Hi,
this was already posted in february, but has not yet been reviewed, so I thought
it is time now to post it again...
As a follow-up for Jakub's c/69669 fix, I'd like to have the enum mode also
honored
in C++ code, because the mode attribute now finally really works in C, but it is
completely
On Sat, 2016-04-30 at 10:58 +0900, Oleg Endo wrote:
> On Fri, 2016-04-29 at 23:11 +0900, Oleg Endo wrote:
> > On Fri, 2016-04-29 at 19:45 +0900, Oleg Endo wrote:
> > > On Thu, 2016-04-28 at 10:27 +0900, Oleg Endo wrote:
> > >
> > > > The removal of SH5 support from GCC has been announced here
> >
On 04/29/2016 02:58 AM, Rainer Orth wrote:
Having a listed libcilkrts maintainer would probably help, too ;-)
That's high on my list to make happen :-) Until then Ilya Verbin and I
will work together the Cilk runtime.
jeff
On Fri, Apr 29, 2016 at 03:32:53PM +0200, Bernd Schmidt wrote:
> On 04/29/2016 03:02 PM, David Edelsohn wrote:
> >As an aside, this change seems to be the source of a new code
> >generation bug affecting the PPC kernel.
>
> Please file a PR.
-frename-registers merely exposed a pre-existing kernel
On Fri, 2016-04-29 at 23:11 +0900, Oleg Endo wrote:
> On Fri, 2016-04-29 at 19:45 +0900, Oleg Endo wrote:
> > On Thu, 2016-04-28 at 10:27 +0900, Oleg Endo wrote:
> >
> > > The removal of SH5 support from GCC has been announced here
> > > https://gcc.gnu.org/ml/gcc/2015-08/msg00101.html
> > >
> >
> Hello.
>
> For every created function_summary, we validate (with flag_checking) that
> all cgraph_nodes have summary_uid > 0. It produces a compile hog.
> It's sufficient to validate that for nodes that really utilize a function
> summary.
>
> Patch can bootstrap®test on ppc64le-linux-gnu.
>
On Fri, Apr 29, 2016 at 10:03:40PM +0200, Rainer Orth wrote:
> Dominik Vogt writes:
>
> > The attached patch cleans up some (mostly unnecessary) dg-do
> > directives in the gcc.dg and gcc.target test cases.
>
> This part
>
> * gcc.dg/spec-options.c: Switch order of the two "dg-do run" so
* Claudiu Zissulescu [2016-04-29 09:03:53
+]:
> I see the next tests failing:
>
> FAIL: gcc.target/arc/movb-1.c scan-assembler movb[ \t]+r[0-5]+, *r[0-5]+,
> *r[0-5]+, *19, *21, *8
> FAIL: gcc.target/arc/movb-2.c scan-assembler movb[ \t]+r[0-5]+, *r[0-5]+,
> *r[0-5]+, *23, *23, *9
> FAIL:
P.S.: Bootstrapped and regression tested on s390, s390x and
x86_64.
--
I'm not sure whether something has to be done about the else
branch in round_push too:
else
{
/* If crtl->preferred_stack_boundary might still grow, use
virtual_preferred_stack_boundary_rtx instead.
The attached patch removes excess stack space allocation with
alloca in some situations. Plese check the commit message in the
patch for details.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/ChangeLog
* explow.c (round_push): Use know adjustment.
(allocate_dynamic_st
On 04/28/2016 10:30 AM, Trevor Saunders wrote:
On Thu, Apr 28, 2016 at 10:28:15AM -0400, David Malcolm wrote:
This is a resend of a patch kit I sent in stage 3; the original post
was here:
https://gcc.gnu.org/ml/gcc-patches/2015-12/msg01933.html
I've rebased the patches against yesterday's t
Here is the updated patch.
I eliminated the blank lines in the ChangeLog files.
I eliminated ALTIVEC_COMPLEX, and folded the result into
ALTIVEC_ARG_MAX_RETURN. Note, due to being able to return homogeneous
aggregates, Elf v2 did not need the change, but Elf v1 needs it to return
complex values i
On 04/28/2016 11:59 AM, Marek Polacek wrote:
3) for the C++ FE I used a macro so that I don't have to change all the
cxx_incomplete_type_error calls now,
How about an inline overload, instead?
It seems sad to discard the location information; could we pass it into
cxx_incomplete_type_diag
Dominik Vogt writes:
> The attached patch cleans up some (mostly unnecessary) dg-do
> directives in the gcc.dg and gcc.target test cases.
