Hi Jason,
I've attached a patch that handles parameter packs in the conversion
op. I thought it best to get this reviewed independently before rolling
up into the 'Support lambda templates' patch.
Do you think it's the right idea? It seems to work okay but I've
reworked the way the templat
Hi HJ,
You were right, and the change from epilogue_size_needed to size_needed was the
rootcause of the bug.
Here is the obvious fix for that, regtested and bootstrapped on i386/x86_64.
Is it OK for trunk?
Changelog:
2013-09-09 Michael Zolotukhin
* config/i386/i386.c (ix86_expand_movm
> Hi HJ,
> You were right, and the change from epilogue_size_needed to size_needed was
> the
> rootcause of the bug.
> Here is the obvious fix for that, regtested and bootstrapped on i386/x86_64.
>
> Is it OK for trunk?
>
> Changelog:
> 2013-09-09 Michael Zolotukhin
>
> * config/i386/i
Hi,
Thanks for the review. Here is the fixed self-contained patch to
enable LRA on AAarch32 and AArch64 (for those who want to give a try).
I'm still working on the issues described previously and will send
rtlanal.c patch separately to prepare the move.
Thanks,
Yvan
On 9 September 2013 01:23
Hi,
ipa-profile does propagation across the CFG in attempt to prove that function is
cold. This is counter-productive when profile is read and we can easilly work
this out from count itself. This patch makes it less agressive in this setting
and it also put ipa-profile into busyness to make coun
> OK,
> Would you mind adding a testcase?
Thanks, here is the patch with Eric's test.
OK to commit?
Changelog:
gcc:
2013-09-09 Michael Zolotukhin
* config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
gcc/testsuite:
2013-09-09 Michael Zolotukhin
* gcc.target/i
On Sun, 8 Sep 2013, Paolo Carlini wrote:
> Hi,
>
> this patchlet fixes the column # of the unused parameter warnings emitted by
> do_warn_unused_parameter by explicitly passing DECL_SOURCE_LOCATION (decl)
> instead of wrongly relying on '+', which in this case ends up meaning the
> location of th
On Mon, Sep 9, 2013 at 9:42 AM, Michael V. Zolotukhin
wrote:
>> OK,
>> Would you mind adding a testcase?
> Thanks, here is the patch with Eric's test.
> OK to commit?
>
> Changelog:
> gcc:
> 2013-09-09 Michael Zolotukhin
>
> * config/i386/i386.c (ix86_expand_movmem): Fix epilogue genera
On Mon, Sep 09, 2013 at 11:42:05AM +0400, Michael V. Zolotukhin wrote:
> --- /dev/null
> +++ b/gcc/testsuite/gcc.target/i386/memcpy-2.c
> @@ -0,0 +1,22 @@
> +/* { dg-do run } */
> +/* { dg-require-effective-target ia32 } */
> +/* { dg-options "-O2 -march=pentiumpro -minline-all-stringops -fno-commo
> I don't see anything i386 specific on the testcase, except the flags,
> and don't see why you need -fno-common in there, there are no global vars.
> So, I think it would be better to stick it into gcc.dg/torture/, drop
> dg-require-* line and instead just add
> /* { dg-additional-options "-march=
On Mon, Sep 09, 2013 at 11:52:49AM +0400, Michael V. Zolotukhin wrote:
> > I don't see anything i386 specific on the testcase, except the flags,
> > and don't see why you need -fno-common in there, there are no global vars.
> > So, I think it would be better to stick it into gcc.dg/torture/, drop
>
Hi,
here are the modifications, discussed in another thread, needed in
rtlanal.c by ARM targets (AArch32 and AArch64) to build GCC with LRA.
Is it ok for trunk ?
Thanks,
Yvan
2013-09-09 Yvan Roux
Vladimir Makarov
* rtlanal.c (must_be_index_p, set_address_index): Add AS
On Sun, 8 Sep 2013, pins...@gmail.com wrote:
> On Sep 8, 2013, at 7:01 PM, "Bin.Cheng" wrote:
>
> > On Wed, Sep 4, 2013 at 5:20 PM, Richard Biener wrote:
> >>
> >> The patch below moves IVOPTs out of the GIMPLE loop pipeline more
> >> closer to RTL expansion. That's done for multiple reasons.
