This patch allows a target to increase the cost of anti-deps to better reflect
the actual cost on the machine.
This gets me get 5% more performance on an important inner loop by exposing the
actual cost of long dep chains that have lots of anti-deps in them. Be
scheduling the longer chain firs
Hi,
this patch further reduce memory use and time of WPA stage, especially without
-g
phase opt and generate : 75.66 (39%) usr 1.78 (14%) sys 77.44 (37%) wall
855644 kB (21%) ggc
phase stream in : 34.62 (18%) usr 1.95 (16%) sys 36.57 (18%) wall
3245604 kB (79%) ggc
phase str
On Fri, 11 Dec 2015, Jan Hubicka wrote:
> Hi,
> this patch saves about 30% of global decl stream size in firefox. While
> implementing the lto sections for initializers I put very stupid heursitcs
> to get_symbol_initial_value deciding whether the initializer is better
> streamed
> inline or off
Hello.
I've just applied suggested change that Richi proposed.
The patch can bootstrap on x86_64-linux-gnu and survives regression tests.
Moreover,
the memory leak/invalid read has gone.
Ready for trunk?
Thanks,
Martin
>From f0eca2d3efd711498fbc8b7e8980fdf32f0abfd0 Mon Sep 17 00:00:00 2001
From:
Hello Jeff:
Sorry for the delay in sending the benchmarks run with Split-Path change.
Here is the Summary of the results.
SPEC CPU 2000 INT benchmarks ( Target i386)
( Geomean Score without Split-Paths changes vs Geomean Score with Split-Path
changes = 3740.789 vs 3745.193).
SPEC CPU 2000 FP
On Fri, 11 Dec 2015, Jan Hubicka wrote:
> Hi,
> this patch further reduce memory use and time of WPA stage, especially
> without -g
> phase opt and generate : 75.66 (39%) usr 1.78 (14%) sys 77.44 (37%)
> wall 855644 kB (21%) ggc
> phase stream in : 34.62 (18%) usr 1.95 (16%) s
> This patch allows a target to increase the cost of anti-deps to better
> reflect the actual cost on the machine.
But it can already do it via the TARGET_SCHED_ADJUST_COST hook, can't it?
--
Eric Botcazou
> On Fri, 11 Dec 2015, Jan Hubicka wrote:
>
> > Hi,
> > this patch saves about 30% of global decl stream size in firefox. While
> > implementing the lto sections for initializers I put very stupid heursitcs
> > to get_symbol_initial_value deciding whether the initializer is better
> > streamed
>
>
> We explicitely do not use debug-info-level tests in free-lang-data
> to allow mixing -g and -g0 objects. Are you sure doing the above
> doesn't mess up tree merging enough to effectively enlarge WPA
> memory use and the merged decl sections?
>
> [I'm quite sure firefox build system manages t
On Fri, 11 Dec 2015, Jan Hubicka wrote:
> Hi,
> this patch makes WPA to copy sections w/o decompressing them. This leads
> to a nice /tmp usage for GCC bootstrap (about 70%) and little for Firefox.
> In GCC about 5% of the ltrans object file is the global decl section, while
> for Firefox it is 8
On Fri, 11 Dec 2015, Jan Hubicka wrote:
> >
> > We explicitely do not use debug-info-level tests in free-lang-data
> > to allow mixing -g and -g0 objects. Are you sure doing the above
> > doesn't mess up tree merging enough to effectively enlarge WPA
> > memory use and the merged decl sections?
On Thu, Dec 10, 2015 at 6:05 PM, Sebastian Pop wrote:
> also update ISL to isl as requested by its author Sven Verdoolaege.
Ok. Please always post ChangeLog entries as well.
Richard.
