RE: [RFC][PATCH] Preferred rename register in regrename pass

Hi Bernd, Thanks for the comments, much appreciated. Comments inlined and a reworked patch attached. > On 09/17/2015 04:38 PM, Robert Suchanek wrote: > > We came across a situation for MIPS64 where moves for sign-extension were > > not converted into a nop because of IRA spilled some of the alloc

[PR67891] drop is_gimple_reg test from set_parm_rtl (was: [PR67766] reorder return value copying from PARALLELs and CONCATs)

On Oct 9, 2015, Alexandre Oliva wrote: > This fixes fallout from the PR64164 expander revamp. > Uroš kindly tested with an alpha-linux-gnu regstrap. The one regression he mentioned from that run was gcc.dg/pr43300.c. The vector parameter there is handled by the emit_block_move case of assign_

Re: Do not use TYPE_CANONICAL in useless_type_conversion

> The type has BLKmode and size 32. DECL_SIZE of the FIELD_DECL is however 24 > (see it printed as Ada size). Yes, no wonder since it's a bitfield, i.e. DECL_BIT_FIELD is set. > The DECL_MODE of the FIELD_DECL is VOIDmode (not printed), while the > TYPE_MODE of type contained is BLKmode. No, the

Re: [PR 67794] Also remap SSA_NAMEs defined in ASMs in IPA-SRA

On Thu, 8 Oct 2015, Martin Jambor wrote: > Hi, > > the following fixes PR 67794 by properly remapping SSA_NAMEs which are > based on PARM_DECLs which are about to be removed as unnecessary. And > by "properly" I mean also when they are defined by a GIMPL_ASM > statement. In fact, it switches to

Re: Use OEP_ADDRESS_OF in emit-rtl.c

On Thu, 8 Oct 2015, Jan Hubicka wrote: > > On Wed, 7 Oct 2015, Jan Hubicka wrote: > > > > > > > > > > Did you audit all callers of mem_attrs_eq_p to see if they really > > > > only care about that? After all MEM_EXPR, via access paths, encode > > > > type-based alias info and thus replacing one

[gomp4.1] OpenMP 4.1 is dead, long live OpenMP 4.5

Hi! Yesterday the upcoming OpenMP specification got renumbered from 4.1 to 4.5 to take into the amount of 4.0 -> 4.5 changes. This patch adjusts the 4.1 occurrences I could find. Not going to rename the development branch at this point, I hope it can be merged RSN into the trunk. 2015-10-09 Ja

[gomp4.1] Change _OPENMP macro to 201511

Hi! This changes it just for C/C++, as Fortran I'm afraid will have to wait with OpenMP 4.5 support until GCC 6.2 (similarly to 4.0 support only appearing in 4.9.1). 2015-10-09 Jakub Jelinek * c-cppbuiltin.c (c_cpp_builtins): Predefine _OPENMP as 201511 instead of 201307. libg

[PATCH] Fix PR67889

This fixes PR67889. Bootstrapped and tested on x86_64-unknown-linux-gnu, applied. Richard. 2015-10-09 Richard Biener PR tree-optimization/67891 * gimple-match.h (gimple_simplified_result_is_gimple_val): New helper. (gimple_resimplify1): Declare. (gimp

[Ada] Make sure generated files are created first

As reported by Alan L., we need to make sure that the generated files are created before gigi is compiled. Tested on x86_64-suse-linux, applied on all branches. 2015-10-09 Eric Botcazou * gcc-interface/Make-lang.in: Make sure that GNAT1_OBJS and not just GNAT1_ADA_OBJS are c

Re: Do not use TYPE_CANONICAL in useless_type_conversion

On Fri, 9 Oct 2015, Jan Hubicka wrote: > > > > Index: expr.c > > === > > --- expr.c (revision 228604) > > +++ expr.c (working copy) > > @@ -6703,7 +6704,7 @@ store_field (rtx target, HOST_WIDE_INT b > > emit_group_store (temp

[gomp4.1] Remove middle argument of omp_target_is_present

Hi! As one can just use pointer-arithmetics on the first argument, and that argument is the host pointer, there is no point to have an OFFSET argument and it got removed from the upcoming 4.5. 2015-10-09 Jakub Jelinek * omp.h.in (omp_target_is_present): Remove OFFSET argument.

