On Fri, Oct 14, 2016 at 8:28 PM, Andrew Pinski wrote:
> On Thu, Sep 22, 2016 at 12:10 PM, Pat Haugen
> wrote:
>> I noticed the loop unroller peels an extra copy of the loop before it enters
>> the switch block code to round the iteration count to a multiple of the
>> unroll factor. This peeled
On Thu, Sep 22, 2016 at 12:10 PM, Pat Haugen
wrote:
> I noticed the loop unroller peels an extra copy of the loop before it enters
> the switch block code to round the iteration count to a multiple of the
> unroll factor. This peeled copy is only needed for the case where the exit
> test is at
On Sat, Oct 15, 2016 at 03:00:20AM +, Segher Boessenkool wrote:
> 2016-10-15 Segher Boessenkool
>
> * config/rs6000/rs6000.c (rs6000_get_separate_components): Do not
> make LR a separately shrink-wrapped component if savres_strategy
> contains any of {SAVE,REST}_INLINE_{GP
All out-of-line register save routines need LR, so we cannot wrap the
LR component if there are out-of-line saves. This didn't show up for
testing on Linux because none of the tests there use out-of-line FPR
saves without also using out-of-line GPR saves, which we did handle.
This patch fixes it,
Hi Bin,
On 15/10/16 00:15, Bin Cheng wrote:
+/* Test for likely overcommitment of vector hardware resources. If a
+ loop iteration is relatively large, and too large a percentage of
+ instructions in the loop are vectorized, the cost model may not
+ adequately reflect delays from unavaila
On 14 October 2016 22:41:25 CEST, Janne Blomqvist
wrote:
>On Fri, Oct 14, 2016 at 8:01 PM, Bernhard Reutner-Fischer
> wrote:
>> On 13 October 2016 22:08:21 CEST, Jerry DeLisle
> wrote:
>>>On 10/13/2016 08:16 AM, Janne Blomqvist wrote:
>>
Regtested on x86_64-pc-linux-gnu. Ok for trunk?
>
On Fri, Oct 14, 2016 at 05:24:53PM -0700, Steve Kargl wrote:
> For the code
>
> program foo
> stop merge(667, 668, .true.)
> end
>
> gfortran with either -std=f95 or -std=f2003 should reject
> this code. My patch does not fix this issue, because it
> would (1) require a complete rewri
This patch to the Go frontend and libgo copies the runtime package
time code from the Go 1.7 runtime.
This tweaks the frontend to fix the handling of function values for
-fgo-c-header to generate FuncVal*, not simply FuncVal.
While we're here change runtime.nanotime to use clock_gettime with
CLOC
The attach patch fixes a number of shortcomings with
STOP codes in gfortran. The updated comment in the
code nicely summarizes the problem.
/* Match a number or character constant after an (ERROR) STOP or PAUSE
- statement. */
+ statement. The requirements for a stop-code differs in the st
On Fri, Oct 14, 2016 at 02:37:42PM -0500, Bill Seurer wrote:
> [PATCH, rs6000] pr65479 Add option to fix failing asan test cases.
>
> This patch adds the -fasynchronous-unwind-tables option to several of the asan
> test cases. The option causes a full strack trace to be produced when the
> saniti
OK.
On Fri, Oct 14, 2016 at 1:29 PM, Jakub Jelinek wrote:
> Hi!
>
> Before early dwarf changes, if we wanted to note some decl property so that
> some corresponding DWARF attribute can be emitted, we had to use some
> generic IL bit for that. Now a langhook can be used instead (hopefully for
> 7
Implementing this proposal was a pretty straightforward matter of
changing the definition of aggregate and treating artificial base
fields as initializable in aggregate initialization. For this to work
with empty bases, I needed to create base fields for them, which we
haven't done in the past. b
Currently GFortran newer reuses unit numbers allocated with NEWUNIT=,
instead having a simple counter that is decremented each time such a
unit is opened. For a long running program which repeatedly opens
files with NEWUNIT= and closes them, the counter can wrap around and
cause an abort. This pa
On 10/14/2016 10:28 PM, David Malcolm wrote:
> I propose that we update our coding conventions to mention the OVERRIDE
> and FINAL macros in the paragraph that discusses virtual funcs.
