On Wed, Mar 18, 2015 at 08:00:33PM +0100, Uros Bizjak wrote:
> 2015-03-18 Uros Bizjak
>
> PR rtl-optimization/60851
> * recog.c (constrain_operands): Accept a pseudo register before reload
> for LRA enabled targets.
>
> --- recog.c (revision 221493)
> +++ recog.c (working copy)
On March 18, 2015 11:36:09 PM GMT+01:00, Jakub Jelinek wrote:
>Hi!
>
>TSAN_FUNC_EXIT internal function is special, because we drop it during
>inlining, so for fnsplit we need to be careful about it, otherwise we
>can
>end up with unbalanced pairs of tsan entry/exit marker functions.
>
>This patch
On 12 Mar 13:09, Ilya Enkovich wrote:
> Hi,
>
> Instrumented function pointer may be propagated into not instrumented
> indirect call and vice versa. It requires additional call modifications
> (either remove bounds or change callee). Bootstrapped and tested on
> x86_64-unknown-linux-gnu. OK
On 12 Mar 13:27, Ilya Enkovich wrote:
> Hi,
>
> Currently cgraph merge has several issues with instrumented code:
> - original function node may be removed => no assembler name conflict is
> detected between function and variable
> - only orig_decl name is privatized for instrumented function =
On Thu, Mar 19, 2015 at 8:44 AM, Jakub Jelinek wrote:
> On Wed, Mar 18, 2015 at 08:00:33PM +0100, Uros Bizjak wrote:
>> 2015-03-18 Uros Bizjak
>>
>> PR rtl-optimization/60851
>> * recog.c (constrain_operands): Accept a pseudo register before reload
>> for LRA enabled targets.
> As
On Wed, Mar 18, 2015 at 6:02 PM, Tom de Vries wrote:
> On 18-03-15 12:18, Richard Biener wrote:
>>
>> On Wed, Mar 18, 2015 at 12:03 PM, Tom de Vries
>> wrote:
>>>
>>> On 18-03-15 11:16, Richard Biener wrote:
On Fri, Mar 13, 2015 at 4:28 PM, Tom de Vries
wrote:
>
>
>>>
Hi,
Currently when we clone alias we also clone its target and create alias
reference during alias target cloning. This doesn't work in case alias clone
is removed and later is re-created. This patch moves alias reference creation
to fix that. Bootstrapped and tested on x86_64-unknown-linux-
On Wed, Mar 18, 2015 at 11:35 PM, Sebastian Pop wrote:
> Hi,
>
> the attached patch fixes PR 65177 in which the code generator of FSM jump
> thread
> would create a diamond on the copied path: see https://gcc.gnu.org/PR65177#c18
> for a detailed description.
>
> The patch is renaming SEME into ju
Hi,
There were some discussion about "x" constraints being too conservative
for some patterns in i386.md.
Patch below fixes it. This is probably stage1 material.
ChangeLog:
gcc/
2015-03-19 Ilya Tocar
* config/i386/i386.h (EXT_SSE_REG_P): New.
* config/i386/i386.md (*cmpi_mix
Hi!
This patch fixes the other part of the PR. On this testcase, ICF creates
a thunk and sets gimple_call_set_tail true on the call in there. No
TSAN_FUNC_EXIT internal call is added, which isn't a big deal, tsan pass
has code to add it if there are none. But, nothing was clearing the tail
call
On Thu, 19 Mar 2015, Jakub Jelinek wrote:
> Hi!
>
> This patch fixes the other part of the PR. On this testcase, ICF creates
> a thunk and sets gimple_call_set_tail true on the call in there. No
> TSAN_FUNC_EXIT internal call is added, which isn't a big deal, tsan pass
> has code to add it if t
On Wed, 18 Mar 2015, Richard Biener wrote:
> On March 18, 2015 4:59:30 PM GMT+01:00, Alan Lawrence
> wrote:
> >Following this patch (r221318), we're seeing what appears to be a
> >miscompile of
> >glibc on AArch64. This causes quite a bunch of tests to fail, segfaults
> >etc., if
> >LD_LIBRARY
On 18-03-15 18:25, Jakub Jelinek wrote:
On Wed, Mar 18, 2015 at 06:21:51PM +0100, Tom de Vries wrote:
this patch fixes PR65458.
