On 20/04/16 13:12, Claudiu Zissulescu wrote:
This patch adds new instruction variants as introduced by the ARCv2
architecture.
You have used groups of 8 spaces at line starts; tabs should be used
instead for indentation.
(arc_dwarf_register_span): Remove enum keyword.
That bit sh
On 11/30/2015 03:35 AM, Bernd Schmidt wrote:
On 11/29/2015 06:14 PM, H.J. Lu wrote:
Is this safe for stage 3?
Is there a reason to do it now? This doesn't include a testcase.
Handling the proposed attribute requires extensions to the current
function_arg capabilities.
I need to go back to t
Hi,
This patch fixes PR70803 by skipping targets that don't support vect_int_mult.
It's an obvious change.
Thanks,
bin
gcc/testsuite/ChangeLog
2016-04-29 Bin Cheng
PR tree-optimization/70803
* gcc.dg/vect/pr56625.c: Require vect_int_mult.diff --git a/gcc/testsuite/gcc.dg/vec
On 12/13/2015 01:19 AM, Markus Trippelsdorf wrote:
Many developers are still using __attribute__((optimize())) in
production code, although it quite broken.
* doc/extend.texi (Common Function Attributes) [optimize]:
Discourage use of the optimize attribute.
I went back and review
On Thu, Apr 28, 2016 at 8:25 AM, Jeff Law wrote:
> On 11/30/2015 03:35 AM, Bernd Schmidt wrote:
>>
>> On 11/29/2015 06:14 PM, H.J. Lu wrote:
>>>
>>> Is this safe for stage 3?
>>
>>
>> Is there a reason to do it now? This doesn't include a testcase.
>
> Handling the proposed attribute requires exte
On 12/15/2015 10:27 AM, Kyrill Tkachov wrote:
Hi all,
This converts the preprocessor checks for WORD_REGISTER_OPERATIONS into
runtime checks
in reload.c.
Since this one is used to guard part of a large condition, I'd
appreciate it if someone
double-checks that the logic is still equivalent.
Bo
On 21/04/16 12:39, Andrew Burgess wrote:
This new iteration of the previous version is largely the same except
that I now no longer use configure time options to build in support
for nps400. Instead support controlled with a -mcpu=nps400 command
line switch. This change was made to mirror a s
Hi!
Optimizing atomic_fetch_add followed by comparison into just testing
the flags of the lock; sub is handled by a peephole2, which works usually
fine, except that for -Os we have another peephole2 that transforms
movl $-1, %reg into orl $-1, %reg and that causes the above mentioned
peephole2 not
Expanders do not have more elements in the operands array than declared
in the pattern. So, we cannot use operands[5] here. Instead just
declare and use another rtx.
Built with a cross compiler; not tested otherwise. Is this okay for trunk?
Segher
2016-04-28 Segher Boessenkool
P
On 12/15/2015 10:28 AM, Kyrill Tkachov wrote:
Hi all,
This converts the preprocessor check for WORD_REGISTER_OPERATIONS into a
runtime check
in reload1.c.
Since this one is used to guard part of a condition, I'd appreciate it
if someone
double-checks that the logic is still equivalent.
Bootstr
On Thu, 28 Apr 2016, Matthew Wahab wrote:
> Hello,
>
> The ARM target supports the half-precision floating point type __fp16
> but does not allow its use as a function return or parameter type. This
> patch removes that restriction and defines the ACLE macro
> __ARM_FP16_ARGS to indicate this. Th
Hi!
As reported in the PR and can be seen on this simplified testcase
everywhere, the FEs sometimes call operand_equal_p e.g. on a SAVE_EXPR
that contains a BIND_EXPR in it, and if arg0 == arg1, operand_equal_p
can return non-zero on it.
The ICE is because inchash::add_expr is unprepared to hash s
On various PowerPC configurations, the top frame is often mentioned
twice in the backtrace, making many asan tests fail. I see no particular
reason the asan tests want to check the frame number, so this patch
makes it check for " #. " instead of " #1 ", etc., in all of the
c-c++-common/asan tests.