On Sun, 8 Sep 2013, Andi Kleen wrote:
2013-06-13 Andi Kleen
* doc/extend.texi: Dont use __atomic_clear in HLE
example. Fix typo.
>> I would like to to backport this to 4.8. Ok too?
> Ping!
That's documentation only, right?
Okay.
Gerald
Hi Richard,
On 09/09/2013 09:46 AM, Richard Biener wrote:
On Sun, 8 Sep 2013, Paolo Carlini wrote:
Hi,
this patchlet fixes the column # of the unused parameter warnings emitted by
do_warn_unused_parameter by explicitly passing DECL_SOURCE_LOCATION (decl)
instead of wrongly relying on '+', whi
On Fri, Sep 6, 2013 at 8:47 PM, Easwaran Raman wrote:
> On Fri, Sep 6, 2013 at 12:05 AM, Richard Biener
> wrote:
>> On Thu, Sep 5, 2013 at 9:23 PM, Easwaran Raman wrote:
>>> On Tue, Aug 27, 2013 at 3:35 AM, Richard Biener
>>> wrote:
On Thu, Aug 1, 2013 at 1:34 AM, Easwaran Raman wrote:
>>
Yvan Roux writes:
> Thanks for the review. Here is the fixed self-contained patch to
> enable LRA on AAarch32 and AArch64 (for those who want to give a try).
> I'm still working on the issues described previously and will send
> rtlanal.c patch separately to prepare the move.
Looks like the rtl
Richard Sandiford writes:
> I think SIGN_EXTRACT from the lsb (i.e. when the third operand is 0 for
> !BITS_BIG_ENDIAN or GET_MODE_PRECISION (mode) - 1 for BITS_BIG_ENDIAN)
Gah, "GET_MODE_PRECISION (mode) - size" for BITS_BIG_ENDIAN.
Richard
On Sat, Sep 7, 2013 at 12:26 AM, Diego Novillo wrote:
> This patch introduces gimple-core.h, which contains just the data
> structures needed to define gimple. I left everything else in
> gimple.h
>
> The addition of alias.h to tree-ssa-alias.h is so that we can
> include tree-ssa-alias.h in isola
On Sat, Sep 7, 2013 at 10:15 AM, Uros Bizjak wrote:
> Hello!
>
> It looks that it is too hard for the compiler to track deref_align
> initialization through dependent deref_base boolean. The patch bellow
> fixes "may be used uninitialized" warning that breaks
> profiledbootstrap.
>
> 2013-09-07 U
On Sat, Sep 7, 2013 at 11:00 PM, Marc Glisse wrote:
> On Sat, 7 Sep 2013, Mike Stump wrote:
>
>> On Sep 7, 2013, at 12:27 PM, Marc Glisse wrote:
>>>
>>> Now flag_check_new should probably disable this optimization…
>>
>>
>> Yes, this why my point.
>
>
> Ok, here it is (again, no proper testing un
Revised patch with testcase. This one also fixes a small problem with
reg_or_add_cint_operand in that any 32-bit value is valid for SImode.
Compare with reg_or_sub_cint_operand.
Bootstrapped and regression tested powerpc64-linux. OK to apply?
gcc/
* config/rs6000/predicates.md (add_cint
> I think it is worthwhile, various targets have many different ways to expand
> memcpy, admittedly i?86/x86_64 probably the biggest number of these, and
> while right now you've encountered it on ia32 with certain options doesn't
> mean that in a few years it couldn't hit some unrelated target, ar
On Mon, Sep 09, 2013 at 01:09:12PM +0400, Michael V. Zolotukhin wrote:
> I've never written torture tests before, could you please check if I did it
> right?
Please don't introduce new *.x files, for tests where you need something
like that just stick it into gcc.dg/torture/ instead and use normal
On Mon, 9 Sep 2013, Richard Biener wrote:
On Sat, Sep 7, 2013 at 11:00 PM, Marc Glisse wrote:
On Sat, 7 Sep 2013, Mike Stump wrote:
On Sep 7, 2013, at 12:27 PM, Marc Glisse wrote:
Now flag_check_new should probably disable this optimization…
Yes, this why my point.