This part
* gcc.dg/spec-options.c: Switch order of the two "dg-do run" so that
the test ist actually "run" on sh*-*-*. Order _does_ matter.
OK.
Jason
On 04/26/2016 07:59 PM, Martin Sebor wrote:
* builtins.c (fold_builtin_FILE): New function.
(fold_builtin_FUNCTION, fold_builtin_LINE): New functions.
(fold_builtin_0): Call them.
Can we now remove the handling for these built-ins from gimplify_call_expr?
+// Verify th
On 04/19/2016 11:50 AM, Patrick Palka wrote:
1. This patch introduces a "regression" in gcc.dg/tree-ssa/ssa-thread-11.c
in that we no longer perform FSM threading during vrp2 but instead we
detect two new jump threading opportunities during vrp1. Not sure if
the new code is better but it is sho
OK, thanks.
Jason
On Apr 29, 2016, at 5:41 AM, Rainer Orth wrote:
> diff --git a/gcc/config/darwin.h b/gcc/config/darwin.h
> --- a/gcc/config/darwin.h
> +++ b/gcc/config/darwin.h
> @@ -179,6 +179,7 @@ extern GTY(()) int darwin_ms_struct;
>%{L*} %(link_libgcc) %o
> %{fprofile-arcs|fprofile-generate*|coverage:-l
On Apr 29, 2016, at 5:41 AM, Rainer Orth wrote:
> diff --git a/gcc/config/darwin.h b/gcc/config/darwin.h
> --- a/gcc/config/darwin.h
> +++ b/gcc/config/darwin.h
> @@ -179,6 +179,7 @@ extern GTY(()) int darwin_ms_struct;
> %{L*} %(link_libgcc) %o
> %{fprofile-arcs|fprofile-generate*|coverage:-
*** index.html 27 Apr 2016 13:06:08 - 1.1005
--- index.html 29 Apr 2016 17:44:45 -
*** mission statement.
*** 47,52
--- 47,56
News
+ http://www.acm.org/awards/2015-technical-awards";>2015
ACM Software System Award
+ [2016-04-29]
+
+
GCC 6.1 rele
On Fri, 29 Apr 2016, Richard Biener wrote:
Another option is to move the enum declaration from flag-types.h to
coretypes.h. I think I like that best.
This works.
2016-05-02 Marc Glisse
gcc/
* flag-types.h (enum warn_strict_overflow_code): Move ...
* coretypes.h: ... here.
This patch by Chris Manghane removes the old escape analysis code that
never quite worked. He has written a series of patches that implement
working escape analysis. I will be working through reviewing those.
This patch bootstrapped and ran Go testsuite on x86_64-pc-linux-gnu.
Committed to mainli
On Fri, Apr 29, 2016 at 10:05 AM, Uros Bizjak wrote:
> On Fri, Apr 29, 2016 at 6:32 PM, H.J. Lu wrote:
>> Some PR target/70155 tests check for nonexistence of the *movdi_internal
>> pattern. Since PIC leads to the *movdi_internal pattern, skip those
>> tests if PIC is used.
>>
>> Tested on x86-6
On Fri, Apr 29, 2016 at 6:32 PM, H.J. Lu wrote:
> Some PR target/70155 tests check for nonexistence of the *movdi_internal
> pattern. Since PIC leads to the *movdi_internal pattern, skip those
> tests if PIC is used.
>
> Tested on x86-64. OK for trunk?
IMO, it would be better to add condition t
On 04/29/2016 02:06 AM, Richard Biener wrote:
>
> The following makes sure to create DECL_EXPRs for VLA types built for
> temporary arrays to properly allow the gimplifier to unshare expression
> in its type fields when required.
>
> This avoids turing those fields into garbage. With a patch to
> Some PR target/70155 tests check for nonexistence of the *movdi_internal
> pattern. Since PIC leads to the *movdi_internal pattern, skip those
> tests if PIC is used.
>
> Tested on x86-64. OK for trunk?
I cannot approve the patch, but I confirm that the tests fail on darwin.
TIA,
Dominique
Some PR target/70155 tests check for nonexistence of the *movdi_internal
pattern. Since PIC leads to the *movdi_internal pattern, skip those
tests if PIC is used.
Tested on x86-64. OK for trunk?
H.J.
---
* gcc.target/i386/pr70155-1.c: Likewise.