The goal of this patch is to improve the location info for the "incomplete
type" error. Turned out this is isn't as trivial as it should be:
1) c_incomplete_type_error is a target hook, so I had to add the location
parameter to the other hooks, too,
2) I had to add the location parameter to siz
On 04/28/2016 09:45 AM, Segher Boessenkool wrote:
Expanders do not have more elements in the operands array than declared
in the pattern. So, we cannot use operands[5] here. Instead just
declare and use another rtx.
Built with a cross compiler; not tested otherwise. Is this okay for trunk?
On Thu, Apr 28, 2016 at 03:57:38PM +, Segher Boessenkool wrote:
> On various PowerPC configurations, the top frame is often mentioned
> twice in the backtrace, making many asan tests fail. I see no particular
> reason the asan tests want to check the frame number, so this patch
> makes it chec
On 04/28/2016 06:57 PM, Segher Boessenkool wrote:
On various PowerPC configurations, the top frame is often mentioned
twice in the backtrace, making many asan tests fail. I see no particular
reason the asan tests want to check the frame number, so this patch
makes it check for " #. " instead of
On 04/28/2016 09:57 AM, Segher Boessenkool wrote:
On various PowerPC configurations, the top frame is often mentioned
twice in the backtrace, making many asan tests fail. I see no particular
reason the asan tests want to check the frame number, so this patch
makes it check for " #. " instead of
On 28.04.2016 16:29, Richard Biener wrote:
>
> Another option would be to try if mini-gmp is enough for our
> (in-tree) use and what the performance impact would be if we'd
> use that (in-tree).
>
Yes, we would certainly never need more than that subset.
But I don't see how mpfr can be built with
On Thu, Apr 28, 2016 at 5:42 PM, Jakub Jelinek wrote:
> Hi!
>
> Optimizing atomic_fetch_add followed by comparison into just testing
> the flags of the lock; sub is handled by a peephole2, which works usually
> fine, except that for -Os we have another peephole2 that transforms
> movl $-1, %reg in
On Thu, Apr 28, 2016 at 06:03:38PM +0200, Jakub Jelinek wrote:
> On Thu, Apr 28, 2016 at 03:57:38PM +, Segher Boessenkool wrote:
> > On various PowerPC configurations, the top frame is often mentioned
> > twice in the backtrace, making many asan tests fail. I see no particular
> > reason the a
On 04/20/2016 07:48 AM, Koval, Julia wrote:
Sorry, here is the right patch.
-Original Message-
From: Koval, Julia
Sent: Wednesday, April 20, 2016 4:42 PM
To: 'gcc-patches@gcc.gnu.org'
Cc: Lu, Hongjiu ; 'vaalfr...@gmail.com' ;
'ubiz...@gmail.com' ; 'l...@redhat.com' ; Zamyatin, Igor
S
No functional changes.
2016-04-28 Uros Bizjak
* config/i386/i386.md (zeroing peephole2): Use general_reg_operand.
(or $-1,reg peephole2): Ditto.
(strict_low_part zeroing peephole2): Use SWI12 mode iterator.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
Committed
On 04/20/2016 03:02 AM, Richard Biener wrote:
On Tue, Apr 19, 2016 at 7:50 PM, Patrick Palka wrote:
This patch makes the jump threader look through the BIT_AND_EXPRs and
BIT_IOR_EXPRs within a condition so that we could find dominating
ASSERT_EXPRs that could help make the overall condition eva
On 28/04/16 16:31, Joern Wolfgang Rennecke wrote:
However, setting defaults and multilib sets at gcc configure time is
also quite useful, as otherwise every user is confronted with building
multilibs for a burgeoning array of variants.
P.S.: One way to do this is to add introduce a new macro
bb_seen is not used outside of tracer.c. Committing as trivial.
2016-04-28 Segher Boesssenkool
* tracer.c (bb_seen): Make static.
