On Mon, 14 Oct 2013, Mike Stump wrote:
> On Sep 25, 2012, at 8:00 AM, Richard Guenther wrote:
> > 2012-09-25 Richard Guenther
> >
> > PR lto/54625
> > * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Do not merge
> > cgraph nodes for builtins.
> >
> > * gcc.dg/lto/pr54702_0.
> -Original Message-
> From: Jakub Jelinek [mailto:ja...@redhat.com]
> Sent: Monday, October 14, 2013 4:49 PM
> To: Zhenqiang Chen
> Cc: gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] Reassociate X == CST1 || X == CST2 if popcount (CST2
-
> CST1) == 1 into ((X - CST1) & ~(CST2 - CST1)) ==
Richard Biener writes:
>>> Honza? For tailr this boils down to symtab_semantically_equivalent_p ().
>>> I suppose we don't want to change that but eventually special case
>>> builtins in tailr, thus
>>>
>>> Index: gcc/tree-tailcall.c
>>> ===
On Tue, Oct 15, 2013 at 03:57:23PM +0800, Zhenqiang Chen wrote:
> Is it OK?
Ok, except two comments apparently still need updating.
+/* Optimize X == CST1 || X == CST2
+ if popcount (CST1 ^ CST2) == 1 into
+ (X & ~(CST1 ^ CST2)) == (CST1 & ~(CST1 ^ CST2)).
+ Similarly for ranges. E.g.
+
On Sun, Oct 6, 2013 at 2:22 PM, Bernd Edlinger
wrote:
> Ping!
>
>
> How I should proceed with this patch, is it OK?
>
> The latest version was posted at:
> http://gcc.gnu.org/ml/gcc-patches/2013-09/msg00234.html
I have now attached a quick draft of a patch that rewrites the moved stmts
to not ex
On Mon, Oct 14, 2013 at 6:10 PM, Paolo Carlini wrote:
> Hi
>
>
> On 10/14/2013 05:43 PM, Paulo Matos wrote:
>>
>> * cp/error.c (code_to_string): Use new wrapper get_tree_code_name.
>> * cp/cxx-pretty-print.c (pp_cxx_assignment_operator): Use new
>> wrapper get_tree_code_name.
>>
On Tue, Oct 15, 2013 at 9:59 AM, Richard Sandiford
wrote:
> Richard Biener writes:
Honza? For tailr this boils down to symtab_semantically_equivalent_p ().
I suppose we don't want to change that but eventually special case
builtins in tailr, thus
Index: gcc/tree-tailcall
> -Original Message-
> From: Richard Biener [mailto:richard.guent...@gmail.com]
> Sent: 15 October 2013 10:12
> To: Paolo Carlini
> Cc: Paulo Matos; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] tree_code_name wrapper
>
>
> Apart from this please make sure to properly use tabs to inden
On Tue, Oct 15, 2013 at 09:42:17AM +, Paulo Matos wrote:
> Thanks, regarding the indentation I was convinced we used 2 space
> indentation with maximum line length of 80 characters.
We do, however 8 consecutive spaces in the indentation should be always
replaced by a tab.
> Should I also subm
The following simplifies code, opportunities exposed by earlier changes.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
Richard.
2013-10-15 Richard Biener
* tree-loop-distribution.c (build_empty_rdg): Inline into
single user.
(rdg_flag_vertex): Inline
This finally reduces the size of the pointer-map we use to track
hash values for canonical type computation. We only need to keep
track of hash values for types that are the canonical type "master".
LTO bootstrap and testing running on x86_64-unknown-linux-gnu, I'll
commit it after that succeede
Hi Eric,
Can you please help to review this patch?
http://gcc.gnu.org/ml/gcc-patches/2013-10/msg00452.html
Thanks,
Kugan
> +2013-09-25 Kugan Vivekanandarajah
> +
> + * dojump.c (do_compare_and_jump): Generate rtl without
> + zero/sign extension if redundant.