* gcc.target/i386/pr70155-2.c: Lik
On Fri, Apr 29, 2016 at 11:00 AM, Jeff Law wrote:
> On 04/28/2016 06:08 PM, Patrick Palka wrote:
>>>
>>>
>>>
>>> The glitch in that plan is there is no easy linkage between the use of
>>> b_5
>>> in bb4 and the ASSERT_EXPR in bb3. That's something Aldy, Andrew and
>>> myself
>>> are looking at in
Hi,
On 29/04/2016 15:58, Jason Merrill wrote:
On 04/28/2016 08:18 PM, Paolo Carlini wrote:
else if (ANON_AGGR_TYPE_P (type))
{
- tree fields;
-
- for (fields = TYPE_FIELDS (type); fields; fields = DECL_CHAIN
(fields))
+ for (tree fields = TYPE_FIELDS (type); fields;
+
On 29/04/16 15:20, Claudiu Zissulescu wrote:
Please find the updated patch.
Ok to commit?
OK.
On 28/04/16 09:26 +0100, Jonathan Wakely wrote:
On 28 April 2016 at 05:14, Chris Gregory wrote:
This doesn't break any of the code.
I made sure that this didn't affect any string literals or character literals.
Cheers,
Chris Gregory!
Thanks, I'll double check it and apply it (there's no nee
I suspected that much. It would be good to have a libcilkrts/README.gcc
describing the rules which changes can go into the gcc tree directly, which
need to go upstream first, and how. libo and libsanitizer already have this.
Hi Rainer,
It is mentioned under the "CONTRIBUTIONS" sections
Hi,
This patch fixes a latent bug in IVOPT. Variable "depends_on" should be
checked before recording invariant expression, otherwise we end up with
constant (even ZERO) invariant expressions. Apparently this results in wrong
register pressure.
Bootstrap and test on x86_64. Is it OK?
Thanks,
Hi,
This fixes a latent bug I introduced. Variable "cbase" shouldn't be modified
since it will be used afterwards. Bootstrap and test on x86_64. I think it's
an obvious change, is it OK?
Thanks,
bin
2016-04-28 Bin Cheng
* tree-ssa-loop-ivopts.c (get_computation_cost_at): Don't cl
The problem here is that some code paths are not prepared to handle a
non-dependent PAREN_EXPR, which my fix for PR c++/70106 introduced. In
particular lvalue_kind() returns clk_none for a PAREN_EXPR which makes
build_x_unary_op() emit a bogus error for an expression like &(A::b).
(If the PAREN_EX
On Fri, Apr 29, 2016 at 4:51 PM, Bin.Cheng wrote:
> On Thu, Apr 28, 2016 at 10:18 AM, Richard Biener
> wrote:
>> On Wed, Apr 27, 2016 at 5:49 PM, Bin Cheng wrote:
>>> Hi,
>>> Currently tree if-conversion only supports PHIs with no more than two
>>> arguments unless the loop is marked with "simd
On Thu, Apr 28, 2016 at 10:18 AM, Richard Biener
wrote:
> On Wed, Apr 27, 2016 at 5:49 PM, Bin Cheng wrote:
>> Hi,
>> Currently tree if-conversion only supports PHIs with no more than two
>> arguments unless the loop is marked with "simd pragma". This patch makes
>> such PHIs supported uncondi
On Fri, Apr 29, 2016 at 8:42 AM, Ilya Enkovich wrote:
> Hi,
>
> As the first part of PR70799 fix I'd like to add constants support for
> DI-STV pass (which is also related to PR70763). This patch adds CONST_INT
> support as insn operands and extends cost model accordingly.
>
> Bootstrapped and re
Hi,
As the first part of PR70799 fix I'd like to add constants support for
DI-STV pass (which is also related to PR70763). This patch adds CONST_INT
support as insn operands and extends cost model accordingly.
Bootstrapped and regtested on x86_64-unknown-linux-gnu{-m32}. OK for trunk?
Thanks,
On 27/04/16 15:17, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On Thursday 17 December 2015 17:32:48 Thomas Preud'homme wrote:
Hi,
We decided to apply the following patch to the ARM embedded 5 branch.
Best regards,
Thomas
-Original Message-
From: gcc-patches-ow...@gcc.gn
Nathan Sidwell writes:
> On 04/27/16 16:59, Aaron Conole wrote:
>> Apologies for the top post. Pinging on this again. It still applies
>> cleanly, so no need to resubmit, I think. Is there anything else missing
>> or required before this can go in?