---
gcc/tracer.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/gcc/tracer.c b/gcc/tracer.c
index 56788a2..477d8b3 100644
--- a/gcc
Please find the updated patch.
Claudiu
gcc/
2016-04-28 Claudiu Zissulescu
* config/arc/arc.h (UNSIGNED_INT12, UNSIGNED_INT16): Define.
* config/arc/arc.md (umulhisi3): Use arc_short_operand predicate.
(umulhisi3_imm): Update predicates and constraint letters.
(
On 21/04/16 12:39, Andrew Burgess wrote:
* config/arc/arc.md (*loadqi_update): Replace use of 'rI'
constraint with separate 'r' and 'Cm2' constraints.
Why don't you use simply rCm2 ?
On 21/04/16 12:39, Andrew Burgess wrote:
The nps400 is an arc700 with a set of extension instructions produced by
Mellanox (formally EZChip). This commit adds support for the nps400
architecture to the arc backend.
After this commit it is possible to compile using -mcpu=nps400 in order
to spe
On 21/04/16 12:39, Andrew Burgess wrote:
* config/arc/constraints.md (Usd): Convert to define_constraint.
(Us<): Likewise.
(Us>): Likewise.
Thanks. I have allied this patch.
In r193072 sbitmap_popcount was removed, so we cannot ask for the popcount
of an sbitmap anymore. Nothing calls sbitmap_alloc_with_popcount either.
This patch removes everything else popcount-related from sbitmap.
Tested on powerpc64-linux; is this okay for trunk?
Segher
2016-04-28 Segher Bo
On 04/28/2016 01:56 AM, Jakub Jelinek wrote:
> On Wed, Apr 27, 2016 at 07:37:17PM -0700, Cesar Philippidis wrote:
>> This patch replaces all of the bool argument to c_finish_omp_clauses and
>> finish_omp_clauses in the c and c++ front ends, respectively. Right now
>> there are three bool arguments,
On 28/04/16 18:10, Claudiu Zissulescu wrote:
Please find the updated patch.
Claudiu
gcc/
2016-04-28 Claudiu Zissulescu
* config/arc/arc.h (UNSIGNED_INT12, UNSIGNED_INT16): Define.
* config/arc/arc.md (umulhisi3): Use arc_short_operand predicate.
(umulhisi3_imm): Up
This makes it clearer this field is only for SPE. Committing.
Segher
2016-04-28 Segher Boessenkool
* config/rs6000/rs6000.c (machine_function): Rename
insn_chain_scanned_p to spe_insn_chain_scanned_p.
(rs6000_stack_info): Adjust.
---
gcc/config/rs6000/rs6000.c | 8
Kyrill Tkachov wrote:
> On 25/04/16 20:21, Wilco Dijkstra wrote:
> > The GCC switch expansion is awful, so
> > even with a good indirect predictor it is better to use conditional
> > branches.
>
> In what way is it awful? If there's something we can do better at
> can you file a bug report with a
On 04/28/2016 07:30 PM, Segher Boessenkool wrote:
In r193072 sbitmap_popcount was removed, so we cannot ask for the popcount
of an sbitmap anymore. Nothing calls sbitmap_alloc_with_popcount either.
This patch removes everything else popcount-related from sbitmap.
Tested on powerpc64-linux; is t
Thank you,
Here is the repost.
Update TARGET_FUNCTION_INCOMING_ARG documentation
On x86, interrupt handlers are only called by processors which push
interrupt data onto stack at the address where the normal return address
is. Since interrupt handlers must access interrupt data via po
Thanks. I have merged this patch.
