On Fri, Sep 22, 2017 at 8:02 AM, Janus Weil wrote:
> Thanks, Steve. I'm attaching the updated ChangeLog and the two test
> cases for the two new flags. Since this appears to be a somewhat
> controversial feature, I'll wait two more days to allow for others to
> contribute their feedback (positive
On Thu, 21 Sep 2017, Matthias Klose wrote:
> On 21.09.2017 17:50, Ian Lance Taylor via gcc-patches wrote:
> > On Thu, Sep 21, 2017 at 4:52 AM, Thomas Schwinge
> > wrote:
> >>
> >> On Fri, 14 Sep 2012 14:20:08 -0700, Ian Lance Taylor
> >> wrote:
> [libbacktrace]
> >>
> >>> I won't commit fo
Hi Daniel,
> On 09/21/2017 05:18 PM, Daniel Santos wrote:
>> So libgcc doesn't use a config.in. :(
>
> Scratch that, I forgot that we're using gcc/config.in via auto-host.h.
> So I only have to add this to gcc/configure.ac and it will be available
> for my libgcc header -- this is what I used to
I've bootstrapped and reg-tested {,-m64} with this on
86_64-pc-linux-gnu, but I'm waiting for a reference test set to finish
to compare them. I've verified that we're getting HAVE_AS_IX86_AVX in
auto-host.h and I've also built and tested w/o to double-verify that the
AVX code is correct. Now that
Dear All,
I seem to have messed up associate_31.f90 totally. Not only is it not
doing the intended test but it is failing at all levels of
optimization. I will remove it first thing tomorrow. Strangely, it did
not fail when I regtested before committing.
Sorry about this.
Paul
On 21 September
Richard Biener writes:
> On Thu, Sep 14, 2017 at 4:05 PM, Richard Sandiford
> wrote:
>> Richard Biener writes:
>>> On Thu, Sep 14, 2017 at 1:23 PM, Richard Sandiford
>>> wrote:
This patch adds a helper function for getting the number of
bytes accessed by a scalar data reference, which
This is a first attempt to make std::future::wait_until and
std::future::wait_for make correct use of
std::chrono::steady_clock/CLOCK_MONOTONIC. It also makes
std::future::wait_until react to changes to CLOCK_REALTIME during the
wait, but only when passed a std::chrono::system_clock time point.
Pr
The futex system call supports waiting for an absolute time if
FUTEX_WAIT_BITSET is used rather than FUTEX_WAIT. Doing so provides two
benefits:
1. The call to gettimeofday is not required in order to calculate a
relative timeout.
2. If someone changes the system clock during the wait then the
The user-visible effect of this change is that std::future::wait_for now
uses std::chrono::steady_clock to determine the timeout. This makes it
immune to changes made to the system clock. It also means that anyone using
their own clock types with std::future::wait_until will have the timeout
conver
The user-visible effect of this change is for std::future::wait_until
to use CLOCK_MONOTONIC when passed a timeout of
std::chrono::steady_clock::time_point type. This makes it immune to
any changes made to the system clock CLOCK_REALTIME.
Add an overload of __atomic_futex_unsigned::_M_load_and_tex
These tests show that changing the system clock has an effect on
std::future::wait_until when using std::chrono::system_clock but not when
using std::chrono::steady_clock. Unfortunately these tests have a number of
downsides:
1. Nothing that is attempting to keep the clock set correctly (ntpd,
On 09/22/2017 02:18 AM, Rainer Orth wrote:
> Hi Daniel,
>
>> On 09/21/2017 05:18 PM, Daniel Santos wrote:
>>> So libgcc doesn't use a config.in. :(
>> Scratch that, I forgot that we're using gcc/config.in via auto-host.h.
>> So I only have to add this to gcc/configure.ac and it will be available
>
Ping. Updated patch posted here:
https://gcc.gnu.org/ml/gcc-patches/2017-09/msg00390.html
Hi Daniel,
> On 09/22/2017 02:18 AM, Rainer Orth wrote:
>> Hi Daniel,
>>
>>> On 09/21/2017 05:18 PM, Daniel Santos wrote:
So libgcc doesn't use a config.in. :(
>>> Scratch that, I forgot that we're using gcc/config.in via auto-host.h.
>>> So I only have to add this to gcc/configure.ac and it
This re-implements it avoding the need to recompute dominators and in
a much simpler way.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress, SPEC CPU
2006 is happy.
Richard.
2017-09-22 Richard Biener
* sese.c: Include cfganal.h.
