Re: [PATCH] Fix unrecognizable insn of pr92865

On Wed, Dec 11, 2019 at 3:54 PM Jakub Jelinek wrote: > > On Wed, Dec 11, 2019 at 09:55:24AM +0800, Hongtao Liu wrote: > > Changelog > > gcc/ > > PR target/92865 > > * config/i386/i386-expand.c (ix86_valid_mask_cmp_mode): Enable > > integer mask cmov when available even with

Re: [PATCH] Fix unrecognizable insn of pr92865

On Wed, Dec 11, 2019 at 09:55:24AM +0800, Hongtao Liu wrote: > Changelog > gcc/ > PR target/92865 > * config/i386/i386-expand.c (ix86_valid_mask_cmp_mode): Enable > integer mask cmov when available even with TARGET_XOP. > > gcc/testsuite > *

Re: [PATCH] OpenACC reference count overhaul

On 2019/12/10 12:04 AM, Julian Brown wrote: I'm citing below the changes introducing 'gomp_remove_var_async', modelled similar to the existing 'gomp_unmap_vars_async'. Also for both these, do I understand correctly, that it's actually not the 'gomp_unref_tgt' that needs to be "delayed" via

[PATCH] Fix gnu-versioned-namespace build

I plan to commit this tomorrow. Note that rather than just adding the missing _GLIBCXX_[BEGIN,END]_VERSION_NAMESPACE I also move anonymous namespace usage outside std namespace. Let me know if it was intentional.     * src/c++11/random.cc: Add _GLIBCXX_BEGIN_NAMESPACE_VERSION and    

Re: [PATCH 0/2] Make C front end share the C++ tree representation of loops and switches

On 12/6/19 3:41 PM, Jeff Law wrote: On Wed, 2019-11-13 at 09:27 -0700, Sandra Loosemore wrote: I bootstrapped and regression-tested this on x86_64-linux-gnu. There are a few regressions involving these tests: gcc.dg/tree-ssa/pr77445-2.c I believe tihs is another instance of the FSA

[PATCH] Rename condition_variable_any::wait_on_* methods

User-agent: mu4e 1.3.4; emacs 26.2 * include/std/condition_variable (condition_variable_any::wait_on(_Lock&, stop_token, _Predicate): Rename to match current draft standard. (condition_variable_any::wait_on_until(_Lock&, stop_token, const

Re: [PATCH] Restore enable_if lost during original import of pstl

Tested x86_64-pc-linux-gnu, committed to trunk, backported to gcc-9-branch. Jonathan Wakely writes: > On 18/11/19 20:54 -0800, Thomas Rodgers wrote: >> >> * include/pstl/glue_numeric_defs.h: Restore enable_if lost during >> original >> import of pstl. >> *

Re: Ping: [GCC][PATCH] Add ARM-specific Bfloat format support to middle-end

On Mon, 2019-12-09 at 13:40 +, Stam Markianos-Wright wrote: > > On 12/3/19 10:31 AM, Stam Markianos-Wright wrote: > > > > On 12/2/19 9:27 PM, Joseph Myers wrote: > > > On Mon, 2 Dec 2019, Jeff Law wrote: > > > > > > > > 2019-11-13 Stam Markianos-Wright < > > > > >

[PATCH] Add abs pattern to handle {si,di} mode abs to avoid pmax/cmove conversion (PR92651)

Hi: Currently smax/smin pattern added by r274481 cause some regression in 525.x264_r by 8% with -O2 -march=corei7. The reason is some IA backends (contain TARGET_SSE4_1) will do transform for simple abs (using rshift, xor and sub) to pmax/pmin if smax/smin pattern exists, which generate

Re: [PATCH] Fix multibyte-related issues in pretty-print.c (PR 91843)

On Tue, 2019-12-10 at 09:36 -0500, Lewis Hyatt wrote: > On Mon, Dec 9, 2019 at 4:58 PM David Malcolm > wrote: > > On Thu, 2019-10-10 at 16:27 -0400, Lewis Hyatt wrote: > > > Hello- > > > > > > This short patch addresses > > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91843 > > > by adding the

Re: [PATCH] Fix unrecognizable insn of pr92865

On Tue, Dec 10, 2019 at 4:11 PM Jakub Jelinek wrote: > > On Tue, Dec 10, 2019 at 01:47:50PM +0800, Hongtao Liu wrote: > > This patch is to enable integer mask cmp/cmov under AVX512F even > > with TARGET_XOP . > > Bootstrap and regression test on i386/x86_64 backend is ok. > > > > Changelog: >

