Il 26/07/2012 04:25, Sandra Loosemore ha scritto:
> On 07/25/2012 01:27 AM, Paolo Bonzini wrote:
>>
>> What I'm worried about is the extra cost of malloc-ing and free-ing
>> the pointer set. Perhaps you can skip the pointer set creation in
>> the common case where find_comparison_args does not it
This patch series removes support for the old POWER CPUs (the RIOS,
RSC, and RIOS2).
It does not do any more than that: it leaves common mode alone,
it does not remove the old assembler mnemonics and the {xx|yy}
insn template syntax, etc.
Bootstrapped and regtested on powerpc64-linux (c,c++,fortr
gcc/
* config/rs6000/constraints.md: Delete "q" constraint.
* config/rs6000/dfp.md (movsd_hardfloat, movsd_softfloat):
Delete the "q" alternative.
* config/rs6000/predicates.md (gpc_reg_operand): Replace
MQ_REGNO with the literal 64.
* config/rs6000/r
Move those parts of rios.md that apply to 601 to a new file
601.md, renaming everything from rios1* to ppc601*.
2012-07-26 Segher Boessenkool
gcc/
* config/rs6000/601.md: New file.
* config/rs6000/aix43.h (ASM_CPU_SPEC): Delete support for
RIOS CPUs.
* config/rs
On 25/07/2012 23:23, Tobias Burnus wrote:
> Tobias Burnus wrote:
>> The following issue was found by Alessandro. (It got triggered by a
>> larger test case, which is required for a larger patch by Alessandro,
>> which is not yet finished.)
>>
>> Accessing the "lower[-1]" is probably not the best id
2012/6/18 Richard Guenther :
> On Fri, 15 Jun 2012, Richard Guenther wrote:
>
>>
>> This tries to completely implement the intersect primitive for
>> VRP (what extract_range_from_assert does at its end when merging
>> new and old knowledge).
>>
>> Bootstrap and regtest pending on x86_64-unknown-lin
On 07/26/2012 10:00 AM, Mikael Morin wrote:
I assume this sits on top of the previous patch? Then OK for both.
Yes. It was supposed to be a complete patch, but as I screwed up (git
diff vs. git diff --cached, combined with git add), it was only an
interdiff.
Thanks for the review. I committ
On Wed, 25 Jul 2012, Andrew Pinski wrote:
> Hi,
> The Linux kernel already supports non-executable stack since around
> February 2010. This patch has GCC emit the notes that are associated
> with non-executable stack.
What does the kernel do when the note isn't present?
> OK? Bootstrapped and
On 25 July 2012 21:29, François Dumont wrote:
> (_Hashtable<>::rehash): Likewise. Set _M_prev_resize to 0 to avoid
> the hashtable to be shrinking on next insertion.
Not "to be shrinking" just "shrinking", but nevermind.
Hello,
Add some explanations because I see a lot of places where the choice
of data structure is less than optimal.
Also some functional changes: sbitmap with popcounts are only used in
ebitmap, and they penalize the "normal" case without popcount a lot.
Bootstrapped&tested on powepc64-unknown-l
On Thu, Jul 26, 2012 at 11:16 AM, Steven Bosscher wrote:
> Hello,
>
> Add some explanations because I see a lot of places where the choice
> of data structure is less than optimal.
>
> Also some functional changes: sbitmap with popcounts are only used in
> ebitmap, and they penalize the "normal" c
Hi,
This adjusts the test for vld1Q_dups/u64 which started failing after
the recent fixes in this area.
Committed to trunk after testing. I've got a few more patches to the
neon intrinsics tests but this leaves these tests in a sane state.
Ramana
2012-07-26 Ramana Radhakrishnan
*
On Thu, Jul 26, 2012 at 11:23 AM, Richard Guenther
wrote:
> Ok! Thanks for adding this exhaustive documentation.
There's more to come! I want to add some explanations to ebitmap,
pointer-set, fibheap, and splay-tree as sets, and add a chapter in the
gccint manual too.
Now if only you'd document
Hello,
+ the set. If an element I is in the set, then sparse[I] is the
+ index of I in the dense vector, and dense[I] == I. The dense
^
This should read "dense[sparse[I]] == I"
Thanks
Alexander
This fixes PR54098 - if the original range was UNDEFINED simply
use the update and iterate once more.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2012-07-26 Richard Guenther
PR tree-optimization/54098
* tree-vrp.c (vrp_visit_phi_node): Iterate once
Hi,
This patch removes the duplicate check on BRANCH_COST in fold_truth_andor.