Re: Do not use TYPE_CANONICAL in useless_type_conversion

> I think a FIELD_DECL with VOIDmode is odd. As I said in my earlier message, the FIELD_DECL does *not* have VOIDmode. > And yes, the type of the COMPONENT_REF should be that of the FIELD_DECL (or > a variant type of it as we share FIELD_DECLs for record variants). But it is of course! Jan is a

[PATCH] Warn about duplicate patterns in match.pd

This teaches genmatch to warn about multiple patterns that match the exact same expressions (you should usually merge those otherwise the first one wins always). It also removes some existing (harmless) duplicates (the first one is the one that should have matched). This reduces the amount of ge

[gomp4.1] Two more array section testcases

Hi! The 4.0 spec required that low-bound in array section must not be negative. The 4.5 spec removes that restriction, for arrays it is obvious that it is invalid anyway (pointer arithmetics disallows that), and for pointers it is meaningful. After looking what I've implemented, it seems I've app

[RFC VTV] Fix VTV for targets that have section anchors.

This started as a Friday afternoon project ... It turned out enabling VTV for AArch64 and ARM was a matter of fixing PR67868 which essentially comes from building libvtv with section anchors turned on. The problem was that the flow of control from output_object_block through to switch_section

Re: Do not use TYPE_CANONICAL in useless_type_conversion

Jan Hubicka writes: > Does the patch in https://gcc.gnu.org/ml/gcc-patches/2015-10/msg00902.html > help? No, it doesn't. #0 fancy_abort (file=0x43f1ce48 "../../gcc/expr.c", line=282, function=0x43f1ec38 "convert_move") at ../../gcc/diagnostic.c:1209 #1 0x41a816e

Re: [Patch PR target/67366 2/2] [gimple-fold.c] Support movmisalign optabs in gimple-fold.c

On Thu, Oct 8, 2015 at 4:10 PM, Ramana Radhakrishnan wrote: > This patch by Richard allows for movmisalign optabs to be supported > in gimple-fold.c. This caused a bit of pain in the testsuite with > strlenopt-8.c > in conjunction with the ARM support for movmisalign_optabs as the test > was code

Re: [PATCH] bb-reorder: Improve the simple algorithm for -Os (PR67864)

On Thu, Oct 8, 2015 at 6:57 PM, Segher Boessenkool wrote: > As the PR points out, the "simple" reorder algorithm makes bigger code > than the STC algorithm did, for -Os, for x86. I now tested it for many > different targets and it turns out to be worse everywhere. > > This simple patch tunes "sim

Re: Move some bit and binary optimizations in simplify and match

On Thu, Oct 8, 2015 at 8:15 PM, Bernd Schmidt wrote: > On 10/08/2015 08:03 PM, Joseph Myers wrote: >> >> On Thu, 8 Oct 2015, Bernd Schmidt wrote: >> >>> On 10/07/2015 11:54 AM, Hurugalawadi, Naveen wrote: Move Fold X & (X ^ Y) as X & ~Y to match.pd. Move Fold X & (Y

Re: Fix use of availability in ipa-icf

On Fri, Oct 9, 2015 at 1:21 AM, Jan Hubicka wrote: > Hi, > this is a bug I noticed while reading the code. We can not assume > interposable > objects to be equivalent unless they are actually the same symbols. > > Bootstrapped/regtested x86_64-linux, comitted. Backport? > Honza > > * i

[gomp4.1] Add negative length check even for pointers

On Fri, Oct 09, 2015 at 11:14:33AM +0200, Jakub Jelinek wrote: > The 4.0 spec required that low-bound in array section must not be negative. > The 4.5 spec removes that restriction, for arrays it is obvious that it is > invalid anyway (pointer arithmetics disallows that), and for pointers it is > m

Re: [PR67828] don't unswitch loops on undefined SSA values (was: Re: [PR64164] drop copyrename, integrate into expand)

On Fri, Oct 9, 2015 at 7:26 AM, Alexandre Oliva wrote: > This patch fixes a latent bug in loop unswitching exposed by the PR64164 > changes. > > We would move a test out of a loop that might never have been executed, > and that accessed an uninitialized variable. The uninitialized SSA > name, due

Re: [PR67766] reorder return value copying from PARALLELs and CONCATs (was: Re: [PR64164] drop copyrename, integrate into expand)

On Fri, Oct 9, 2015 at 7:36 AM, Alexandre Oliva wrote: > This fixes fallout from the PR64164 expander revamp. On alpha, PARALLEL > hard return values may be modeless, and this confuses the code that > wants to copy the pseudo/s in the returned value to the return hard > regs. > > It used to work