>
> I'm not convinced this is a desirable featur
On Fri, Apr 29, 2016 at 12:16 PM, Richard Biener
wrote:
> On Thu, Apr 28, 2016 at 2:56 PM, Bin Cheng wrote:
>> Hi,
>> Tree if-conversion sometimes cannot convert conditional array reference into
>> unconditional one. Root cause is GCC conservatively assumes newly
>> introduced array reference
On 04/28/2016 06:08 PM, Patrick Palka wrote:
The glitch in that plan is there is no easy linkage between the use of b_5
in bb4 and the ASSERT_EXPR in bb3. That's something Aldy, Andrew and myself
are looking at independently for some of Aldy's work.
I see.. One other deficiency I noticed in
On Fri, Apr 29, 2016 at 4:30 PM, Eric Botcazou wrote:
>> Eh, then the build yawns about the missing mode of the input operand.
>
> Every good back-end has at least an example of this. ;-)
>
>> This is getting a bit frustrating, but attached patch should solve
>> this failure. Again lightly tested,
This was supposed to go to the binutils list :(
Please ignore.
Kyrill
On 29/04/16 15:37, Kyrill Tkachov wrote:
Hi all,
I recently upgraded my host compiler to GCC 6.1.0 and while trying to build a
cross toolchain for arm-none-eabi I've encountered some -Werror errors in
tc-arm.c
This patch fi
Hi all,
I recently upgraded my host compiler to GCC 6.1.0 and while trying to build a
cross toolchain for arm-none-eabi I've encountered some -Werror errors in
tc-arm.c
This patch fixes them.
Some static const variables that are unused are marked with ATTRIBUTE_UNUSED.
In parse_neon_el_struct_l
No functional changes.
2016-04-29 Uros Bizjak
* config/i386/i386.md
(operations with memory inputs setting flags peephole2):
Remove uneeded REG_P checks. Cleanup pattern generation.
Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}.
Committed to mainline SVN.
Uros.
> Eh, then the build yawns about the missing mode of the input operand.
Every good back-end has at least an example of this. ;-)
> This is getting a bit frustrating, but attached patch should solve
> this failure. Again lightly tested, regtest in progress.
Everything is back to normal with this
Please find the updated patch.
Ok to commit?
Claudiu
gcc/
2016-04-20 Claudiu Zissulescu
* config/arc/arc-protos.h (compact_memory_operand_p): Declare.
* config/arc/arc.c (arc_output_commutative_cond_exec): Consider
bmaskn instruction.
(arc_dwarf_register_span):
> > (arc_dwarf_register_span): Remove enum keyword.
> That bit should be separate.
About this small one, I will make a new patch and commit it as obvious.
//Claudiu
On April 29, 2016 3:48:37 PM GMT+02:00, David Edelsohn
wrote:
>On Fri, Apr 29, 2016 at 9:44 AM, Bernd Schmidt
>wrote:
>>
>>
>> On 04/29/2016 03:42 PM, David Edelsohn wrote:
>>>
>>> On Fri, Apr 29, 2016 at 9:32 AM, Bernd Schmidt
>>> wrote:
On 04/29/2016 03:02 PM, David Edelsohn wrote:
On Fri, 2016-04-29 at 19:45 +0900, Oleg Endo wrote:
> On Thu, 2016-04-28 at 10:27 +0900, Oleg Endo wrote:
>
> > The removal of SH5 support from GCC has been announced here
> > https://gcc.gnu.org/ml/gcc/2015-08/msg00101.html
> >
> > The attached patch removes support for SH5 in the compiler back
Jakub,
currently automatic loop partitioning assigns from the innermost loop outwards
-- that was the simplest thing to implement. A better algorithm is to assign
the outermost loop to the outermost available axis, and then assign from the
innermost loop outwards. That way we (generally) get
On 04/28/2016 08:18 PM, Paolo Carlini wrote:
else if (ANON_AGGR_TYPE_P (type))
{
- tree fields;
-
- for (fields = TYPE_FIELDS (type); fields; fields = DECL_CHAIN (fields))
+ for (tree fields = TYPE_FIELDS (type); fields;
+ fields = DECL_CHAIN (fields))
if
On 04/29/2016 03:42 PM, David Edelsohn wrote:
On Fri, Apr 29, 2016 at 9:32 AM, Bernd Schmidt wrote:
On 04/29/2016 03:02 PM, David Edelsohn wrote:
How has this show general benefit for all architectures to deserve
enabling it by default at -O2?