The BRANCH_COST condition removed is a duplicate of the default definition
of LOGICAL_OP_NON_SHORT_CIRCUIT.
All current targets (mips and rs6000) that provide non-default definitions
of LOGICAL_OP_SHORT_CIRCUIT set it to
Hi,
This patch defines LOGICAL_OP_NON_SHORT_CIRCUIT in arm back-end by calling a
new hook function(logical_op_non_short_circuit") in tune_params structure.
For most cases the value of the macro is same as the default version in
fold-const.c, while it is "FALSE" to prefer short circuit when optimizi
On Thu, Jul 26, 2012 at 3:20 AM, Bin Cheng wrote:
> Hi,
> This patch removes the duplicate check on BRANCH_COST in fold_truth_andor.
> The BRANCH_COST condition removed is a duplicate of the default definition
> of LOGICAL_OP_NON_SHORT_CIRCUIT.
> All current targets (mips and rs6000) that provide
> No regression introduced, is it OK?
By testing in what configuration ?
Ramana
> -Original Message-
> From: Ramana Radhakrishnan [mailto:ramana.radhakrish...@linaro.org]
> Sent: Thursday, July 26, 2012 6:45 PM
> To: Bin Cheng
> Cc: gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH arm]Define LOGICAL_OP_NON_SHORT_CIRCUIT in ARM back
end
>
> > No regression introduced, i
On 07/25/2012 08:04 PM, Tobias Burnus wrote:
Committed as obvious (Rev. 189859).
I just saw that I missed the testsuite changes. I have now committed
them as Rev. 189887.
Tobias
Index: testsuite/gfortran.dg/contiguous_1.f90
===
On Thu, Jul 26, 2012 at 12:14 PM, Alexander Monakov wrote:
> Hello,
>
> + the set. If an element I is in the set, then sparse[I] is the
> + index of I in the dense vector, and dense[I] == I. The dense
> ^
> This should read "dense[sparse[I]
Attached patch applied on 4.7 branch.
Tested under Linux x86_64.
I will fix this small english issue in trunk ChangeLog.
François
On 07/26/2012 11:11 AM, Jonathan Wakely wrote:
On 25 July 2012 21:29, François Dumont wrote:
(_Hashtable<>::rehash): Likewise. Set _M_prev_resize to 0 t
_tune, arm_cortex_a9_tune, arm_fa726te_tune): Set the
> field
> logical_op_non_short_circuit to
> arm_default_logical_op_non_short_circuit.
> (arm_v6m_tune): New tune_params struct.
> * config/arm/arm.h (LOGICAL_OP_NON_SHORT_CIRCUIT): Use the hook
> logical_o
On 26/07/12 11:27, Andrew Pinski wrote:
> On Thu, Jul 26, 2012 at 3:20 AM, Bin Cheng wrote:
>> Hi,
>> This patch removes the duplicate check on BRANCH_COST in fold_truth_andor.
>> The BRANCH_COST condition removed is a duplicate of the default definition
>> of LOGICAL_OP_NON_SHORT_CIRCUIT.
>> All
TS29113 allows also non interoperable procedures with
c_funloc/c_f_procpointer; hence, this patch allows them with -std=f2008ts:
"The function C F PROCPOINTER from the intrinsic module ISO C BINDING
has the restriction in ISO/IEC 1539-1:2010 that CPTR and FPTR shall not
be the C address and in
On Wed, 25 Jul 2012, Andrew Pinski wrote:
> * config/mips/linux-common.h (TARGET_ASM_FILE_END): Define.
Do you also need to change libgcc/config/mips/*.S to include the
appropriate note?
--
Joseph S. Myers
jos...@codesourcery.com
On 21/07/2012 13:08, Tobias Burnus wrote:
> Only failing are:
> lbound(x) / ubound(x) / shape(x)
>
Here is a draft for those.
Lightly tested with print *, ...
Mikael
Index: trans-array.c
===
--- trans-array.c (révision 189883)
++
Some of the hosts were we run this script are still using Python 2.4.
