarf2cfi.c b/gcc/dwarf2cfi
The regression seems to be a side effect of the -fwhole-file effort.
The BT_UNKNOWN of the second argument of SIGNAL(NUMBER,HANDLER) triggers
an ICE in trans-types.c, when generating the decl for the external
function "signal" (or rather for __gfortran_signal_func{,_int}).
As both handler ==
Hello!
gfc_get_corank returns integer value, not bool. This problem was
triggered by --enable-build-with-cxx configured build.
2011-07-09 Uros Bizjak
PR fortran/48926
* expr.c (gfc_get_corank): Change return value to int.
* gfortran.h (gfc_get_corank): Update function
My patch for 49528 broke this testcase because build_vec_init_elt was
returning what looks like a temporary of non-literal type. There isn't
actually a temporary, so let's not return something that looks like one.
Tested x86_64-pc-linux-gnu, applying to trunk and 4.6.
commit 3f54d9656d3e65b396
On Jul 8, 2011, at 7:57 AM, Georg-Johann Lay wrote:
> These tests are too big for AVR: .text (128 KiB) overflows and ld
> complains.
>
> Ok to commit?
Ok. If people feel they have a nice design for `too big', let us know... I
think it would have to be ld message scan based, to allow things tha
On Jul 8, 2011, at 7:25 AM, Georg-Johann Lay wrote:
> There are tests in suite that set -fschedule-insns[2] and FAIL because of
>
> warning: instruction scheduling not supported on this target machine
> [enabled by default]
>
> As there is no individual switch do disable the warning, the patch
On 07/09/2011 06:30 AM, Paul Koning wrote:
>
> On Jul 8, 2011, at 6:13 PM, Richard Henderson wrote:
>
>> This appears to do the right thing. I didn't bother with
>> markers for unwind info, since pdp11 is limited to a.out
>> and thus will never use dwarf2.
>>
>> There are improvements that could
On 07/07/2011 10:35 PM, Ralf Wildenhues wrote:
> Hi Matthias,
>
> On Thu, Jul 07, 2011 at 10:26:59PM +0200, Jakub Jelinek wrote:
>> On Thu, Jul 07, 2011 at 10:22:37PM +0200, Matthias Klose wrote:
>>> +AC_PROG_LIBTOOL
>
> This tests the wrong compiler and toolchain. The compiler you want
> to tes
OK.
Jason
On 23/06/11 15:37, Andrew Stubbs wrote:
This patch should have no effect on the compiler output. It merely
replaces one way to represent widening operations with another, and
refactors the other parts of the compiler to match. The rest of the
patch set uses this new framework to implement the opt
The attached patch makes flat nesting the default, while still
supporting closed nesting on demand (for user-controlled aborts via
__transaction_cancel).
Previously, a new transaction object was created for each nested
transaction. The patch changes this to using one transaction object
during the w
This adds the new header. Currently only
std::vector has the necessary support for C++0x allocators to be
usable with std::scoped_allocator_adaptor.
* include/Makefile.am: Add new header.
* include/Makefile.in: Regenerate.
* include/std/scoped_allocator: New.
* doc
On Sat, Jul 9, 2011 at 5:02 AM, Tobias Burnus wrote:
Tobias Burnus wrote:
This patch adds a run-time error function to mpi.c, which gives a
proper error message including the image number. Additionally, it
allows to clean up the error handling, avoiding the duplicated
declaration of strings.
On Jul 8, 2011, at 6:13 PM, Richard Henderson wrote:
> This appears to do the right thing. I didn't bother with
> markers for unwind info, since pdp11 is limited to a.out
> and thus will never use dwarf2.
>
> There are improvements that could be made. I added some
> comments to that effect but
This patch adds full C++0x allocator support to std::vector, building
on the alloc_traits work I've been committing recently.
This is the easiest of the allocator-aware containers, I'm planning to
do the same for std::deque soon.
* include/bits/stl_vector.h: Use new allocator model in C++
* include/ext/alloc_traits.h (__allocator_always_compares_equal): New
trait, provide partial specializations for known allocators.
(__alloc_traits::construct, __alloc_traits::destroy): Overload for
non-standard pointer types.
(__alloc_traits::_S_always_equal)
Tobias Burnus wrote:
This patch adds a run-time error function to mpi.c, which gives a
proper error message including the image number. Additionally, it
allows to clean up the error handling, avoiding the duplicated
declaration of strings.
I have not touched the SYNC functions
Well, I did n
This patch adds a run-time error function to mpi.c, which gives a proper
error message including the image number. Additionally, it allows to
clean up the error handling, avoiding the duplicated declaration of strings.
I have not touched the SYNC functions; they can be handled either after
the
Daniel Carrera wrote:
I'd like to submit the attached patch. This patch was just discussed
in the gfortran list. It fixes a couple of TODO items in the MPI
library. It is a simple patch.
OK with the following nits fixed.
* As you have now an SVN account, can you add yourself to the
./MAINTAI
* include/ext/cast.h: Fix typo in include guard.
* include/ext/pointer.h (_Unqualified_type): Remove redundant
partial specializations for volatile types. Fix typos in comments.
(pointer_traits<_Pointer_adaptor>::pointer_to): Define.
Tested x86_64-linux, committed t
This removes a couple of unused macros and moves another around.
Tested on SPARC/Solaris, applied on the mainline.
2011-07-09 Eric Botcazou
* config/sparc/sparc.h (STACK_SAVEAREA_MODE): Move around.
(FP_REG_P): Delete.
(IN_OR_GLOBAL_P): Likewise.
--
Eric Botcazou
I
std::bind forwards the bound arguments as lvalues, so we shouldn't be
using it to implement the
INVOKE(DECAY_COPY(f), DECAY_COPY(args)...)
pseudo-expressions in the threads clauses. Using std::bind we can't
invoke functions which require rvalue arguments.
This patch adds a bind-like helper that ca
56 matches
Mail list logo