>
> The attached patch (to the website) does so.
>
Good idea, I like it.
GDB is following GCC's C++ coding co
On Fri, 2016-10-14 at 16:27 +0100, Pedro Alves wrote:
> On 10/12/2016 03:13 PM, Bernd Schmidt wrote:
> > On 10/12/2016 04:09 PM, Pedro Alves wrote:
> > >
> > > Thanks. Here's a follow up patch that I was just testing.
> > >
> > > Need this if building with "g++ -std=gnu++11", with gcc < 4.7.
> >
On Fri, Oct 14, 2016 at 8:01 PM, Bernhard Reutner-Fischer
wrote:
> On 13 October 2016 22:08:21 CEST, Jerry DeLisle wrote:
>>On 10/13/2016 08:16 AM, Janne Blomqvist wrote:
>
>>>
>>> Regtested on x86_64-pc-linux-gnu. Ok for trunk?
>>>
>>
>>Yes, OK, clever! Thanks!
>
> Is 32 something a typical prog
On Fri, 2016-10-14 at 15:50 -0400, Nathan Sidwell wrote:
> On 10/14/16 15:17, David Malcolm wrote:
>
> > "Limits the maximum number of error messages to @var{n}, at which
> > point
> > GCC bails out rather than attempting to continue processing the
> > source
> > code. If @var{n} is 0 (the defaul
On Thu, Oct 13, 2016 at 09:45:22AM -0600, Kelvin Nilsen wrote:
> 3. Replace magic number 74 with CR6_REGNO in vsx.md (2 occurrences)
>and vector.md (3 occurrences).
Some remain, see below.
> +moves bytes 16 - @code{len} to 15 of the corresponding vector. For the
> +the element to be extract
On 10/14/2016 10:12 PM, David Malcolm wrote:
gcc/ChangeLog:
* print-rtl-function.c (print_edge): Omit "(flags)" when none are
set.
(print_rtx_function): Update example in comment for...
* print-rtl.c (print_rtx_operand_code_r): In compact mode, print
non-vi
On 10/14/16 15:17, David Malcolm wrote:
"Limits the maximum number of error messages to @var{n}, at which point
GCC bails out rather than attempting to continue processing the source
code. If @var{n} is 0 (the default), there is no limit on the number
of error messages produced. If @option{-Wf
Hi!
Another set of GNU extensions that were accepted into DWARF5, so we should
emit them even for -gstrict-dwarf -gdwarf-5, and for -gdwarf-5 should use
the accepted standard opcodes instead of the corresponding GNU ones.
Bootstrapped/regtested on x86_64-linux and i686-linux on top of the
dwarf2.
On 02/09/16 20:53 +0200, Eelis wrote:
On 2016-09-02 20:20, Eelis van der Weegen wrote:
On 2016-08-31 14:45, Jonathan Wakely wrote:
Is this significantly faster than just using
uniform_int_distribution<_IntType>{0, __bound - 1}(__g) so we don't
need to duplicate the logic? (And people maintainin
On Thu, 2016-10-13 at 16:18 +0200, Bernd Schmidt wrote:
> On 10/13/2016 04:08 PM, David Malcolm wrote:
> > I thought it might be useful to brainstorm [1] some ideas on this,
> > so here are various possible ways it could be printed for this use
> > -case:
> >
> > * Offset by LAST_VIRTUAL_REGISTER
With --target_board=unix/-std=gnu++17 this test fails, because
std::sample is defined too.
* testsuite/experimental/algorithm/sample.cc: Qualify calls to
resolve ambiguity between std::sample and std::experimental::sample.
Tested x86_64-linux, committed to trunk.
commit 8129c87
Hi!