This patch replaces the use of 'with ... as ...' to avoid syntax errors.
2012-07-26 Diego Novillo
* testsuite-management/validate_failures.py: Do not use
'with ... as ...' constructs.
diff --git a/contrib
On 26/07/2012 16:53, Mikael Morin wrote:
> On 21/07/2012 13:08, Tobias Burnus wrote:
>> Only failing are:
>> lbound(x) / ubound(x) / shape(x)
>>
> Here is a draft for those.
> Lightly tested with print *, ...
>
Better with the tests.
$ ./test1
1 1
3 8
On 07/26/2012 05:12 PM, Mikael Morin wrote:
On 26/07/2012 16:53, Mikael Morin wrote:
Here is a draft for those. Lightly tested with print *, ...
Looks rather nice. The output for test1 is also good:
integer :: a(1:3,-2:5)
gives
lbound(arg) == [1, 1]
ubound(arg) == [3, 8]
shape(arg) =
Hello
I am friendly pinging this patch that felt below my radar.
Dodji Seketeli a écrit:
> Hello,
>
> Consider this short test snippet:
>
> -8---
> #define STRINGIFY(x) #x
> #define TEST(x) \
> _Pragma(STRINGIFY(GCC diagnostic ignored "-Wunu
Ping.
Teresa
On Wed, Jul 18, 2012 at 8:48 AM, Teresa Johnson wrote:
> Ping (retrying ping in plain text mode so that it goes through properly).
>
> Thanks,
> Teresa
>
> On Wed, Jul 11, 2012 at 10:42 AM, Teresa Johnson wrote:
>> Ports some patches related to improving FDO program summary informa
Hello,
here is a new version of the ifcombine patch, which handles Andrew's
examples (though only with -O2 for one of them, same_phi_args_p returns
false otherwise).
Note that the new patch never calls maybe_fold_or_comparisons, only
maybe_fold_and_comparisons. It would be possible to call
Do you handle how, in certain syntactic locations, a C++11 attribute binds
differently to a GNU attribute? (I haven't studied the patch, so feel
free to point me to testcases it adds that verify such differences, if
applicable.)
--
Joseph S. Myers
jos...@codesourcery.com
Hello world,
the attached, rather obvious patch emits warnings for several
cases where there is something wrong with include directories.
No test case because I couldn't figure out how to test for a
warning with no line number.
OK for trunk?
Thomas
2012-07-26 Thomas König
P
On 07/26/2012 10:16 AM, Thomas Koenig wrote:
> No test case because I couldn't figure out how to test for a
> warning with no line number.
Try using line number 0.
Janis
http://codereview.appspot.com/6351086/diff/1/gcc/gcov-io.h
File gcc/gcov-io.h (right):
http://codereview.appspot.com/6351086/diff/1/gcc/gcov-io.h#newcode396
gcc/gcov-io.h:396: gcov_unsigned_t num_hot_counters;/* number of
counters to reach a given
Should it be made into an array accessed with pr
On Thu, Jul 26, 2012 at 12:38:47AM -0700, Segher Boessenkool wrote:
> This patch series removes support for the old POWER CPUs (the RIOS,
> RSC, and RIOS2).
>
> It does not do any more than that: it leaves common mode alone,
> it does not remove the old assembler mnemonics and the {xx|yy}
> insn t
Note, you will need to update the copyright to include 2012 for the
files that
have not been modified in this year.
I think you looked at the old version of the patch I mailed
to you privately; this is fixed already (there are some other
small differences as well).
In running parallel tests f
On Thu, Jul 26, 2012 at 09:18:09PM +0200, Segher Boessenkool wrote:
> >Note, you will need to update the copyright to include 2012 for
> >the files that
> >have not been modified in this year.
>
> I think you looked at the old version of the patch I mailed
> to you privately; this is fixed already
On 07/26/2012 01:28 AM, Paolo Bonzini wrote:
> Il 26/07/2012 04:25, Sandra Loosemore ha scritto:
>> On 07/25/2012 01:27 AM, Paolo Bonzini wrote:
>>>
>>> What I'm worried about is the extra cost of malloc-ing and free-ing
>>> the pointer set. Perhaps you can skip the pointer set creation in
>>> the
On 07/26/2012 01:22 PM, Sandra Loosemore wrote:
> 2012-07-26 Andrew Jenner
> Sandra Loosemore
>
> gcc/
> * cse.c (find_comparison_args): Check for cycles of any length.
