On Sun, 8 Apr 2018, Martin Sebor wrote:
> While updating the -Wrestrict option to mention that it works
> best not just with -O2 but also at higher optimization levels
> I looked around for other options that might benefit from
> a similar clarification. I found a few inlining options that
> only
On 2018-04-07 22:08, Gerald Pfeifer wrote:
On Thu, 1 Feb 2018, Daniel Hellstrom wrote:
I would like to commit the following comment about LEON3-FT errata now
available in GCC-7.3.
Thanks, I have now committed it.
Apparently the ultimate URL is the one below (which I applied for
now)?
Gerald
Hi Richard,
I do not see the said patch applied in ToT yet. When do you expect it
to be available in ToT?
- Thanks and regards,
Sameera D.
On 30 March 2018 at 17:01, Sameera Deshpande
wrote:
> Hi Richard,
>
> The testcase is working with the patch you suggested, thanks for
> pointing that out
On 06.04.2018 18:55, Alexander Monakov wrote:
> On Tue, 3 Apr 2018, Andrey Belevantsev wrote:
>
>> Hello,
>>
>> This issue is when we cannot correctly make insn tick data for the
>> unscheduled code (but processed by the modulo scheduler). Fixed by closely
>> following usual scheduling process as
On Mon, 9 Apr 2018, Daniel Hellstrom wrote:
> I received feedback from the webmaster that the preferred link is
> "http://www.gaisler.com/notes"; so I updated it just be pushing.
Maybe, but then they should reconfigure their server such that
it does not redirect.
We've had to many cases in the pa
On 06.04.2018 18:59, Alexander Monakov wrote:
> On Tue, 3 Apr 2018, Andrey Belevantsev wrote:
>
>> Hello,
>>
>> This issues is about the correct order in which we need to call the
>> routines that fix up the control flow for us.
>
> OK with formatting both in the new comment and the Changelog fix
On 06.04.2018 19:10, Alexander Monakov wrote:
> On Tue, 3 Apr 2018, Andrey Belevantsev wrote:
>
>> Hello,
>>
>> This issue ended up being fixed the way different from described in the PR.
>> We do not want to walk away from the invariant "zero SCHED_TIMES -- insn
>> is not scheduled" even for boo
On 06.04.2018 19:18, Alexander Monakov wrote:
> On Tue, 3 Apr 2018, Andrey Belevantsev wrote:
>
>> Hello,
>>
>> In these PRs we deal with the dependencies we are forced to make between a
>> debug insn and its previous insn (unless bb head). In the latter case, if
>> such an insn has been moved, w
On Fri, 6 Apr 2018, Jakub Jelinek wrote:
> On Fri, Apr 06, 2018 at 06:48:52AM -0700, Cesar Philippidis wrote:
> > 2018-04-06 Cesar Philippidis
> >
> > PR middle-end/84955
> >
> > gcc/
> > * cfgloop.c (flow_loops_find): Add assert.
> > * omp-expand.c (expand_oacc_for): Add dumm
Somehow ended up off-list only.
-- Forwarded message --
Date: Mon, 9 Apr 2018 09:30:15 +0200 (CEST)
From: Richard Biener
To: Gerald Pfeifer
Subject: Re: [nvptx,
PR81352] Add exit insn after noreturn call for neutered threads in warp
On Sun, 8 Apr 2018, Gerald Pfeifer wrote:
Hi Paul,
Apologies for the delay in reply, my @imgtec.com address has since become
defunct and I have only come across your e-mail in my ever-lagging routine
mailing traffic review.
On Wed, 14 Mar 2018, Paul Hua wrote:
> I noticed that data-sym-pool.c fails on -O0 flags.
>
> -O0 output :
> -
Hi.
I'm sending updated version of the patch.
Patch can bootstrap on ppc64le-redhat-linux and x86_64-linux and survives
regression tests.