(if_region_set_false_region): Rem
2017-09-22 9:11 GMT+02:00 Janne Blomqvist :
> On Fri, Sep 22, 2017 at 8:02 AM, Janus Weil wrote:
>> Thanks, Steve. I'm attaching the updated ChangeLog and the two test
>> cases for the two new flags. Since this appears to be a somewhat
>> controversial feature, I'll wait two more days to allow fo
> On 18 Sep 2017, at 22:08, Simon Wright wrote:
>
> On 18 Sep 2017, at 21:09, Iain Sandoe wrote:
>>
>>> If I propose this alternative patch, should it be a new post, or should I
>>> continue this thread?
>>
>> thanks for the patch.
>>
>> The basic idea seems sound - as a workaround (as note
On 09/22/2017 03:28 AM, Rainer Orth wrote:
> Hi Daniel,
>
>> On 09/22/2017 02:18 AM, Rainer Orth wrote:
>>> Hi Daniel,
>>>
On 09/21/2017 05:18 PM, Daniel Santos wrote:
> So libgcc doesn't use a config.in. :(
Scratch that, I forgot that we're using gcc/config.in via auto-host.h.
On Mon, Sep 4, 2017 at 2:54 PM, Richard Biener
wrote:
> On Wed, Aug 30, 2017 at 6:32 PM, Bin.Cheng wrote:
>> On Wed, Aug 30, 2017 at 3:19 PM, Richard Biener
>> wrote:
>>> On Wed, Aug 30, 2017 at 3:18 PM, Bin Cheng wrote:
Hi,
This patch implements a simple loop interchange pass in GCC,
On Thu, Sep 21, 2017 at 10:56 PM, Will Schmidt
wrote:
> Hi,
>
> Folding of vector stores in GIMPLE.
>
> - Add code to handle gimple folding for the vec_st (vector store) builtins.
> - Remove the now obsoleted folding code for vec_st from rs6000-c-c.
>
> There are two spots that I could use some fe
On Fri, Sep 22, 2017 at 10:15 AM, Richard Sandiford
wrote:
> Richard Biener writes:
>> On Thu, Sep 14, 2017 at 4:05 PM, Richard Sandiford
>> wrote:
>>> Richard Biener writes:
On Thu, Sep 14, 2017 at 1:23 PM, Richard Sandiford
wrote:
> This patch adds a helper function for getting
On Fri, Sep 22, 2017 at 05:28:00AM -0500, Daniel Santos wrote:
> +/* If the assembler doesn't support AVX then directly emit machine code
> + for the instructions above directly. */
Just a nit: too many "directly" words.
Jakub
On Tue, 19 Sep 2017, Alexander Monakov wrote:
> * haifa-sched.c (autopref_rank_for_schedule): Order 'irrelevant' insns
> first, always call autopref_rank_data otherwise.
May I apply this patch now to unblock qsort checking? Further changes or
adjustments can then go in independently a
On Fri, Sep 22, 2017 at 12:28 PM, Daniel Santos wrote:
> On 09/22/2017 03:28 AM, Rainer Orth wrote:
>> Hi Daniel,
>>
>>> On 09/22/2017 02:18 AM, Rainer Orth wrote:
Hi Daniel,
> On 09/21/2017 05:18 PM, Daniel Santos wrote:
>> So libgcc doesn't use a config.in. :(
> Scratch tha
Hi,
This is the V2 patch fixing PR82163. It rewrites verify_loop_closed_ssa by
checking
uses of all definitions inside of loop. This gives advantage that we can check
loop
closed ssa form for a specific loop, rather than for whole function. The
interface
is used in fixing this issue.
Bootstra
On Fri, Sep 22, 2017 at 1:27 PM, Uros Bizjak wrote:
> On Fri, Sep 22, 2017 at 12:28 PM, Daniel Santos
> wrote:
>> On 09/22/2017 03:28 AM, Rainer Orth wrote:
>>> Hi Daniel,
>>>
On 09/22/2017 02:18 AM, Rainer Orth wrote:
> Hi Daniel,
>
>> On 09/21/2017 05:18 PM, Daniel Santos wrot
The following fixes a goof in if-conversion which did nothing to
false predicated blocks. But we have to at least remove all stores.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2017-09-22 Richard Biener
PR tree-optimization/82291
* tree-i
On Fri, Sep 22, 2017 at 1:37 PM, Bin Cheng wrote:
> Hi,
> This is the V2 patch fixing PR82163. It rewrites verify_loop_closed_ssa by
> checking
> uses of all definitions inside of loop. This gives advantage that we can
> check loop
> closed ssa form for a specific loop, rather than for whole f
This simplifies canonicalize_loop_closed_ssa and does other minimal
TLC. It also adds a testcase I reduced from a stupid mistake I made
when reworking canonicalize_loop_closed_ssa.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
SPEC CPU 2006 is happy with it, current sta
Hi!