The GNU DW_AT_GNU_macros and .debug_macro* extensions were added to DWARF5,
so we should emit those even in -gstrict-dwarf -gdwarf-5 mode.
Bootstrapped/regtested on x86_64-linux and i686-linux on top of the
dwarf2.{def,h} patch, ok for trunk?
2016-10-14 Jakub Jelinek
* dwarf2out.
[PATCH, rs6000] pr65479 Add option to fix failing asan test cases.
This patch adds the -fasynchronous-unwind-tables option to several of the asan
test cases. The option causes a full strack trace to be produced when the
sanitizer detects an error. Without the full trace the 3 test cases fail.
S
On Fri, 2016-10-14 at 21:27 +0200, Bernd Schmidt wrote:
> On 10/14/2016 09:25 PM, David Malcolm wrote:
> >
> > The behavior probably should be that it runs the remainder of the
> > RTL
> > passes from some specified point, and generates valid assembler (so
> > that we can have dg-do DejaGnu tests)
On October 14, 2016 7:20:43 PM GMT+02:00, Jakub Jelinek
wrote:
>Hi!
>
>The following (invalid) testcase ICEs, because we try to store into
>CONST_DECL's FIELD. Normally in GIMPLE we have MEM_REF[&C.1234] and
>writes to that expand gracefully into a MEM, but as soon as we use
>get_inner_reference
On 10/14/2016 09:25 PM, David Malcolm wrote:
The behavior probably should be that it runs the remainder of the RTL
passes from some specified point, and generates valid assembler (so
that we can have dg-do DejaGnu tests).
Actually I had imagined that tests would specify before and after RTL so
On Fri, 2016-10-14 at 11:33 +0200, Richard Biener wrote:
> On Thu, Oct 13, 2016 at 3:51 PM, Bernd Schmidt
> wrote:
> > On 10/13/2016 03:49 PM, Richard Biener wrote:
> > >
> > > Does it really run a single pass only? Thus you can't do a { dg
> > > -do run }
> > > test
> > > with __RTL?
> >
> >
On Thu, 2016-10-13 at 15:49 +0200, Richard Biener wrote:
> On Fri, Oct 7, 2016 at 5:58 PM, David Malcolm
> wrote:
> > On Wed, 2016-10-05 at 16:09 +, Joseph Myers wrote:
> > > On Wed, 5 Oct 2016, David Malcolm wrote:
> > >
> > > > @@ -1752,6 +1759,35 @@ c_parser_declaration_or_fndef (c_parser
On Thu, 2016-10-13 at 06:48 -0400, Nathan Sidwell wrote:
> On 10/11/16 16:07, David Malcolm wrote:
>
> > This logic is running when the next diagnostic is about to be
> > emitted.
> > But what if the user has selected -Wfatal-errors and there's a
> > single
> > error and no further diagnostics? C
Use new FPU instruction sequences documented in the ISA book to
implement __divsf3, __divdf3, __recipsf2, __recipdf2, __rsqrtsf2,
__rsqrtdf2 and __ieee754_sqrtf and __ieee754_sqrt.
2013-02-12 Ding-Kai Chen
libgcc/
* config/xtensa/ieee754-df.S (__recipdf2, __rsqrtdf2,
__ieee754_s
Tested on SPARC/Solaris, applied on the mainline.
2016-10-14 Eric Botcazou
* config/sparc/sparc-passes.def: New file.
* config/sparc/t-sparc (PASSES_EXTRA): Add sparc-passes.def.
* config/sparc/sparc-protos.h (make_pass_work_around_errata): New.
* config/sparc/
I think this is the last place where we weren't using std::__invoke()
for something the standard defines in terms of INVOKE. This should
make all our call wrappers and functional stuff consistent in terms of
support for invoking a pointer-to-member, reference_wrapper etc.
This means we can remove
Hi,
> Split-stack prologue on function entry is as follow (this goes before the
> usual function prologue):
> mrsx9, tpidr_el0
> movx10, -
As Jiong already remarked, the nop won't work. Do we know the maximum adjustment
that the linker is allowed to make? If so, and we can li
OK.