> + * cfgexpand.c (exp
This restructures the recursion through copies and conversions
that forwprop does to ADDR_EXPRs. First this avoids doing forward
copy-propagation, that is,
a_1 = &x[i_0];
a_2 = a_1;
is no longer replaced with
a_2 = &x[i_0];
as that performs code sinking. Second, the conversion case
was
This patch corrects the analysis of an initialization item with an input_list
to catch cases where input items come from the same related package.
-- Source --
-- init.ads
package Init
with Abstract_State => S,
Initializes=> (S => (Emb.SS, Emb.XX, Emb.YY
Valgrind complains:
==27870== Conditional jump or move depends on uninitialised value(s)
==27870==at 0x557CDC: cgraph_create_edge_1(cgraph_node*, cgraph_node*,
gimple_statement_d*, long, int) (cgraph.c:695)
==27870==by 0x55882E: cgraph_create_edge(cgraph_node*, cgraph_node*,
gimple_statem
This patch modifies the legality checks of aspect/pragma Refined_State to
ensure that when a state acts as a constituent of another state, the said state
has a Part_Of dependency in its corresponding aspect/pragma Abstract_State.
-- Source --
-- abstract_state_illegal.a
We now give a better message for a redundant ELSE part
in an IF expression.
1. package Elseelse is
2.function F return Boolean is
3. (if 1 + 1 = 2 then True
4. else False
5. else True);
|
>>> only one "else" part is allowed
This change improves the circuitry that is responsible for eliminating
checks in the right hand side of short circuit operators, when the left
hand side contains a test that guarantees the check can't fail.
The following test must show no "raise containt_error" nodes in expanded
code:
gcc -c k.ad
On Tue, Oct 15, 2013 at 12:31 PM, Markus Trippelsdorf
wrote:
> Valgrind complains:
> ==27870== Conditional jump or move depends on uninitialised value(s)
> ==27870==at 0x557CDC: cgraph_create_edge_1(cgraph_node*, cgraph_node*,
> gimple_statement_d*, long, int) (cgraph.c:695)
> ==27870==by
We have a testcase where we have this insn:
(insn:HI 53 55 56 2 (set (reg:SI 214 [ D.1303 ])
(mem:SI (plus:SI (mult:SI (reg/v:SI 219 [ p2 ])
(const_int 4 [0x4]))
(reg/v/f:SI 218 [ p1 ])) [3 S4 A32])) 163
which produces the following reloads:
Reload 0:
If a library level package declaration or generic package declaration
contains a non-null abstract state, and does not otherwise require a
body, then in accorance with SPARK (LRM 7.1.5(4)), the package must
contain a pragma Elaborate_Body. This implements this requirement,
as shown in the following
Hi,
The historical neon_type attribute formed groups over the Neon
instructions which were well suited for modelling the Cortex-A8
pipeline, but were cumbersome for other processor models.
The AArch64 has another classification "simd_type". This is, with a
few exceptions, and when augmented by si
Hi,
This patch updates all the types used by the "types" attribute
to describe neon instructions.
These have, in the past, been designed to support modelling the
Cortex-A8 pipeline. We can split these types, and use them to better
model other pipelines.
This patch adds the new types we plan to
Hi,
This patch transforms:
neon_ldm_2 to neon_load1_2reg
neon_str_2 to neon_store1_2reg
in the aarch64 backend. This is just an administrative change, as
we have no cores consuming these types in the aarch64 backend.
Tested on aarch64-none-elf with no regressions.
Thanks,
James
---
gcc/
Hi,
Now that we have proted the pipeline models, this patch removes the
old neon types.
Bootstrapped on a chromebook in series and sanity checked.
Thanks,
James
---
gcc/
2013-10-15 James Greenhalgh
* config/arm/types: Remove old neon types.
diff --git a/gcc/config/arm/types.md b/g
Hi,
After refactoring all the Neon "type" attributes, neon-schedgen.ml is
out of date and only serves to distract.
This patch removes the script.
I've run a bootstrap for arm just to ensure that no funky Make
machinery remains.
OK?
Thanks,
James
---
2013-10-15 James Greenhalgh
*
Hi,
This patch updates the A7 pipeline for the new Neon types.