An overflow isn't detected on multiplication of signed integer
with all ones in the upper half of bits and all zeros in the lower half
by itself, either with -ftrapv, or __builtin_mul_overflow{,_p},
or signed-integer-overflow sanitization.
For libgcc, we have various cases we handle earlier (
On Fri, Sep 22, 2017 at 4:37 AM, Richard Biener wrote:
>
> This re-implements it avoding the need to recompute dominators and in
> a much simpler way.
>
> Bootstrapped on x86_64-unknown-linux-gnu, testing in progress, SPEC CPU
> 2006 is happy.
>
> Richard.
>
> 2017-09-22 Richard Biener
>
>
This patch to libbacktrace provides dummy versions of lstat and
readlink if they are not available on the system. Bootstrapped and
ran libbacktrace tests on x86_64-pc-linux-gnu both normally and with a
hand-edited config.h. Committed to mainline.
Ian
2017-09-22 Ian Lance Taylor
PR sanitizer
On Fri, Sep 22, 2017 at 10:11:55AM +0300, Janne Blomqvist wrote:
> On Fri, Sep 22, 2017 at 8:02 AM, Janus Weil wrote:
> > Thanks, Steve. I'm attaching the updated ChangeLog and the two test
> > cases for the two new flags. Since this appears to be a somewhat
> > controversial feature, I'll wait t
On Fri, Sep 22, 2017 at 8:03 AM, Richard Biener wrote:
>
> This simplifies canonicalize_loop_closed_ssa and does other minimal
> TLC. It also adds a testcase I reduced from a stupid mistake I made
> when reworking canonicalize_loop_closed_ssa.
>
> Bootstrapped and tested on x86_64-unknown-linux-
Hi!
I'd like to ping the
http://gcc.gnu.org/ml/gcc-patches/2017-09/msg00937.html
- fix compile time hog in replace_placeholders
patch.
Thanks
Jakub
On Thu, 2017-09-21 at 09:20 -0700, Nathan Sidwell wrote:
> On 09/20/2017 12:52 PM, David Malcolm wrote:
> > When we have a type mismatch in a C++ function call, e.g.
> > whereas underlining the mismatching things would make the messages
> > easier to comprehend:
> >
> >test.c: In function 'int
On Falkor, because of an idiosyncracy of how the pipelines are designed, a
quad-word store using a reg+reg addressing mode is almost twice as slow as an
add followed by a quad-word store with a single reg addressing mode. So we
get better performance if we disallow addressing modes using register
Hello,
This patch by Richard Earnshaw was reverted earlier as it was breaking
cross-native builds. Respinning now with a minor change that fixes the build
issue - adding arm-isa.h to GTM_H. Also remove a redundant dependency (TM_H
includes GTM_H).
Tested by building cross/cross-native arm-none-li
On 22/09/17 16:54, Vidya Praveen wrote:
Hello,
This patch by Richard Earnshaw was reverted earlier as it was breaking
cross-native builds. Respinning now with a minor change that fixes the build
issue - adding arm-isa.h to GTM_H. Also remove a redundant dependency (TM_H
includes GTM_H).
Tested
Hello,
This patch by Richard Earnshaw was reverted along with the commit that preceded
it as the preceding commit was causing cross-native builds to fail and I
presumed this patch was related too. Now I am respinning as the issue that
caused the cross-native failure have been fixed. This patch
On Fri, Sep 22, 2017 at 8:49 AM, Jim Wilson wrote:
> On Falkor, because of an idiosyncracy of how the pipelines are designed, a
> quad-word store using a reg+reg addressing mode is almost twice as slow as an
> add followed by a quad-word store with a single reg addressing mode. So we
> get better
On 9/22/17 4:58 PM, Vidya Praveen wrote:
Hello,
This patch by Richard Earnshaw was reverted along with the commit that preceded
it as the preceding commit was causing cross-native builds to fail and I
presumed this patch was related too. Now I am respinning as the issue that
caused the cross-
On Thu, Sep 21, 2017 at 01:08:01PM -0700, Kevin Buettner wrote:
> Ping.
Ok, thanks.
