Re: [PATCHv2] do not throw in std::make_exception_ptr

2016-08-05 6:49 GMT+02:00 Gleb Natapov : > Instead of throwing an exception allocate its memory and initialize it > explicitly. Makes std::make_exception_ptr more efficient since no stack > unwinding is needed. [..] > +#ifndef _CXXABI_INIT_EXCEPTION_H > +#define

RE: [ARM] mno-pic-data-is-text-relative & msingle-pic-base

Nathan, Sorry for jumping in a relatively old thread. I saw this in mailing list archives during a web search (I wasn't on this mailing list before). I decided to speak up as I would be an affected party if this patch (or some similar future patch) gets accepted or worse yet, the feature

[RS6000] PR72802 part 2, reload ICE

After fixing the constraint problem, we hit an "insn does not satisfy its constraints" with -mno-lra on the following insn, a vector load from mem which has an invalid offset: (insn 631 630 1122 12 (set (reg:SF 108 31 [orig:260 pretmp_44 ] [260]) (mem:SF (plus:DI (reg:DI 30 30 [orig:338

[PATCHv2] do not throw in std::make_exception_ptr

Instead of throwing an exception allocate its memory and initialize it explicitly. Makes std::make_exception_ptr more efficient since no stack unwinding is needed. v1->v2: - fix indentation - drop static from __dest_thunk - use static_cast instead of reinterpret_cast * libsupc++/exception

Re: [PATCH] Improve forward jump threading of switch statements (PR18046)

On Thu, Jul 28, 2016 at 11:46 PM, Patrick Palka wrote: > This patch improves the forward jump threader's ability to thread > GIMPLE_SWITCHes by making the VRP simplification callback attempt to > determine which case label will be taken. > > For example, if the index operand

Re: protected alloca class for malloc fallback

On 08/04/2016 05:30 AM, Aldy Hernandez wrote: Howdy! As part of my -Walloca-larger-than=life work, I've been running said pass over gcc, binutils, and gdb, and trying to fix things along the way. Particularly irritating and error prone is having to free malloc'd pointers on every function exit

[RS6000] PR72802 part 1, fix constraints for lxssp/stxssp

We can't use "o" constraint for lsxxp/stxssp since those insns have a DS-form offset field, ie. the bottom two bits of the offset must be 0. So use "wY" instead, but that leads to finding another problem.. mem_operand_gpr is only suitable for gpr loads/stores since it does not enforce

[patch] New libstdc++ docs on testing and library versioning

I've been working on some changes to the libstdc++ manual recently. A lot of the changes are just using DocBook markup with more semantic meaning (e.g. or instead of using for everything that should use a monospace font) but there are some changes to content too. I've added a new subsection

Re: Implement C _FloatN, _FloatNx types

On Thu, 4 Aug 2016, Michael Meissner wrote: > It brings up a general complaint that I had when I started the whole PowerPC > __float128 adventure is that we have multiple places in the compiler it goes > through a list of floating point types, and only for the blessed types does it > do

Re: Ping Re: Implement C _FloatN, _FloatNx types [version 5]

On Thu, 4 Aug 2016, Michael Meissner wrote: > Or is the check at the end whether ieee_bits > (for extended tests) or == (for > normal tests) sufficient to say that SFmode, DFmode, really is IEEE? Yes. That's the semantics of ieee_bits as detailed in the comment in real.h. Note that the

Re: Implement C _FloatN, _FloatNx types

In gcc/c-family/c-prettyprint.c in the function that you modified pp_c_floating_constant: It brings up a general complaint that I had when I started the whole PowerPC __float128 adventure is that we have multiple places in the compiler it goes through a list of floating point types, and only for

Re: Ping Re: Implement C _FloatN, _FloatNx types [version 5]

On Fri, Jul 29, 2016 at 05:37:26PM +, Joseph Myers wrote: > Ping. This patch > > (non-C-front-end parts) is pending review. I'm starting to look at the patches. It will take some time to come up to speed, particularly with the

Re: [RFC] ipa bitwise constant propagation

Hi Honza, On 04/08/16 23:05, Jan Hubicka wrote: I didn't look at the propagation part but eventually the IPA-CP lattice gets quite big. Also the alignment lattice is very similar to the bits lattice so why not merge those two? But This was always the original idea to replace alignment

Re: [PATCH v2,rs6000] Add built-in function support for Power9 binary floating point operations

On Thu, Aug 04, 2016 at 09:27:55AM -0600, Kelvin Nilsen wrote: > The patch has been bootstrapped and tested on powerpc64le-unknown-linux > and on powerpc64-unknown-linux (big-endian) with no regressions. Is > this ok for the trunk? Just a couple more nits.. Okay for trunk with those fixed /