>
> gcc/testsuite/
> * gcc.c-torture/compile/pr50380.c: Add code to cause cycle of length
Resending in plain text mode...sigh.
Teresa
On Thu, Jul 26, 2012 at 1:32 PM, Teresa Johnson wrote:
>
>
>
>
> On Thu, Jul 26, 2012 at 11:26 AM, wrote:
>>
>>
>> http://codereview.appspot.com/6351086/diff/1/gcc/gcov-io.h
>> File gcc/gcov-io.h (right):
>>
>> http://codereview.appspot.com/6351086/d
On 07/26/2012 08:19 AM, Dodji Seketeli wrote:
> + attributes_table->scoped = XRESIZEVEC (struct scoped_attributes,
> + attributes_table->scoped,
> + attributes_table->len + 1);
A good clue that you want VEC's in
Hello,
This adjusts the cost calculations for shifts on SH.
I tried out Richard's advice
( http://gcc.gnu.org/ml/gcc-patches/2012-07/msg01206.html )
and it seems to work OK to just leave out the mentioned CONST_INT_P
case.
Tested on rev 189870 with
make -k check RUNTESTFLAGS="--target_board=sh-sim
On Wed, Jul 25, 2012 at 5:06 PM, Cary Coutant wrote:
> This patch is for the google/gcc-4_7 branch.
>
> Compute a dwo_id and write the DW_AT_GNU_dwo_id attribute to both the
> skeleton compile unit DIE and the DWO compile unit DIE.
>
This is OK for google 4.7.
Sterling
This is a patch I should have gone ahead and committed when I wrote it
against gcc 4.6, but... better late than never I suppose.
The Idea is to have some mechanism akin to __builtin_expect that can apply
to the edges created by asm goto. We tossed around possible syntax additions
to the asm const
On 07/26/2012 02:41 PM, Richard Henderson wrote:
> This is a patch...
... that I should have attached. Bah.
r~
gcc/
* doc/extend.texi (attribute): Document hot/cold for labels.
* predict.c (tree_estimate_probability_bb): Handle hot/cold
attributes on user labels.
This is a libiberty patch for Windows that we've had in our local tree for a
couple of years now. The problem it addresses showed up in Windows-hosted GDB;
we were getting a crash inside _close() in the case where creation of the child
process has failed.
Apparently Windows gets very unhappy
Test gcc.dg/pr45259 fails on targets that don't support fpic because it
skips using -w and therefore gets pedantic warnings that are enabled by
default. This patch provides an alternate dg-option command to supply
options other than -fpic for targets that don't support fpic.
Checked in on trunk,
On 07/25/2012 04:55 PM, Jonathan Wakely wrote:
Yes, I think so, it's a regression from 4.6.
Thanks for dealing with it so quickly.
Thanks indeed.
However, looks like we have another issue, unrelated to reserve, pure
performance not correctness, with the sheer number of rehashes when the
map
I wonder if registering a handler for invalid parameters, at least
around those calls, so that we can enforce the posix-like "return an
error" semantics?
> OK for mainline?
Ok.
> 2012-07-26 Kazu Hirata
> Sandra Loosemore
>
> libiberty/
> * pex-win32.c (pex_win32_exec
Oleg Endo wrote:
> This adjusts the cost calculations for shifts on SH.
> I tried out Richard's advice
> ( http://gcc.gnu.org/ml/gcc-patches/2012-07/msg01206.html )
> and it seems to work OK to just leave out the mentioned CONST_INT_P
> case.
> Tested on rev 189870 with
> make -k check RUNTESTFLAG
Updated patch based on review feedback.
Bootstrapped and tested on x86_64-unknown-linux-gnu. Ok for trunk?
Teresa
Original patch description:
Ports some patches related to improving FDO program summary information
and using it to guide loop unrolling from google branches to mainline.
The patch
Il 26/07/2012 22:22, Sandra Loosemore ha scritto:
> Aha, I honestly couldn't figure out that was what you were trying to
> catch with the version you posted previously.
>
> How about this one? Tested as before.
Yeah, that's cleaner.
Paolo
56 matches
Mail list logo