> On Mon, 7 Aug 2017 17:51:38 -0700
> Kevin Buettner wrote:
>
> > On Wed, 10 May 2017 17:24:27 +0200
> > Jakub Jelinek wrote:
> >
> > > What I don't like is that the patch is inconsistent, it sets DECL_CONTEX
This patch makes TARGET_VECTORIZE_VEC_PERM_CONST_OK take the permute
vector in the form of a vec_perm_indices instead of an unsigned char *.
It follows on from the recent patch that did the same in target-independent
code.
It was easy to make ARM and AArch64 use vec_perm_indices internally
as well
This patch changes the element type of (auto_)vec_perm_indices from
unsigned char to unsigned short. This is needed for fixed-length
2048-bit SVE. (SVE is variable-length by default, but it's possible
to ask for specific vector lengths if you want to.)
Tested on aarch64-linux-gnu, x86_64-linux-g
Richard Biener writes:
> On Thu, Sep 21, 2017 at 2:56 PM, Richard Sandiford
> wrote:
>> Richard Biener writes:
>>> On September 20, 2017 2:36:03 PM GMT+02:00, Richard Sandiford
>>> wrote:
When forcing a constant of mode MODE into memory, force_const_mem
asks the frontend to provide the
Hi!
While fixing libgcc2.c (__mulvDI3), I've noticed that while we
optimize x == 0 && y == 0 into (x | y) == 0, we don't optimize
analogical x == -1 && y == -1 into (x & y) == -1.
The following patch does that, bootstrapped/regtested on x86_64-linux and
i686-linux, ok for trunk?
2017-09-22 Jaku
Richard Biener writes:
> On Wed, Sep 20, 2017 at 2:06 PM, Richard Sandiford
> wrote:
>> extract_range_from_binary_expr_1 had:
>>
>> if (range_int_cst_p (&vr0)
>> && range_int_cst_p (&vr1)
>> && TYPE_OVERFLOW_WRAPS (expr_type))
>> ...
>> ...
>> extract
This PR shows that we weren't filtering out irrelevant stmts in
vect_get_peeling_costs_all_drs (unlike related loops in which
we iterate over all datarefs).
Tested on aarch64-linux-gnu, x86_64-linux-gnu and powerpc64le-linux-gnu.
Installed as obvious.
Richard
2017-09-22 Richard Sandiford
gc
Enough changed since V3 that I think another review round on the ppc
bits is warranted.
wrap_frame_mem is gone. Allocation, attribute setting and
notes/scheduler barriers are now in rs6000_emit_allocate_stack_1.
handle_residual is gone, it's folded into rs6000_emit_allocate_stack_1.
rs6000_emi
> In my view, the ideal message would use the same form as in
> the source code. But I suspect that's not easy to determine at
> all sites where the message is being formatted, so the next best
> thing is to avoid using the keyword and fall back on the general
> term.
>
> Martin
>
> PS If possi
OK.
On Thu, Sep 14, 2017 at 4:12 PM, Jakub Jelinek wrote:
> Hi!
>
> When the expression replace_placeholders is called on contains
> many SAVE_EXPRs that appear more than once in the tree, we hang walking them
> over and over again, while it is sufficient to just walk it without
> duplicates (not
On Fri, Sep 22, 2017 at 8:59 AM, Jim Wilson wrote:
> On Fri, Sep 22, 2017 at 8:49 AM, Jim Wilson wrote:
>> On Falkor, because of an idiosyncracy of how the pipelines are designed, a
>> quad-word store using a reg+reg addressing mode is almost twice as slow as an
>> add followed by a quad-word sto
On Thu, 2017-09-21 at 12:37 +, Joseph Myers wrote:
> On Tue, 5 Sep 2017, Steve Ellcey wrote:
>
> >
> > 2017-09-05 Steve Ellcey
> >
> > * config.gcc: Add new case statement to set
> > default_gnu_indirect_function. Remove it from x86_64-*-linux*,
> > i[34567]86-*,
On September 22, 2017 6:59:48 PM GMT+02:00, Jakub Jelinek
wrote:
>Hi!
>
>While fixing libgcc2.c (__mulvDI3), I've noticed that while we
>optimize x == 0 && y == 0 into (x | y) == 0, we don't optimize
>analogical x == -1 && y == -1 into (x & y) == -1.