Ok, here it is (a
> Please don't introduce new *.x files, for tests where you need something
> like that just stick it into gcc.dg/torture/ instead and use normal dg stuff
> in there.
Thanks, fixed. Ok to commit?
Changelog:
gcc:
2013-09-09 Michael Zolotukhin
* config/i386/i386.c (ix86_expand_movmem): F
Bootstrapped & regression-tested on x86_64-linxu-gnu. Applied as obvious.
Thanks,
Richard
gcc/
* asan.c (asan_shadow_cst): Use gen_int_mode.
Index: gcc/asan.c
===
--- gcc/asan.c 2013-09-08 16:54:50.485108772 +0100
+++ gcc
On Mon, Sep 9, 2013 at 11:23 AM, Michael V. Zolotukhin
wrote:
>> Please don't introduce new *.x files, for tests where you need something
>> like that just stick it into gcc.dg/torture/ instead and use normal dg stuff
>> in there.
> Thanks, fixed. Ok to commit?
No -march in runtime tests, please
On Sun, 8 Sep 2013, Jeff Law wrote:
This fixes the problem noted by Zhendong in c#4. I'll be working
through other reports of issues related to the tree-ssa-threadedge.c
to see if there's any lingering problems.
It also fixes my original report, PR bootstrap/58340 (on
x86_64-unknown-freebsd)
On Mon, Sep 9, 2013 at 1:09 AM, Kugan wrote:
>
> On 06/09/13 16:16, Richard Biener wrote:
>>
>> On 9/3/13 2:15 PM, Kugan wrote:
>>>
>>> Thanks Richard for reviewing.
>>>
>>> On 02/09/13 22:15, Richard Biener wrote:
On Wed, Jul 3, 2013 at 2:25 PM, Kugan
wrote:
>
> On 17/06/1
On Mon, 9 Sep 2013, Paolo Carlini wrote:
> Hi Richard,
>
> On 09/09/2013 09:46 AM, Richard Biener wrote:
> > On Sun, 8 Sep 2013, Paolo Carlini wrote:
> >
> > > Hi,
> > >
> > > this patchlet fixes the column # of the unused parameter warnings emitted
> > > by
> > > do_warn_unused_parameter by ex
On Mon, 9 Sep 2013, Richard Biener wrote:
> On Mon, 9 Sep 2013, Paolo Carlini wrote:
>
> > Hi Richard,
> >
> > On 09/09/2013 09:46 AM, Richard Biener wrote:
> > > On Sun, 8 Sep 2013, Paolo Carlini wrote:
> > >
> > > > Hi,
> > > >
> > > > this patchlet fixes the column # of the unused parameter
On Mon, Sep 09, 2013 at 11:37:31AM +0200, Richard Biener wrote:
> > > > > this patchlet fixes the column # of the unused parameter warnings
> > > > > emitted
> > > > > by
> > > > > do_warn_unused_parameter by explicitly passing DECL_SOURCE_LOCATION
> > > > > (decl)
> > > > > instead of wrongly re
Thanks for noticing it Richard, I made a refactoring mistake and addr
was supposed to be used instead of x. In fact on AArch64 it occurs
that we don't have stripped rtxes at this step and we have some of the
form below, this if why I added the strip.
(insn 29 27 5 7 (set (mem:SI (plus:DI (sign_ex
On Mon, 9 Sep 2013, Jakub Jelinek wrote:
> On Mon, Sep 09, 2013 at 11:37:31AM +0200, Richard Biener wrote:
> > > > > > this patchlet fixes the column # of the unused parameter warnings
> > > > > > emitted
> > > > > > by
> > > > > > do_warn_unused_parameter by explicitly passing DECL_SOURCE_LOCATI
Hi,
On 09/09/2013 11:33 AM, Richard Biener wrote:
How is '+' in %q+D defined? I failed to find documentation of the
diagnostic formats (but only searched for like two minutes).