Sanity checked and tested with some neon intrinsics code to see
schedule quality.
Thanks,
James
---
gcc/
2013-10-15 James Greenhalgh
* config/arm/cortex-a7.md
(cortex_a7_neon_type): New.
(cortex_a7_neon
On 1 October 2013 12:40, Marcus Shawcroft wrote:
> On 30/09/13 13:40, Marcus Shawcroft wrote:
>
>>> Well, I thought this patch would work for me, but it does not. It looks
>>> like gcc_no_link is set to 'no' on my target because, technically, I can
>>> link even if I don't use a linker script. I
This makes sure that we do not record a signed GIV that we do not
know whether it overflows or not. For the testcase IVOPTs else
can end up replacing an unsigned computation with a signed one.
Note that alternatively we may decide that it is not desirable
for SCEV to return signed { 2147483643,
On Tue, 15 Oct 2013, Richard Biener wrote:
>
> This makes sure that we do not record a signed GIV that we do not
> know whether it overflows or not. For the testcase IVOPTs else
> can end up replacing an unsigned computation with a signed one.
>
> Note that alternatively we may decide that it i
i added the assertion that richard requested and tested this on x86-64.
committed as revision 203602.
On 10/06/2013 05:13 AM, Richard Sandiford wrote:
Kenneth Zadeck writes:
On 10/04/2013 01:00 PM, Richard Sandiford wrote:
I was hoping Richard would weigh in here. In case not...
Kenneth Z
Hi!
Especially on i?86/x86_64 if-conversion pass seems to be often
a pessimization, but the vectorization relies on it and without it we can't
vectorize a lot of the loops.
Here is a prototype of a patch that will by default (unless explicit
-ftree-loop-if-convert) only if-convert loops internall
Ping^2. Jason, Joseph, are you fine with the C++/C FE changes?
Thanks.
On Mon, Oct 07, 2013 at 10:17:38PM +0200, Marek Polacek wrote:
> Ping.
>
> On Wed, Sep 25, 2013 at 02:41:32PM +0200, Marek Polacek wrote:
> > On Thu, Sep 12, 2013 at 02:26:55PM +0200, Marek Polacek wrote:
> > > This patch ad
> -Original Message-
> From: Jakub Jelinek [mailto:ja...@redhat.com]
> Sent: 15 October 2013 10:51
> To: Paulo Matos
> Cc: Richard Biener; Paolo Carlini; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] tree_code_name wrapper
>
> On Tue, Oct 15, 2013 at 09:42:17AM +, Paulo Matos wrote:
>
Gerald> - When I do a gmake at the top level of the build tree, nothing is
Gerald>rebuilt at all. This only happens during `gmake install`
[...]
Gerald> Is nobody else seeing this? Or is everyone just lucky enough to have
Gerald> a recent version of GCC as the default compiler?
How exactly
On Tue, 15 Oct 2013, Kugan wrote:
> Hi Eric,
>
> Can you please help to review this patch?
> http://gcc.gnu.org/ml/gcc-patches/2013-10/msg00452.html
I think that gimple_assign_is_zero_sign_ext_redundant and its
description is somewhat confused. You seem to have two cases
here, one being NOP_EXP
Hey guys,
could please someone look at this small patch? It blocks approved MPX
ISA support on i386 target.
Thanks,
Ilya
2013/10/2 Ilya Enkovich :
> Ping
>
> 2013/9/17 Ilya Enkovich :
>> Hi,
>>
>> Here is a patch introducing new type and mode for bounds. It is a part of
>> MPX ISA support patch
On Tue, Oct 15, 2013 at 3:19 PM, Paulo Matos wrote:
>> -Original Message-
>> From: Jakub Jelinek [mailto:ja...@redhat.com]
>> Sent: 15 October 2013 10:51
>> To: Paulo Matos
>> Cc: Richard Biener; Paolo Carlini; gcc-patches@gcc.gnu.org
>> Subject: Re: [PATCH] tree_code_name wrapper
>>
>> On
On Sat, Sep 28, 2013 at 3:28 AM, Cong Hou wrote:
> The current GCC vectorizer requires the following pattern as a simple
> reduction computation:
>
>loop_header:
> a1 = phi < a0, a2 >
> a3 = ...