Re: [PATCH] [3/n] Fix minor SSA_NAME leaks

On Thu, Oct 8, 2015 at 10:47 PM, Jeff Law wrote: > And other minor leak. This time in tree-stdarg. Unlike other cases, we're > dropping just the virtual definition, other definitions on the statement > need to be preserved (they're going to be re-used). Additionally, this one > is missing the ca

Re: [PR67891] drop is_gimple_reg test from set_parm_rtl (was: [PR67766] reorder return value copying from PARALLELs and CONCATs)

On Fri, Oct 9, 2015 at 9:33 AM, Alexandre Oliva wrote: > On Oct 9, 2015, Alexandre Oliva wrote: > >> This fixes fallout from the PR64164 expander revamp. > >> Uroš kindly tested with an alpha-linux-gnu regstrap. > > The one regression he mentioned from that run was gcc.dg/pr43300.c. The > vecto

[Patch] [x86_64]: Add bdver4 for multi versioning and fix AMD cpu model detection.

Hi Uros, Please find below patch that adds bdver4 target for multi versioning. Also I while computing model, the extended_model is incorrectly left shifted by 4. I have removed it now. Is below patch Ok for trunk ? GCC bootstrap and regressions passed. diff --git a/libgcc/ChangeLog b/libgcc/

Re: [Patch] [x86_64]: Add bdver4 for multi versioning and fix AMD cpu model detection.

On Fri, Oct 9, 2015 at 11:50 AM, Kumar, Venkataramanan wrote: > Hi Uros, > > Please find below patch that adds bdver4 target for multi versioning. > Also I while computing model, the extended_model is incorrectly left shifted > by 4. I have removed it now. > > Is below patch Ok for trunk ? > GCC

RE: [Patch] [x86_64]: Add bdver4 for multi versioning and fix AMD cpu model detection.

Thank you Uros, I will test and commit model selection change in all release branches as well. Regards, Venkat. > -Original Message- > From: Uros Bizjak [mailto:ubiz...@gmail.com] > Sent: Friday, October 09, 2015 3:25 PM > To: Kumar, Venkataramanan > Cc: gcc-patches@gcc.gnu.org > Subjec

[PR fortran/63858] Fix mix of OpenACC and OpenMP sentinels in continuations

Hi! On Mon, 27 Jul 2015 16:14:17 +0200, I wrote: > On Tue, 30 Jun 2015 03:39:42 +0300, Ilmir Usmanov wrote: > > 08.06.2015, 17:59, "Cesar Philippidis" : > > > On 06/07/2015 02:05 PM, Ilmir Usmanov wrote: > > >> 08.06.2015, 00:01, "Ilmir Usmanov" : > > This patch fixes checks of OpenMP and

Re: [nvptx] fix some c++ tests

On 10/09/2015 07:55 AM, Thomas Schwinge wrote: Hi Nathan! Thanks for looking into this. When I reported this, , quite a lot of testcases had been failing -- with recent GCC trunk, the number is smaller (because

Re: [3/7] Optimize ZEXT_EXPR with tree-vrp

On Thu, Oct 8, 2015 at 1:40 AM, Kugan wrote: > > > On 07/10/15 19:20, Richard Biener wrote: >> On Wed, Oct 7, 2015 at 1:12 AM, kugan >> wrote: >>> >>> Hi Richard, >>> >>> Thanks for the review. >>> >>> On 15/09/15 23:08, Richard Biener wrote: On Mon, Sep 7, 2015 at 4:58 AM, Kugan

Re: RFA: PATCH to store_field for storing a CONSTRUCTOR into a base subobject

On Thu, Oct 8, 2015 at 4:42 AM, Jason Merrill wrote: > While looking at another issue I noticed that in g++.dg/init/vbase1.C the > Diamond(int) constructor was unnecessarily storing a CONSTRUCTOR into a > stack temporary and then copying it into the SubB base subobject rather than > directly stori

Re: [PATCH] bb-reorder: Improve the simple algorithm for -Os (PR67864)

On 10/08/2015 06:57 PM, Segher Boessenkool wrote: As the PR points out, the "simple" reorder algorithm makes bigger code than the STC algorithm did, for -Os, for x86. I now tested it for many different targets and it turns out to be worse everywhere. That's somewhat disappointing. Wasn't it su