You see, this is an issue ;) No seriously, in C++ I understand it by
looking at error.c:3438: when '+' is there, cp_p
On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
> Well, in this case the patch should IMHO be a no-op.
>
> - warning (OPT_Wunused_parameter, "unused parameter %q+D", decl);
> + warning_at (DECL_SOURCE_LOCATION (decl), OPT_Wunused_parameter,
> + "unused param
[Resending, since I was away and not pinging it]
Hi all,
In PR58088 the constant folder goes into an infinite recursion and runs out of
stack space because of two conflicting optimisations:
(X * C1) & C2 plays dirty when nested inside an IOR expression like so: ((X *
C1) & C2) | C4. One can undo
Hi,
On 09/09/2013 11:45 AM, Richard Biener wrote:
Well, in this case the patch should IMHO be a no-op.
- warning (OPT_Wunused_parameter, "unused parameter %q+D", decl);
+ warning_at (DECL_SOURCE_LOCATION (decl), OPT_Wunused_parameter,
+ "unused parameter %qD", decl);
On Mon, 9 Sep 2013, Jakub Jelinek wrote:
> On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
> > Well, in this case the patch should IMHO be a no-op.
> >
> > - warning (OPT_Wunused_parameter, "unused parameter %q+D", decl);
> > + warning_at (DECL_SOURCE_LOCATION (decl), OP
On 09/09/2013 12:04 PM, Richard Biener wrote:
On Mon, 9 Sep 2013, Jakub Jelinek wrote:
On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
Well, in this case the patch should IMHO be a no-op.
- warning (OPT_Wunused_parameter, "unused parameter %q+D", decl);
+ warning_at
> No -march in runtime tests, please.
Is mtune ok here?
Michael
> Uros.
On 09/09/2013 12:08 PM, Paolo Carlini wrote:
Everything happens via call backs. Thus from the generic diagnostic
machinery, you go to cp_printer for C++, thus location_of for C++. In
C is different, but again there is, evidently, a mechanism which uses
DECL_CONTEXT for PARM_DECLs which leads to
On Mon, 9 Sep 2013, Paolo Carlini wrote:
> On 09/09/2013 12:04 PM, Richard Biener wrote:
> > On Mon, 9 Sep 2013, Jakub Jelinek wrote:
> >
> > > On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
> > > > Well, in this case the patch should IMHO be a no-op.
> > > >
> > > > - warn
On Mon, Sep 9, 2013 at 12:12 PM, Michael V. Zolotukhin
wrote:
>> No -march in runtime tests, please.
> Is mtune ok here?
Yes.
Uros.
On Mon, Sep 09, 2013 at 02:12:35PM +0400, Michael V. Zolotukhin wrote:
> > No -march in runtime tests, please.
> Is mtune ok here?
Please verify that with the s/-march/-mtune/ change without the i386.c fix the
testcase
will fail with
make check-gcc RUNTESTFLAGS='--target_board=unix\{-m32,-m64\}
On Mon, Sep 9, 2013 at 12:17 PM, Jakub Jelinek wrote:
> On Mon, Sep 09, 2013 at 02:12:35PM +0400, Michael V. Zolotukhin wrote:
>> > No -march in runtime tests, please.
>> Is mtune ok here?
>
> Please verify that with the s/-march/-mtune/ change without the i386.c fix
> the testcase
It will fail,
Hi,
On 09/09/2013 12:13 PM, Richard Biener wrote:
Everything happens via call backs. Thus from the generic diagnostic machinery,
you go to cp_printer for C++, thus location_of for C++. In C is different, but
again there is, evidently, a mechanism which uses DECL_CONTEXT for PARM_DECLs
which lead
These four patches are the result of going through gcc/* looking for
places where there was an obvious mode associated with a GEN_INT and
where gen_int_mode could therefore be used instead. Some of the GEN_INTs
did create noncanoical rtl before, and were trapped by Kenny's assert on
the wide-int b
> > Is mtune ok here?
> Yes.
Thanks, fixed. Ok to commit?
I verified that the test fails without fix in i386.c and passes with it.