> ---
> gcc/doc/install.texi | 9 +
> gcc/doc/invoke.texi | 4 ++--
> 2 files changed, 7 insertions(+
Hi,
This patch adds Pointer Bounds Checker support for VA_ARG calls. I added
bndret call for VA_ARG and corresponding bndret replacement when VA_ARG is
expanded. This fixes all vararg tests from MPX testsuite. Bootstrapped and
tested on x86_64-pc-linux-gnu. Will commit after additional test
> > For now I added the information if section is compressed into
> > decl_state. I am not thrilled by this but it is only way I found w/o
> > wasting 4 bytes per every lto section (because the lto header is not
> > really extensible and the stream is assumed to be aligned).
>
> So this trick
On Thu, Dec 10, 2015 at 6:05 PM, Sebastian Pop wrote:
> we now check the isl version, as there are no real differences in existing
> files
> in between isl 0.14 and isl 0.15.
I thought ISL 0.15 has some new features you could check? Also using a run test
is bad for cross compiling.
I'd simply
On Thu, Dec 10, 2015 at 8:33 PM, David Edelsohn wrote:
> On Thu, Dec 10, 2015 at 2:23 PM, Bill Schmidt
> wrote:
>> Hi,
>>
>> The subject test case has been failing as follows:
>>
>> FAIL: gcc.dg/vect/costmodel/ppc/costmodel-fast-math-vect-pr29925.c
>> scan-tree-dump-times vect "vectorization not
Ping.
https://gcc.gnu.org/ml/gcc-patches/2015-12/msg00526.html
Thanks,
Kyrill
On 04/12/15 09:30, Kyrill Tkachov wrote:
Hi all,
We don't handle properly the patterns for the [us]bfiz and [us]bfx instructions
when they
have an extend+ashift form. For example, the
*_ashl pattern.
This leads to
Hi,
Currently when MASK_LOAD and MASK_STORE is vectorized we check
scalar type of a mask but don't check its vector mask. It means
we may vectorize it when mask was just loaded from an array of
booleans. This happens e.g. for following test:
SUBROUTINE TEST (x, y, z, mask, ims, ime)
IMPLICIT
On Fri, Dec 11, 2015 at 12:03:26AM +, Joseph Myers wrote:
> On Thu, 10 Dec 2015, Manuel López-Ibáñez wrote:
> > On 12/09/2015 03:53 PM, Tobias Burnus wrote:
> > > In principle, %<%c%> and %<%d%> should be convertable to %qc and
> > > %qd (as the code is more readable), but the current function
Hi,
this patch removes a TERMINATE_PASS_LIST from passes.def that is not
paired with any INSERT_PASSES_AFTER.
Bootstrapped and reg-tested on x86_64.
OK for stage3 trunk?
Thanks,
- Tom
Remove unpaired TERMINATE_PASS_LIST in passes.def
2015-12-11 Tom de Vries
* passes.def: Remove unpaire
On Thu, Dec 10, 2015 at 9:08 PM, Jeff Law wrote:
> On 12/03/2015 07:38 AM, Richard Biener wrote:
>>
>> This pass is now enabled by default with -Os but has no limits on the
>> amount of
>> stmts it copies.
>
> The more statements it copies, the more likely it is that the path spitting
> will turn
On Thu, Dec 10, 2015 at 11:03 PM, Jason Merrill wrote:
> The C++ front end uses a temporary hash table to remember specializations of
> local variables during template instantiations. In a nested function such
> as a lambda or local class member function, we need to retain the elements
> from the
H Joern,
> Or have some target hook to make it not even bother filling delay slots
> speculatively; for targets that can fully unexpose the delay slot, like SH and
> ARC >= ARC700, this aspect of fill_eager_delay_slots only mucks up
> schedules and increases code size.
I propose to solve the dwar
On Fri, 11 Dec 2015, Shiva Chen wrote:
> Hi all,
>
> Loop unrolling would decrease performance in some case due to unrolling high
> register pressure
> loops and produce lots of spill code.
>
> I try to implement register pressure aware loop
> unrolling(-funroll-loops-pressure-aware)
> to avoi
On Fri, Dec 11, 2015 at 10:10 AM, Martin Liška wrote:
> Hello.
>
> I've just applied suggested change that Richi proposed.