The patch marks omp thread functions as parallelized, which means the
parloops pass no longer attempts to modify that function.
Bootstrapped and reg-tested on x86_64.
On 18-03-15 18:22, Tom de Vries wrote:
Hi,
this patch fixes PR65460.
The patch marks offloaded functions as parallelized, which means the parloops
pass no longer attempts to modify that function.
Updated patch to postpone mark_parallelized_function until the corresponding
cgraph_node is avai
The following patch re-orders elements to avoid 8 bytes of padding.
Bootstrapped on x86_64-unknown-linux-gnu, ok?
Thanks,
Richard.
2015-03-19 Richard Biener
* partition.h (struct partition_elem): Re-order elements to
avoid padding.
Index: include/partition.h
===
On Thu, Mar 19, 2015 at 12:02:01PM +0100, Tom de Vries wrote:
> +void
> +mark_parallelized_function (tree fndecl)
> +{
> + cgraph_node *node = cgraph_node::get (fndecl);
> + gcc_assert (node != NULL);
> + node->parallelized_function = 1;
> }
I'm not convinced we need this wrapper, I'd just use
On Thu, Mar 19, 2015 at 12:09:10PM +0100, Richard Biener wrote:
>
> The following patch re-orders elements to avoid 8 bytes of padding.
>
> Bootstrapped on x86_64-unknown-linux-gnu, ok?
>
> Thanks,
> Richard.
>
> 2015-03-19 Richard Biener
>
> * partition.h (struct partition_elem): Re-
On 19-03-15 12:11, Jakub Jelinek wrote:
On Thu, Mar 19, 2015 at 12:02:01PM +0100, Tom de Vries wrote:
+void
+mark_parallelized_function (tree fndecl)
+{
+ cgraph_node *node = cgraph_node::get (fndecl);
+ gcc_assert (node != NULL);
+ node->parallelized_function = 1;
}
I'm not convinced we
On Thu, Mar 19, 2015 at 12:27:04PM +0100, Tom de Vries wrote:
> Sure, I can update that, I'll retest and repost.
Yes, please. Probably the tree-parloops.h include will not be needed either
then.
> Indeed, it's not done here, but it is still done, only later.
>
> The function we create in parloo
Hi Dominique, Hi all,
please find attached a new version of the patch to fix pr64787 after processing
Dominique's comments. Thank you very much for your work, Dominique.
The patch now also fixes:
pr63230 - allocation of deferred length character as derived type component
causes internal compiler
On Wed, Mar 18, 2015 at 9:25 PM, Xinliang David Li wrote:
> get_local_tls_init_fn can be called from other contexts other than
> 'handle_tls_init' -- is the added new assertion safe ?
In fact there is still an issue here, for file static TLS vars. Will
work on a fix and send the original test cas
Hi all,
This patches fixes the ICE reported at
https://gcc.gnu.org/ml/gcc-patches/2015-03/msg00949.html
The problem is that gcse generates a (set (reg:OI) (const_int 0)) that
doesn't match anything
in the arm backend. That SET was created through processing an insn
generated by the neon_movoi
On Thu, Mar 19, 2015 at 01:45:19PM +, Kyrill Tkachov wrote:
> Bootstrapped and tested on arm, x86, aarch64.
> This ICE doesn't happen with 4.9 and 4.8 so it's only a regression for GCC
> 5.
> The currently ICE'ins testcase passes now, so no new testcase is added.