On 16-04-28 15:14:20, Jakub Jelinek wrote:
> On Thu, Apr 28, 2016 at 03:10:26PM +0200, Bernd Schmidt wrote:
> > On 04/28/2016 12:35 PM, Jakub Jelinek wrote:
> > >On Thu, Apr 28, 2016 at 12:31:40PM +0200, Bernd Schmidt wrote:
> > >>I really don't see anything in that function that looks like a huge
On 01/18/2016 10:22 AM, Richard Biener wrote:
See also https://gcc.gnu.org/bugzilla/show_bug.cgi?id=52171 - the
inline expansion
for small sizes and equality compares should be done on GIMPLE. Today the
strlen pass might be an appropriate place to do this given its
superior knowledge
about strin
On 21/04/16 12:39, Andrew Burgess wrote:
Add support for nps400 bit operation instructions. There's a new flag
-mbitops that turns this feature on. There are new instructions, some
changes to existing instructions, a new register class to support the
new instructions, and some new expand and
I'm sorry I'm a little late but I have a couple of minor comments
on the patch:
+ epoch = strtoll (source_date_epoch, &endptr, 10);
+ if ((errno == ERANGE && (epoch == LLONG_MAX || epoch == LLONG_MIN))
+ || (errno != 0 && epoch == 0))
+fatal_error (UNKNOWN_LOCATION, "environment varia
C++17 adds a [[nodiscard]] attribute, which is similar to the GNU
warn_unused_result attribute, except that it applies to function
declarations and class/enum types rather than to function types and is
suppressed by an explicit conversion to void. I considered just treating
it as warn_unused_re
On 04/25/2016 01:28 PM, Jason Merrill wrote:
There doesn't seem to be any need to step through the is-a inline
functions. OK for trunk?
Likewise line-map.h.
Jason
commit 59d1faf71bda301f6ac608534d3a7208f99cb8a5
Author: Jason Merrill
Date: Mon Apr 25 21:45:33 2016 -0400
* gdbinit.in:
Thanks. I have applied this patch.
This patch to the Go frontend by Chris Manghane makes
String_index_expression accessible outside of expressions.cc. This is
a step toward future patches. Bootstrapped and ran Go testsuite on
x86_64-pc-linux-gnu. Committed to mainline.
Ian
Index: gcc/go/gofrontend/MERGE
=
Hi!
The PR reported one ICE caused by a builtin for __x86_64__ guarded
intrinsics to be mistakenly available in -m32 too, I've looked for
INT64 substrings in the various i386.c builtin tables and for each
that has been missing OPTION_MASK_ISA_64BIT in the mask looked at
whether the uses of the bui
On 21/04/16 12:39, Andrew Burgess wrote:
* gcc.target/arc/nps400-1.c: New file.
Thanks. I have applied this patch.
Hello!
Attached patch extends TARGET_READ_MODIFY{,_WRITE} peepholes to handle
all integer modes, while also taking care not to introduce additional
QImode register stalls.
While looking at the insn enable condition, I noticed that we don't
use "probe_stack" pattern any more, as the stack check lo
On Thu, Apr 28, 2016 at 9:14 PM, Jakub Jelinek wrote:
> Hi!
>
> The PR reported one ICE caused by a builtin for __x86_64__ guarded
> intrinsics to be mistakenly available in -m32 too, I've looked for
> INT64 substrings in the various i386.c builtin tables and for each
> that has been missing OPTIO
On Thu, Apr 28, 2016 at 11:22 AM, Yulia Koval wrote:
> Thank you,
> Here is the repost.
>
> Update TARGET_FUNCTION_INCOMING_ARG documentation
>
> On x86, interrupt handlers are only called by processors which push
> interrupt data onto stack at the address where the normal return address
>
This PR was fixed earlier with a patch that was deemed safe for GCC 6
through the removal of splitters for zero extend and sign extend to
HImode.
Now that trunk has opened for GCC 7 development, the following patch
restores the splitters and fixes the bug in the more aggressive manner
originally p
On 03/22/2016 03:37 AM, Richard Biener wrote:
On Mon, Mar 21, 2016 at 9:32 PM, Jeff Law wrote:
On 03/21/2016 01:10 PM, Bernd Schmidt wrote:
On 03/21/2016 08:06 PM, Jeff Law wrote:
As noted last week, find_removable_extensions initializes several
bitmaps, but doesn't clear them.