Re: [PATCH 1/9] Change default of non-overlapping address space conversion

On 10/08/2015 06:59 AM, Richard Henderson wrote: The current default of making all undefined coversions being set to null is not useful. It has caused all users to lie and say that spaces are subsets when they are not, just so that they can override the conversion. * expr.c (expand_exp

[COMMITTED] Add myself to MAINTAINERS (Write After Approval)

Hi All, I've just added myself to Write After Approval maintainers. Committed r228642. Regards, Alexander Index: ChangeLog === diff --git a/trunk/ChangeLog b/trunk/ChangeLog --- a/trunk/ChangeLog (revision 228641) +++ b/trunk/Chan

Re: [RFC][PATCH] Preferred rename register in regrename pass

Hi Robert, gcc/ * regrename.c (create_new_chain): Initialize terminated_dead, renamed and tied_chain. (find_best_rename_reg): Pick and check register from the tied chain. (regrename_do_replace): Mark head as renamed. (scan_rtx_reg): Tie chains in move insns

Re: PING: [PATCH] PR target/35514: Gcc shoud generate symbol type for undefined symbol

On 10/08/2015 01:10 PM, H.J. Lu wrote: On Wed, Jul 8, 2015 at 1:35 PM, H.J. Lu wrote: On Sun, Jul 5, 2015 at 2:54 PM, H.J. Lu wrote: On Sun, Jul 5, 2015 at 11:14 AM, H.J. Lu wrote: Update default_elf_asm_output_external to also output symbol type to help ELF linker to properly issue diagnos

Re: Additional headers under include/experimental/

On 06/10/15 15:12 +0100, Jonathan Wakely wrote: On 03/07/14 11:35 +0100, Jonathan Wakely wrote: I'm working on parts of the Filesystem TS and think it makes sense to split it up into separate headers for path, directory utils and the free functions, then include those from , otherwise it's going

RE: [RFC][PATCH] Preferred rename register in regrename pass

Hi Bernd, > Hi Robert, > > gcc/ > > * regrename.c (create_new_chain): Initialize terminated_dead, > > renamed and tied_chain. > > (find_best_rename_reg): Pick and check register from the tied chain. > > (regrename_do_replace): Mark head as renamed. > > (scan_rtx_reg): Tie chain

Re: [COMMITTED] Add myself to MAINTAINERS (Write After Approval)

On Fri, Oct 09, 2015 at 01:56:23PM +0300, Alexander Fomin wrote: > I've just added myself to Write After Approval maintainers. Welcome. > Index: ChangeLog > === > diff --git a/trunk/ChangeLog b/trunk/ChangeLog > --- a/trunk/ChangeLog

Re: [Patch ifcvt costs 0/3] Introduce a new target hook for ifcvt costs.

On 10/01/2015 11:37 AM, Bernd Schmidt wrote: On 09/29/2015 04:31 PM, James Greenhalgh wrote: On the other side of the equation, we want a cost for the converted sequence. We can build a cost of the generated rtl sequence, but for targets like AArch64 this is going to be wildly off. AArch64 will

Re: [patch 0/6] scalar-storage-order merge (2)

On 10/06/2015 12:57 PM, Eric Botcazou wrote: this is a repost of the diff of the scalar-storage-order branch vs mainline. It contains the fixes suggested by Joseph for the C front-end and the doc, fixes for the handling of complex types, the new pragma scalar_storage_order and associated -fsso-st

Re: [PATCH] 2015-10-02 Benedikt Huber Philipp Tomsich

On Sat, Jan 03, 1970 at 03:27:04AM +, Benedikt Huber wrote: > * config/aarch64/aarch64-builtins.c: Builtins for rsqrt and > rsqrtf. > * config/aarch64/aarch64-protos.h: Declare. > * config/aarch64/aarch64-simd.md: Matching expressions for > frsqrte and fr

[COMMITTED] Add myself to MAINTAINERS (Write After Approval)

Hi All, I've just added myself to Write After Approval maintainers. Committed 228649. Martin Index: ChangeLog === --- ChangeLog (revision 228647) +++ ChangeLog (working copy) @@ -1,3 +1,7 @@ +2015-10-09 Martin Liska + + * MAINTAIN

Re: libgomp: Guard all devices/num_devices/num_devices_openmp access by register_lock