[patch] factor out common files for compare_exclusions

the toplevel configure.ac repeats common exclusion files for specific targets. Just factor those out. Maybe not required, but gm2 is adding more files to be ignored on every target, so make it easy to only have these files mentioned in one place. Ok for the trunk? Matthias 2019-12-11 Matthias

Go patch committed: Build type descriptor for alias in other package

This patch to the Go frontend by Cherry Zhang generates type descriptors for pointers to aliases defined in other packages. When a type descriptor is needed (for e.g. interface conversion), if the type is a pointer to a named type defined in another package, we don't generate the definition of

[PATCH] libstdc++: Correct noexcept-specifiers on span constructors

As discussed at https://github.com/cplusplus/draft/issues/3534 two std::span constructors specify incorrect conditions for throwing exceptions. This patch makes those constructors have correct noexcept-specifiers that accurately reflect what can actually throw. (span(ContiguousIterator,

Re: [PATCH] Fix ICE in compute_objsize (PR tree-optimization/92891)

On Wed, 2019-12-11 at 00:23 +0100, Jakub Jelinek wrote: > Hi! > > The following testcase ICEs, because gimple_call_alloc_size doesn't > always > return a sizetype typed INTEGER_CST, which the callers rely on > (compare > those converted to wide_int with other wide_ints with the sizetype >

[PATCH] Fix ICE in compute_objsize (PR tree-optimization/92891)

Hi! The following testcase ICEs, because gimple_call_alloc_size doesn't always return a sizetype typed INTEGER_CST, which the callers rely on (compare those converted to wide_int with other wide_ints with the sizetype precision). If alloc_size attribute has two arguments, gimple_call_alloc_size

Re: [PATCH] avoid invoking assignment on uninitialized objects (PR 92761, 92762)

On 12/10/19 3:07 PM, David Malcolm wrote: On Tue, 2019-12-03 at 15:41 -0700, Martin Sebor wrote: After allocating a new chunk of memory hash_table::expand() copy- assigns elements from the current array over the newly allocated elements without constructing them. Similarly, after destroying

Re: [PATCH] PR 92846: [ARC] generate signaling FDCMPF for hard float comparisons

On 12/10/19 1:12 AM, Claudiu Zissulescu wrote: > Hi, > > Thank you for your contribution, I'll push it asap. As far as I understand, > you need this patch both in gcc9 branch and mainline. > > Cheers, > Claudiu Indeed both mainline and gcc9 Thx -Vineet > >> -Original Message- >> From:

Re: [Patch, Fortran] PR92898 - [9/10 Regression] ICE in gfc_check_is_contiguous, at fortran/check.c:7157

Hello Harald, Index: gcc/fortran/check.c === --- gcc/fortran/check.c (Revision 279183) +++ gcc/fortran/check.c (Arbeitskopie) @@ -7154,7 +7154,9 @@ bool gfc_check_is_contiguous (gfc_expr *array) { if (array->expr_type ==

patch to fix PR92796

The following patch fixes https://gcc.gnu.org/bugzilla/show_bug.cgi?id=92796 The patch was successfully bootstrapped and tested on x86-64. Committed as r279204 Index: ChangeLog === --- ChangeLog (revision 279200) +++ ChangeLog

Re: [PATCH] avoid invoking assignment on uninitialized objects (PR 92761, 92762)

On Tue, 2019-12-03 at 15:41 -0700, Martin Sebor wrote: > After allocating a new chunk of memory hash_table::expand() copy- > assigns elements from the current array over the newly allocated > elements without constructing them. > > Similarly, after destroying existing elements, hash_table:: >

[Patch, Fortran] PR92898 - [9/10 Regression] ICE in gfc_check_is_contiguous, at fortran/check.c:7157

The testcase in the PR exhibits a corner case in a check on invalid code that was not handled appropriately and in turn ICEs. The patch below enhances that check. Instead of adding a new testcase, I modified the related one that came with the 'introduction' of the regression when fixing PR91641.

[patch, fortran, committed] Fix PR 91643, repacking of assumed rank argument

Hello world, this fixes a regression introduced by my inline repacking patch. With the test case, it is simple and obvious enough - do not repack an assumed rank argument (which makes no sense). Committed as obvious and simple as r279203 after regression-testing. 2019-12-10 Thomas Koenig

Re: [patch, fortran] Introduce -finline-pack

Am 09.12.19 um 17:30 schrieb Thomas Koenig: Maybe -finline-repack would be a better name? -finline-internal-pack? Steve made an excellent suggestion: -finline-arg-packing . So, OK with that change? Regards Thomas

[C++ PATCH] Improve C++ error recovery (PR c++/59655)