>
>The following patch does that, bootstrapped
On Fri, Sep 22, 2017 at 10:58 AM, Andrew Pinski wrote:
> Two overall comments:
> * What about splitting register_offset into two different elements,
> one for non 128bit modes and one for 128bit (and more; OI, etc.) modes
> so you get better address generation right away for the simd load
> cases
This patch by Tony Reix adds XCOFF support to libgo and to the go and
cgo commands. Bootstrapped and ran Go tests on x86_64-pc-linux-gnu.
Committed to mainline.
Ian
Index: gcc/go/gofrontend/MERGE
===
--- gcc/go/gofrontend/MERGE (
2017-09-22 11:44 GMT+02:00 Janus Weil :
> 2017-09-22 9:11 GMT+02:00 Janne Blomqvist :
>> And since the standard requires that double precision variables are
>> twice as big as reals, in the absence of an explicit -fdefault-double=
>> flag, would it make sense to have -fdefault-real=N imply
>> -fdef
On 09/22/2017 11:54 AM, Steve Ellcey wrote:
In my view, the ideal message would use the same form as in
the source code. But I suspect that's not easy to determine at
all sites where the message is being formatted, so the next best
thing is to avoid using the keyword and fall back on the gener
On Sep 22, 2017, at 5:27 AM, Richard Biener wrote:
>
> On Thu, Sep 21, 2017 at 10:56 PM, Will Schmidt
> wrote:
>> Hi,
>>
>> Folding of vector stores in GIMPLE.
>>
>> - Add code to handle gimple folding for the vec_st (vector store) builtins.
>> - Remove the now obsoleted folding code for vec_s
Hi Jim,
This looks like a general issue with reg+reg addressing modes being generated in
cases where it is not correct. I haven't looked at lmbench for a while, but it
generated
absolutely horrible code like:
add x1, x0, #120
ldr v0, [x2, x1]
add x1, x0, #128
ldr v1, [x2, x1]
If this is still h
> This looks good. But let's call it debug_type_hash instead. OK with
> that change.
Thanks. It occured to me that we don't need to look up the type twice when we
need to insert it so I have installed the attached patchlet as obvious after
boostrapping/regtesting on x86-64/Linux.
PR
On Sep 21, 2017, Richard Biener wrote:
> On Tue, 12 Sep 2017, Richard Biener wrote:
>>
>> The following avoids adding DW_AT_alignment twice by not doing it
>> for incomplete types.
>>
>> Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
>>
>> Alex, is that ok or do we want DW_AT_a
On Fri, Sep 22, 2017 at 11:39 AM, Jim Wilson wrote:
> On Fri, Sep 22, 2017 at 10:58 AM, Andrew Pinski wrote:
>> Two overall comments:
>> * What about splitting register_offset into two different elements,
>> one for non 128bit modes and one for 128bit (and more; OI, etc.) modes
>> so you get bett
Hi Carl,
On Mon, Sep 18, 2017 at 11:31:07AM -0700, Carl Love wrote:
> * gcc.target/powerpc/builtins-5-p9-runnable.c: Add new runable test file
> for the new built-ins and the existing built-ins.
Typo ("runable").
> (define_expand "altivec_lvsr"
> - [(use (match_operand:V16QI 0 "reg
On Thu, 21 Sep 2017 09:40:49Richard Biener wrote:
> On Wed, Sep 20, 2017 at 10:35 PM, Damian Rouson
> wrote:
> >> Attached is a patch that adds the downloading of gfortran prerequisites
> OpenCoarrays and MPICH in the contrib/download_prerequisites script. The
> patch also provides a useful error
On Thu, Sep 21, 2017 at 01:11:11PM -0500, Will Schmidt wrote:
> Add testcase coverage for the vec_st (vector store)
> intrinsic builtins.
>
> Tested across power platforms (p6 and newer). OK for trunk?
Yep, looks fine. Thanks!
Segher
> 2017-09-21 Will Schmidt
>
> * gcc.target
On Fri, Sep 22, 2017 at 02:58:54PM -0500, Bill Schmidt wrote:
> OK. Will, for now, let's again use the (void *) solution for the time being,
> and
> add commentary recording this improvement for future work. Same would
> go for the vec_vsx_ld/st variations once you get to those.
>
> Otherwise t
Hi Carl,
Some comments, mostly trivial:
On Thu, Sep 21, 2017 at 04:09:11PM -0700, Carl Love wrote:
> --- a/gcc/config/rs6000/rs6000-builtin.def
> +++ b/gcc/config/rs6000/rs6000-builtin.def
> @@ -608,6 +608,16 @@
> CODE_FOR_ ## ICODE) /* ICODE */
>
>
> +/* Mis
69 matches
Mail list logo