On 09/28/2015 10:52 AM, Thomas Schwinge wrote: On Fri, 25 Sep 2015 19:49:50 +0300, Ilya Verbin wrote: Looks like you're right, and this scenario is possible. Thanks for your review! Jakub, OK to commit the patch I had posted? Then, in context of a similar scenario, I think we'll also want

Re: [PATCH GCC]Improve rtl loop inv cost by checking if the inv can be propagated to address uses

On Wed, Sep 30, 2015 at 11:33 AM, Bin.Cheng wrote: > On Tue, Sep 29, 2015 at 1:21 AM, Jeff Law wrote: >> On 09/28/2015 05:28 AM, Bernd Schmidt wrote: >>> >>> On 09/28/2015 11:43 AM, Bin Cheng wrote: Bootstrap and test on x86_64 and x86_32. Will test it on aarch64. So any com

Re: [PATCH GCC]Improve rtl loop inv cost by checking if the inv can be propagated to address uses

On 10/09/2015 02:00 PM, Bin.Cheng wrote: I further bootstrap and test attached patch on aarch64. Also three cases in spec2k6/fp are improved by 3~6%, two cases in spec2k6/fp are regressed by ~2%. Overall score is improved by ~0.8% for spec2k6/fp on aarch64 of my run. I may later analyze the re

Re: Add -foffload-abi support for PPC

David, On 10/08/2015 11:53 AM, David Edelsohn wrote: On Thu, Oct 8, 2015 at 12:19 PM, James Norris wrote: I've revised the patch from the review comments (thank you) and is attached. Regtested on x86_64 and powerpcle64. OK for trunk? What is the goal? Do you want this to return the correc

Re: Add -foffload-abi support for PPC

On Fri, Oct 9, 2015 at 8:20 AM, James Norris wrote: > David, > > > On 10/08/2015 11:53 AM, David Edelsohn wrote: >> >> On Thu, Oct 8, 2015 at 12:19 PM, James Norris >> wrote: >> >>> I've revised the patch from the review comments (thank you) and >>> is attached. >>> >>> Regtested on x86_64 and po

Re: [PR c/64765, c/64880] Support OpenACC Combined Directives in C, C++

On 10/08/15 12:39, Thomas Schwinge wrote: Hi! Some bits extracted out of gomp-4_0-branch, and some other bits rewritten; here is a patch to support OpenACC Combined Directives in C, C++. (The Fortran front end already does support these.) As far as I know, Jakub is not available at this time,

Re: PING: [PATCH] PR target/35514: Gcc shoud generate symbol type for undefined symbol

On Fri, Oct 9, 2015 at 4:11 AM, Bernd Schmidt wrote: > On 10/08/2015 01:10 PM, H.J. Lu wrote: >> >> On Wed, Jul 8, 2015 at 1:35 PM, H.J. Lu wrote: >>> >>> On Sun, Jul 5, 2015 at 2:54 PM, H.J. Lu wrote: On Sun, Jul 5, 2015 at 11:14 AM, H.J. Lu wrote: > > Update default_elf_asm_

Re: PING: [PATCH] PR target/35514: Gcc shoud generate symbol type for undefined symbol

On 10/09/2015 02:45 PM, H.J. Lu wrote: What kind of diagnostic message? The ELF linker seems to have managed to do without for a long time. Is there some discussion of this on the binutils list that you can point to? https://groups.google.com/forum/#!topic/generic-abi/_ZPPq_c8FSQ Hmm, near t

Re: [PR c/64765, c/64880] Support OpenACC Combined Directives in C, C++

Hi! Thanks to Joseph and you for your review! On Fri, 9 Oct 2015 08:26:37 -0400, Nathan Sidwell wrote: > On 10/08/15 12:39, Thomas Schwinge wrote: > > Some bits extracted out of gomp-4_0-branch, and some other bits > > rewritten; here is a patch to support OpenACC Combined Directives in C, > > C

Re: [PR c/64765, c/64880] Support OpenACC Combined Directives in C, C++

On 10/09/15 09:26, Thomas Schwinge wrote: Hi! You mean the cp_parser_oacc_loop and cp_parser_oacc_kernels_parallel functions need documentation? I agree it's a bit terse, but documenting these by just listing the accepted parsing tokens "# pragma acc loop" etc., followed by the *_CLAUSE_MASKs