Re: [PATCH build/doc] Replacing libiberty with gnulib

On 08/02/2016 12:38 AM, Manuel López-Ibáñez wrote: > > If there is something wrong or missing, ideally we would like to know > so that Ayush can work on fixing it before the Summer of Code is over > in less than two weeks. I couldn't see anywhere gnulib's config.h file is included. gnulib's

Re: [PATCH] Replacing gcc's dependence on libiberty's fnmatch to gnulib's fnmatch

On 07/30/2016 02:10 AM, Manuel López-Ibáñez wrote: > What about my suggestion of forcing GCC to use the gnulib functions by > temporarily removing the system-wide functions? Would that be > equivalent testing to building on a host that requires the libiberty > version of a function? I don't

Re: [PATCH] Replacing gcc's dependence on libiberty's fnmatch to gnulib's fnmatch

On 07/31/2016 11:50 PM, Manuel López-Ibáñez wrote: > Oh well, Ayush does not have access to different hosts, thus I > guess it is better that he focuses his limited time on functions that > he can test. There are plenty of functions that are both in gnulib and > libiberty but not in glibc. > My

[PATCH, i386]: Fix wrong cast in _mm512_{,mask_}cmp_ep{i,u}32_mask intrinsics

2016-08-04 Uros Bizjak PR target/72805 * config/i386/avx512fintrin.h (_mm512_cmp_epi32_mask) [!__OPTIMIZE__]: Cast builtin function result to __mmask16 instead of __mmask8. (_mm512_cmp_epu32_mask) [!__OPTIMIZE__]: Ditto. (_mm512_mask_cmp_epi32_mask)

Re: [PATCH, ARM] Add a new target hook to compute the frame layout

Hi, this patch introduces a new target hook that allows the target's INITIAL_ELIMINATION_OFFSET function to use cached values instead of re-computing the frame layout every time. I have updated the documentation a bit and hope it is clearer this time. It still needs a review by ARM port

Re: [PATCH] - improve sprintf buffer overflow detection (middle-end/49905)

On 07/21/2016 03:48 PM, Jeff Law wrote: Sorry about the delay in responding... I saw a few places in GCC itself where you increased buffer sizes. Were any of those level 1 failures? Yes. With optimization, even level 1 uses range information when it's available. The idea is that when

Re: [PATCH 4/4] c-format.c: suggest the correct format string to use (PR c/64955)

On Thu, 2016-08-04 at 13:55 -0600, Jeff Law wrote: > On 08/03/2016 09:45 AM, David Malcolm wrote: > > This adds fix-it hints to c-format.c so that it can (sometimes) > > suggest > > the format string the user should have used. > > > > The patch adds selftests for the new code in c-format.c.

Re: [PING**2] [PATCH] Fix asm X constraint (PR inline-asm/59155)

On 07/21/2016 10:29 AM, Bernd Edlinger wrote: But I think we have a use case where "X" means really more possible registers (i.e. includes ss2, mmx etc.) than "g" (only general registers). Otherwise, in the test cases of pr59155 we would not have any benefit for using "+X" instead of "+g" or

Re: [PATCH 3/4] Use class substring_loc in c-format.c (PR c/52952)

On 08/04/2016 01:24 PM, David Malcolm wrote: Do you realize that this isn't used for ~700 lines after this point? Is there any sensible way to factor some code here to avoid the coding disconnect. I realize the function was huge before you got in here, but if at all possible, I'd like to see

Re: [PATCH 2/4] (v3) On-demand locations within string-literals

On 08/04/2016 01:21 PM, David Malcolm wrote: + +static enum cpp_ttype +get_cpp_ttype_from_string_type (tree string_type) +{ + gcc_assert (string_type); + if (TREE_CODE (string_type) != ARRAY_TYPE) +return CPP_OTHER; + + tree element_type = TREE_TYPE (string_type); + if (TREE_CODE

make streaming routines for widest_int non-static

Hi, This patch makes streamer_read_wi and streamer_write_wi non-static, and exports them from lto-streamer.h. I suppose this hunk could be committed independently of the ipa-bitwise-cp patch ? Bootstrap+test in progress on x86_64-unknown-linux-gnu. OK for trunk ? Thanks, Prathamesh 2016-08-05

Re: [tree-ssa-ccp] modify extend_mask to extend bits based on signop

Ah, the mail failed to be delivered to gcc-patches, sorry for the double-post. On 5 August 2016 at 01:26, Prathamesh Kulkarni wrote: > Hi, > Is the attached patch OK ? > Since we want to extend based on signop, I removed ORing with wi::mask(). > Bootstrap+test