Hi! On the following testcase, we emit 2 errors and 1 warning, when the user really should see one error. The desirable error is static_assert failure, the bogus error is during error recovery, complaining that a no_linkage template isn't defined when it really is defined, but we haven't

[PATCH] Fix vect rotate pattern recog (PR target/92723)

Hi! Unlike x86, where the last operand of vector by scalar shift is DImode for V[248]DImode shifts, on aarch64 they seem to be SImode. vect_recog_rotate_pattern when the rotate amount is not constant casts the amount to the element type of the vector, so for V[248]DImode vectors to DImode, but

[C++ PATCH] Fix -std=c++17 and earlier handling of CLASSTYPE_NON_AGGREGATE (PR c++/92869)

Hi! In C++20, types with user-declared constructors are not aggregate types, while in C++17 only types with user-provided or explicit constructors. In check_bases_and_members we handle it properly: CLASSTYPE_NON_AGGREGATE (t) |= ((cxx_dialect < cxx2a ?

Re: [PATCH] Improve -fstack-protector-strong (PR middle-end/92825)

On Tue, 2019-12-10 at 10:27 +0100, Jakub Jelinek wrote: > Hi! > > As mentioned in the PR, -fstack-protector-strong in some cases > instruments > even functions which except for the stack canary setup and later > testing of > that don't touch the stack at all. > > This is because for

Re: [PATCH] Fix propagate_vr_across_jump_function (PR ipa/92883)

On Tue, 2019-12-10 at 21:33 +0100, Jakub Jelinek wrote: > Hi! > > The ipa vr hash table apparently intentionally doesn't differentiate > between > value ranges with different types, all that matters are the values of > min and max, so before using it ipa_vr_operation_and_type_effects > needs to >

Re: [PATCH] Fix ICE in regstat_bb_compute_calls_crossed (PR rtl-optimization/92882)

On Tue, 2019-12-10 at 21:40 +0100, Jakub Jelinek wrote: > Hi! > > The following testcase ICEs on the newly added asserts. > Some comments hint that maybe it is fine if CODE_LABEL additions > don't > trigger df recomputations, but even if it is not ok, > regstat_bb_compute_calls_crossed doesn't

[PATCH] Fix ICE in regstat_bb_compute_calls_crossed (PR rtl-optimization/92882)

Hi! The following testcase ICEs on the newly added asserts. Some comments hint that maybe it is fine if CODE_LABEL additions don't trigger df recomputations, but even if it is not ok, regstat_bb_compute_calls_crossed doesn't look like an IL verification routine and for !NONDEBUG_INSN_P it really

[committed] Fix some comment typos

Hi! While working on the PR92883 patch I've noticed some typos, fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, committed to trunk as obvious. 2019-12-10 Jakub Jelinek * ipa-param-manipulation.c (ipa_param_body_adjustments::register_replacement): Fix

[PATCH] Fix propagate_vr_across_jump_function (PR ipa/92883)

Hi! The ipa vr hash table apparently intentionally doesn't differentiate between value ranges with different types, all that matters are the values of min and max, so before using it ipa_vr_operation_and_type_effects needs to be called to convert the value_range to the right type. Most of the

[C++ PATCH] PR c++/92847 - C++20 comparison ambiguity with class template.

This testcase demonstrates that looking at cand->template_decl is not a good starting place for finding the most general template, as it is only set for primary templates. Tested x86_64-pc-linux-gnu, applying to trunk. * call.c (cand_parms_match): Handle all templated functions. ---

[C++ PATCH] Fix C++20 structural type vs. private base.

In my patch to implement C++20 "structural type" I tried to set the access flags on the artificial base fields appropriately, but failed. I was copying TREE_PRIVATE from the binfo, but TREE_PRIVATE on binfo is just a temporary cache for dfs_access_in_type; we really need to get the inheritance

[C++ PATCH] PR c++/92560 - ICE with decltype and rewritten operator.

A call as the immediate operand of decltype is handled differently; we don't create an object of the return type as we do normally. But in the case of a rewritten operator, we're adding another call as a wrapper, so the inner call doesn't get the special handling. Tested x86_64-pc-linux-gnu,

Re: [PATCH] rs6000: Fix 2 for PR92661, Do not define builtins that overload disabled builtins

Peter Bergner wrote: On 12/10/19 12:27 PM, Peter Bergner wrote: Ok, how about the patch below? If Iain and David could test this on Darwin and AIX respectively, that would be great. I'm currently testing this on powerpc64le-linux, with and without --disable-decimal-float. So my

Re: [PATCH 28/49] analyzer: new files: analyzer.{cc|h}

On 12/9/19, David Malcolm wrote: > On Fri, 2019-12-06 at 22:38 -0500, Eric Gallager wrote: >> On 11/15/19, David Malcolm wrote: > [...] >> > diff --git a/gcc/analyzer/analyzer.h b/gcc/analyzer/analyzer.h >> > new file mode 100644 >> > index 000..ace8924 >> > --- /dev/null >> > +++

[C++ PATCH] c++/92878 - Parenthesized init of aggregates in new-expression.