> a2 = operation (a3, a1)
>
> But a3 can also be defined outside of the loop. For ex
This patch fixes a profile count insanity introduced by ssa tail
merging. When replacing bb1 with bb2, which has the same successors,
the bb counts were being merged, but the successor edge weights
were not.
Bootstrapped and tested on x86_64-unknown-linux-gnu. Ok for trunk?
Thanks,
Teresa
2013-1
Hello,
On 11 Oct 10:30, Richard Henderson wrote:
> On 10/09/2013 03:29 AM, Kirill Yukhin wrote:
> > +(define_insn "avx512f_vec_dup_mem"
> > + [(set (match_operand:VI48F_512 0 "register_operand" "=x")
> > + (vec_duplicate:VI48F_512
> > + (match_operand: 1 "nonimmediate_operand" "xm")))]
> > +
This is an alternate fix (see
http://gcc.gnu.org/ml/gcc-patches/2013-09/msg00234.html for the other
one) for the various PRs that show that LIM exposes undefined
signed overflow on paths where it wasn't executed before LIM
ultimately leading to a miscompilation.
For this fix we rewrite invariant
Hello,
On 11 Oct 11:21, Richard Henderson wrote:
> On 10/09/2013 03:30 AM, Kirill Yukhin wrote:
> > +;; Return true if OP is either -1 constant or stored in register.
> > +(define_predicate "register_or_constm1_operand"
> > + (ior (match_operand 0 "register_operand")
> > + (match_test "op ==
Hello,
On 14 Oct 13:10, Richard Henderson wrote:
> On 10/09/2013 03:31 AM, Kirill Yukhin wrote:
> > +(define_mode_attr ssefixupmode
> > + [(V16SF "V16SI") (V4SF "V4SI") (V8DF "V8DI") (V2DF "V2DI")])
> > +
>
> Oh, I forgot. How is this different from sseintvecmode?
It is definetely a bug. Ok with
Hi,
On Tue, 15 Oct 2013 15:57:16, Richard Biener wrote:
>
> This is an alternate fix (see
> http://gcc.gnu.org/ml/gcc-patches/2013-09/msg00234.html for the other
> one) for the various PRs that show that LIM exposes undefined
> signed overflow on paths where it wasn't executed before LIM
> ultimat
On Fri, Jul 05, 2013 at 09:25:52AM -0700, Mike Stump wrote:
> On Jul 4, 2013, at 9:17 AM, Marcus Shawcroft wrote:
> >* gcc.dg/pr57518.c: Adjust scan-rtl-dump-not pattern.
>
> [ If you want a review or need an approval, be sure to ask Ok? Just in case
> you forgot... ] Ok.
>
> Thanks.
On 15/10/13 12:31, Marcus Shawcroft wrote:
> On 1 October 2013 12:40, Marcus Shawcroft wrote:
>> On 30/09/13 13:40, Marcus Shawcroft wrote:
>>
Well, I thought this patch would work for me, but it does not. It looks
like gcc_no_link is set to 'no' on my target because, technically, I can
On 15/10/13 12:23, James Greenhalgh wrote:
> Hi,
>
> The historical neon_type attribute formed groups over the Neon
> instructions which were well suited for modelling the Cortex-A8
> pipeline, but were cumbersome for other processor models.
>
> The AArch64 has another classification "simd_type".
On 10/15/2013 06:57 AM, Kirill Yukhin wrote:
> Hello,
> On 14 Oct 13:10, Richard Henderson wrote:
>> On 10/09/2013 03:31 AM, Kirill Yukhin wrote:
>>> +(define_mode_attr ssefixupmode
>>> + [(V16SF "V16SI") (V4SF "V4SI") (V8DF "V8DI") (V2DF "V2DI")])
>>> +
>>
>> Oh, I forgot. How is this different
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-ow...@gcc.gnu.org] On
> Behalf Of Richard Biener
> Sent: 15 October 2013 14:34
> To: Paulo Matos
> Cc: Jakub Jelinek; Paolo Carlini; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] tree_code_name wrapper
>
> On T
The ChangeLog is still wrong:
On Tue, Oct 15, 2013 at 02:38:31PM +, Paulo Matos wrote:
> 2013-10-15 Paulo Matos
>
> gcc/
> * tree-core.h: Remove extern declaration of tree_code_name.