Re: [PATCH 4/4] c-format.c: suggest the correct format string to use (PR c/64955)

On 08/03/2016 09:45 AM, David Malcolm wrote: This adds fix-it hints to c-format.c so that it can (sometimes) suggest the format string the user should have used. The patch adds selftests for the new code in c-format.c. These selftests are thus lang-specific. This is the first time we've had

Re: protected alloca class for malloc fallback

On 08/04/2016 08:26 PM, Jeff Law wrote: > On 08/04/2016 01:21 PM, Pedro Alves wrote: >> Most allocas I've seen in my life were written to simply build >> strings at run time, while lazily avoiding to think about writing >> a "free" call, rather than having been written for optimizing >> some

Re: protected alloca class for malloc fallback

On 08/04/2016 01:21 PM, Pedro Alves wrote: On 08/04/2016 08:16 PM, Jeff Law wrote: On 08/04/2016 01:06 PM, Pedro Alves wrote: How wedded are we to alloca? I would think only in the sense of existing codebase usage. Based on how often alloca usage has resulted in a security vulnerability it's

Re: [PATCH 3/4] Use class substring_loc in c-format.c (PR c/52952)

On Thu, 2016-08-04 at 12:08 -0600, Jeff Law wrote: > On 08/03/2016 09:45 AM, David Malcolm wrote: > > This patch updates c-format.c to use the new class substring_loc, > > added > > in the previous patch, replacing location_column_from_byte_offset. > > Hence with this patch, Wformat can underline

Re: protected alloca class for malloc fallback

On 08/04/2016 09:19 AM, Aldy Hernandez wrote: Particularly irritating and error prone is having to free malloc'd pointers on every function exit point. We end up with a lot of: foo(size_t len) { void *p, *m_p = NULL; if (len < HUGE) p = alloca(len); else p = m_p =

Re: protected alloca class for malloc fallback

On 08/04/2016 08:16 PM, Jeff Law wrote: > On 08/04/2016 01:06 PM, Pedro Alves wrote: >> How wedded are we to alloca? > I would think only in the sense of existing codebase usage. > > Based on how often alloca usage has resulted in a security vulnerability > it's clear we as developers can't use

Re: [PATCH 2/4] (v3) On-demand locations within string-literals

On Thu, 2016-08-04 at 11:37 -0600, Jeff Law wrote: > On 08/03/2016 09:45 AM, David Malcolm wrote: > > Changes in v3: > > - Avoid including cpplib.h from input.h > > - Properly handle stringified macro arguments (with tests for this) > > - Minor whitespace fixes > > - Move selftest.h changes to a

Re: fix fallout of pr22051-2.c on arm

On 4 August 2016 at 12:39, Richard Biener wrote: > On Thu, 4 Aug 2016, Prathamesh Kulkarni wrote: > >> Hi, >> The attached patch fixes pr22051-2.c which regressed due to >> r238754. Matthew, could you please confirm if this patch fixes the >> test-case for you ? >> >>

Re: protected alloca class for malloc fallback

On 08/04/2016 01:06 PM, Pedro Alves wrote: How wedded are we to alloca? I would think only in the sense of existing codebase usage. Based on how often alloca usage has resulted in a security vulnerability it's clear we as developers can't use it correctly on a consistent basis, thus I'd like

Re: protected alloca class for malloc fallback

How wedded are we to alloca? On 08/04/2016 12:30 PM, Aldy Hernandez wrote: > Is there a better blessed C++ way? std::string or if I'm reading gcc's vec.h correctly: auto_vec ? (or some typedef of the latter) The latter would always allocate the stack space, but

Re: [PATCH] do not throw in std::make_exception_ptr

On 04/08/16 21:36 +0300, Gleb Natapov wrote: On Thu, Aug 04, 2016 at 07:01:45PM +0100, Jonathan Wakely wrote: The Doxygen group that starts with @{ ends later in this file, but you haven't moved the corresponding @} to the new file. I can take care of fixing that up though (we want the

Re: [PATCH 1/4] selftest.h: Add ASSERT_TRUE_AT and ASSERT_FALSE_AT

On Wed, 2016-08-03 at 10:06 -0600, Jeff Law wrote: > On 08/03/2016 09:45 AM, David Malcolm wrote: > > I split out the selftest.h changes from v2 of the kit for ease of > > review; > > here they are. > > > > Successfully bootstrapped in conjunction with the rest > > of the > > patch kit on

Re: [C++ Patch] PR 72800

OK. On Thu, Aug 4, 2016 at 1:02 PM, Paolo Carlini wrote: > Hi, > > when back in 2014 I sent the patch for c++/61088 I noticed some cases where > we wanted to return early error_mark_node from add_capture to avoid ICEs > during error recovery when COMPLETE_TYPE_P is used

Re: [PR63240] generate debug info for defaulted member functions

On Mon, Aug 1, 2016 at 10:57 PM, Alexandre Oliva wrote: > * langhooks.h (struct lang_hooks_for_decls): Add > function_decl_defaulted_inclass_p and > function_decl_defaulted_outofclass_p. Maybe one hook that returns unsigned rather than two bool hooks?