On Fri, Oct 14, 2016 at 1:23 PM, Jakub Jelinek wrote:
> Hi!
>
> We weren't implementing this DR, in the past all non-extern const vars
> (and non-inline) at namespace scope had internal linkage, but now only
> non-volatile const var.
>
> Fixed thusly, bootstrapped/regtested on x86_64-linux an
On 10/12/2016 09:27 PM, Denys Vlasenko wrote:
Yes, something like "if max_skip >= func_size, temporarily lower
max_skip to func_size-1" (because otherwise we can create padding
bigger-or-equal to the entire function in size, which is stupid
- it's better to just put the function in that space).
This removes redundant switches, reorders them and adds missing final -margs.
Tested on x86_64-suse-linux and SPARC/Solaris, applied on the mainline.
2016-10-14 Eric Botcazou
* gnat.dg/debug7.adb (dg-options): Remove -g.
* gnat.dg/debug8.adb (dg-options): Add -margs.
On Wed, 2016-10-12 at 22:57 +0100, Richard Sandiford wrote:
> Sorry, haven't had time to read the full series yet, but:
>
> David Malcolm writes:
> > On Wed, 2016-10-05 at 17:51 +0200, Bernd Schmidt wrote:
> > > On 10/05/2016 06:14 PM, David Malcolm wrote:
> > > > The selftests for the RTL fronte
Hi!
Now that DWARF5 public review draft has been released, I went through
the document looking for double dagger marked constants and added them to
dwarf2.{def,h}.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
2016-10-14 Jakub Jelinek
* dwarf2.h (enum dwarf_cal
Hi!
This also uses the infrastructure of the langhook patch I've sent earlier.
It emits (if not strict dwarf) DW_AT_inline on explicit or implicit inline
variables, and also tweaks dwarf2out so that for inline static data members
we consider in-class declarations as definitions (emit DW_AT_linkage
Hi!
This relies on the previous langhook patch (which greatly simplifies it).
I'm only handling variables for now, DW_AT_const_expr is just weird on
functions/methods, it is supposed to appear only on
DW_TAG_inlined_subroutine?
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
Hi!
Before early dwarf changes, if we wanted to note some decl property so that
some corresponding DWARF attribute can be emitted, we had to use some
generic IL bit for that. Now a langhook can be used instead (hopefully for
7.x even with LTO), but having a single langhook for each such bit looks
Hi!
We weren't implementing this DR, in the past all non-extern const vars
(and non-inline) at namespace scope had internal linkage, but now only
non-volatile const var.
Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for
trunk?
2016-10-14 Jakub Jelinek
DR 151
In order to port stack backtraces to Go, we need the ability to look
up file/line information for PC values without allocating memory.
This libgo patch moves the handling of Func from C code to Go code,
and simplifies the C code to just look up function/file/line/entry
information for a PC. Bootst
Hi!
The following (invalid) testcase ICEs, because we try to store into
CONST_DECL's FIELD. Normally in GIMPLE we have MEM_REF[&C.1234] and
writes to that expand gracefully into a MEM, but as soon as we use
get_inner_reference in expand_assignment (even if the MEM is just reverse
order, or we jus
This patch by James Clarke adds support for SPARC64/ELF relocs to the
debug/elf package in libgo. This is a backport of
https://golang.org/cl/30870 from the master library. Bootstrapped and
ran Go testsuite on x86_64-pc-linux-gnu. Committed to mainline.
Ian
Index: gcc/go/gofrontend/MERGE
==
On Fri, Sep 30, 2016 at 06:03:57PM +0100, James Greenhalgh wrote:
> Hi,
>
> Finally, this patch adds the back-end wiring to get AArch64 support for
> the _Float16 type working.
>
> Bootstrapped on AArch64 with no issues.
>
> OK?