It should improve postreload scheduling in g
Hello.
For every created function_summary, we validate (with flag_checking) that
all cgraph_nodes have summary_uid > 0. It produces a compile hog.
It's sufficient to validate that for nodes that really utilize a function
summary.
Patch can bootstrap®test on ppc64le-linux-gnu.
Ready for trunk?
M
On Fri, Apr 29, 2016 at 9:44 AM, Bernd Schmidt wrote:
>
>
> On 04/29/2016 03:42 PM, David Edelsohn wrote:
>>
>> On Fri, Apr 29, 2016 at 9:32 AM, Bernd Schmidt
>> wrote:
>>>
>>> On 04/29/2016 03:02 PM, David Edelsohn wrote:
How has this show general benefit for all architectures to
Hi Bill,
On Mon, Apr 25, 2016 at 09:09:03AM -0500, Bill Schmidt wrote:
> Here's the fix for the obvious pasto separated out. CCing Richi and
> Jakub as I'd appreciate release manager approval to include this in
> gcc-6-branch. This fixes some cases where built-in functions are
> connected to the
On Fri, Apr 29, 2016 at 9:32 AM, Bernd Schmidt wrote:
> On 04/29/2016 03:02 PM, David Edelsohn wrote:
>>
>> How has this show general benefit for all architectures to deserve
>> enabling it by default at -O2?
>
>
> It should improve postreload scheduling in general, and it can also help
> clear up
On 04/29/2016 03:02 PM, David Edelsohn wrote:
How has this show general benefit for all architectures to deserve
enabling it by default at -O2?
It should improve postreload scheduling in general, and it can also help
clear up bad code generation left behind by register allocation.
As an asi
On 29.04.2016 09:46, Richard Biener wrote:
> On Thu, 28 Apr 2016, Bernd Edlinger wrote:
>
>> 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).
>>>
How has this show general benefit for all architectures to deserve
enabling it by default at -O2?
As an aside, this change seems to be the source of a new code
generation bug affecting the PPC kernel.
Thanks, David
Hi Dominique,
> Any reason why the dg-additional-options "*-lcilkrts" have been removed?
> AFAICT they are needed for darwin.
-fcilkplus does (and should) include -lcilkrts when linking. It
certainly does on Solaris and Linux. Everything else is a usability
nightmare: you don't need to link For
On Fri, Apr 29, 2016 at 5:30 AM, Richard Sandiford
wrote:
> Following on from the comparison patch, I think it makes sense to
> support << and >> for offset_int (int128_t) and widest_int (intNNN_t),
> with >> being arithmetic shift. It doesn't make sense to use
> logical right shift on a potentia
On Thu, Apr 28, 2016 at 04:15:41PM +0200, Jakub Jelinek wrote:
> On Thu, Apr 28, 2016 at 04:10:01PM +0200, Marek Polacek wrote:
> > That works too, though it of course affects all users, not just ubsan.
> > Here's
>
> Of course, but I think that is a good thing ;)
>
> > the patch with your sugg
While looking at the use of cst_and_fits_in_hwi in tree-ssa-loop-ivopts.c,
I had difficulty working out what the function actually tests. The
final NUNITS check seems redundant, since it asks about the number of
HWIs in the _unextended_ constant. We've already checked that the
unextended constant
As Richard says, we ought to have a convenient way of converting
an INTEGER_CST to a wide_int of a particular precision without
having to extract the sign of the INTEGER_CST's type each time.
This patch adds a wi::to_wide helper for that, alongside the
existing wi::to_offset and wi::to_widest.
Tes
Following on from the comparison patch, I think it makes sense to
support << and >> for offset_int (int128_t) and widest_int (intNNN_t),
with >> being arithmetic shift. It doesn't make sense to use
logical right shift on a potentially negative offset_int, since
the precision of 128 bits has no mea
offset_int and widest_int are supposed to be at least one bit wider
than all the values they need to represent, with the extra bits
being signs. Thus offset_int is effectively int128_t and widest_int
is effectively intNNN_t, for target-dependent NNN.