Changelog:
gcc:
2013-09-09 Michael Zolotukhin
* config/i386/i386.c (ix86_expand_movmem): Fix epilogue generation.
gcc/testsuite:
2013-09-09 Michael Zolo
Similar to patch 1, but here the calls are all to expand_* routines.
Tested in the same way as before. OK to install?
Thanks,
Richard
gcc/
* asan.c (asan_clear_shadow): Use gen_int_mode with the mode
of the associated expand_* call.
(asan_emit_stack_protection): Likewis
Similar to patch 1, but here the calls are all to simplify_* routines.
In the 7414 combine.c hunk, the code is PLUS, AND, IOR or XOR.
In the hunk after that, the "cval |= ..." stuff is effectively doing
a trunc_int_for_mode by hand.
Tested in the same way as before. OK to install?
Thanks,
Richar
Similar to patch 1, but here the calls are involved in a SET instruction
(or equivalent) and are cases where the operand must have the same mode as
the SET destination.
Tested in the same way as before. OK to install?
Thanks,
Richard
gcc/
* asan.c (asan_emit_stack_protection): Use gen_
Hi,
On 09/09/2013 06:35 AM, Jason Merrill wrote:
On 09/05/2013 06:44 PM, Paolo Carlini wrote:
+ && warning (0, "possible problem detected in invocation of "
+ "delete [] operator:"))
The warning should probably be suppressible by some flag.
In fact clang has such a flag - I
expmed.c:lshift_value has:
val = double_int::from_uhwi (INTVAL (value)).zext (bitsize);
val = val.llshift (bitpos, HOST_BITS_PER_DOUBLE_INT);
but its only caller has already zero-extended INTVAL (value) from BITSIZE,
so we might as well pass that instead of the (value, bitsize) pair.
This is
On 09/09/2013 11:37 AM, Richard Biener wrote:
That said, grepping for %q+D reveals quite some uses and it looks like
all of them expect the location being used to be that of the decl passed
to the diagnostic call, not some random other location.
If the decl is *not* a PARM_DECL, I expect %q+D to
On Mon, Sep 09, 2013 at 12:38:46PM +0200, Paolo Carlini wrote:
> On 09/09/2013 11:37 AM, Richard Biener wrote:
> >That said, grepping for %q+D reveals quite some uses and it looks like
> >all of them expect the location being used to be that of the decl passed
> >to the diagnostic call, not some ra
On 09/09/2013 12:41 PM, Jakub Jelinek wrote:
On Mon, Sep 09, 2013 at 12:38:46PM +0200, Paolo Carlini wrote:
On 09/09/2013 11:37 AM, Richard Biener wrote:
That said, grepping for %q+D reveals quite some uses and it looks like
all of them expect the location being used to be that of the decl pass
On 09/09/2013 12:44 PM, Paolo Carlini wrote:
I understand that. It seems to me a much bigger project and must be
done for the C front-end too (I don't know the name of the equivalent
of location_of, but the location is wrong for it too, there must be
the equivalent of t = DECL_CONTEXT (t) for i
Hello,
On 04 Sep 22:45, Kirill Yukhin wrote:
> Hello,
>
> PING.
PING.
--
Thanks, K
Hello,
On 06 Sep 17:41, Kirill Yukhin wrote:
> Hello,
>
> PING.
PING.
--
Thanks, K
On Mon, Sep 9, 2013 at 12:27 PM, Richard Sandiford
wrote:
> Similar to patch 1, but here the calls are involved in a SET instruction
> (or equivalent) and are cases where the operand must have the same mode as
> the SET destination.
>
> Tested in the same way as before. OK to install?
Ok.
Thank
On Mon, Sep 9, 2013 at 12:21 PM, Richard Sandiford
wrote:
> These four patches are the result of going through gcc/* looking for
> places where there was an obvious mode associated with a GEN_INT and
> where gen_int_mode could therefore be used instead. Some of the GEN_INTs
> did create noncanoic
On Mon, Sep 9, 2013 at 12:24 PM, Richard Sandiford
wrote:
> Similar to patch 1, but here the calls are all to expand_* routines.