Ville pointed out that our paren init of aggregates doesn't work for auto a = new A(1, 2, 3); and I think it should: A new-expression that creates an object of type T initializes that object as follows: ... -- Otherwise, the new-initializer is interpreted according to the initialization

Re: [PATCH] rs6000: Fix 2 for PR92661, Do not define builtins that overload disabled builtins

On 12/10/19 12:27 PM, Peter Bergner wrote: > Ok, how about the patch below? If Iain and David could test this on Darwin > and AIX respectively, that would be great. I'm currently testing this on > powerpc64le-linux, with and without --disable-decimal-float. So my --enable-decimal-float builds

[patch, fortran, committed] Fix PR 92863, 'ICE in gfc_typename

Hello world, I have just committed the attached patch as obvious and simple as r279180. The ICE for the test case occurred because a previous error had left the derived field of the typespec NULL. Just returning "invalid type" or "invalid class" in such a case is enough to cure the ICE.

Re: [PATCH v2 2/2] sched-deps.c: Avoid replacing address if it increases address cost

Hi Jim, On 31/10/2019 00:03, Jim Wilson wrote: > On Fri, Oct 25, 2019 at 10:40 AM Craig Blackmore > wrote: >> The sched2 pass undoes some of the addresses generated by the RISC-V >> shorten_memrefs code size optimization (patch 1/2) and consequently increases >> code size. This patch prevents

Re: [PATCH v2 1/2] RISC-V: Add shorten_memrefs pass

Hi Jim, Thank you for your review. I have posted an updated patch below which I think addresses your comments. On 30/10/2019 23:57, Jim Wilson wrote: > On Fri, Oct 25, 2019 at 10:40 AM Craig Blackmore > wrote: >> This patch aims to allow more load/store instructions to be compressed by >>

Re: [PATCH] rs6000: Fix 2 for PR92661, Do not define builtins that overload disabled builtins

On 12/4/19 5:03 PM, Segher Boessenkool wrote: > On Wed, Dec 04, 2019 at 03:53:30PM -0600, Peter Bergner wrote: >> Right. I'll come up with a patch and hopefully Iain and David can test >> on Darwin and AIX and I can test on Linux with --enable-decimal-float >> and --disable-decimal-float. That

Re: [PATCH] Freestanding proposal P0829

Den ons. 4. dec. 2019 kl. 22.43 skrev Jonathan Wakely : > > On 03/12/19 23:38 +0100, Paul M. Bendixen wrote: > >Hello > >I've made an implementation of P0829 and tested it with gcc for avr. > > This is great, thanks! > > What's the status of libstdc++ for AVR? If I understand correctly, it >

Fix function profile computation

Hi, this patch fixes compute_function_frequency to work well with profiles that was downgraded from IPA to local profiles. Bootstrapped/regtested x86_64-linux, comitted. honza * predict.c (compute_function_frequency): Check for presence of IPA profile. Index: predict.c

Re: Fix overflows in -fprofile-reorder-functions

> On Tue, 10 Dec 2019, Jan Hubicka wrote: > > > > I would recommend to make these variables uint64_t, then you can simply do > > > > > > tp_first_run_a--; > > > tp_first_run_b--; > > > > > > making 0 wrap around to UINT64_MAX. Then they will naturally sort after > > > all > > > other

[Patch, Fortran] OpenMP/OpenACC – fix more issues with OPTIONAL

Nonallocatable, nonpointer array arguments (of assumed shape) are special as they get a get an array descriptor ('arg') as argument but create a local variable which accesses the actual data ('arg.0 = arg->data'). With OPTIONAL, there are/were two outstanding issues: (A) If the argument is

Fix hot/startup partitioning with LTO

Hi, as noticed by Martin's new heatmaps default_function_sections disables test.unlikely and text.startup sections during LTO. This was meant to be only with tp_first_run profiling. The hot section is still useful since it does "poor man's clustering" so I kept only the test to disable startup

[PATCH, libstdc++] Two parts of C++20 p1032...