I spotted a bug in the way I'd written aarch64_promoted_type. We wer
On 13 October 2016 22:08:21 CEST, Jerry DeLisle wrote:
>On 10/13/2016 08:16 AM, Janne Blomqvist wrote:
>>
>> Regtested on x86_64-pc-linux-gnu. Ok for trunk?
>>
>
>Yes, OK, clever! Thanks!
Is 32 something a typical program uses?
I'd have started at 8 and had not doubled but += 16 fwiw.
Cheers
On Fri, Sep 30, 2016 at 05:32:01PM +, Joseph Myers wrote:
> On Fri, 30 Sep 2016, James Greenhalgh wrote:
>
> >/* float.h needs to know this. */
> > + /* We already have the option -fno-fp-int-builtin-inexact to ensure
> > + certain built-in functions follow TS 18661-1 semantics. It
On Fri, Sep 30, 2016 at 11:28:28AM -0600, Jeff Law wrote:
> On 09/30/2016 11:01 AM, James Greenhalgh wrote:
> >
> >Hi,
> >
> >This patch ports the logic from m68k's TARGET_FLT_EVAL_METHOD to the new
> >target hook TARGET_C_EXCESS_PRECISION.
> >
> >Patch tested by building an m68k-none-elf toolchai
Hi,
On Fri, Oct 07, 2016 at 10:34:25AM +0200, Andreas Krebbel wrote:
> On 10/04/2016 03:42 PM, Joseph Myers wrote:
> > On Tue, 4 Oct 2016, Andreas Krebbel wrote:
> >
> >>> (b) Handling EXCESS_PRECISION_TYPE_IMPLICIT like
> >>> EXCESS_PRECISION_TYPE_FAST would accurately describe what the back end
On Fri, Sep 30, 2016 at 05:56:53PM +0100, James Greenhalgh wrote:
>
> This patch introduces TARGET_C_EXCESS_PRECISION. This hook takes a tri-state
> argument, one of EXCESS_PRECISION_TYPE_IMPLICIT,
> EXCESS_PRECISION_TYPE_STANDARD, EXCESS_PRECISION_TYPE_FAST. Which relate to
> the implicit extra p
This patch wires up generation of predicated instruction forms in nvptx.md and
fixes their handling in nvptx.c. This is a prerequisite for the following
patch. On its own it doesn't affect generated code because COND_EXEC
instructions are created by if-conversion only after register allocation,
w
This patch implements in nvptx.md a few new instruction patterns that are used
for OpenMP SIMD code.
* config/nvptx/nvptx-protos.h (nvptx_shuffle_kind): Move enum
declaration from nvptx.c.
(nvptx_gen_shuffle): Declare.
* config/nvptx/nvptx.c (nvptx_shuffle_kind): Mo
This patch adds a new target hook and implements it in a straightforward
manner on NVPTX to indicate that the target is running in SIMT fashion with 32
threads in a synchronous group ("warp"). For use in OpenMP transforms.
* config/nvptx/nvptx.c (nvptx_simt_vf): New.
(TARGET_SIMT_
Hi,
I'm resending the patch series with backend prerequisites for OpenMP
offloading to the NVIDIA PTX ISA. The patches are rebased on trunk.
Could a global reviewer have a look at patch 6 (new TARGET_SIMT_VF hook) please?
Documentation changes in doc/invoke.texi have already been reviewed
by Sa
This patch implements emission of OpenMP target region entrypoints: the
compiler emits the target function with '$impl' appended to the name, and
under the original name it emits a short entry sequence that sets up shared
memory arrays and calls the target function via 'gomp_nvptx_main' (which is
i
This patch wires up use of alternative -mgomp multilib for OpenMP offloading
via nvptx mkoffload. It makes OpenACC and OpenMP incompatible for
simultaneous offloading compilation, so I've added a diagnostic for that.
* config/nvptx/mkoffload.c (main): Check that either OpenACC or OpenMP
This patch implements '-msoft-stack' code generation variant for NVPTX. The
goal is to avoid relying on '.local' memory space for placement of automatic
data, and instead have an explicitely-maintained stack pointer (which can be
set up to point to preallocated global memory space). This allows t
This patch adds option -mgomp which enables -msoft-stack plus -muniform-simt,
and wires up the corresponding multilib variant. This codegen convention is
used for OpenMP offloading.