Re: [PATCH] do not throw in std::make_exception_ptr

On Thu, Aug 04, 2016 at 07:01:45PM +0100, Jonathan Wakely wrote: > On 04/08/16 20:01 +0300, Gleb Natapov wrote: > > Instead of throwing an exception allocate its memory and initialize it > > explicitly. Makes std::make_exception_ptr more efficient since no stack > > unwinding is needed. > > > >

Re: [PATCH 3/4] Use class substring_loc in c-format.c (PR c/52952)

On 08/03/2016 09:45 AM, David Malcolm wrote: This patch updates c-format.c to use the new class substring_loc, added in the previous patch, replacing location_column_from_byte_offset. Hence with this patch, Wformat can underline the precise erroneous format string in many more cases. The patch

[PATCH, Fortran] New flag -finit-derived to initialize components of derived types

All, With many other compilers, local variables are automatically initialized to zero (or some other user-specified value) by default. GNU Fortran allows this with the options -finit-local-zero, -finit-real=, -finit-integer=, etc... However several other compilers also initialize structure

[PATCH] Define std::is_callable and std::is_nothrow_callable

Two new traits for C++17, as well as identical __is__callable and __is_nothrow_callable traits defined for C++11 mode (so I can use the latter to add an exception specification to std::__invoke). * doc/xml/manual/status_cxx2017.xml: Update status table. * include/std/functional

Re: [PATCH] do not throw in std::make_exception_ptr

On 04/08/16 20:01 +0300, Gleb Natapov wrote: Instead of throwing an exception allocate its memory and initialize it explicitly. Makes std::make_exception_ptr more efficient since no stack unwinding is needed. In this version I hopefully addressed all Jonathan comments. * libsupc++/exception

Re: [PATCH 2/4] (v3) On-demand locations within string-literals

On 08/03/2016 09:45 AM, David Malcolm wrote: Changes in v3: - Avoid including cpplib.h from input.h - Properly handle stringified macro arguments (with tests for this) - Minor whitespace fixes - Move selftest.h changes to a separate patch Changes in v2: - Tweaks to substring location selftests

Re: [PATCH 1/3] (v2) On-demand locations within string-literals

On 08/04/2016 08:27 AM, David Malcolm wrote: As for test coverage, v2 and v3 of the kit add over a thousand lines of selftest code that heavily exercise string lexing, using the line_table_case machinery to run the tests with various interesting boundary conditions with line_table (e.g. near

Re: Implement -Wimplicit-fallthrough (take 2): fix missing breaks

On Thu, Jul 28, 2016 at 8:57 AM, Marek Polacek wrote: > On Wed, Jul 27, 2016 at 10:05:25AM -0700, Mike Stump wrote: >> On Jul 27, 2016, at 9:52 AM, Marek Polacek wrote: >> > >> > This is what the new warning pointed out. I think all these are bugs. >> >

Re: [PATCH, rs6000] Switch the rs6000 port over to LRA

On 8/3/16 6:03 PM, David Edelsohn wrote: Please open a Bugzilla for the rs6000 backend about the vsx-timode performance regression. The vsx-timode regression needs to be fixed for GCC 7. Ok, I opened https://gcc.gnu.org/PR72804 and will start debugging the problem. Peter

Re: [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch

On 08/04/16 12:43, Nathan Sidwell wrote: How about: gcov_t expected; atomic_load ([0], val, ...); gcov_t delta = val == value ? 1 : -1; atomic_add ([1], delta); <-- or atomic_add_fetch if (delta < 0) { /* can we set counter[0]? */ atomic_load ([1], , ...); if (expected < 0) {

Re: Use correct location information for OpenACC shape and simple clauses in C/C++

On 07/27/2016 09:17 AM, Thomas Schwinge wrote: Hi! I found that for a lot of OpenACC (and potentially also OpenMP) clauses (in C/C++ front ends; didn't look at Fortran), we use wrong location information. The problem is that c_parser_oacc_all_clauses/c_parser_omp_all_clauses calls