* tree-core.h (tree_code_name): Remove.
(or Remove external declaration. etc.).
> * t
This memory leak is because forgetting virtual destructor of the base
class _Executor.
Thanks!
--
Tim Shen
a.patch
Description: Binary data
The patch updated:
Index: gcc/cgraph.c
===
--- gcc/cgraph.c (revision 203609)
+++ gcc/cgraph.c (working copy)
@@ -877,6 +877,10 @@ cgraph_create_edge_1 (struct cgraph_node *caller,
if (call_stmt && caller->call_site_hash)
cgra
On Tue, Oct 15, 2013 at 2:18 AM, Richard Biener
wrote:
>
> Ok for the tailcall parts and the testcase - I'd prefer someone else to
> ack the libgcc change. CCing maintainer.
The libgcc patch is missing the updates to the comments. This code is
confusing enough as it is, having incorrect comment
On Tue, 15 Oct 2013, Marek Polacek wrote:
> Ping^2. Jason, Joseph, are you fine with the C++/C FE changes?
The C changes are fine with me.
--
Joseph S. Myers
jos...@codesourcery.com
On Mon, Oct 14, 2013 at 3:26 PM, Dehao Chen wrote:
> On Mon, Oct 14, 2013 at 2:50 PM, Jan Hubicka wrote:
>>> For my test case, the entire inline instance is optimized away, so
>>> there is no info about it in the profile. I can do some fixup in the
>>> rebuild_cgraph_edge though.
>>
>> Yep, I und
On 10/15/2013 04:56 PM, Tim Shen wrote:
This memory leak is because forgetting virtual destructor of the base
class _Executor.
Of course.
Thanks,
Paolo.
On Tue, Oct 15, 2013 at 11:01 AM, Jonathan Wakely wrote:
> Great, if it passes the testsuite please commit it.
It surely passed -m32 and -m64, and committed :)
--
Tim Shen
> -Original Message-
> From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-ow...@gcc.gnu.org] On
> Behalf Of Jakub Jelinek
> Sent: 15 October 2013 15:45
> To: Paulo Matos
> Cc: Richard Biener; Paolo Carlini; gcc-patches@gcc.gnu.org
> Subject: Re: [PATCH] tree_code_name wrapper
>
> The
Hi,
in order to fix this parsing issue with '>' between square brackets, it
seems we have simply to use the parser->greater_than_is_operator_p
mechanism. Tested x86_64-linux.
Thanks,
Paolo.
PS: Even if the testcase uses a constexpr array I don't think we need to
special case c++11 vs c++98.
Hi Tobias,
Have you checked that:
subroutine sub(a)
class(*),pointer :: a
a => null()
end subroutine
does not give an error? I think that it is why the check was introduced.
Cheers
Paul
On 13 October 2013 09:51, Tobias Burnus wrote:
> *PING*: http://gcc.gnu.org/ml/fortran/2013-10/msg00
On 15 October 2013 15:56, Tim Shen wrote:
> This memory leak is because forgetting virtual destructor of the base
> class _Executor.
>
> Thanks!
Great, if it passes the testsuite please commit it.
Thanks for the quick fix, and to Paolo for identifying the cause!
Thanks for the pointer to Honza's patch. The patch does exactly what I
need. But it only resides in the instrumentation based FDO path. Can
we move the code to more common place (like rebuild_cgraph_edges)?
Thanks,
Dehao
On Tue, Oct 15, 2013 at 7:59 AM, Teresa Johnson wrote:
> On Mon, Oct 14, 20
I'm planning to move it to ipa_profile (pass ipa-profile_estimate) and
doing it iteratively. Would that location work?