* config/nvptx/nvptx.c (diagnose_openacc_conflict): New. Use it...
(nvptx_option_override): ...her
This patch implements -muniform-simt code generation option, which is used to
emit code for OpenMP offloading. The goal is to emit code that can either
execute "normally", or can execute in a way that keeps all lanes in a given
warp active, their local state synchronized, and observable effects fr
While looking at LWG 2511 I realised that we can prevent
scoped_allocator_adaptor::construct(pair*, ...) from making any
copies internally. The transformed tuples that get passed to the
std::pair constructor (with additional allocator arguments) can be
tuples of references even if the incoming tu
On 10/12/2016 03:13 PM, Bernd Schmidt wrote:
> On 10/12/2016 04:09 PM, Pedro Alves wrote:
>>
>> Thanks. Here's a follow up patch that I was just testing.
>>
>> Need this if building with "g++ -std=gnu++11", with gcc < 4.7.
>
> Lovely. That's ok too if the other one goes in.
FYI, I pushed these i
2016-10-12 14:32 GMT+03:00 Georg-Johann Lay :
> This patch introduces a new variable attribute "absdata".
>
> Reduced Tiny cores have only a limited range of addresses 0x40..0xbf which
> can be handled by LDS / STS directly. The attribute allows the user to
> assert that it is legitimate to use ab
Hi Adhemerval,
On 06/10/16 22:54, Adhemerval Zanella wrote:
+ bool split_stack_arg_pointer_used = split_stack_arg_pointer_used_p ();
if (flag_stack_usage_info)
current_function_static_stack_size = frame_size;
@@ -3220,6 +3264,10 @@ aarch64_expand_prologue (void)
aarch64_em
I've committed this patch to fix failures in g++.dg/pch.exp and gcc.dg/pch.exp
tests when compiling with -mmicromips. Invalid data was being read from the
precompiled-header causing a segfault while trying to switch compression
context.
This patch fixes that segfault by initializing micromips_
Markus, thanks for fixing this during my night. Appreciate it!
Bill
> On Oct 14, 2016, at 6:08 AM, Markus Trippelsdorf
> wrote:
>
> On 2016.10.13 at 22:04 +0200, Rainer Orth wrote:
>> Hi Bill,
>>
>>> Here are torture test cases for
>>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77937. Mar
Hi,
the below, once again trivial, patch takes care of three additional
directories. Tested x86_64-linux.
Thanks, Paolo.
///
2016-10-14 Paolo Carlini
* testsuite/26_numerics/accumulate/1.cc: Remove 'test' variables.
* testsuite/26_numerics/adjacent_differe
Ping?
Best regards,
Thomas
On 03/10/16 17:46, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:50, Thomas Preudhomme wrote:
Hi,
This patch is part of a patch series to add support for atomic operations on
ARMv8-M Baseline targets in GCC. This specific patch enables atomi
Hi Richard,
> On Oct 14, 2016, at 4:19 AM, Richard Biener
> wrote:
>
> On Thu, Oct 13, 2016 at 5:38 PM, Bill Schmidt
> wrote:
>> The previous patch for
>> https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77937 is necessary, but not
>> sufficient in all cases. It allows -1 to be used with a pointe
Ping?
Best regards,
Thomas
On 03/10/16 17:45, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:47, Thomas Preudhomme wrote:
Hi,
This patch is part of a patch series to add support for atomic operations on
ARMv8-M Baseline targets in GCC. This specific patch adds support
Ping?
Best regards,
Thomas
On 03/10/16 17:44, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:44, Thomas Preudhomme wrote:
Hi,
This patch is part of a patch series to add support for atomic operations on
ARMv8-M Baseline targets in GCC. This specific patch refactors the
Ping?