Not an expert on this, but it
On Thu, Mar 19, 2015 at 2:45 PM, Kyrill Tkachov wrote:
> As pointed out by James Greenhalgh offline the correct thing would have been
> to do an
> emit_move_insn to let the backend expanders do the right thing (especially
> in the concerned
> testcase gcc.c-torture/execute/pr65427.c that uses 256-b
Hi,
When thunk is cloned, its callee is temporarily set to original thunks's target
which is later also cloned. Thus we should be more careful when clonning
node's thunks. Bootstrapped and tested on x86_64-unknown-linux-gnu. Applied
to trunk.
Thanks,
Ilya
--
2015-03-19 Ilya Enkovich
Hi all,
This patch fixes PR 65358. For details look at the excellent write-up
by Honggyu in bugzilla. The problem is that we're trying to pass a struct
partially on the stack and partially in regs during a tail-call optimisation
but the struct we're passing is also a partial incoming arg though t
On 19/03/15 13:56, Steven Bosscher wrote:
On Thu, Mar 19, 2015 at 2:45 PM, Kyrill Tkachov wrote:
As pointed out by James Greenhalgh offline the correct thing would have been
to do an
emit_move_insn to let the backend expanders do the right thing (especially
in the concerned
testcase gcc.c-tortu
On 19/03/15 13:56, Steven Bosscher wrote:
On Thu, Mar 19, 2015 at 2:45 PM, Kyrill Tkachov wrote:
As pointed out by James Greenhalgh offline the correct thing would have been
to do an
emit_move_insn to let the backend expanders do the right thing (especially
in the concerned
testcase gcc.c-tortu
Hi all,
please find attached the parts missing to stop valgrind's complaining about the
use of uninitialized memory. The issue was, that when constructing a temporary
class-object to call a routine with unlimited polymorphic arguments, the _len
component was never set. This is fixed by this patch
Second ping.
Also, have attached updated patchset for the current gcc. Thanks.
--
James Bowman
FTDI Open Source Liaison
From: Joseph Myers [jos...@codesourcery.com]
Sent: Tuesday, February 17, 2015 2:06 AM
To: James Bowman
Cc: gcc-patches@gcc.gnu.org
Subj
Hello Jakub,
Jakub Jelinek writes:
> __has_{cpp_,}attribute builtin macros are effectively function-like macros
> taking one argument (and the ISO preprocessor expands macros in the argument
> which is IMHO desirable), but the traditional preprocessor has been crashing
> on them or reporting err
Hello.
Following patch is fix for the PR. Problem is caused if we fill up
cgraph_thunk_info with some values (e.g. virtual_value != 0)
and further analysis set thunk_p = false. In all situations IPA ICF needs to
reset all fields of the struct as it sets
thunk_p = true.
Tested on x86_64-linux-p
On Thu, Mar 19, 2015 at 06:08:03PM +0100, Martin Liška wrote:
> >From 1b0416658cf59348664d44b14518c994075fd9bd Mon Sep 17 00:00:00 2001
> From: mliska
> Date: Thu, 19 Mar 2015 15:36:34 +0100
> Subject: [PATCH] Fix for PR ipa/65465.
>
> gcc/ChangeLog:
>
> 2015-03-19 Martin Liska
>
> PR
On 03/19/2015 06:13 PM, Jakub Jelinek wrote:
On Thu, Mar 19, 2015 at 06:08:03PM +0100, Martin Liška wrote:
>From 1b0416658cf59348664d44b14518c994075fd9bd Mon Sep 17 00:00:00 2001
From: mliska
Date: Thu, 19 Mar 2015 15:36:34 +0100
Subject: [PATCH] Fix for PR ipa/65465.
gcc/ChangeLog:
2015-03-1
Hello.
This is fix for the issue as introduced by Honza. I've just finished testing on
x86_64-linux-pc and
the patch is pre-approved by Honza.
Thanks,
Martin
>From f3c845d74b7a1edbbba9ecefa081feb58700d515 Mon Sep 17 00:00:00 2001
From: mliska
Date: Thu, 19 Mar 2015 10:52:44 +0100
Subject: [PAT
On 03/19/2015 02:08 AM, Richard Biener wrote:
On Wed, Mar 18, 2015 at 10:28 PM, Jason Merrill wrote:
If you move the call to rest_of_decl_compilation we could go through and
prune the debug info for unused decls at dwarf2out_finish time, the way we
do with unused types.