> The patch can bootstrap on x86_64-linux-gnu and survives regression tests.
> Moreover,
> the memory leak/invalid read has gone.
>
> Ready for trunk?
Ok.
Thanks,
Richard.
On Fri, Dec 11, 2015 at 10:58 AM, Ilya Enkovich wrote:
> Hi,
>
> Currently when MASK_LOAD and MASK_STORE is vectorized we check
> scalar type of a mask but don't check its vector mask. It means
> we may vectorize it when mask was just loaded from an array of
> booleans. This happens e.g. for fol
On Fri, Dec 11, 2015 at 11:00 AM, Tom de Vries wrote:
> Hi,
>
> this patch removes a TERMINATE_PASS_LIST from passes.def that is not paired
> with any INSERT_PASSES_AFTER.
>
> Bootstrapped and reg-tested on x86_64.
>
> OK for stage3 trunk?
Ok.
Richard.
> Thanks,
> - Tom
LTO bootstrapped on x86_64-unknown-linux-gnu, applied.
Richard.
2015-12-11 Richard Biener
* lto-streamer.h (lto_simple_header_with_strings): Remove
main_size field already in lto_simple_header.
Index: gcc/lto-streamer.h
===
Hi,
This patch adds a parameter to TERMINATE_PASS_LIST, that should match
the pass list it's supposed to terminate.
The intention of the patch is that it:
- makes it easier to understand the top-level hierarchy of the pass
list (given that the top-level list may be quite long).
- ensures tha
Hi,
PR 68064 has been fixed by Richi's revision 231246. I would still
like to add the testcase to the testsuite and add a checking assert
so that if ever get zero alignment again, we catch it in the analysis
part of IPA-CP (which with LTO means in compilation and not linking
phase which makes a b
On 12/04/2015 06:15 PM, Dominik Vogt wrote:
> Version 5 with the latest requested changes. Seems to work now.
> I've dropped the extra patch and rather marked the failing tests
> as "xfail".
>
> Ciao
>
> Dominik ^_^ ^_^
>
Patch applied with minor changes:
> + ; Convert Pmode to BLKmode
> +
On 12/04/2015 06:12 PM, Dominik Vogt wrote:
> Version 6 with another fix. This should work now.
Applied. Thanks!
-Andreas-
This breaks bootstrap on darwin:
../../work/gcc/config/darwin.c: In function 'bool
darwin_use_anchors_for_symbol_p(const_rtx)':
../../work/gcc/config/darwin.c:3016:9: error: statement is indented as if it
were guarded by... [-Werror=misleading-indentation]
return default_use_anchors_for
On 04/12/15 12:39, Szabolcs Nagy wrote:
As described in pr other/67627, the all-multi target can be
built in parallel with the %_.lo targets which generate make
dependencies that are parsed during the build of all-multi.
gcc -MD does not generate the makefile dependencies in an
atomic way so mak
> On 11 Dec 2015, at 11:25, Dominique d'Humières wrote:
>
> This breaks bootstrap on darwin:
>
> ../../work/gcc/config/darwin.c: In function 'bool
> darwin_use_anchors_for_symbol_p(const_rtx)':
> ../../work/gcc/config/darwin.c:3016:9: error: statement is indented as if it
> were guarded by...
On 11/12/15 09:53 +0200, Ville Voutilainen wrote:
On 11 December 2015 at 09:52, Marc Glisse wrote:
/libstdc++-v3
* libsupc++/nested_exception.h (_S_rethrow): Use __std::addressof.
Typo.
I must be blind, but I don't see what you mean. :)
Shouldn't the underscores apply to addressof, not
On 10/12/15 17:58 -0500, DJ Delorie wrote:
I can't really judge this one. Either DJ or Jon would need to some in
on this.
Looks OK to me, although in the default configuration (plain DJGPP)
the #ifdefs will always be false (omitted), which is harmless.