[C++ Patch] PR 72800

Hi, when back in 2014 I sent the patch for c++/61088 I noticed some cases where we wanted to return early error_mark_node from add_capture to avoid ICEs during error recovery when COMPLETE_TYPE_P is used on an error_mark_node. The new testcase noticed one additional case, where the

[PATCH] do not throw in std::make_exception_ptr

Instead of throwing an exception allocate its memory and initialize it explicitly. Makes std::make_exception_ptr more efficient since no stack unwinding is needed. In this version I hopefully addressed all Jonathan comments. * libsupc++/exception (std::exception): Move... *

Re: Fix fir PR71696 in Libiberty Demangler (6)

On 08/04/2016 01:07 AM, Marcel Böhme wrote: Hi Jeff, Can you take care of the minor issues above, retest & repost? Sure. I removed the whitespace nits, used XDUPVEC instead of XNEWVEC+memcpy, and adjusted the growing heuristics of the new array proctypevec. The revised patch is attached

Re: [PATCH] gcov tool: Implement Hawick's algorithm for cycle detection, (PR gcov-profile/67992)

On 08/04/16 12:10, Martin Liška wrote: On 08/04/2016 05:13 PM, Nathan Sidwell wrote: On 08/04/16 10:42, Martin Liška wrote: I decided to use a new enum, hope it's better? that's fine. But you know, if you set the enum values appropriately you could use the | trick rather than the compare

Re: [PATCH] Fix wrong code on aarch64 due to paradoxical subreg

On Wed, Aug 03, 2016 at 04:08:30PM -0600, Jeff Law wrote: > On 08/03/2016 11:41 AM, Bernd Edlinger wrote: > >On 08/03/16 17:38, Jeff Law wrote: > >>cse.c changes look good, but I'd really like to see a testcase for each > >>issue in the dejagnu framework. Extra points if you tried to build a >

Re: libgo patch committed: Update to 1.7rc3

On Thu, Aug 4, 2016 at 1:11 AM, Uros Bizjak wrote: > On Thu, Aug 4, 2016 at 12:53 AM, Ian Lance Taylor wrote: >> On Thu, Jul 28, 2016 at 4:24 AM, Uros Bizjak wrote: >>> >>> A new testsuite failure is introduced: >>> >>> FAIL: text/template

Re: Implement -Wimplicit-fallthrough (take 2): the rest

On 08/04/2016 06:36 AM, Michael Matz wrote: Hi, On Wed, 27 Jul 2016, Marek Polacek wrote: And this is the rest. Either I just adjusted a falls through comment, or I added __builtin_fallthrough (). These were the cases where I was fairly sure that the fall through is intentional. I saw one

Re: [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch

On 08/04/16 11:34, Martin Liška wrote: On 08/04/2016 04:48 PM, Nathan Sidwell wrote: diff --git a/libgcc/libgcov-profiler.c b/libgcc/libgcov-profiler.c +static inline void +__gcov_one_value_profiler_body_atomic (gcov_type *counters, gcov_type value) +{ ... The body looks to have data races.

Re: PING: new pass to warn on questionable uses of alloca() and VLAs

On 07/27/2016 03:01 AM, Aldy Hernandez wrote: Just in case this got lost in noise, since I know there was a lot of back and forth between Martin Sebor and I. This is the last iteration. Tested on x86-64 Linux. OK for trunk? curr gcc/ * Makefile.in (OBJS): Add

Re: Go patch committed: add escape analysis debugging

On Thu, Aug 4, 2016 at 6:47 AM, Rainer Orth wrote: > >> This patch by Chris Manghane adds debugging to the escape analysis >> code. This debugging is designed to generate the same sort of output >> as the gc Go compiler, for easier comparison of results. Escape

Re: [PR55641] drop spurious const_type from reference_type variables

OK. On Mon, Aug 1, 2016 at 11:12 PM, Alexandre Oliva wrote: > Although C++ reference types, denoted by DW_TAG_reference_type in > DWARFv2+ debug info, are unchangeable, we output names of reference type > with DW_TAG_const_type, because internally we mark such variables as >

Re: [PATCH] gcov tool: Implement Hawick's algorithm for cycle detection, (PR gcov-profile/67992)

On 08/04/2016 05:13 PM, Nathan Sidwell wrote: > On 08/04/16 10:42, Martin Liška wrote: > >> I decided to use a new enum, hope it's better? > > that's fine. But you know, if you set the enum values appropriately you > could use the | trick rather than the compare you've done (c++ enum type >

C++ PATCH for c++/72415 (concepts ICE with fold-expression constraint)