These patches are modest reworkings of previous patches. The patch for char_traits was pretty much approved already.  But in response to a comment from François I at least added __copy_backwards, and used ptrdiff_t instead of size_t parameters and I put all the mem* wrappers in a __detail

Re: [PATCH 28/49] analyzer: new files: analyzer.{cc|h}

On 11/15/19 6:23 PM, David Malcolm wrote: gcc/ChangeLog: * analyzer/analyzer.cc: New file. * analyzer/analyzer.h: New file. Here are a few more comments/suggestions as I'm slowly making my way through the patch, partly for my own benefit. Hopefully they're at least moderately

Re: [PING][PATCH][GCC][ARM] Arm generates out of range conditional branches in Thumb2 (PR91816)

Hi Stam, On 11/15/19 5:26 PM, Stam Markianos-Wright wrote: Pinging with more correct maintainers this time :) Also would need to backport to gcc7,8,9, but need to get this approved first! Sorry for the delay. Thank you, Stam Forwarded Message Subject: Re:

[PATCH] libstdc++: Fix description of std::ios::trunc (PR 92886)

PR libstdc++/92886 * include/bits/ios_base.h (std::ios_base::trunc): Fix comment. Committing to trunk as obvious. commit 9acff881dea3a00ae588f8ebc4c5289b98f7ee52 Author: Jonathan Wakely Date: Tue Dec 10 17:00:30 2019 + libstdc++: Fix description of std::ios::trunc

Re: [PATCH][RFC] Add new ipa-reorder pass

> * Makefile.in: Add ipa-reorder.o. > * cgraph.c (cgraph_node::dump): Dump > text_sorted_order. > (cgraph_node_cmp_by_text_sorted): > New function that sorts functions based > on text_sorted_order. > * cgraph.h (cgraph_node): Add text_sorted_order. >

[committed][AArch64] Don't allow partial SVE modes in GPRs

With -msve-vector-bits=N, the payload of some partial SVE modes can be 16 bytes or smaller, which makes them small enough to fit in a pair of GPRs. We specifically don't want that, because the payload is distributed evenly across the SVE register rather than collected at one end. Marshalling it

[committed][AArch64] Fix INDEX patterns for partial VNx2 modes

The INDEX patterns handle partial modes by choosing the container size rather than the element size, so that the number of lanes (and thus number of additions) matches the mode. This means that all VNx4 modes use .s and all VNx2 modes use .d, etc. When adding this, I'd forgotten that the choice

Re: [PATCH 0/2] Add tests to verify OpenACC clause locations

Hi Thomas, On 10.12.19 15:44, Thomas Schwinge wrote: > Thanks, yes, with my following remarks considered, and acted on per your > preference. To record the review effort, please include "Reviewed-by: > Thomas Schwinge " in the commit log, see > . Committed

[PATCH] libstdc++: Define __cpp_lib_constexpr_complex macro

This is LWG issue 3349. * include/std/complex (__cpp_lib_constexpr_complex): Define. * include/std/version (__cpp_lib_constexpr_complex): Likewise. * testsuite/26_numerics/complex/1.cc: New test. * testsuite/26_numerics/complex/2.cc: New test. Tested

[PATCH] libstdc++: Reduce header dependencies in

* include/std/span: Do not include and . (tuple_size, tuple_element): Declare. Tested powerpc64le-linux, committed to trunk. commit a32163a1de2a84dde15df80dd33e54a57746d6eb Author: Jonathan Wakely Date: Tue Dec 10 15:59:54 2019 + libstdc++: Reduce header

[PATCH] libstdc++: Fix bug in std::indirect_result_t

The alias template wasn't working because it applied iter_reference_t to the pack of iterators before and after passing the pack to the __indeirect_result helper. * include/bits/iterator_concepts.h (indirect_result_t): Do not apply iter_reference_t to parameter pack. *

Re: Fix overflows in -fprofile-reorder-functions

On Tue, 10 Dec 2019, Jan Hubicka wrote: > > I would recommend to make these variables uint64_t, then you can simply do > > > > tp_first_run_a--; > > tp_first_run_b--; > > > > making 0 wrap around to UINT64_MAX. Then they will naturally sort after all > > other nodes. > > Then we would

Re: Fix overflows in -fprofile-reorder-functions

> 2On Sun, 8 Dec 2019, Jan Hubicka wrote: > > > Other explanation would be that our new qsort with broken comparator due to > > overflow can actualy remove some entries in the array, but that sounds bit > > crazy. > > gcc_qsort only reorders elements, making it possible for gcc_qsort_chk (that >

Re: copy/copy_backward/fill/fill_n/equal rework

On 09/12/19 10:32 +0100, François Dumont wrote: After completing this work and running more tests I realized that the declaration of algos was still not ideal. So here is another version where algos are not re-declare in stl_deque.h, I rather include stl_algobase.h in deque.tcc. The problem