Re: Do not describe -std=c11 etc. as experimental in c.opt help text

Jason: ping. On Fri, Oct 02, 2015 at 05:35:39PM +0200, Marek Polacek wrote: > On Thu, Oct 01, 2015 at 05:01:26PM +, Joseph Myers wrote: > > I noticed that c.opt still described -std=c11 and related options as > > experimental in the --help text. This patch fixes this. > > > > Jason, note tha

Re: [PR c/64765, c/64880] Support OpenACC Combined Directives in C, C++

On Fri, Oct 09, 2015 at 09:34:51AM -0400, Nathan Sidwell wrote: > On 10/09/15 09:26, Thomas Schwinge wrote: > >Hi! > > >You mean the cp_parser_oacc_loop and cp_parser_oacc_kernels_parallel > >functions need documentation? I agree it's a bit terse, but documenting > >these by just listing the acce

Re: [PR c/64765, c/64880] Support OpenACC Combined Directives in C, C++

Hi Nathan! On Fri, 9 Oct 2015 09:34:51 -0400, Nathan Sidwell wrote: > On 10/09/15 09:26, Thomas Schwinge wrote: > > You mean the cp_parser_oacc_loop and cp_parser_oacc_kernels_parallel > > functions need documentation? I agree it's a bit terse, but documenting > > these by just listing the accep

[gomp4] Revert patch

Hi, The attached patch reverts a previously attached patch. 2015-10-06 James Norris * common.opt (OFFLOAD_ABI_PPC64): New enum. * config/nvptx/mkoffload.c (compile_native): Handle new enum. (main): Handle new option. * config/rs6000/rs6000.c (rs6000_off

Re: [PATCH v2 0/3] [ARM] PR63870 vldN_lane/vstN_lane error messages

On 7 October 2015 at 00:59, wrote: > From: Charles Baylis > > This patch series fixes up the error messages for single lane vector > load/stores, similarly to AArch64. > > make check on arm-linux-gnueabihf/qemu completes with no new regressions. > > Changes since the last version: > . removed th

[gomp4]

I've applied this to gomp4 branch. 1) ports the break fix in gimple-fold from trunk 2) fixes missing tab in ptx output. nathan 2015-10-09 Nathan Sidwell * config/nvptx/nvptx.c (nvptx_init_axis_predicate): Fix output formatting. PR 67861 * gimple-fold.c (gimple_fold_builtin): Add break

Re: [gomp4.1] fix dependency scheduling problem

On Fri, Oct 02, 2015 at 03:46:59PM -0700, Aldy Hernandez wrote: > commit 6d8f6db0583326d803c7c7abd8ea26cc842643fc > Author: Aldy Hernandez > Date: Fri Oct 2 15:40:30 2015 -0700 > > * task.c (gomp_task_maybe_wait_for_dependencies): Fix scheduling > problem such that the first non par

Re: [PATCH, i386, AVX512] PR target/67895: Fix position of embedded rounding/SAE mode in AVX512 vrangep* and vcvt?si2s* instructions.

Hello, On 08 Oct 20:31, Alexander Fomin wrote: > Hi All, > > This patch addresses PR target/67895. For some AVX512 instructions > we've used to emit embedded rounding/SAE specifier in a wrong place. > The patch fixes its position for vrange* and vcvt?si2s* instructions. > I've also updated regula

Re: Move sqrt and cbrt simplifications to match.pd

On 8 October 2015 at 18:55, Richard Sandiford wrote: > Marc Glisse writes: >> On Mon, 5 Oct 2015, Richard Sandiford wrote: >> >>> + /* cbrt(sqrt(x)) -> pow(x,1/6). */ >>> + (simplify >>> + (sqrts (cbrts @0)) >>> + (pows @0 { build_real_truncate (type, dconst<1, 6> ()); })) >>> + /* sqrt(c

Re: libgomp: Guard all devices/num_devices/num_devices_openmp access by register_lock

On Fri, Oct 09, 2015 at 13:58:32 +0200, Bernd Schmidt wrote: > One oddity I noticed in target.c is that there are two different num_devices > variables: > > /* Total number of available devices. */ > static int num_devices; > > /* Number of GOMP_OFFLOAD_CAP_OPENMP_400 devices. */ > stat

Re: [PATCH, rs6000] Add memory barriers to tbegin, tend, etc.