Substituting into the fold-expression was producing an EXPR_PACK_EXPANSION, but it would be better to keep it as a fold-expression. tsubst_unary_left_fold et al try to do this, but weren't recognizing the case where tsubst_pack_expansion returns a TREE_VEC containing a single pack expansion. In

Re: [PATCH][AArch64] Improve stack adjustment

On 04/08/16 12:06, Wilco Dijkstra wrote: > Improve stack adjustment by reusing a temporary move immediate > from the epilog if the register is still valid in the epilog. This generates > smaller code for leaf functions: > > mov x16, 4 > sub sp, sp, x16 > ldr

C++ PATCH to rename TYPE_ANONYMOUS_P to TYPE_UNNAMED_P

Martin pointed out that the macro name TYPE_ANONYMOUS_P is confusing because it doesn't identify anonymous structs/unions; the C++ standard consistently refers to classes with no name as "unnamed", not "anonymous", so let's use that term internally as well. Tested x86_64-pc-linux-gnu, applying to

C++ PATCH for c++/72796 (wrong resolution of scoped call)

On this testcase, finish_class_member_access_expr unpacked the SCOPE_REF, changing the name variable, and then used that variable to return a dependent COMPONENT_REF. We are already holding onto the original value of name, we should use that instead. Tested x86_64-pc-linux-gnu, applying to

Re: [PING] Use correct location information for OpenACC shape and simple clauses in C/C++

On Thu, 2016-08-04 at 16:54 +0200, Thomas Schwinge wrote: > Hi! > > On Wed, 27 Jul 2016 17:09:38 -0400, David Malcolm < > dmalc...@redhat.com> wrote: > > On Wed, 2016-07-27 at 17:17 +0200, Thomas Schwinge wrote: > > > I found that for a lot of OpenACC (and potentially also OpenMP) > > > clauses >

Re: [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch

On 08/04/2016 04:48 PM, Nathan Sidwell wrote: > diff --git a/libgcc/libgcov-profiler.c b/libgcc/libgcov-profiler.c > +static inline void > +__gcov_one_value_profiler_body_atomic (gcov_type *counters, gcov_type value) > +{ > ... > > The body looks to have data races. Some kind of cmp_store needed

Re: [PATCH][AArch64] Add legitimize_address_displacement hook

On 04/08/16 12:00, Wilco Dijkstra wrote: > This patch adds legitimize_address_displacement hook so that stack accesses > with large offsets are split into a more efficient sequence. Byte and > halfword > accesses use a 4KB range, wider accesses use a 16KB range to maximise the > available

[PATCH v2,rs6000] Add built-in function support for Power9 binary floating point operations

This patch adds built-in support for the following fourteen new binary floating point instructions introduced with the Power9 architecture: VSX Scalar Extract Exponent Double-Precision VSX Scalar Extract Significand Double-Precision VSX Scalar Insert Exponent Double-Precision VSX Scalar Compare

Re: protected alloca class for malloc fallback

On 08/04/2016 08:57 AM, Richard Biener wrote: On Thu, Aug 4, 2016 at 1:30 PM, Aldy Hernandez wrote: Howdy! As part of my -Walloca-larger-than=life work, I've been running said pass over gcc, binutils, and gdb, and trying to fix things along the way. Particularly irritating

Re: [PATCH] gcov tool: Implement Hawick's algorithm for cycle detection, (PR gcov-profile/67992)

On 08/04/16 10:42, Martin Liška wrote: I decided to use a new enum, hope it's better? that's fine. But you know, if you set the enum values appropriately you could use the | trick rather than the compare you've done (c++ enum type safety would require an overloaded | operator though). I

Re: [RS6000] rs6000_preferred_reload_class

On Thu, Aug 04, 2016 at 02:44:34PM +0930, Alan Modra wrote: > On Wed, Aug 03, 2016 at 03:30:53PM -0400, Michael Meissner wrote: > > On Mon, Aug 01, 2016 at 09:00:43AM +0930, Alan Modra wrote: > > > Hi Mike, > > > I've been looking at a lot of reload/lra code lately in trying to fix > > > pr71680,

Re: [PATCH], Improve vector int/long initialization on PowerPC

Hi Mike, On Thu, Aug 04, 2016 at 12:33:44AM -0400, Michael Meissner wrote: > I built spec 2006 with these patches on a little endian power8 system, and at > least 18 of the benchmarks had vector initializations replaced. Most > benchmarks only used the initialization in a few places, but gamess,

Re: [PATCH/AARCH64] Add ThunderX vector cost model

On 03/08/16 23:42, Andrew Pinski wrote: > Hi, > This patch adds to the thunderx model, the vector cost model. I > benchmarked this on SPEC CPU INT 2006 and got a small speed up. I > have a few more cost model patches that I am going upstream but they > are going to be split up. > > OK?