Re: [PATCH 0/2] Add tests to verify OpenACC clause locations

Hi Thomas, On 10.12.19 15:44, Thomas Schwinge wrote: >> Frederik Harwath (2): >> Use clause locations in OpenACC nested reduction warnings >> Add tests to verify OpenACC clause locations > > I won't insist, but suggest (common practice) to merge that into one > patch: bug fix plus test

[PATCH] Fix libstdc++ testsuite to handle VxWorks gthreads implementation

Hello there ! When implementing the support for gthreads in VxWorks, we stumbled on a problem in the testsuite. In the libstdc++ testsuite, we indiscriminately add the `-pthread` switch to the tests that require linking against the pthread library. In certain cases, such as VxWorks, the

Re: [PATCH 0/2] Add tests to verify OpenACC clause locations

Hi Frederik! On 2019-12-10T15:23:01+0100, Frederik Harwath wrote: > On 09.12.19 16:58, Harwath, Frederik wrote: >> Tobias has recently fixed a problem with the column information in gfortran >> locations >> [...] >> I have tested the patch manually by adapting the validity check for nested >>

Re: [PATCH] Fix multibyte-related issues in pretty-print.c (PR 91843)

On Mon, Dec 9, 2019 at 4:58 PM David Malcolm wrote: > > On Thu, 2019-10-10 at 16:27 -0400, Lewis Hyatt wrote: > > Hello- > > > > This short patch addresses > > https://gcc.gnu.org/bugzilla/show_bug.cgi?id=91843 > > by adding the needed multibyte awareness to pretty-print.c. > > Together with my

[committed] Add myself to MAINTAINERS

Thanks for the sponsorship David. 2019-12-10 Lewis Hyatt * MAINTAINERS (Write After Approval): Add myself. Index: MAINTAINERS === --- MAINTAINERS (revision 279166) +++ MAINTAINERS (revision 279167) @@ -429,6 +429,7 @@

[PATCH 2/2] Add tests to verify OpenACC clause locations

Check that the column information for OpenACC clauses is communicated correctly to the middle-end, in particular by the Fortran front-end (cf. PR 92793). 2019-12-10 Frederik Harwath gcc/testsuite/ * gcc.dg/goacc/clause-locations.c: New test. *

[PATCH 0/2] Add tests to verify OpenACC clause locations

Hi, On 09.12.19 16:58, Harwath, Frederik wrote: > Tobias has recently fixed a problem with the column information in gfortran > locations > [...] > I have tested the patch manually by adapting the validity check for nested > OpenACC reductions (see omp-low.c) > to include the location of

[PATCH 1/2] Use clause locations in OpenACC nested reduction warnings

Since the Fortran front-end now sets the clause locations correctly, we can emit warnings with more precise locations if we encounter conflicting operations for a variable in reduction clauses. 2019-12-10 Frederik Harwath gcc/ * omp-low.c (scan_omp_for): Use clause location in

Re: [PATCH] Bail out in gfc_dep_compare_expr for a NULL argument.

Hi Martin, On 12/10/19 2:04 PM, Martin Liška wrote: The patch is about early bail out in gfc_dep_compare_expr function when one of the arguments (and not both at the same time) is null. I'm not sure, it's a proper fix however. I think it is the best one can do. Technically, one could replace

Re: [PATCH, Modula-2 (C/C++/D/F/Go/Jit)] (Register spec fn) (v3)

On 09.12.19 17:41, Gaius Mulley wrote: > Matthias Klose writes: > >> On 17.11.19 07:49, Gaius Mulley wrote: >>> >>> Hello, >>> >>> while spending the weekend on the Howland and Baker islands :-) I >>> thought I'd post version three of the patches which introduce Modula-2 >>> into the GCC trunk.

[PATCH][vect] Keep track of DR_OFFSET advance in dr_vec_info rather than data_reference

Hi, This patch aims at refactoring the vectorizer code to no longer need to reset DR_OFFSET for epilogue vectorization and instead keep track of DR_OFFSET changes per dr_vec_info and just update it as needed. I added a member of type 'tree' called 'offset' to dr_vec_info, which keeps track

Re: [RFC, vectorizer] Fix ICE with masked vectors

On 09/12/2019 15:59, Richard Sandiford wrote: No, the assumption's correct even there. The assert usually triggers because something elsewhere is getting confused about the vector types. The attached patch fixes the ICE in the testcase, but I suspect does not go far enough. Can it happen that

[PATCH] Bail out in gfc_dep_compare_expr for a NULL argument.