>
> Tested in the same way as before. OK to install?
Ok.
Thanks,
Richard.
> Thanks,
> Richard
>
>
> gcc/
> * asan.c (asan_clear_shadow): Use gen_int_mode wit
On Mon, Sep 9, 2013 at 12:27 PM, Richard Sandiford
wrote:
> Similar to patch 1, but here the calls are all to simplify_* routines.
> In the 7414 combine.c hunk, the code is PLUS, AND, IOR or XOR.
> In the hunk after that, the "cval |= ..." stuff is effectively doing
> a trunc_int_for_mode by hand.
On Mon, Sep 9, 2013 at 12:31 PM, Richard Sandiford
wrote:
> expmed.c:lshift_value has:
>
> val = double_int::from_uhwi (INTVAL (value)).zext (bitsize);
> val = val.llshift (bitpos, HOST_BITS_PER_DOUBLE_INT);
>
> but its only caller has already zero-extended INTVAL (value) from BITSIZE,
> so we
On Mon, Sep 9, 2013 at 12:22 PM, Michael V. Zolotukhin
wrote:
>> > Is mtune ok here?
>> Yes.
> Thanks, fixed. Ok to commit?
> I verified that the test fails without fix in i386.c and passes with it.
>
> Changelog:
> gcc:
> 2013-09-09 Michael Zolotukhin
>
> * config/i386/i386.c (ix86_ex
Ok to commit? Attached is the doc fix patch. I'll send the TREE_TYPE
patch shortly.
Andrew
Andrew Sutton
On Sat, Sep 7, 2013 at 1:00 PM, Jason Merrill wrote:
> On 09/06/2013 12:03 PM, Andrew Sutton wrote:
>>
>> +// Returns the template type of the class scope being entered. If we're
>> +// ente
This patch prevents the powerpc backend from combining a 64-bit
volatile load or store with a bswap insn when the resulting combined
insn will be implemented as two lwbrx or stwbrx machine insns.
Bootstrapped and regression tested powerpc64-linux.
PR target/58330
* config/rs6000/rs
Andrew Sutton writes:
| Ok to commit? Attached is the doc fix patch. I'll send the TREE_TYPE
| patch shortly.
Yes, please -- that is what Jason earlier. Let me know when youve
committed so that I can synchronize with trunk.
-- Gaby
On Mon, Sep 9, 2013 at 4:06 AM, Richard Biener
wrote:
> On Sat, Sep 7, 2013 at 11:00 PM, Marc Glisse wrote:
>> On Sat, 7 Sep 2013, Mike Stump wrote:
>>
>>> On Sep 7, 2013, at 12:27 PM, Marc Glisse wrote:
Now flag_check_new should probably disable this optimization…
>>>
>>>
>>> Yes, thi
On Mon, Sep 9, 2013 at 4:19 AM, Marc Glisse wrote:
> On Mon, 9 Sep 2013, Richard Biener wrote:
>
>> On Sat, Sep 7, 2013 at 11:00 PM, Marc Glisse wrote:
>>>
>>> On Sat, 7 Sep 2013, Mike Stump wrote:
>>>
On Sep 7, 2013, at 12:27 PM, Marc Glisse wrote:
>
>
> Now flag_check_new shou
Hi all, hi Gaby,
thus a more sensible try at fixing this issue:
function.c:do_warn_unused_parameter uses "%q+D" for the error call. That
means that cp/error.c:cp_printer does:
if (set_locus && t != NULL)
*text->locus = location_of (t);
and it's important that location_of (t) is correct
On Mon, Sep 9, 2013 at 4:46 AM, Jakub Jelinek wrote:
> On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
>> Well, in this case the patch should IMHO be a no-op.