Teresa
On Tue, Oct 15, 2013 at 8:40 AM, Dehao Chen wrote:
> Thanks for the pointer to Honza's patch. The patch does exactly what I
> need. But it only resides in the instrumenta
On 10/09/2013 03:31 AM, Kirill Yukhin wrote:
> + rtx op1 = operands[1];
> + if (REG_P (op1))
> +op1 = gen_rtx_REG (V16HImode, REGNO (op1));
> + else
> +op1 = gen_lowpart (V16HImode, op1);
The IF case is incorrect. You need to use gen_lowpart always.
> +(define_insn "*avx512f_unpcklpd5
Martin put his entry in the wrong ChangeLog. (toplevel instead of
inside the gcc subdirectory).
Fixed in the obvious way.
diff --git a/ChangeLog b/ChangeLog
index 63c6cd8..0d3c199 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,9 +1,3 @@
-2013-10-15 Martin Jambor
-
- * ipa-utils.h (ipa
On 10/09/2013 03:31 AM, Kirill Yukhin wrote:
> + else if (TARGET_AVX512PF && (write || !TARGET_PREFETCH_SSE))
> +operands[2] = GEN_INT (1);
I don't believe you want the TARGET_PREFETCH_SSE check there.
That was really to select between SSE and 3dNow prefetch. If we have AVX,
we're guaranteed
On 10/11/13 14:49, David Malcolm wrote:
On Fri, 2013-10-11 at 20:45 +, Joseph S. Myers wrote:
On Fri, 11 Oct 2013, David Malcolm wrote:
On Thu, 2013-10-10 at 01:05 +, Joseph S. Myers wrote:
On Wed, 9 Oct 2013, David Malcolm wrote:
This patch adds an "--enable-host-shared" option thr
OK.
Jason
On 10/09/13 18:25, David Malcolm wrote:
My JIT branch requires embedding GCC's code as a shared library on the
host. To do requires building the host code as position-independent,
which unfortunately incurs a small speed hit.
Obviously this is an opt-in right now, so I'm not terribly concerned.
Yes, that would work. So let's discard this patch because the fix for
comdat can also fix this problem.
Thanks,
Dehao
On Tue, Oct 15, 2013 at 8:42 AM, Teresa Johnson wrote:
> I'm planning to move it to ipa_profile (pass ipa-profile_estimate) and
> doing it iteratively. Would that location work?
On 10/11/13 20:11, Zhenqiang Chen wrote:
-Original Message-
From: Jeff Law [mailto:l...@redhat.com]
Sent: Friday, October 11, 2013 1:20 PM
To: Zhenqiang Chen
Cc: gcc-patches@gcc.gnu.org
Subject: Re: [PATCH] Reassociate X == CST1 || X == CST2 if popcount (CST2
-
CST1) == 1 into ((X -
On 10/15/13 02:12, Jakub Jelinek wrote:
On Tue, Oct 15, 2013 at 03:57:23PM +0800, Zhenqiang Chen wrote:
Is it OK?
Ok, except two comments apparently still need updating.
+/* Optimize X == CST1 || X == CST2
+ if popcount (CST1 ^ CST2) == 1 into
+ (X & ~(CST1 ^ CST2)) == (CST1 & ~(CST1 ^ CS
On Tue, Oct 15, 2013 at 10:50:39AM -0600, Jeff Law wrote:
> I noticed that we're now including rtl.h and tm_p.h in
> tree-ssa-reassoc.c, which seems wrong.
Isn't that required for BRANCH_COST use?
Other option would be to add some dummy wrapper around
BRANCH_COST, put that wrapper into some file t
On 10/15/13 10:53, Jakub Jelinek wrote:
On Tue, Oct 15, 2013 at 10:50:39AM -0600, Jeff Law wrote:
I noticed that we're now including rtl.h and tm_p.h in
tree-ssa-reassoc.c, which seems wrong.
Isn't that required for BRANCH_COST use?