Martin
>From 6d19b1bf0c28a683e1458e16943e34bb547d36d6 Mon Sep 17 00:00:00 2001
From: marxin
Date: Wed, 14 Mar 2018 09:44:18 +0100
Subject: [PATCH] Introduce
Hi,
we've been having hanging OpenMP tests for nvptx offloading:
for-{3,5,6}.c and the corresponding C++ test-cases.
The failures have now been analyzed down to gomp_ptrlock_get in
libgomp/config/nvptx/ptrlock.h:
...
static inline void *gomp_ptrlock_get (gomp_ptrlock_t *ptrlock)
{
uintptr
The following fixes the new capability of analyzing a wrapping IV as
non-wrapping under a condition that ensures we do not wrap by
restricting it to the case where the exit test must be executed
when it reaches that bound. Otherwise we may not analyze the
IV with the assumption that the exit test
Hi!
On Wed, Apr 04, 2018 at 10:32:58AM -0500, Kelvin Nilsen wrote:
> 2018-04-03 Kelvin Nilsen
>
> * config/rs6000/rs6000-c.c (altivec_overloaded_builtins): Remove
> erroneous entries for
> "vector int vec_ldl (int, long int *)", and
> "vector unsigned int vec_ldl (int, unsigned
Hi.
Following patch adds checking of noreturn attribute when we merge DECLs in LTO.
Note that proper fix is probably
https://gcc.gnu.org/bugzilla/attachment.cgi?id=43884&action=diff
but that would be applicable after MPX is removed. Thus in next stage1.
Patch can bootstrap on x86_64-linux-gnu an
On Mon, Apr 9, 2018 at 3:55 PM, Martin Liška wrote:
> Hi.
>
> Following patch adds checking of noreturn attribute when we merge DECLs in
> LTO.
> Note that proper fix is probably
> https://gcc.gnu.org/bugzilla/attachment.cgi?id=43884&action=diff
> but that would be applicable after MPX is remove
Hi,
this patch fixes most offending artifacts across the function partition
bounary with -freorder-blocks-and-partitions which is now on by default
for i386 (no other targets). The problem is that most of cfgcleanup and
cfgrtl is disabled on crossing edges and thus we produce pretty ugly
code.
Th
Hi Ramana,
On 06/04/18 17:17, Thomas Preudhomme wrote:
On 06/04/18 17:08, Ramana Radhakrishnan wrote:
On 06/04/2018 16:54, Thomas Preudhomme wrote:
Instruction pattern for setting the FPSCR expects the input value to be
in a register. However, __builtin_arm_set_fpscr expander does not ensure
Committed. Thank you for your review,
Claudiu
> -Original Message-
> From: Andrew Burgess [mailto:andrew.burg...@embecosm.com]
> Sent: Saturday, April 07, 2018 11:07 AM
> To: Claudiu Zissulescu
> Cc: gcc-patches@gcc.gnu.org; claudiu.zissule...@synopsys.com;
> francois.bed...@synopsys.com
Accepted & committed, thank you for your contribution,
Claudiu
> -Original Message-
> From: Alexey Brodkin [mailto:abrod...@synopsys.com]
> Sent: Friday, April 06, 2018 6:02 PM
> To: gcc-patches@gcc.gnu.org
> Cc: Claudiu Zissulescu ; Alexey Brodkin
>
> Subject: [PATCH] [ARC] Fix stack usa
On April 9, 2018 4:17:45 PM GMT+02:00, Jan Hubicka wrote:
>Hi,
>this patch fixes most offending artifacts across the function partition
>bounary with -freorder-blocks-and-partitions which is now on by default
>for i386 (no other targets). The problem is that most of cfgcleanup
>and
>cfgrtl is dis
Here if we return immediately after the diagnostic about unsupported
variably-modified type capture, we avoid the crash.
I'm also changing the diagnostic from error to sorry, since it would
make sense to support this usage and apparently Clang already does.