This is no
This patch by Chris Manghane marks concurrent calls in the Go
frontend. This is a small change that prepares for future patches.
Bootstrapped and ran Go testsuite on x86_64-pc-linux-gnu. Committed
to mainline.
Ian
Index: gcc/go/gofrontend/MERGE
===
Hi,
only when Jakub bumped some bugs in preparation for the release I noted
that this one remained assigned to me for way too much time...
Roughly speaking, the problem is caused by the fact that when we have a
GNU anonymous struct inside a union the fields are flattened out and
appear to be
On Wed, 27 Apr 2016, Richard Biener wrote:
--- trunk4/gcc/fold-const.h (revision 235452)
+++ trunk4/gcc/fold-const.h (working copy)
@@ -13,20 +13,22 @@ WARRANTY; without even the implied warra
FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License
for more details.
You
Otherwise, I'd suggest using a traditional integer letter. 'J' is free.
Thanks for the suggestion, I will use 'J'.
Why do you remove half of the indentation?
Unwanted reformatting, sorry for this, I will revert it.
I have the feeling you are happy with my new patch. Is there anything to
be
On 28/04/16 21:31, Claudiu Zissulescu wrote:
Otherwise, I'd suggest using a traditional integer letter. 'J' is free.
Thanks for the suggestion, I will use 'J'.
Why do you remove half of the indentation?
Unwanted reformatting, sorry for this, I will revert it.
I have the feeling you are h
In GCC 6.x, I was not able to get complex __float128 to work before the cut off
period for stage1 submissions. This patch enables complex support for PowerPC
__float128. Note, it does not address adding complex support in libgcc.
Note, similar to the x86_64, you cannot say:
typedef _Compl
I would expect this to cause a false negative on a union of two
anonymous structs, both of which have initialized members.
I think better would be to have a local any_default_members rather than
passing the same pointer through all levels.
Also, you can look at 'type' rather than DECL_CONTEXT
On 03/31/2016 07:00 AM, Andrew MacLeod wrote:
Another potential buglet I stumbled across whilst testing the tree-type
work:
in c/c-array-notation.c::fix_builtin_array_notation_fn()
<...>
if (list_size > 1)
{
new_yes_ind = build_modify_expr
(location, *new_var,
As Andrew pointed out a while ago, build_modify_expr's last argument is
supposed to be a tree type. However in two cases within
fix_builtin_array_notation_fn we end passing down an tree expression.
Thankfully in the contexts were this happens, the last argument isn't
actually used, so it ha
On Thu, Apr 28, 2016 at 05:06:14PM -0400, Michael Meissner wrote:
Hi Mike,
> * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Add
> support for __float128 complex datatypes.
> (rs6000_hard_regno_mode_ok): Likewise.
> (rs6000_setup_reg_addr_masks): Likewise.
>
On 04/28/2016 09:50 AM, Jakub Jelinek wrote:
Hi!
As reported in the PR and can be seen on this simplified testcase
everywhere, the FEs sometimes call operand_equal_p e.g. on a SAVE_EXPR
that contains a BIND_EXPR in it, and if arg0 == arg1, operand_equal_p
can return non-zero on it.
The ICE is be
On 04/28/2016 02:49 AM, Richard Biener wrote:
The following prototype patch re-uses cc1-checksum.c from the
previous stage when compiling stage-final. This eventually
allows to compare cc1 from the last two stages to fix the
lack of a true comparison when doing LTO bootstrap (it
compiles LTO by
On 04/26/2016 05:59 PM, Martin Sebor wrote:
PR c++/66639 asked to declare __func__ , __FUNCTION__ and
__PRETTY_FUNCTION__ as constexpr​. With the request fulfilled
sometime in the 6.0 time frame (possibly as a result of fixing
c++/70353), the attached patch implements the corresponding
change su
Ping.