[PING] Use correct location information for OpenACC shape and simple clauses in C/C++

Hi! On Wed, 27 Jul 2016 17:09:38 -0400, David Malcolm wrote: > On Wed, 2016-07-27 at 17:17 +0200, Thomas Schwinge wrote: > > I found that for a lot of OpenACC (and potentially also OpenMP) > > clauses > > (in C/C++ front ends; didn't look at Fortran), we use wrong location >

Re: [PATCH] gcov tool: Implement Hawick's algorithm for cycle detection, (PR gcov-profile/67992)

On 08/04/2016 03:15 PM, Nathan Sidwell wrote: > On 08/04/16 06:41, Martin Liška wrote: >> On 08/03/2016 04:22 PM, Nathan Sidwell wrote: >>> Martin, As I've going through all PRs related to gcov-profile, I've noticed this PR. Current implementation of cycle detection in gcov is very

Re: libgo patch committed: Update to 1.7rc3

On Thu, Aug 4, 2016 at 5:12 AM, Rainer Orth wrote: > Hi Uros, > >>> I have committed a patch to update libgo to the 1.7rc3 release >>> candidate. This is very close to the upcoming 1.7 release. As usual >>> with libgo updates, the patch is too large to include in

[GCC Steering Committee attention] [PING] [PING] [PING] libgomp: In OpenACC testing, cycle though $offload_targets, and by default only build for the offload target that we're actually going to test

Hi! Ping. It has now been more than three months (!) that I first submitted this, without receiving any meaningful review. (Apart from one initial "deprecative" comment by Jakub, which I then repeatedly detailed on, without receiving any further response.)

Re: [PATCH 1/4] Cherry-pick fprofile-generate-atomic from google/gcc-4_9 branch

On 08/01/16 09:29, Martin Liška wrote: I also added a small hunk that describes problematic of app having not-joined (or detached) threads, can you please take a look at documentation change, maybe it would need some transformation? sorry for the tady response,thanks for the ping. In

[PATCH][PR64971]Convert function pointer to Pmode when emit call

Hi all, In the case of PR64971 (https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64971), the compiler ICE when compiling gcc.c-torture/compile/pr37433.c with ilp32 abi. As we know, in aarch64 ilp32, the ptr_mode is SImode while Pmode is still DImode. It means all address should be DImode, and

Re: [PATCH 1/3] (v2) On-demand locations within string-literals

On Wed, 2016-08-03 at 09:59 -0600, Jeff Law wrote: > On 07/29/2016 03:42 PM, Joseph Myers wrote: > > On Tue, 26 Jul 2016, David Malcolm wrote: > > > > > This patch implements precise tracking of source locations for > > > the > > > individual chars within string literals, so that we can e.g. > >

Re: [PTX] fix worker propagation ICE

Hi! On Wed, 3 Aug 2016 13:30:10 -0400, Nathan Sidwell wrote: > --- libgomp/testsuite/libgomp.oacc-c-c++-common/crash-1.c (nonexistent) > +++ libgomp/testsuite/libgomp.oacc-c-c++-common/crash-1.c (working copy) > @@ -0,0 +1,28 @@ > +/* { dg-do compile } */ > +/* {

Re: Use "oacc kernels" attribute for OpenACC kernels

Hi! Ping. On Wed, 27 Jul 2016 12:06:59 +0200, I wrote: > On Mon, 25 Jan 2016 16:09:14 +0100, Jakub Jelinek wrote: > > On Mon, Jan 25, 2016 at 10:06:50AM -0500, Nathan Sidwell wrote: > > > On 01/04/16 10:39, Nathan Sidwell wrote: > > > >There's currently no robust predicate to

Re: Test cases to check OpenACC offloaded function's attributes and classification

Hi! Ping. On Wed, 27 Jul 2016 10:59:02 +0200, I wrote: > Hi! > > OK for trunk? > > commit 8200af082db5438be18bc60f721fcf21641c0d86 > Author: Thomas Schwinge > Date: Tue Jul 26 17:18:21 2016 +0200 > > Test cases to check OpenACC offloaded function's attributes

Re: Rework C/C++ OpenACC routine parsing

Hi! On Fri, 22 Jul 2016 16:22:18 +0200, Jakub Jelinek wrote: > On Wed, Jul 13, 2016 at 04:10:31PM +0200, Thomas Schwinge wrote: > > @@ -14029,29 +14032,32 @@ c_parser_oacc_kernels_parallel (location_t loc, > > c_parser *parser, > > static void > > c_parser_oacc_routine