Is there a non-default configuration
On Thu, Dec 10, 2015 at 05:56:26PM +0100, Jan Hubicka wrote:
> > Is this really necessary, is it not enough to remove the assignment to
> > ret below? If the parameter is not used, devirtualization time bonus,
> > which you then rely on estimate_local_effects, should be zero for it.
> >
> > It is
Hi,
while testing the oacc kernels patch series on top of trunk, using the
optimal handling of BUILTIN_IN_GOACC_PARALLEL in fipa-pta I ran into a
failure where the stores to the omp_data_sizes array were removed by dse.
The call bb in the failing testcase normally looks like this:
...
:
On Fri, 11 Dec 2015, Tom de Vries wrote:
> Hi,
>
> while testing the oacc kernels patch series on top of trunk, using the optimal
> handling of BUILTIN_IN_GOACC_PARALLEL in fipa-pta I ran into a failure where
> the stores to the omp_data_sizes array were removed by dse.
>
> The call bb in the f
On 13/11/15 12:39, Jakub Jelinek wrote:
We simply have some compiler internal interface between the caller and
callee of the outlined regions, each interface in between those has
its own structure type used to communicate the info;
we can attach attributes on the fields, or some flags to indicate
On 10/12/15 14:02 +, Jonathan Wakely wrote:
This adjusts some Doxygen comments and updates the Doxygen config file
to ensure all headers are processed (previously doxygen was ignoring
filenames without an extension, which is a lot of the library!)
There's a workaround in include/std/bitset f
On 11/12/15 12:47 +, Jonathan Wakely wrote:
* doc/doxygen/user.cfg.in: Use EXTENSION_MAPPING tag. Add new headers
to INPUT. Remove obsolete XML_SCHEMA and XML_DTD tags. Update
PREDEFINED macros. Set BRIEF_MEMBER_DESC for man-pages.
Oops, that changelog is not quite r
On Fri, 11 Dec 2015, Tom de Vries wrote:
> On 13/11/15 12:39, Jakub Jelinek wrote:
> > We simply have some compiler internal interface between the caller and
> > callee of the outlined regions, each interface in between those has
> > its own structure type used to communicate the info;
> > we can
Function (or more narrow) scope static variables (as well as others not
placed on the stack) should also not have any effect on the stack
alignment. I noticed the issue first with Linux'es dynamic_pr_debug()
construct using an 8-byte aligned sub-file-scope local variable.
According to my checking
On 12/11/2015 02:48 PM, Jan Beulich wrote:
Function (or more narrow) scope static variables (as well as others not
placed on the stack) should also not have any effect on the stack
alignment. I noticed the issue first with Linux'es dynamic_pr_debug()
construct using an 8-byte aligned sub-file-sco
I noticed a C++ test ICE the compiler trying to output a 128bit enum
initializer. The fix is crazy simple -- have nvptx_assemble_initializer return
false rather than ICE. Defining SUPPORTS_WIDE_INT is helpful, but not necessary.
While adding the testcase, I noticed I'd missed the opening '{'
Richard.
Thanks for your review.
I re-designed fix for assert by adding additional checks for vector
comparison with boolean result to fold_binary_op_with_conditional_arg
and remove early exit to combine_cond_expr_cond.
Unfortunately, I am not able to provide you with test-case since it is
in my se
On 12/11/2015 02:22 AM, Eric Botcazou wrote:
This patch allows a target to increase the cost of anti-deps to better
reflect the actual cost on the machine.
But it can already do it via the TARGET_SCHED_ADJUST_COST hook, can't it?
And can't this be done with define_bypass as well?
Jeff
The attached patch introduces wide int support to S/390 in order
to resolve a test case failure in gcc.dg/pr68129_1.c that is
caused by an assertion in
simplify-rtx.c:simplify_const_binary_operation().
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/ChangeLog
* config/s390/s390.
Dominik Vogt wrote:
> +; Note: Although CONST_INT and CONST_DOUBLE are not handled in this
> predicate,
> +; at least one of them needs to appear or otherwise safe_predicate_mode will
> +; assume that a DImode LABEL_REF is not accepted either (see genrecog.c).