The second patch below is the beginning
final_1 already sets insn_current_address for each instruction, making
it possible to use some of the address functions in final.c during
assembly generation. This patch also sets insn_last_address, since
as the comment says, we can treat final as a shorten_branches pass that
does nothing. It's t
In this PR we used WIDEN_SUM_EXPR to vectorise:
short i, y;
int sum;
[...]
for (i = x; i > 0; i--)
sum += y;
with 4 ints and 8 shorts per vector. The problem was that we set
the VF based only on the ints, then calculated the number of vector
copies based on the shorts, giving 4/8. P
On Mon, Apr 09, 2018 at 06:47:45PM +0100, Richard Sandiford wrote:
> In this PR we used WIDEN_SUM_EXPR to vectorise:
>
> short i, y;
> int sum;
> [...]
> for (i = x; i > 0; i--)
> sum += y;
>
> with 4 ints and 8 shorts per vector. The problem was that we set
> the VF based only on th
On Fri, Apr 6, 2018 at 2:17 PM, Marek Polacek wrote:
> On Mon, Mar 26, 2018 at 01:17:22PM -0400, Jason Merrill wrote:
>> On Sat, Mar 24, 2018 at 6:59 AM, Marek Polacek wrote:
>> > Recently the code in finish_static_assert was changed to use
>> > perform_implicit_conversion_flags followed by fold_
On Mon, Apr 09, 2018 at 01:30:12PM +0300, Andrey Belevantsev wrote:
> I think that should be fine, that is, as long as the insn moved up through
> all those debug insns, the copy will do that as well. It's that
> problematic conditional in sched-deps.c that we should take care of.
>
> I've reword
Hi!
As the following testcase shows, we emit an incorrect PTR prefix in a
vpbroadcastb instruction in -masm=intel mode; gas accepts and the manual
documents that the input operand is xmm2/m8 for vpbroadcastb and
xmm2/m16 for vpbroadcastw, so we need to use BYTE PTR and WORD PTR instead
of XMMWORD
OK.
On Fri, Apr 6, 2018 at 3:05 PM, Paolo Carlini wrote:
> Hi,
>
> On 06/04/2018 19:04, Jason Merrill wrote:
>>
>> On Fri, Apr 6, 2018 at 5:01 AM, Paolo Carlini
>> wrote:
>>>
>>> here, for an incomplete type we ICE pretty soon in
>>> find_decomp_class_base.
>>> Comparing to other cases too, I co
OK.
On Fri, Apr 6, 2018 at 3:54 PM, David Malcolm wrote:
> On Thu, 2018-03-29 at 09:20 -0400, Jason Merrill wrote:
>> On Wed, Mar 28, 2018 at 4:44 PM, David Malcolm
>> wrote:
>> > This followup patch updates the specific error-handling path
>> > to add a note showing the pertinent parameter decl
On Mon, Apr 9, 2018 at 11:37 AM, Jakub Jelinek wrote:
> Hi!
>
> I've tested make check-gcc
> RUNTESTFLAGS='--target_board=unix\{-m32/-masm=intel,-m64/-masm=intel\}
> i386.exp vect.exp'
> testing and looked solely at assembly Error: (there are many scan-assembler*
> directives that just fail, and
cp_parser_check_template_parameters is supposed to catch when we have
the wrong number of template parameter lists, but it wasn't diagnosing
this case. Fixed by checking whether the thing we're declaring used a
template-id; the case of declaring a primary template, when we allow
one more template
OK.
On Wed, Apr 4, 2018 at 3:48 PM, Jakub Jelinek wrote:
> Hi!
>
> The maybe_range_for_decl argument is documented on
> cp_parser_simple_declaration as:
>If MAYBE_RANGE_FOR_DECL is not NULL, the pointed tree will be set to the
>parsed declaration if it is an uninitialized single declarato
Hello world,
the attached patch removes the parallel annotation from DO CONCURRENT.
As discussed in the PR, the autoparallellizer currently generates
wrong code. The only feasible way is to disable the annotation for
gcc-8 and work on the wrong-code issues for gcc-9. This is an 8
regression.