Cesar
On 04/15/2016 02:30 PM, Cesar Philippidis wrote:
> This patch makes the c, c++ and fortran FEs duplicate the reduction
> clauses in a combined 'acc parallel loop' directive when it splits that
> directive into separate parallel and loop directives. So given something
> like
>
> #pr
This patch fixes PR tree-optimization/51513, namely the generation of
wild branches due to switch case statements that only contain calls to
__builtin_unreachable(). For example, compiling using -O2 -fjump-tables
--param case-values-threshold=1 (to easily expose the bug), we see:
switch (which)
On Thu, Apr 28, 2016 at 05:10:07PM -0500, Segher Boessenkool wrote:
> On Thu, Apr 28, 2016 at 05:06:14PM -0400, Michael Meissner wrote:
> Hi Mike,
>
> > * config/rs6000/rs6000.c (rs6000_hard_regno_nregs_internal): Add
> > support for __float128 complex datatypes.
> > (rs6000_hard_regno
+/* Fold __builtin_FILE() to a constant string. */
NIT: When we refer to functions, we don't have the trailing(). So drop
it from the comment.
+
+/* Fold __builtin_FUNCTION() to a constant string. */
Similarly.
+
+/* Fold __builtin_LINE() to an integer constant. */
Similarly.
Okay, I'
On Thu, Apr 28, 2016 at 12:52 PM, Jeff Law wrote:
> On 04/20/2016 03:02 AM, Richard Biener wrote:
>>
>> On Tue, Apr 19, 2016 at 7:50 PM, Patrick Palka
>> wrote:
>>>
>>> This patch makes the jump threader look through the BIT_AND_EXPRs and
>>> BIT_IOR_EXPRs within a condition so that we could find
Hi Jason,
On 28/04/2016 23:45, Jason Merrill wrote:
I would expect this to cause a false negative on a union of two
anonymous structs, both of which have initialized members.
I think better would be to have a local any_default_members rather
than passing the same pointer through all levels.
I had to break this patch up into multiple edits.
Here is the one for ``libstdc++-v3/config''.
--
Chris Gregory
Index: libstdc++-v3/config/abi/compatibility.h
===
--- libstdc++-v3/config/abi/compatibility.h (revision 235619)
+++ li
Hi,
The lack of certain swap optimizations added in GCC 6 has shown up as a
performance issue in some customer code, where the customer is unable to
move off of GCC 5. To accommodate this, I would like to backport these
changes to GCC 5. They have all been burned in on trunk for many
months. Th
On Wed, Apr 27, 2016 at 01:30:54PM -0500, Bill Schmidt wrote:
> 2016-04-27 Bill Schmidt
>
> * config/rs6000/altivec.h: Change definitions of vec_xl and
> vec_xst.
> * config/rs6000/rs6000-builtin.def (LD_ELEMREV_V2DF): New.
> (LD_ELEMREV_V2DI): New.
> (LD_ELEMREV_V
Thanks,
Here is the patch. Is it ok?
Update TARGET_FUNCTION_INCOMING_ARG documentation
On x86, interrupt handlers are only called by processors which push
interrupt data onto stack at the address where the normal return address
is. Since interrupt handlers must access interrupt d
I refactored out all the macros at the end of ``gcc/config/i386.c'' into
``gcc/config/i386-target-macros.c''. Included are the diff on
``i386.c'' and the new file.
--
Chris Gregory
2016-04-28 Chris Gregory
* config/i386/i386.c: Move TARGET macros to separate file (and
#inclu
Adding the following to the beginning of ``i386-target-macros.c'' would
be appreciated:
/* Macros used for code generation on IA-32. Only used in i386.c
Copyright (C) 1988-2016 Free Software Foundation, Inc.
This file is part of GCC.
GCC is free software; you can redistribute it and/or modif
Hello!
2016-04-29 Uros Bizjak
* config/i386/i386.md (Load+RegOp to Mov+MemOp peephole2):
Use SWI mode iterator. Use general_reg_operand predicate.
(Load+RegOp to Mov+MemOp peephole2 with vector regs): Split
peephole to MMX and SSE part. Use mmx_reg_operand and sse_reg_operand
101 - 179 of 179 matches
Mail list logo