Because the types are signed, there's not real
Hi all,
I'm getting a warning when building genrecog that 'label' may be used
uninitialised in:
uint64_t label = 0;
if (d->test.kind == rtx_test::CODE
&& d->if_statement_p (&label)
&& label == CONST_INT)
This is because if_statement_p looks like this:
inline bool
dec
On Fri, 2016-04-29 at 11:56 +0200, Richard Biener wrote:
> Your testcase passes '2' where it passes just fine. If I pass 3 as which
> I indeed get an abort () but you can't reasonably expect it to return 13 then.
Bah, I added an extra case and didn't change the argument. :-(
Let me fix that and
Any reason why the dg-additional-options "*-lcilkrts" have been removed? AFAICT
they are needed for darwin.
TIA
Dominique
On 29/04/16 12:58, Andrew Burgess wrote:
* config/arc/arc.md (*loadqi_update): Replace use of 'rI'
constraint with 'rCm2' constraints to limit possible immediate
size.
(*load_zeroextendqisi_update): Likewise.
(*load_signextendqisi_update): Likewise.
On 04/29/2016 02:01 PM, Marek Polacek wrote:
Here, we segv because we're checking TYPE_MAXVAL of domain that is actually
null for "extern int a[];". The fix is trivial.
Bootstrapped/regtested on x86_64-linux, ok for trunk?
2016-04-29 Marek Polacek
PR c/70852
* c-common.c (w
Hi Ilya,
>> >> 2016-04-07 Rainer Orth
>> >>
>> >> * runtime/except-gcc.cpp (DEBUG_EXCEPTIONS): Allow redefinition.
>> >> * runtime/cilk_fiber.h (FIBER_DEBUG): Likewise.
>> >> * runtime/scheduler.h (REDPAR_DEBUG): Likewise.
>> > Ilya will have to chime in here -- we're a downstream consume
Brown-paper bag bug ...
>> Do I misuse constm1_operand?
>
> No, it is just a typo that crept in constm1_operand:
>
> ;; Match exactly -1.
> (define_predicate "constm1_operand"
> (and (match_code "const_int")
>(match_test "op = constm1_rtx")))
>
> There should be a test, not an assignment
Here, we segv because we're checking TYPE_MAXVAL of domain that is actually
null for "extern int a[];". The fix is trivial.
Bootstrapped/regtested on x86_64-linux, ok for trunk?
2016-04-29 Marek Polacek
PR c/70852
* c-common.c (warn_for_memset): Check domain before accessing
* Joern Wolfgang Rennecke [2016-04-28 18:06:42 +0100]:
> 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 ?
You are absolutely correc
gcc/ChangeLog:
2016-04-25 Martin Liska
* tree-ssa-loop-ivopts.c(comp_cost::operator=): New function.
(comp_cost::infinite_cost_p): Likewise.
(operator+): Likewise.
(comp_cost::operator+=): Likewise.
(comp_cost::operator-=): Likewise.
(comp_cost::
Hello.
As profile-guided optimization can provide very useful information
about basic block frequencies within a loop, following patch set leverages
that information. It speeds up a single benchmark from upcoming SPECv6
suite by 20% (-O2 -profile-generate/-fprofile use) and I think it can
also imp
gcc/ChangeLog:
2016-04-25 Martin Liska
* tree-ssa-loop-ivopts.c (struct ivopts_data): Add inv_expr_map.
(tree_ssa_iv_optimize_init): Initialize it.
(get_expr_id): Assign expressions to the map.
(iv_ca_dump): Dump invariant expressions.
(create_new_ivs):
gcc/ChangeLog:
2016-04-25 Martin Liska
* tree-ssa-loop-ivopts.c (struct comp_cost): Introduce
m_cost_scaled and m_frequency fields.
(comp_cost::operator=): Assign to m_cost_scaled.
(operator+): Likewise.
(comp_cost::operator+=): Likewise.
(comp_c
On Fri, Apr 29, 2016 at 1:26 PM, Ilya Enkovich wrote:
> 2016-04-29 12:48 GMT+03:00 Uros Bizjak :
>> On Thu, Apr 28, 2016 at 12:36 PM, Ilya Enkovich
>> wrote:
>>
>>> That's what I have in my draft for DImode immediates:
>>>
>>> @@ -3114,6 +3123,20 @@ scalar_chain::build (bitmap candidates,
>>> un
On Fri, Apr 29, 2016 at 1:23 PM, Eric Botcazou wrote:
>> Attached patch was actually tested on a couple of cases. It generates
>> the same assembly as before.
>
> Note that you could just remove the second ":W" in the define_insn pattern.