>>
>> - warning (OPT_Wunused_parameter, "unused parameter %q+D", decl);
>> + warning_at (DECL_SOURCE_LOCATION (de
Hi Richard,
> On 29/07/13 14:58, Kyrylo Tkachov wrote:
> > Hi all,
> >
> > Now that combine emits the canonical form for (eq (neg x) (y)) instead
> of (eq
> > (x) (neg y)), this patch fixes up the corresponding pattern in aarch64
> to
> > match that. This enables combine to properly generate the c
On Mon, Sep 9, 2013 at 5:04 AM, Richard Biener wrote:
> On Mon, 9 Sep 2013, Jakub Jelinek wrote:
>
>> On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
>> > Well, in this case the patch should IMHO be a no-op.
>> >
>> > - warning (OPT_Wunused_parameter, "unused parameter %q+D",
On Mon, Sep 9, 2013 at 5:13 AM, Richard Biener wrote:
> On Mon, 9 Sep 2013, Paolo Carlini wrote:
>
>> On 09/09/2013 12:04 PM, Richard Biener wrote:
>> > On Mon, 9 Sep 2013, Jakub Jelinek wrote:
>> >
>> > > On Mon, Sep 09, 2013 at 11:45:08AM +0200, Richard Biener wrote:
>> > > > Well, in this case
On Mon, Sep 9, 2013 at 5:38 AM, Paolo Carlini wrote:
> On 09/09/2013 11:37 AM, Richard Biener wrote:
>>
>> That said, grepping for %q+D reveals quite some uses and it looks like
>> all of them expect the location being used to be that of the decl passed
>> to the diagnostic call, not some random o
Hi guys!
PING for both C and C++.
Thanks.
Original Message
Subject: Re: PING: Fwd: Re: [patch] implement Cilk Plus simd loops on trunk
Date: Tue, 27 Aug 2013 15:03:26 -0500
From: Aldy Hernandez
To: Richard Henderson
CC: jason merrill , gcc-patches
On 08/26/13 12:22, Rich
On Mon, Sep 9, 2013 at 5:41 AM, Jakub Jelinek wrote:
> On Mon, Sep 09, 2013 at 12:38:46PM +0200, Paolo Carlini wrote:
>> On 09/09/2013 11:37 AM, Richard Biener wrote:
>> >That said, grepping for %q+D reveals quite some uses and it looks like
>> >all of them expect the location being used to be tha
On Mon, Sep 9, 2013 at 7:49 AM, Paolo Carlini wrote:
> Hi all, hi Gaby,
>
> thus a more sensible try at fixing this issue:
> function.c:do_warn_unused_parameter uses "%q+D" for the error call. That
> means that cp/error.c:cp_printer does:
>
> if (set_locus && t != NULL)
> *text->locus = loca
On Sun, Sep 8, 2013 at 3:26 PM, Paolo Carlini wrote:
> Hi all, Gaby,
>
> I was having a look to c++/58363 and besides the main issue, that is we
> probably want to help the user and tell him/her something about destructors,
> etc, I noticed that we aren't able to pretty print the pseudo destructor
at revision 202396. No conflict.
-- Gaby
Enabling the CONST_INT assert showed that this branch-local code was
passing the wrong mode to std::make_pair. The mode of X is OLDMODE,
and we're supposed to be converting it to MODE.
In the case where OLDMODE is VOIDmode, I think we have to assume
that every bit of the input is significant.
Te
This fixes PR58326, fix_bb_placements via unloop does not properly
mark all blocks with uses that are now eventually subject to
loop-closed SSA handling.
Fixed with the following, bootstrapped and tested on
x86_64-unknown-linux-gnu, applied.
Richard.
2013-09-09 Richard Biener
PR mi
OK, thanks.
Jason
On Mon, Sep 9, 2013 at 3:11 PM, Richard Sandiford
wrote:
> Enabling the CONST_INT assert showed that this branch-local code was
> passing the wrong mode to std::make_pair. The mode of X is OLDMODE,
> and we're supposed to be converting it to MODE.
>
> In the case where OLDMODE is VOIDmode, I thin
On 09/09/2013 03:26 AM, Gerald Pfeifer wrote:
On Sun, 8 Sep 2013, Jeff Law wrote:
This fixes the problem noted by Zhendong in c#4. I'll be working
through other reports of issues related to the tree-ssa-threadedge.c
to see if there's any lingering problems.