Other option would be to add some dummy wrapper around
BRANCH
Ian Lance Taylor writes:
> On Tue, Oct 15, 2013 at 2:18 AM, Richard Biener
> wrote:
>>
>> Ok for the tailcall parts and the testcase - I'd prefer someone else to
>> ack the libgcc change. CCing maintainer.
>
> The libgcc patch is missing the updates to the comments. This code is
> confusing eno
I have corrected the ChangeLog format, and committed this patch.
Thank you!
Cong
On Tue, Oct 15, 2013 at 6:38 AM, Richard Biener
wrote:
> On Sat, Sep 28, 2013 at 3:28 AM, Cong Hou wrote:
>> The current GCC vectorizer requires the following pattern as a simple
>> reduction computation:
>>
>>
On Tue, Oct 15, 2013 at 10:18 AM, Richard Sandiford
wrote:
> Ian Lance Taylor writes:
>> On Tue, Oct 15, 2013 at 2:18 AM, Richard Biener
>> wrote:
>>>
>>> Ok for the tailcall parts and the testcase - I'd prefer someone else to
>>> ack the libgcc change. CCing maintainer.
>>
>> The libgcc patch
On 10/15/13 02:12, Jakub Jelinek wrote:
On Tue, Oct 15, 2013 at 03:57:23PM +0800, Zhenqiang Chen wrote:
Is it OK?
Ok, except two comments apparently still need updating.
+/* Optimize X == CST1 || X == CST2
+ if popcount (CST1 ^ CST2) == 1 into
+ (X & ~(CST1 ^ CST2)) == (CST1 & ~(CST1 ^ CS
This patch add a new flag to let user to tell compiler that the
AutoFDO profile is accurate. So the compiler will assume function
without any sample is UNLIKELY_EXECUTED. This could save 10%~20% text
section size.
Bootstrapped and passed regression test.
OK for google-4_8 branch?
Thanks,
Dehao
Thanks for doing this.
Kenneth Zadeck writes:
> @@ -1204,11 +1204,11 @@ wide_int_to_tree (tree type, const wide_
> }
>
>wide_int cst = wide_int::from (pcst, prec, sgn);
> - int len = int (cst.get_len ());
> - int small_prec = prec & (HOST_BITS_PER_WIDE_INT - 1);
> + unsigned int len
Richard Sandiford writes:
> if (small_prec)
> ;
> else if (precision == xprecision)
> while (len >= 0 && val[len - 1] == -1)
> len--;
Err, len > 0 obviously.
On 09/24/13 11:49, David Malcolm wrote:
Classes containing access-control keywords such as "public:"
confuse struct_field_seq, leading it to call consume_until_eos
i.e. ignore text until after the next semicolon.
This leads to the first field after an access-control keyword
being ignored by geng
On 09/24/13 11:49, David Malcolm wrote:
Extend gengtype (and gtype.state reading/writing) so that it is able to
parse base classes in simple cases, and only attempt to do it for
GTY-marked types.
* gengtype-parse.c (require_without_advance): New.
(type): For GTY-marked types that
On 09/24/13 11:49, David Malcolm wrote:
gcc/
* doc/gty.texi (GTY Options): Add note about inheritance to
description of desc and tag.
(Inheritance and GTY): New.
So what happens if I have a class hierarchy without a gty-user marker
which violates the assumptions made by y
Treat GTY structs that have a "desc" as being the root of an inheritance
hierarchy. Generate a switch on desc within the marking function with
cases for each subclass, visiting all fields of the type (including
inherited ones).
Don't create marking functions for subclasses, instead using the bas
On 10/11/13 17:21, Marc Glisse wrote:
The gcc-specific part now. Bootstrap+testsuite on
x86_64-unknown-linux-gnu together with the libiberty patch (well,
libgomp.graphite/force-parallel-4.c failed, but that randomly happens).