Hi. The patch is about early bail out in gfc_dep_compare_expr function when one of the arguments (and not both at the same time) is null. I'm not sure, it's a proper fix however. Patch can bootstrap on x86_64-linux-gnu and survives regression tests. Ready to be installed? Thanks, Martin

Re: Make dwarf2out punt for MODE_VECTOR_BOOL

On December 10, 2019 12:44:01 PM GMT+01:00, Richard Sandiford wrote: >The dwarf2 handling of vector constants currently divides the vector >into a length (number of elements) and byte element size. This doesn't >work well for MODE_VECTOR_BOOL, where several elements are packed into >the same

Re: Record the loop masks needed for EXTRACT_LAST_REDUCTIONs

On December 10, 2019 12:34:33 PM GMT+01:00, Richard Sandiford wrote: >The analysis phase of vectorizable_condition wasn't recording the >loop masks needed by the transform phase. This meant that the masks >wouldn't be created in the (rare) case that no other statement needed >them. > >Tested on

Re: Add missing conversion in vect_create_epilog_for_reduction

On December 10, 2019 12:39:53 PM GMT+01:00, Richard Sandiford wrote: >The direct_slp_reduc code in vect_create_epilog_for_reduction was >still assuming that all types involved in a reduction are the same >(up to types_compatible_p), whereas we now support differences in >sign. This was causing

Re: Fix EXTRACT_LAST_REDUCTION handling of pattern stmts

On December 10, 2019 12:28:24 PM GMT+01:00, Richard Sandiford wrote: >Unlike most vector ops, extract-last reductions replace the original >scalar code in-situ rather than adding an adjacent vector >implementation. >I.e.: > > dest_1 = COND_EXPR <...>; > >becomes: > > dest_1 = .EXTRACT_LAST

Make dwarf2out punt for MODE_VECTOR_BOOL

The dwarf2 handling of vector constants currently divides the vector into a length (number of elements) and byte element size. This doesn't work well for MODE_VECTOR_BOOL, where several elements are packed into the same byte. We should probably add a way of encoding this in future, but for now

Add missing conversion in vect_create_epilog_for_reduction

The direct_slp_reduc code in vect_create_epilog_for_reduction was still assuming that all types involved in a reduction are the same (up to types_compatible_p), whereas we now support differences in sign. This was causing an ICE in gcc.dg/vect/pr92324-4.c for SVE. Tested on aarch64-linux-gnu and

Record the loop masks needed for EXTRACT_LAST_REDUCTIONs

The analysis phase of vectorizable_condition wasn't recording the loop masks needed by the transform phase. This meant that the masks wouldn't be created in the (rare) case that no other statement needed them. Tested on aarch64-linux-gnu and x86_64-linux-gnu. OK to install? Richard

[committed] Disallow EXTRACT_LAST_REDUCTION for reduction chains

gcc.dg/vect/vect-cond-reduc-5.c was ICEing for SVE because we tried to use an extract-last reduction for a chain of COND_EXPRs. Adding support for the chained case would be too invasive for stage 3 so this patch explicitly forbids it instead. I've filed PR92884 for the possible future work.

Fix EXTRACT_LAST_REDUCTION handling of pattern stmts

Unlike most vector ops, extract-last reductions replace the original scalar code in-situ rather than adding an adjacent vector implementation. I.e.: dest_1 = COND_EXPR <...>; becomes: dest_1 = .EXTRACT_LAST (...); gcc.dg/vect/vect-cond-reduc-4.c was ICEing for SVE because we tried to

Re: [RFC] ipa-cp: Fix PGO regression caused by r278808

> Hi, > > On Tue, Dec 10 2019, Jan Hubicka wrote: > > Hi, > > I think the updating should treat self recursive edges as loops: that is > > calculate SUM of counts incomming edges which are not self recursive, > > calculate probability of self recursing and then set count as > > SUM *

Re: Fwd: [PATCH, GCC, Vect] Fix costing for vector shifts

Hi Christophe On 10/12/2019 09:01, Christophe Lyon wrote: > Hi, > > On Mon, 9 Dec 2019 at 11:23, Sudakshina Das wrote: >> >> Hi Jeff >> >> On 07/12/2019 17:44, Jeff Law wrote: >>> On Fri, 2019-12-06 at 14:05 +, Sudakshina Das wrote: Hi While looking at the vectorization for

[patch, committed, Fortran] PR 92872 – Fix get_CFI_desc

Fix the condition; the DECL_LANG_SPECIFIC can be used for much more than just to save the decl. Committed as Rev. 279160. Tobias 2019-12-10 Tobias Burnus PR fortran/92872 * trans-array.c (get_CFI_desc): Fix cond whether saved desc exists. PR fortran/92872 *