It also fixes my original report, P
On 09/09/13 13:50, Kyrylo Tkachov wrote:
> Hi Richard,
>
>> On 29/07/13 14:58, Kyrylo Tkachov wrote:
>>> Hi all,
>>>
>>> Now that combine emits the canonical form for (eq (neg x) (y)) instead
>> of (eq
>>> (x) (neg y)), this patch fixes up the corresponding pattern in aarch64
>> to
>>> match that.
Richard Biener writes:
> On Mon, Sep 9, 2013 at 3:11 PM, Richard Sandiford
> wrote:
>> Enabling the CONST_INT assert showed that this branch-local code was
>> passing the wrong mode to std::make_pair. The mode of X is OLDMODE,
>> and we're supposed to be converting it to MODE.
>>
>> In the case
On 13-08-22 9:37 AM, Andrey Turetskiy wrote:
Hi Vladimir,
I'm trying to apply KNF and KNC changes for GCC 4.9 (they have been
applied for 4.7 before), but I have some problems with LRA.
I've tried to build cross compiler for KNC. During the compiling of
libgcc/libgcc2.c the compiler becomes lock
On 09/09/2013 03:22 AM, Adam Butcher wrote:
I've attached a patch that handles parameter packs in the conversion
op. I thought it best to get this reviewed independently before rolling
up into the 'Support lambda templates' patch.
Do you think it's the right idea? It seems to work okay but I'v
On 09/09/2013 09:56 AM, Richard Sandiford wrote:
Richard Biener writes:
On Mon, Sep 9, 2013 at 3:11 PM, Richard Sandiford
wrote:
Enabling the CONST_INT assert showed that this branch-local code was
passing the wrong mode to std::make_pair. The mode of X is OLDMODE,
and we're supposed to be c
> "Tom" == Tom Tromey writes:
Tom> This is version 3 of my series to resurrect automatic dependencies for
Tom> GCC. Version 2 is here:
Tom> http://gcc.gnu.org/ml/gcc-patches/2013-07/msg01386.html
Tom> Ordinarily I would simply ping the existing patches, but Alexandre
Tom> asked me to re
The last merge didn't compile for me. Apparently some pretty printing
functions have disappeared that were being called in the concept
stuff. This patch just replaces the broken calls to pp_cxx_type_id
with pp->type_id.
Ok to commit?
Andrew Sutton
fix.patch
Description: Binary data
Hi,
vrsqrte_f64 is currently defined to take a float64x2_t, but it should
take a float64x1_t.
I've committed the attached, obvious fix as revision 202407.
James
---
2013-09-09 James Greenhalgh
* config/aarch64/arm_neon.h (vrsqrte_f64): Fix parameter type.
diff --git a/gcc/config/aa
On Mon, Sep 9, 2013 at 5:07 AM, Alan Modra wrote:
> Revised patch with testcase. This one also fixes a small problem with
> reg_or_add_cint_operand in that any 32-bit value is valid for SImode.
> Compare with reg_or_sub_cint_operand.
>
> Bootstrapped and regression tested powerpc64-linux. OK to
On Mon, 2013-09-09 at 14:25 +0800, bin.cheng wrote:
> Thanks for reviewing, I will correct all stupid spelling problem in the next
> version of patch.
>
> On Mon, Sep 9, 2013 at 8:15 AM, Bill Schmidt
> wrote:
> >
> >>>+ int (i * S).
> >>>+ Otherwise, just return double int zero. */
> >
> >
Hi!
It is not safe to always handle a = ~b; for bool a and b as
in_p = !in_p; exp = arg0. If the current range is say
+[-,-] or -[-,-] or similar (i.e. unconditional false or true),
then the value of exp doesn't matter and thus the fact that exp
has been negated is irrelevant.
Fixed by just maki
On Mon, Sep 9, 2013 at 8:21 AM, Alan Modra wrote:
> This patch prevents the powerpc backend from combining a 64-bit
> volatile load or store with a bswap insn when the resulting combined
> insn will be implemented as two lwbrx or stwbrx machine insns.
> Bootstrapped and regression tested powerpc64
1 - 100 of 145 matches
Mail list logo