2013-10-12 Marc Glisse
PR tree-optimization/58689
* syst
On 10/10/13 10:25, Jakub Jelinek wrote:
On Thu, Oct 10, 2013 at 07:26:43AM -0700, Cesar Philippidis wrote:
This patch addresses an ICE when building qemu for a Mips target in
Yocto. Both gcc-trunk, gcc-4.8 and all of the targets are potentially
affected. The problem occurs because the instructio
On Tue, 15 Oct 2013, Jeff Law wrote:
On 10/11/13 17:21, Marc Glisse wrote:
The gcc-specific part now. Bootstrap+testsuite on
x86_64-unknown-linux-gnu together with the libiberty patch (well,
libgomp.graphite/force-parallel-4.c failed, but that randomly happens).
2013-10-12 Marc Glisse
On 10/14/13 17:31, Cong Hou wrote:
Any comment on this patch?
Richi replied in the BZ you opened.
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=58508
Essentially he said emit the load on the edge rather than in the block
itself.
jeff
Marcus Shawcroft wrote:>> 2013-10-01 Marcus Shawcroft
>>
>> * configure.ac (AC_CHECK_FUNCS_ONCE): Add for exit() then make
>> existing AC_CHECK_FUNCS_ONCE dependent on outcome.
>
> Ping^2
For configure patches, I am never quite sure whether they should be
reviewed by a build
Hi Paul,
Paul Richard Thomas wrote:
Have you checked that:
subroutine sub(a)
class(*),pointer :: a
a => null()
end subroutine
does not give an error? I think that it is why the check was introduced.
I haven't checked it in particular, but was relying that some test in
the library wou
Hi Jakub, hi all,
the attached patch updated the references in libgomp to OpenMP 4.0 and
documents the new OpenMPv4 library functions and environment variables.
It does *not* update the section about the libgomp API.
Additionally, I fixes some bugs (-> omp_get_schedule, omp_set_schedule),
ad
On 10/03/13 12:24, Wei Mi wrote:
Thanks,
Wei Mi.
2013-10-03 Wei Mi
* gcc/config/i386/i386.c (memory_address_length): Extract a part
of code to rip_relative_addr_p.
(rip_relative_addr_p): New Function.
(ix86_macro_fusion_p): Ditto.
(ix86_macro_fusi
On Tue, Oct 15, 2013 at 10:13:32PM +0200, Tobias Burnus wrote:
> I used texinfo 4.14a to create the PDF and info file, which didn't
> show any warning. Also the output looks okay. (However, I believe
> some newer texinfo is picker.)
> OK for the trunk?
Thanks. Just a few nits:
> +@node omp_set_d
On Tue, 2013-10-15 at 10:29 -0600, Jeff Law wrote:
> On 10/09/13 18:25, David Malcolm wrote:
[...]
> Presumably other host libraries we depend on such as gmp, mpc, etc are
> available in shared (or at least PIC) form as well? Obviously these are
> out of our source tree and largely out of our c
On Oct 15, 2013, at 12:45 AM, Richard Biener wrote:
> On Mon, 14 Oct 2013, Mike Stump wrote:
>> On Sep 25, 2012, at 8:00 AM, Richard Guenther wrote:
>>> 2012-09-25 Richard Guenther
>>>
>>> PR lto/54625
>>> * lto-symtab.c (lto_symtab_merge_cgraph_nodes_1): Do not merge
>>> cgraph n
I noticed a testsuite entry went into gcc/ChangeLog rather than
gcc/testsuite/ChangeLog while looking to see if Paulo's recent patch had
already been reviewed & installed.
Fixed in the obvious way.
commit c7b05d9a220e67fdaaab74aa51c81d14284ed99f
Author: Jeff Law
Date: Tue Oct 15 14:43:27
Thank you for your reminder, Jeff! I just noticed Richard's comment. I
have modified the patch according to that.
The new patch is attached.
thanks,
Cong
On Tue, Oct 15, 2013 at 12:33 PM, Jeff Law wrote:
> On 10/14/13 17:31, Cong Hou wrote:
>>
>> Any comment on this patch?
>
> Richi replied i
> Yes, that would work. So let's discard this patch because the fix for
> comdat can also fix this problem.
Unforutnately ipa-profile-estimate is an IPA pass and as such it does
not have access to profile_status_to_function.
You can probably just factor this out into a function that can be called
1 - 100 of 126 matches
Mail list logo