Eh, then the build yawns about the missing mode of the inp
Hi Rainer!
On Fri, Apr 29, 2016 at 10:58:25 +0200, Rainer Orth wrote:
> > On 04/26/2016 08:04 AM, Rainer Orth wrote:
> >> When working on a couple of Cilk Plus issues lately (PRs target/60290,
> >> target/68945), I noticed that you have to modify the libcilkplus sources
> >> to enable various debu
On Thu, 28 Apr 2016, Jeff Law wrote:
> 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 tr
2016-04-29 12:48 GMT+03:00 Uros Bizjak :
> On Thu, Apr 28, 2016 at 12:36 PM, Ilya Enkovich
> wrote:
>
>> That's what I have in my draft for DImode immediates:
>>
>> @@ -3114,6 +3123,20 @@ scalar_chain::build (bitmap candidates,
>> unsigned insn_uid)
>>BITMAP_FREE (queue);
>> }
>>
>> +/* Retu
> Attached patch was actually tested on a couple of cases. It generates
> the same assembly as before.
Note that you could just remove the second ":W" in the define_insn pattern.
That's better, but not quite it because this segfaults at -O2:
#0 memory_operand (op=0xabababababababab, mode=mode@e
On Thu, Apr 28, 2016 at 2:56 PM, Bin Cheng wrote:
> Hi,
> Tree if-conversion sometimes cannot convert conditional array reference into
> unconditional one. Root cause is GCC conservatively assumes newly introduced
> array reference could be out of array bound and thus trapping. This patch
> i
On Fri, Apr 29, 2016 at 12:17 PM, Eric Botcazou wrote:
>> I'm testing the attached patch. Does it fix your ada failures?
>
> No, it totally breaks stack checking. :-(
Eh, I was trying to be too clever.
Attached patch was actually tested on a couple of cases. It generates
the same assembly as bef
> > It should do the job, at least for EM where the jump takes 2 cycle, and by
> means of using delay slots we can make all the cycles count. HS has a branch
> prediction mechanism, hence, filling up the delay slot doesn't have such a big
> impact like in EM or even earlier cpus.
> No, the alternat
On Sun, Apr 17, 2016 at 1:14 AM, kugan
wrote:
> As explained in PR61839,
>
> Following difference results in extra instructions:
> - c = b != 0 ? 486097858 : 972195717;
> + c = a + 972195718 >> (b != 0);
>
> As suggested in PR, attached patch converts CST BINOP COND_EXPR to COND_EXPR
> ? (CST BI
On Thu, 2016-04-28 at 10:27 +0900, Oleg Endo wrote:
> The removal of SH5 support from GCC has been announced here
> https://gcc.gnu.org/ml/gcc/2015-08/msg00101.html
>
> The attached patch removes support for SH5 in the compiler back end.
> There are still some leftovers and new simplification o
> On Apr 29, 2016, at 10:34 , Jakub Jelinek wrote:
>
>
> No, for these files we aren't upstream and just periodically merge stuff
> from there. So, you should try to discuss this in asan upstream and get
> a fix committed there and we can then merge it and/or cherry-pick it.
> Please see libsa
On 29/04/16 11:31, Claudiu Zissulescu wrote:
It should do the job, at least for EM where the jump takes 2 cycle, and by
means of using delay slots we can make all the cycles count. HS has a branch
prediction mechanism, hence, filling up the delay slot doesn't have such a big
impact like in E
On 28 Apr 15:59, Richard Biener wrote:
> On Thu, Apr 28, 2016 at 3:26 PM, Ilya Enkovich wrote:
> > On 27 Apr 16:05, Richard Biener wrote:
> >> >>
> >> >> I'd like to see testcases covering the corner-cases - have them have
> >> >> upper bound estimates by adjusting known array sizes and also cover
It should do the job, at least for EM where the jump takes 2 cycle, and by
means of using delay slots we can make all the cycles count. HS has a branch
prediction mechanism, hence, filling up the delay slot doesn't have such a big
impact like in EM or even earlier cpus.
//Claudiu
> -Origin
P.S.: the .d suffix on the branch was there just for scheduling purposes -
not sure if that actually helped any chip's pipeline, or if it was just
a bug
in the documentation.
On 29/04/16 11:16, Claudiu Zissulescu wrote:
This is the updated patch on handling FPX NaNs.
Ok to apply?
Claudiu
OK.
1 - 100 of 140 matches
Mail list logo