Reg
On Mon, Apr 09, 2018 at 10:10:13PM +0200, Thomas Koenig wrote:
> Hello world,
>
> the attached patch removes the parallel annotation from DO CONCURRENT.
> As discussed in the PR, the autoparallellizer currently generates
> wrong code. The only feasible way is to disable the annotation for
> gcc-8
Here, the code was improperly assuming that the member argument must
be a decl at the point of the error, but since we stopped folding
foo[0] to foo, in this case it's an INDIRECT_REF.
And the existing error for INDIRECT_REF is inaccurate for this case,
so let's stick to complaining about asking f
On Mon, Apr 09, 2018 at 10:10:13PM +0200, Thomas Koenig wrote:
>
> the attached patch removes the parallel annotation from DO CONCURRENT.
> As discussed in the PR, the autoparallellizer currently generates
> wrong code. The only feasible way is to disable the annotation for
> gcc-8 and work on the
Hi!
I've tested make check-gcc
RUNTESTFLAGS='--target_board=unix\{-m32/-masm=intel,-m64/-masm=intel\} i386.exp
vect.exp'
testing and looked solely at assembly Error: (there are many scan-assembler*
directives that just fail, and some tests use e.g. only att inline asm
etc.).
The following patch
A::A names the constructor, but you can't call the constructor
directly, you should just write A(args). Our error recovery was
failing in a template because we'd wrapped the arguments with
NON_DEPENDENT_EXPR, and then kept the result as part of the template
tree, whereas we're supposed to throw aw
Hi Steve,
On Mon, Apr 09, 2018 at 10:10:13PM +0200, Thomas Koenig wrote:
the attached patch removes the parallel annotation from DO CONCURRENT.
As discussed in the PR, the autoparallellizer currently generates
wrong code. The only feasible way is to disable the annotation for
gcc-8 and work on
For some reason, cp_parser_name_lookup_error uses %E for printing the
parsing scope, so dump_expr needs to know how to handle some _TYPEs.
DECLTYPE_TYPE was missing.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit 4d380464839058b3d70e1262339ed973642d6741
Author: Jason Merrill
Date: Mon Ap
On Mon, Apr 09, 2018 at 10:58:13PM +0200, Thomas Koenig wrote:
> > On Mon, Apr 09, 2018 at 10:10:13PM +0200, Thomas Koenig wrote:
> >>
> >> the attached patch removes the parallel annotation from DO CONCURRENT.
> >> As discussed in the PR, the autoparallellizer currently generates
> >> wrong code.
Hi,
changes in predictors caused that LTO IPA-CP no longer clones the qsort
implementation for both of the comparators used in 505.mcf_r from Spec
2017 to be quite and this resulted in quite some slowdown of the
benchmark.
I have tried various way of teaching IPA-CP that the remaining contexts
al
On 04/05/2018 08:20 AM, Kyrill Tkachov wrote:
> Hi all,
>
> In this PR the expansion code emits an invalid memory address for the
> stack probe, which the backend fails to recognise.
> The address is created explicitly in
> anti_adjust_stack_and_probe_stack_clash in explow.c and passed down to
> g
On 04/05/2018 06:32 AM, Richard Biener wrote:
>
> This fixes exponential time spent in find_base_term by extending the
> existing fix of NULL-ifying VALUE locations during their processing
> to keep them NULL-ified during the whole recursion.
>
> As Jakub figured this has the chance of returning
On 04/04/2018 06:43 AM, Peter Bergner wrote:
> On 4/4/18 4:06 AM, Tamar Christina wrote:
>> Now that I know how the loads are done, I have a patch should be both
>> correct and generate better code in most cases.