The problem is not DImode LABEL_REF
On 08 Dec 13:53, Aleksandra Tsvetkova wrote:
> Wrong version of patch was attached.
>
> On Tue, Dec 8, 2015 at 1:46 PM, Aleksandra Tsvetkova
> wrote:
> > gcc/testsuite/ChangeLog
> > 2015-10-27 Tsvetkova Alexandra
> >
> > * gcc.target/i386/mpx/memmove-1.c: New test for __mpx_wrapper_memmov
On 12/11/2015 03:20 PM, Dominik Vogt wrote:
> The attached patch introduces wide int support to S/390 in order
> to resolve a test case failure in gcc.dg/pr68129_1.c that is
> caused by an assertion in
> simplify-rtx.c:simplify_const_binary_operation().
Applied with the change suggested by Uli. T
On Thu, Dec 10, 2015 at 4:43 PM, Patrick Palka wrote:
> This patch fixes name-lookup of operators in template definitions whose
> operands are non-dependent expressions, i.e. PR c++/21802 (and
> incidentally 53223).
>
> The approach that this patch takes is to detect when build_new_op()
> returns
Hi,
PR 66616 happens because in find_more_scalar_values_for_callers_subset
we do not do the same thunk checks like we do in
propagate_constants_accross_call. I am in the process of
bootstrapping and testing the following patch to fix it. OK if it
passes?
Thanks,
Martin
2015-12-11 Martin Jam
This test can timeout on PTX during execution. Reducing the number of
iterations as-if ptxx was a simulator, resolves that.
nathan
2015-12-11 Nathan Sidwell
* gcc.dg/pr59605-1.c: Reduce iterations for nvptx.
Index: gcc/testsuite/gcc.dg/pr59605-2.c
=
Good points.
I will send an updated patch following all your recommendations.
Sebastian
-Original Message-
From: Richard Biener [mailto:richard.guent...@gmail.com]
Sent: Friday, December 11, 2015 3:42 AM
To: Sebastian Pop
Cc: Sebastian Pop; GCC Patches; hiradi...@msn.com
Subject: Re: [PA
David Malcolm wrote:
> gcc/ChangeLog:
> * graphite-optimize-isl.c (scop_get_domains): Fix indentation.
> ---
> gcc/graphite-optimize-isl.c | 2 +-
> 1 file changed, 1 insertion(+), 1 deletion(-)
>
> diff --git a/gcc/graphite-optimize-isl.c b/gcc/graphite-optimize-isl.c
> index 8727e39..0c6a
Uros Bizjak wrote:
> > Finally the mechanical changes necessary due to the API change in the
> > walker.
>
> You forgot to change the graphite part, as in the attached patch.
>
> 2015-12-10 Uros Bizjak
>
> PR tree-optimization/68619
> * graphite-scop-detection.c (gather_bbs::before_d
Revision r231571 with Jan-Benedict Glaw’s fix for trailing whitespace.
Index: gcc/ChangeLog
===
--- gcc/ChangeLog (revision 231570)
+++ gcc/ChangeLog (working copy)
@@ -1,3 +1,13 @@
+2015-12-11 Jan-Benedict Glaw
+
On Thu 2015-12-10 19:12:57 -0500, Daniel Kahn Gillmor wrote:
> On Thu 2015-12-10 18:59:33 -0500, Joseph Myers wrote:
>> On Thu, 10 Dec 2015, Daniel Kahn Gillmor wrote:
>>
>>> Specifically, if the first character of the "old" argument is a
>>> literal $, then gcc will treat it as an environment vari
The attached patch enables using r1 to r4 as the literal pool base pointer if
one of them is unused in a leaf function. The unpatched code supports only r5
and r13.
Ciao
Dominik ^_^ ^_^
--
Dominik Vogt
IBM Germany
gcc/ChangeLog
* config/s390/s390.c (s390_init_frame_layout): Allow to
> Hi,
>
> PR 66616 happens because in find_more_scalar_values_for_callers_subset
> we do not do the same thunk checks like we do in
> propagate_constants_accross_call. I am in the process of
> bootstrapping and testing the following patch to fix it. OK if it
> passes?