Best regards,
Thomas
On 03/10/16 17:45, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:46, Thomas Preudhomme wrote:
Hi,
This patch is part of a patch series to add support for atomic operations on
ARMv8-M Baseline targets in GCC. This specific patch makes the nec
Ping?
Best regards,
Thomas
On 03/10/16 17:42, Thomas Preudhomme wrote:
Ping?
Best regards,
Thomas
On 22/09/16 14:41, Thomas Preudhomme wrote:
Hi,
This patch is part of a patch series to add support for atomic operations on
ARMv8-M Baseline targets in GCC. This specific patch adapts atomic
Hi,
It is suspected that for some large loops with too many vect_insns after
vectorization, the performance will be regressed because of various reasons.
Two possible reasons are:
1) Suggested by powerpc backend, large loops with too many vect_insns may jam
vector unit's pipeline.
2) Vectorizer
On Fri, Oct 14, 2016 at 1:12 AM, kugan
wrote:
> Hi Richard,
>
>
> On 13/10/16 20:44, Richard Biener wrote:
>>
>> On Thu, Oct 13, 2016 at 6:49 AM, kugan
>> wrote:
>>>
>>> Hi Richard,
>>>
what does this try to do? Preserve info VRP computed across PTA?
I think we didn't yet sor
The following fixes PR77979, yet another case where the jump-threading
code in VRP makes things difficult.
Bootstrapped on x86_64-unknown-linux-gnu, testing in progress.
Richard.
2016-10-14 Richard Biener
PR tree-optimization/77979
* tree-vrp.c (compare_name_with_value): Han
On 10/14/2016 01:33 PM, Markus Trippelsdorf wrote:
Thanks. ChangeLog entries for the testsuite are not mandatory.
They are actually.
Bernd
Richard,
Here is updated patch with the changes proposed by you.
Bootstrapping and regression testing did not show any new failures.
Is it OK for trunk?
ChangeLog:
2016-10-14 Yuri Rumyantsev
* dominance.c (dom_info::dom_info): Add new constructor for region
presented by vector of basic block
I've encountered another case where the assumption that only a
bind vars and its block vars may share a common chain -- for
-finstrument-functions or TSAN the gimplifier violates this,
making the chain shared between different binds vars and block.
This can cause BLOCK_VARs to be dropped (I exper
Now that the warning in PR77742 is fixed we can make std::allocator
use aligned new for types with extended alignment.
This only changes the allocators that actually allocate memory
themselves. The others (array, extptr, the testsuite ones) don't need
any changes because they simply forward (de)
On Thu, Oct 13, 2016 at 5:26 PM, Martin Liška wrote:
> Simplified version of tests, where I added tests for side effects.
>
> Patch can bootstrap on ppc64le-redhat-linux and survives regression tests.
>
> Ready to be installed?
Ok.
Richard.
> Martin
On Thu, Oct 13, 2016 at 5:26 PM, Martin Liška wrote:
> Simplified version that supports only valid null-terminated string constants.
> Apart from that, I added checking for constant folding of expressions that
> have side effects.
>
> Patch can bootstrap on ppc64le-redhat-linux and survives regres
On Thu, Oct 13, 2016 at 5:24 PM, Martin Liška wrote:
> Simplified version that just supports only null-terminated strings.
> Patch can bootstrap on ppc64le-redhat-linux and survives regression tests.
>
> Ready to be installed?
Ok.
Richard.
> Martin
On Fri, Oct 14, 2016 at 1:12 PM, Martin Liška wrote:
> On 10/13/2016 11:39 AM, Richard Biener wrote:
>> Yes, as said, the patch is ok as-is.
>>
>> Richard.
>
> Installed, I've just tested the patch on 5 and 6 branch.
> May I install it as well?
Yes please.
Ricahrd.
> M.