>> It just calculates bitsize inside the loop and does the copying in the
>> largest
Hi,
On 09/04/2018 21:39, Jason Merrill wrote:
cp_parser_check_template_parameters is supposed to catch when we have
the wrong number of template parameter lists, but it wasn't diagnosing
this case. Fixed by checking whether the thing we're declaring used a
template-id; the case of declaring a p
Define a new exception type derived from std::ios::failure[abi:cxx11]
which also aggregates an object of the gcc4-compatible ios::failure
type. Make __throw_ios_failure throw this new type for iostream errors
that raise exceptions. Provide custom type info for the new type so that
it can be caught
Ping: I will go ahead and commit this patch sometime this
week unless I hear objections:
https://gcc.gnu.org/ml/gcc-patches/2018-04/msg00036.html
Martin
On 04/02/2018 03:55 PM, Martin Sebor wrote:
Attached is a patch to remove the hyphen from uses of side-effect
in the manual as per the com
On 04/09/2018 05:58 PM, Martin Sebor wrote:
Ping: I will go ahead and commit this patch sometime this
week unless I hear objections:
https://gcc.gnu.org/ml/gcc-patches/2018-04/msg00036.html
This patch is mostly OK, except for these hunks:
Index: gcc/doc/invoke.texi
==
I would like to queue this patch for stage1 review.
In DSE, while in dse_classify_store, as soon as we see a PHI use
statement that is part of the loop, we are immediately giving up.
As far as I understand, this can be improved. Attached patch is trying
to walk the uses of the PHI statement (by r
On 04/09/2018 01:36 AM, Richard Biener wrote:
On Sun, 8 Apr 2018, Martin Sebor wrote:
While updating the -Wrestrict option to mention that it works
best not just with -O2 but also at higher optimization levels
I looked around for other options that might benefit from
a similar clarification. I
>>> "H.J. Lu" 04/09/18 9:09 PM >>>
>On Mon, Apr 9, 2018 at 11:37 AM, Jakub Jelinek wrote:
>> BTW, -masm=intel seems to be in quite bad shape even in the assembler, in
>> various testcases I'm getting errors like on the following reduced one:
>> int k1, xmm0;
>> int foo (void) { return k1; }
>> in
On 09.04.2018 21:23, Jakub Jelinek wrote:
> On Mon, Apr 09, 2018 at 01:30:12PM +0300, Andrey Belevantsev wrote:
>> I think that should be fine, that is, as long as the insn moved up through
>> all those debug insns, the copy will do that as well. It's that
>> problematic conditional in sched-deps.
Hi,
this patch prevent lto streaming ice where devirtualization pulls out symbol
that is never seen by free lang data becuase type inheritnace graph contains
types that are no longer reference by IL.
Patch also fixes binfo walking - we need to walk vtables and there is no need
to walk virtuals becu
On Mon, 9 Apr 2018, Martin Sebor wrote:
> On 04/09/2018 01:36 AM, Richard Biener wrote:
> > On Sun, 8 Apr 2018, Martin Sebor wrote:
> >
> > > While updating the -Wrestrict option to mention that it works
> > > best not just with -O2 but also at higher optimization levels
> > > I looked around for
On Mon, Apr 9, 2018 at 10:58 PM, Thomas Koenig wrote:
> Hi Steve,
>
>> On Mon, Apr 09, 2018 at 10:10:13PM +0200, Thomas Koenig wrote:
>>>
>>>
>>> the attached patch removes the parallel annotation from DO CONCURRENT.
>>> As discussed in the PR, the autoparallellizer currently generates
>>> wrong c
On Mon, Apr 9, 2018 at 7:43 PM, Richard Sandiford
wrote:
> final_1 already sets insn_current_address for each instruction, making
> it possible to use some of the address functions in final.c during
> assembly generation. This patch also sets insn_last_address, since
> as the comment says, we can
On Mon, Apr 9, 2018 at 7:47 PM, Richard Sandiford
wrote:
> In this PR we used WIDEN_SUM_EXPR to vectorise:
>
> short i, y;
> int sum;
> [...]
> for (i = x; i > 0; i--)
> sum += y;
>
> with 4 ints and 8 shorts per vector. The problem was that we set
> the VF based only on the ints, the
60 matches
Mail list logo