>
> Thanks,
>
> Martin
>
On 12/11/2015 05:49 PM, Daniel Kahn Gillmor wrote:
I've re-rolled the patch (attached below, here) to use the ENV: prefix
instead of the $.
It might be irrelevant at this point, but the "ENV:" prefix is used in
AmigaOS and could be part of a filename.
+ if (0 == strncmp(ENV_PREFIX, arg, EN
Hi,
this is the regression of c-c++-common/opaque-vector.c on 32-bit targets where
'long double' is 128-bit large, for example PowerPC and SPARC, with an ICE in
the RTL expander because emit_store_flag is invoked with TImode.
As noted by Ilya, the underlying issue (ICE because emit_store_flag i
Hi,
atm, we dump ssa-name info for lhs-es of statements. That leaves out the
ssa names with default defs.
This proof-of-concept patch prints the ssa-name info for default defs,
in the following format:
...
__attribute__((noclone, noinline))
bar (intD.6 * cD.1755, intD.6 * dD.1756)
# PT = non
On Tue, Dec 08, 2015 at 05:45:59PM +0300, Ilya Verbin wrote:
> --- a/libgomp/oacc-init.c
> +++ b/libgomp/oacc-init.c
> @@ -306,10 +306,11 @@ acc_shutdown_1 (acc_device_t d)
> {
>struct gomp_device_descr *acc_dev = &base_dev[i];
>gomp_mutex_lock (&acc_dev->lock);
> - if (ac
On Thu, Dec 10, 2015 at 06:52:07PM +0100, Martin Jambor wrote:
> Good catch. I have modified this code so that it never leaves any
> holes in offload_names[i].
>
> > names[i] is null-terminated, so it looks like you're deliberately
> > allowing anything that starts with "hsa" here, but:
>
> Righ
On Fri, 2015-12-11 at 10:47 +0100, Richard Biener wrote:
> On Thu, Dec 10, 2015 at 8:33 PM, David Edelsohn wrote:
> > On Thu, Dec 10, 2015 at 2:23 PM, Bill Schmidt
> > wrote:
> >> Hi,
> >>
> >> The subject test case has been failing as follows:
> >>
> >> FAIL: gcc.dg/vect/costmodel/ppc/costmodel-
On Fri, Dec 11, 2015 at 18:27:13 +0100, Jakub Jelinek wrote:
> On Tue, Dec 08, 2015 at 05:45:59PM +0300, Ilya Verbin wrote:
> > --- a/libgomp/oacc-init.c
> > +++ b/libgomp/oacc-init.c
> > @@ -306,10 +306,11 @@ acc_shutdown_1 (acc_device_t d)
> > {
> >struct gomp_device_descr *acc_dev =
On Thu, Dec 10, 2015 at 06:52:23PM +0100, Martin Jambor wrote:
> I see, I prefer the clean approach, even if it is more work, this
> interface looks like it is going to be extended in the future. But I
> am wondering whether embedding the value into the identifier element
> is actually worth it.
Jan,
it looks like your recent changes to function_and_variable_visibility and
friends causes regressions in targets that do not support aliases (PTX for example).
specifically, we get a segfault in update_vtable_references (ipa-visibility.c)
at
*tp = symtab_node::get (*tp)->noninterp
This patch:
1) removes unused hard register defines
2) removes unused ASM_OUTPUT_COMMON, ASM_OUTPUT_LOCAL defines (we define the
aligned variants)
3) name the static chain registers.
4) formatting and simplification to bunch of function argument/return hooks. In
particular I found nvptx_fun
> Jan,
> it looks like your recent changes to
> function_and_variable_visibility and friends causes regressions in
> targets that do not support aliases (PTX for example).
>
> specifically, we get a segfault in update_vtable_references
> (ipa-visibility.c) at
> *tp = symtab_node::get (*tp)
The terminology within rich_location has become muddled, and
with the simplifications of the previous patch, I'd like to
rename things to better reflect what's going on.