True. Note that the v
On Wed, Mar 18, 2015 at 11:08:15AM +0100, Richard Biener wrote:
> On Fri, Mar 13, 2015 at 3:40 PM, Marek Polacek wrote:
> > We started to reject this (IMHO valid) testcase with r214941 that did away
> > with
> > try_move_mult_to_index -- meaning that we are no longer able to fold
> > *(&s[0] + 1
On Thu, Mar 19, 2015 at 07:05:36PM +0100, Jakub Jelinek wrote:
> On Wed, Mar 18, 2015 at 11:08:15AM +0100, Richard Biener wrote:
> > On Fri, Mar 13, 2015 at 3:40 PM, Marek Polacek wrote:
> > > We started to reject this (IMHO valid) testcase with r214941 that did
> > > away with
> > > try_move_mul
On Thu, Mar 19, 2015 at 07:13:47PM +0100, Marek Polacek wrote:
> On Thu, Mar 19, 2015 at 07:05:36PM +0100, Jakub Jelinek wrote:
> > I believe cxx_fold_indirect_ref result is not passed through to the
> > middle-end, unless it can be folded into a constant.
> >
> > Though, a question is if we do (o
On Thu, Mar 19, 2015 at 07:17:23PM +0100, Jakub Jelinek wrote:
> On Thu, Mar 19, 2015 at 07:13:47PM +0100, Marek Polacek wrote:
> > On Thu, Mar 19, 2015 at 07:05:36PM +0100, Jakub Jelinek wrote:
> > > I believe cxx_fold_indirect_ref result is not passed through to the
> > > middle-end, unless it ca
This patch makes some significant changes to attribute abi_tag.
First, it allows explicit naming of tags on inline namespaces, which
previously always had a tag with the same name as the namespace itself;
this is still the default if no tag is specified.
It also introduces automatic tagging o
Hi.
Following patch improves performance by adding hash values of references that
are not
candidates for IPA ICF. Tested on x86_64-linux-pc w/o any new regression
observed.
Ready for trunk?
Thanks,
Martin
>From 6c04cc4283d08a8aa9829574dd91579a918cb508 Mon Sep 17 00:00:00 2001
From: marxin
Dat
On 03/19/2015 02:03 PM, Aldy Hernandez wrote:
I have moved the debug early generation for _symbols_ to
rest_of_decl_compilation
I think you mean "variables". Functions are also symbols. :)
Other than that, makes sense to me.
Jason
Richard Biener wrote:
> please instead fixup after copy_bbs in duplicate_seme_region.
>
Thanks for the review.
Attached patch that does not modify copy_bbs.
Fixes make check in hmmer and make check RUNTESTFLAGS=tree-ssa.exp
Full bootstrap and regtest in progress on x86_64-linux. Ok for trunk?
>
The following patch fixes
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=63491
The patch was bootstrapped on x86-64, power64, and aarch64.
Committed as rev. 221552.
2015-03-19 Vladimir Makarov
PR rtl-optimization/63491
* lra-constraints.c (check_and_process_move): Use src i
On 03/19/2015 12:43 PM, Jason Merrill wrote:
On 03/19/2015 02:03 PM, Aldy Hernandez wrote:
I have moved the debug early generation for _symbols_ to
rest_of_decl_compilation
I think you mean "variables". Functions are also symbols. :)
Oops, yes I did :).