On Tue, Sep 20, 2016 at 2:31 PM, Robin Dapp wrote:
> Hi,
>
>> I meant to do sth like
>>
>> Index: tree-ssa-propagate.c
>> ===
>> --- tree-ssa-propagate.c(revision 240133)
>> +++ tree-ssa-propagate.c(working copy)
>> @@
Hi Markus,
>> You should have added a ChangeLog entry for that change. Unfortunately,
>> neither the original testcases nor the fixed ones can ever have been
>> properly tested: there's no such thing as dg-do options, leading to
>> horrendous errors in mail-report.log:
>>
>> ERROR: gcc.dg/tortur
On 2016.10.14 at 13:28 +0200, Rainer Orth wrote:
> Hi Markus,
>
> > On 2016.10.13 at 22:04 +0200, Rainer Orth wrote:
> >> Hi Bill,
> >>
> >> > Here are torture test cases for
> >> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77937. Markus Trippelsdorf
> >> > kindly provided the source for the
Hi Markus,
> On 2016.10.13 at 22:04 +0200, Rainer Orth wrote:
>> Hi Bill,
>>
>> > Here are torture test cases for
>> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77937. Markus Trippelsdorf
>> > kindly provided the source for the tests and verified the correct
>> > dejagnu options on x86_64-pc-
On 10/13/2016 11:39 AM, Richard Biener wrote:
> Yes, as said, the patch is ok as-is.
>
> Richard.
Installed, I've just tested the patch on 5 and 6 branch.
May I install it as well?
M.
On 10/14/2016 11:38 AM, Richard Biener wrote:
> On Thu, Oct 13, 2016 at 5:23 PM, Martin Liška wrote:
>> Hello.
>>
>> After receiving feedback from Richi and Wilco Dijkstra, I decided to fully
>> not
>> support not null-terminated strings. It brings more complications and the
>> code has started
On 2016.10.13 at 22:04 +0200, Rainer Orth wrote:
> Hi Bill,
>
> > Here are torture test cases for
> > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=77937. Markus Trippelsdorf
> > kindly provided the source for the tests and verified the correct
> > dejagnu options on x86_64-pc-linux-gnu. Committe
On Fri, Oct 14, 2016 at 12:57 PM, Bernd Schmidt wrote:
> On 10/14/2016 11:26 AM, Richard Biener wrote:
>>
>> On Fri, Oct 14, 2016 at 11:15 AM, Bernd Schmidt
>> wrote:
>>>
>>> So maybe it should just call clear_bb_flags instead of doing the loop
>>> itself? Ok if that works.
>>
>>
>> That doesn't
On Fri, Oct 14, 2016 at 1:00 PM, Nathan Sidwell wrote:
> On 10/14/16 05:28, Richard Biener wrote:
>
>> The BB_VISITED flag has indetermined state at the beginning of a pass.
>> You have to ensure it is cleared yourself.
>
>
> In that case the openacc (&nvptx?) passes should be modified to clear th
On 10/14/16 05:28, Richard Biener wrote:
The BB_VISITED flag has indetermined state at the beginning of a pass.
You have to ensure it is cleared yourself.
In that case the openacc (&nvptx?) passes should be modified to clear the flags
at their start, rather than at their end.
nathan
On 10/14/2016 11:26 AM, Richard Biener wrote:
On Fri, Oct 14, 2016 at 11:15 AM, Bernd Schmidt wrote:
So maybe it should just call clear_bb_flags instead of doing the loop
itself? Ok if that works.
That doesn't generally work, it clears too many flags (it was appearantly
designed for RTL).
O
> That said - why do we need to remove pcommit support on the branches at all?
I think it is better to have old branches updated because pcommit will
not be implemented in hardware.
--
WBR,
Andrew
On 10/13/16 18:10, Andrew Pinski wrote:
/home/jenkins/workspace/BuildToolchainAARCH64_thunder_elf_upstream/toolchain/scripts/../src/libgcc/libgcov-driver.c:53:0:
/home/jenkins/workspace/BuildToolchainAARCH64_thunder_elf_upstream/toolchain/scripts/../src/libgcc/../gcc/gcov-io.c:
In function ‘__gc
1 - 100 of 125 matches
Mail list logo