Sorry for the much delayed response. I've been sick and am slowly catching up. On Thu, 2015-09-03 at 16:58 -0500, Peter Bergner wrote: > On a glibc thread discussing this issue, Torvald also asked that I add > documention describing the memory consistency semantics the HTM instructions > should h

Re: PING: [PATCH] PR target/35514: Gcc shoud generate symbol type for undefined symbol

On Fri, Oct 9, 2015 at 5:50 AM, Bernd Schmidt wrote: > On 10/09/2015 02:45 PM, H.J. Lu wrote: >> >> What kind of diagnostic message? The ELF linker seems to have managed to >> do >>> >>> without for a long time. Is there some discussion of this on the binutils >>> list that you can point to? >> >>

RE: [PATCH 1/4] [MIPS] Add support for MIPS SIMD Architecture (MSA)

Hi Robert, Next batch of comments. This set covers the rest of mips-msa.md. >+++ b/gcc/config/mips/mips-msa.md >+(define_expand "vec_perm" >+ [(match_operand:MSA 0 "register_operand") >+ (match_operand:MSA 1 "register_operand") >+ (match_operand:MSA 2 "register_operand") >+ (match_operand:

[HSA] Fix usage of an uninitialized memory

Hello. In the following patch, I fix usage of an uninitialized variable: hsa_cfun->decl. Installed as revision 228659. Thanks, Martin >From 7f85ed1c1d3f2fe6132117f58c84af33cb709689 Mon Sep 17 00:00:00 2001 From: marxin Date: Fri, 9 Oct 2015 15:06:46 +0200 Subject: [PATCH 1/7] HSA: fix usage of

Re: Move sqrt and cbrt simplifications to match.pd

Christophe Lyon writes: > On 8 October 2015 at 18:55, Richard Sandiford > wrote: >> Marc Glisse writes: >>> On Mon, 5 Oct 2015, Richard Sandiford wrote: >>> + /* cbrt(sqrt(x)) -> pow(x,1/6). */ + (simplify + (sqrts (cbrts @0)) + (pows @0 { build_real_truncate (type, d

libgomp testsuite: Remove some explicit acc_device_nvidia usage

Hi! On Wed, 22 Jul 2015 16:39:54 +0200, I wrote: > [...] cleanup; committed to > gomp-4_0-branch in r226072: [...] OK for trunk? commit a8bd7234d7b6b0111a1e3b3dc032ea5eead7e258 Author: Thomas Schwinge Date: Fri Oct 9 16:40:56 2015 +0200 libgomp testsuite: Remove some explicit acc_device_

[libgomp/66518] Resolve XFAIL in libgomp.oacc-c-c++-common/lib-3.c

Hi! On Wed, 22 Jul 2015 16:32:17 +0200, I wrote: > [...] in r226070, I committed the > following to gomp-4_0-branch: [...] OK for trunk? commit 8ba8b058417afc2e7979ef25fdb943ae4e8838dd Author: Thomas Schwinge Date: Fri Oct 9 13:20:52 2015 +0200 [libgomp/66518] Resolve XFAIL in libgomp.oa

libgomp: Additional acc_shutdown bug fixing and testing

Hi! On Wed, 22 Jul 2015 16:37:46 +0200, I wrote: > [...] found an inconsistency in libgomp; committed > to gomp-4_0-branch in r226071: [...] OK for trunk? commit dc0c73e9ac5decf5b36be25497ed5f4b99136fba Author: Thomas Schwinge Date: Fri Oct 9 13:22:52 2015 +0200 libgomp: Additional acc_s

[PR libgomp/65437, libgomp/66518] Initialize runtime in acc_update_device, acc_update_self

Hi! On Fri, 19 Jun 2015 09:47:41 +0200, I wrote: > On Tue, 5 May 2015 11:43:20 +0200, I wrote: > > On Mon, 4 May 2015 10:20:14 -0400, John David Anglin > > wrote: > > > FAIL: libgomp.oacc-c/../libgomp.oacc-c-c++-common/lib-42.c > > > -DACC_DEVICE_TYPE_hos > > > t=1 -DACC_MEM_SHARED=1 output pat

Re: [PATCH] [3/n] Fix minor SSA_NAME leaks

On 10/09/2015 03:37 AM, Richard Biener wrote: On Thu, Oct 8, 2015 at 10:47 PM, Jeff Law wrote: And other minor leak. This time in tree-stdarg. Unlike other cases, we're dropping just the virtual definition, other definitions on the statement need to be preserved (they're going to be re-used).