I will adjust the comments in my p
> diff --git a/gcc/ipa-icf.c b/gcc/ipa-icf.c
> index f68d23c..b8e3aa4 100644
> --- a/gcc/ipa-icf.c
> +++ b/gcc/ipa-icf.c
> @@ -557,6 +557,69 @@ sem_function::equals_wpa (sem_item *item,
>return true;
> }
>
> +/* Update hash by address sensitive references. */
> +
> +void
> +sem_item::updat
On 18/03/2015 19:16, Jonathan Wakely wrote:
Preparing this patch reminded me that we currently have two copies of
the Catalog_info and Catalogs code in the unnamed namespace in
config/locale/gnu/messages_members.cc, one using the old string and
one using the new. We should really alter the code
On 03/13/2015 05:42 AM, Jan Hubicka wrote:
2015-03-09 Martin Liska
* config/i386/i386.c (def_builtin): Collect union of all
possible masks.
(ix86_add_new_builtins): Do not iterate over all builtins
in cases that isa value has no intersection with possible masks
[ was: Re: [PATCH, stage1] Make parloops gate more strict ]
On 19-03-15 10:00, Richard Biener wrote:
Yeah - it makes the -fdump-passes "hack" more pervasive throughout
>>the compiler.
>>
>>I suppose it should instead build & push a "dummy" sturct function.
>>
>
>Like this?
Looks good to me.
New patch below. Passes regression tests plus internal application build.
2015-03-19 Teresa Johnson
gcc/
Google ref b/19618364.
* cp/decl2.c (get_local_tls_init_fn): Assert on aux module.
(get_tls_init_fn): Promote non-public init functions if necessary
does generate_tls_wrapper also need to be suppressed for aux module?
David
On Thu, Mar 19, 2015 at 4:11 PM, Teresa Johnson wrote:
> New patch below. Passes regression tests plus internal application build.
>
> 2015-03-19 Teresa Johnson
>
> gcc/
> Google ref b/19618364.
>
> Hi.
>
> Explanation of the patch is introduced.
thanks
>
> >>
> >>mask &= ~OPTION_MASK_ISA_64BIT;
> >>if (mask == 0
> >>@@ -30670,6 +30673,14 @@ def_builtin_const (HOST_WIDE_INT mask, const char
> >>*name,
> >> static void
> >> ix86_add_new_builtins (HOST_WIDE_INT isa)
> >>
On Thu, Mar 19, 2015 at 4:45 PM, Xinliang David Li wrote:
> does generate_tls_wrapper also need to be suppressed for aux module?
No, we generate the wrapper in the aux module and use it to access the
TLS variable and invoke it's init function (which is defined in the
variable's own module). Presu
ok -- then there is an over assertion in get_local_tls_init_fn. The
method can be called for aux module -- the decl created will also be
promoted.
Also can we rely on late promotion (special case of artificial
function __tls_init? This can avoid duplicated logic here.
David
On Thu, Mar 19, 2015
On Thu, Mar 19, 2015 at 8:00 PM, Xinliang David Li wrote:
> ok -- then there is an over assertion in get_local_tls_init_fn. The
> method can be called for aux module -- the decl created will also be
> promoted.
No, it won't ever be called for an aux module. Both callsites are
guarded (handle_tls_
On Thu, Mar 19, 2015 at 9:57 PM, Teresa Johnson wrote:
> On Thu, Mar 19, 2015 at 8:00 PM, Xinliang David Li wrote:
>> ok -- then there is an over assertion in get_local_tls_init_fn. The
>> method can be called for aux module -- the decl created will also be
>> promoted.
>
> No, it won't ever be c
Hi,
this patch fixes accounting error in inliner growth calculation. We have
CIF_FINAL_NORMAL and CIF_FINAL_ERROR inline_failed codes. The FINAL_ERROR
means that it can not be revisited and inlining will fail.
Problem is that CIF_FINAL_ERROR is set at link time and not revisited
during LTO mergi
Hi,
this patch fixes ICE in where ODR violation merge non-polymorphic and
polymorphic type.
I will keep the PR open as I would like to improve the ODR violation warning
Bootstrapped/regtested x86_64-linux, will commit it tomorrow.
Honza
PR ipa/65475
* ipa-devirt.c (add_type_dupl
59 matches
Mail list logo