A rich_location can contain one more more nested locations, each
of which can have a start, a finish, and optionally a caret.
Th
Previously, source_location/location_t values passed to
rich_location were immediately expanded (to expanded_location
instances stored inside the rich_location).
This patch updates the insides of class rich_location to delay
this expansion until the insides of diagnostic_show_locus.
This simplifi
Changes in v2:
- add two more testcases, based on additional reproducers
reported (one to the PR, another to a dup); now adds 17 PASS
results to gcc.sum.
- update for delayed expansion of locations within rich_location
and the terminology fixes from the other patches in this kit.
A
On Tue, 2015-11-24 at 13:43 +0100, Bernd Schmidt wrote:
> On 11/23/2015 07:26 PM, David Malcolm wrote:
> >
> > In theory we could attempt to try to handle this kind of thing by
> > looking at the macro expansions, and to print something like:
> >
> > 13 TEST_EQ (fmin);
> >
On 12/11/2015 05:10 AM, Richard Biener wrote:
On Thu, Dec 10, 2015 at 11:03 PM, Jason Merrill wrote:
The C++ front end uses a temporary hash table to remember specializations of
local variables during template instantiations. In a nested function such
as a lambda or local class member function
On Fri 2015-12-11 12:03:28 -0500, Bernd Schmidt wrote:
> On 12/11/2015 05:49 PM, Daniel Kahn Gillmor wrote:
>> I've re-rolled the patch (attached below, here) to use the ENV: prefix
>> instead of the $.
>
> It might be irrelevant at this point, but the "ENV:" prefix is used in
> AmigaOS and could
Maybe not the most important PR in the database, but we might as well
fix and close it. Count the number of alternatives in a MATCH_SCRATCH
against the max.
Bootstrapped and tested on x86_64-linux (one testcase timed out, almost
certainly because another test went heavily into swap at one poin
> Unfortunately this patch doesn't work properly on operator overloads
> that are defined as friend functions. E.g. the following now fails to
> compile:
>
> struct A
> {
> friend int operator* (A);
> };
>
> template
> void func(T t)
> {
> A x;
> int y = *x;
> }
>
> int main()
> {
> func(0);
On 12/11/2015 12:29 PM, Bernd Schmidt wrote:
Maybe not the most important PR in the database, but we might as well
fix and close it. Count the number of alternatives in a MATCH_SCRATCH
against the max.
Bootstrapped and tested on x86_64-linux (one testcase timed out, almost
certainly because anot
Hi,
The Linux kernel calls lse as atomics in /proc/cpuinfo. We should
change aarch64-option-extensions.def to take that into account.
OK? Bootstrapped and tested on aarch64-linux-gnu with no regressions
and tested with -mcpu=native on ThunderX T88 pass 2 with Linux 4.4 to
see if lse gets enabl
On Thu, Dec 10, 2015 at 08:30:37AM +0100, Jan Hubicka wrote:
> I am bootstrapping/regtesting this on x86_64-linux, does it seem OK?
I think this patch (just a guess, but certainly ipa-cp related during last
24 hours) significantly regressed guality/pr36728-*.c on x86_64.
Previously we have not tur
On Fri, Dec 11, 2015 at 07:30:59AM +0100, Richard Biener wrote:
> >So, to fix ICE on the following testcase, we can either do what the
> >patch
> >does, or could conditionalize both the calculate_dominance_info and
> >free_dominance_info in the convert_scalars_to_vector function (stv
> >pass)
> >on
Hi!