Re: C/C++: Simplify handling of location information for OpenACC routine directives

Hi! On Wed, 13 Jul 2016 11:25:46 +0200, I wrote: > C/C++: Simplify handling of location information for OpenACC routine > directives Without changes, committed to trunk in r239127: commit 5f429ee2993ea1795d88c5589251c500e6e9062a Author: tschwinge

Re: Tighten syntax checking for OpenACC routine construct in C

Hi! On Tue, 24 May 2016 16:02:39 +0200, I wrote: > Committed without changes in r236639: > > commit c9d624bd2672463771546e73bf3d6446d64e43c0 > Author: tschwinge > Date: Tue May 24 14:00:39 2016 + > > Tighten syntax checking for OpenACC

Re: [Fortran, Patch, pr70524, v1] [5/6/7 Regression] ICE when using -frepack-arrays -Warray-temporaries

The patch works as advertised. It would be nice to have it reviewed and committed. TIA Dominique

Re: C++ OpenACC routine directive testing: templated, and "auto", trailing return type syntax

Hi! On Tue, 12 Jul 2016 15:55:37 +0200, Jakub Jelinek wrote: > On Tue, Jul 12, 2016 at 03:50:14PM +0200, Thomas Schwinge wrote: > > Templated, and "auto", trailing return type syntax with the C++ OpenACC > > routine directive all works, but doesn't have test coverage. OK for >

Re: [PATCH] Teach VRP to truncate the case ranges of a switch

On Thu, Aug 4, 2016 at 2:14 PM, Patrick Palka wrote: > On Thu, 4 Aug 2016, Richard Biener wrote: > >> On Thu, Aug 4, 2016 at 4:30 AM, Patrick Palka wrote: >> > On Wed, 3 Aug 2016, David Malcolm wrote: >> > >> >> On Wed, 2016-08-03 at 15:47 +0200,

Re: Go patch committed: add escape analysis debugging

Hi Ian, > This patch by Chris Manghane adds debugging to the escape analysis > code. This debugging is designed to generate the same sort of output > as the gc Go compiler, for easier comparison of results. Escape > analysis is still not enabled by default. Bootstrapped and ran Go > testsuite

Re: split test cases pr71078-1.c and pr71078-2.c

On Thu, 4 Aug 2016, Prathamesh Kulkarni wrote: > Hi, > The attached patch splits each test-case into three, one for float, > double and long-double. > I verified that the long double tests are unsupported now for arm target. > OK to commit ? Ok. Richard. > Thanks, > Prathamesh > -- Richard

[PATCH] Update C++17 library implementation status table

This adds all the features from the latest draft of SD-6, and reformats the table to be closer to the tables at https://gcc.gnu.org/projects/cxx-status.html (including adding the feature-test macros). * doc/xml/manual/status_cxx2017.xml: Update C++17 status table. *

Re: [PATCH] gcov tool: Implement Hawick's algorithm for cycle detection, (PR gcov-profile/67992)

> On 08/04/16 06:41, Martin Liška wrote: > >On 08/03/2016 04:22 PM, Nathan Sidwell wrote: > >>Martin, > >>>As I've going through all PRs related to gcov-profile, I've noticed this > >>>PR. > >>>Current implementation of cycle detection in gcov is very poor, leading to > >>>extreme run time >

split test cases pr71078-1.c and pr71078-2.c

Hi, The attached patch splits each test-case into three, one for float, double and long-double. I verified that the long double tests are unsupported now for arm target. OK to commit ? Thanks, Prathamesh 2016-08-04 Prathamesh Kulkarni testsuite/ *

Re: [PATCH] gcov tool: Implement Hawick's algorithm for cycle detection, (PR gcov-profile/67992)

On 08/04/16 06:41, Martin Liška wrote: On 08/03/2016 04:22 PM, Nathan Sidwell wrote: Martin, As I've going through all PRs related to gcov-profile, I've noticed this PR. Current implementation of cycle detection in gcov is very poor, leading to extreme run time for cases like mentioned in the

Re: [RFC] ipa bitwise constant propagation

> I didn't look at the propagation part but eventually the IPA-CP > lattice gets quite big. Also the alignment lattice is very > similar to the bits lattice so why not merge those two? But This was always the original idea to replace alignment propagation by bitwise ccp. I suppose we only have

Re: protected alloca class for malloc fallback

On Thu, Aug 4, 2016 at 1:30 PM, Aldy Hernandez wrote: > Howdy! > > As part of my -Walloca-larger-than=life work, I've been running said pass > over gcc, binutils, and gdb, and trying to fix things along the way. > > Particularly irritating and error prone is having to free

  1   2   >