Re: [PATCH] Oprimize stack_protect_set_1_ followed by a move to the same register (PR target/92841)

On Tue, Dec 10, 2019 at 11:02:39AM +0100, Andreas Schwab wrote: > On Dez 10 2019, Jakub Jelinek wrote: > > > --- gcc/testsuite/gcc.target/i386/pr92841.c.jj 2019-12-09 > > 19:38:29.572759215 +0100 > > +++ gcc/testsuite/gcc.target/i386/pr92841.c 2019-12-09 19:40:59.642492417 > > +0100 > > @@

Re: [PATCH] Oprimize stack_protect_set_1_ followed by a move to the same register (PR target/92841)

On Dez 10 2019, Jakub Jelinek wrote: > --- gcc/testsuite/gcc.target/i386/pr92841.c.jj2019-12-09 > 19:38:29.572759215 +0100 > +++ gcc/testsuite/gcc.target/i386/pr92841.c 2019-12-09 19:40:59.642492417 > +0100 > @@ -0,0 +1,17 @@ > +/* PR target/92841 */ > +/* { dg-do compile { target

[PATCH] Oprimize stack_protect_set_1_ followed by a move to the same register (PR target/92841)

Hi! The stack_protect_set_1_ pattern intentionally clears the register it used as a temporary to read the canary from the register and push it back on the stack for security reasons, to make sure the stack canary isn't spilled somewhere else etc. On the following testcase, we end up with a

[PATCH] Improve -fstack-protector-strong (PR middle-end/92825)

Hi! As mentioned in the PR, -fstack-protector-strong in some cases instruments even functions which except for the stack canary setup and later testing of that don't touch the stack at all. This is because for -fstack-protector-strong we setup the canary whenever stack_protect_decl_p returns

Re: [RFC] ipa-cp: Fix PGO regression caused by r278808

Hi, On Tue, Dec 10 2019, Jan Hubicka wrote: > Hi, > I think the updating should treat self recursive edges as loops: that is > calculate SUM of counts incomming edges which are not self recursive, > calculate probability of self recursing and then set count as > SUM *

RE: [PATCH] PR 92846: [ARC] generate signaling FDCMPF for hard float comparisons

Hi, Thank you for your contribution, I'll push it asap. As far as I understand, you need this patch both in gcc9 branch and mainline. Cheers, Claudiu > -Original Message- > From: Vineet Gupta [mailto:vgu...@synopsys.com] > Sent: Monday, December 09, 2019 8:02 PM > To:

[committed] Fix up i386/avx512f-vmovntp?-2.c tests

Hi! I've noticed these two tests FAIL execution with -fstack-protector-strong, without it we are just lucky. _mm512_stream_* requires the destination to be 64-byte aligned and without stack protector the s vars were allocated next to res and ensured the right alignment, but with

Re: [PATCH] Fix unrecognizable insn of pr92865

> No gcc/ or gcc/testsuite/ prefixes in ChangeLog. And 2 different ChangeLogs: ChangeLog and testsuite/ChangeLog. -- Eric Botcazou

[committed] Fix some i386.c comment typos

Hi! Last night when working on -fstack-protector* improvement to be posted, I've noticed a few typos. Fixed thusly, committed to trunk as obvious. 2019-12-10 Jakub Jelinek * config/i386/i386.c (IX86_LEA_PRIORITY): Fix comment typos. --- gcc/config/i386/i386.c.jj 2019-12-09

Re: Fwd: [PATCH, GCC, Vect] Fix costing for vector shifts

Hi, On Mon, 9 Dec 2019 at 11:23, Sudakshina Das wrote: > > Hi Jeff > > On 07/12/2019 17:44, Jeff Law wrote: > > On Fri, 2019-12-06 at 14:05 +, Sudakshina Das wrote: > >> Hi > >> > >> While looking at the vectorization for following example, we > >> realized > >> that even though

Re: [PATCH] Refactor IPA devirt a bit.

> > > > 2019-12-09 Richard Sandiford > > > > gcc/ > > * ipa-utils.h (get_odr_name_for_type): Check for a TYPE_DECL. > > * ipa-devirt.c (warn_types_mismatch): Don't call xstrdup for the > > second demangled name. > > > > gcc/testsuite/ > > * gcc.dg/lto/tag-1_0.c,

Re: [RFC] ipa-cp: Fix PGO regression caused by r278808

Hi, I think the updating should treat self recursive edges as loops: that is calculate SUM of counts incomming edges which are not self recursive, calculate probability of self recursing and then set count as SUM * (1/(1-probability_of_recursion)) we do same thing when computing counts withing

  1   2   >