On Tue, Dec 08, 2015 at 12:56:20PM +0100, Bernd Schmidt wrote:
> On 12/07/2015 11:41 PM, Jakub Jelinek wrote:
> >On Mon, Dec 07, 2015 at 04:11:48PM +0100, Bernd Schmidt wrote:
> >>Let's document arguments; for the ones identical to read_cmdline_option an
> >>explicit pointer there is sufficien
On 11/26/2015 07:34 AM, Pierre-Marie de Rodat wrote:
On 11/25/2015 07:35 PM, Jason Merrill wrote:
Actually, even though my patches introduce DWARF procedures for only
one
case (size functions from stor-layout.c), they don’t necessarily come
from code generation (GENERIC): they are just a way to
On 12/11/2015 03:59 AM, Tom de Vries wrote:
Hi,
This patch adds a parameter to TERMINATE_PASS_LIST, that should match
the pass list it's supposed to terminate.
The intention of the patch is that it:
- makes it easier to understand the top-level hierarchy of the pass
list (given that the top-
> On Thu, Dec 10, 2015 at 08:30:37AM +0100, Jan Hubicka wrote:
> > I am bootstrapping/regtesting this on x86_64-linux, does it seem OK?
>
> I think this patch (just a guess, but certainly ipa-cp related during last
> 24 hours) significantly regressed guality/pr36728-*.c on x86_64.
> Previously we
On 12/11/2015 02:43 PM, Patrick Palka wrote:
+ if (processing_template_decl
+ && result != NULL_TREE
+ && result != error_mark_node
+ && DECL_HIDDEN_FRIEND_P (cand->fn))
+ {
+ tree call = result;
+ if (REFERENCE_REF_P (
On 12/11/2015 12:28 AM, Jan Beulich wrote:
gcc/c/
2015-12-10 Jan Beulich
* c-fold.c (c_fully_fold_internal): Also emit shift count
warnings for vector types.
* c-typeck.c (build_binary_op): Likewise.
Needs testcases for the added warnings.
My additional concern here
On 12/11/2015 01:13 PM, Jakub Jelinek wrote:
On Fri, Dec 11, 2015 at 07:30:59AM +0100, Richard Biener wrote:
So, to fix ICE on the following testcase, we can either do what the
patch
does, or could conditionalize both the calculate_dominance_info and
free_dominance_info in the convert_scalars_to
On 12/09/2015 10:08 AM, Paolo Bonzini wrote:
Left shifts into the sign bit is a kind of overflow, and the
standard chooses to treat left shifts of negative values the
same way.
However, the -fwrapv option modifies the language to one where
integers are defined as two's complement---which also de
On 12/11/2015 10:02 AM, Eric Botcazou wrote:
Hi,
this is the regression of c-c++-common/opaque-vector.c on 32-bit targets where
'long double' is 128-bit large, for example PowerPC and SPARC, with an ICE in
the RTL expander because emit_store_flag is invoked with TImode.
As noted by Ilya, the un
Actually I added
if (!ipa_is_param_used (info, i))
continue;
shortcut to gather_context_independent_values which prevents
us from recording context_independent_aggregate_values for
My attempt to implement LWG 2219 wasn't very good and resulted in
ambiguous overloads for the testcase in PR 59768. This fixes it, and
makes sure that const reference_wrappers can be used too.
Tested powerpc64le-linux, committed to trunk.
commit 9082b4be70fe2e4993ae0c765608fcd8da544bc5
Author:
> Presumably we know the code we're generating here is always gimple.
In fact all the callers of tree_vec_extract pass the result to gimplify_build
routines, so everything is gimplified if need be; it was over-gimplification.
> OK for the trunk.
Thanks.
--
Eric Botcazou
Patch ping.
Steve Ellcey
sell...@imgtec.com
On Tue, 2015-11-10 at 15:57 -0800, Steve Ellcey wrote:
> This patch removes the definition of TARGET_PROMOTE_PROTOTYPES from MIPS,
> where it was defined as true, so that it now defaults to false.
>
> Currently MIPS does prototype promotion in the cal
As is detailed thoroughly in the BZ entry; DOM was changed to not rely
on the jump threader to handle trivial conditionals.
Those changes twiddle the number of discovered & realized jump threads
on some targets. I've gone through the resultant dumps and verified
that we're doing the right t
1 - 100 of 105 matches
Mail list logo