Re: [Patch ifcvt costs 0/3] Introduce a new target hook for ifcvt costs.

On 10/09/2015 05:28 AM, Bernd Schmidt wrote: I don't know whether you've done any more work on the patch series, but I think I've made up my mind that optimizing the sequence before computing its cost would be a good thing to try first. Either with a better expander interface which generates the

Re: [PATCH][PR tree-optimization/67816] Fix jump threading when DOM removes conditionals in jump threading path

On 10/08/2015 03:56 AM, Richard Biener wrote: On Thu, Oct 8, 2015 at 12:00 AM, Jeff Law wrote: On 10/07/2015 02:26 AM, Richard Biener wrote: Hmm, other passes avoid all this by not removing edges or blocks themselves but leaving that to cfgcleanup. They simply replace the condition in GIMPL

Re: [PATCH] bb-reorder: Improve the simple algorithm for -Os (PR67864)

On Fri, Oct 09, 2015 at 11:29:16AM +0200, Richard Biener wrote: > I think the patch makes sense but it also raises a question for me - how > did we decide what edge gets EDGE_FALLTHRU when going out-of-cfglayout? Good question. I think it just tries to make "natural" control flow; I'll investigat

Re: [PATCH V3][GCC] Algorithmic optimization in match and simplify

On Thu, Oct 08, 2015 at 01:29:34PM +0100, Richard Biener wrote: > > Thanks again for the comments Richard! > > > > A new algorithmic optimisation: > > > > ((X inner_op C0) outer_op C1) > > With X being a tree where value_range has reasoned certain bits to always be > > zero throughout its computed

[PATCH][COMMITTED] PR target/67895: Fix embedded rounding/SAE specifier position for some AVX512 instructions.

Committed to trunk in r228660. Thanks, Alexander --- gcc/ PR target/67895 * config/i386/sse.md (define_insn "sse_cvtsi2ss"): Adjust embedded rounding/SAE specifier position. (define_insn "sse_cvtsi2ssq"): Likewise. (define_insn "cvtusi232"): Likewise.

Re: [AArch64_be] Fix vtbl[34] and vtbx4

On 8 October 2015 at 11:12, James Greenhalgh wrote: > On Wed, Oct 07, 2015 at 09:07:30PM +0100, Christophe Lyon wrote: >> On 7 October 2015 at 17:09, James Greenhalgh >> wrote: >> > On Tue, Sep 15, 2015 at 05:25:25PM +0100, Christophe Lyon wrote: >> > >> > Why do we want this for vtbx4 rather th

Re: Move sqrt and cbrt simplifications to match.pd

Richard Sandiford writes: > Christophe Lyon writes: >> On 8 October 2015 at 18:55, Richard Sandiford >> wrote: >>> Marc Glisse writes: On Mon, 5 Oct 2015, Richard Sandiford wrote: > + /* cbrt(sqrt(x)) -> pow(x,1/6). */ > + (simplify > + (sqrts (cbrts @0)) > + (p

Re: [PATCH] bb-reorder: Improve the simple algorithm for -Os (PR67864)

On Fri, Oct 09, 2015 at 12:35:46PM +0200, Bernd Schmidt wrote: > On 10/08/2015 06:57 PM, Segher Boessenkool wrote: > >As the PR points out, the "simple" reorder algorithm makes bigger code > >than the STC algorithm did, for -Os, for x86. I now tested it for many > >different targets and it turns o

[PATCH] [6/n] Fix minor SSA_NAME leaks

free_stmt_vec_info was releasing the LHS of a statement from the *wrong* statement. Thankfully, the LHS it was actually releasing was just released by some earlier code, so nothing bad happened (the name manager explicitly allows multiple releases of the same node -- something we may want to c

Re: [gomp4.1] fix more scheduling inconsistencies and add verification routines

As per our IRC discussion. I am conditionally compiling the verification code because you mentioned that the GPGPUs may not having a working printf. Also, I removed the code caching the workgroup since it may contain the incorrect workgroup as I had suggested. Now instead we look in child_t

Re: [gomp4.1] fix more scheduling inconsistencies and add verification routines

On Fri, Oct 09, 2015 at 09:38:40AM -0700, Aldy Hernandez wrote: > As per our IRC discussion. > > I am conditionally compiling the verification code because you mentioned > that the GPGPUs may not having a working printf. Both that and code size being important there. > Also, I removed the code c

Re: [PATCH, rs6000] Add memory barriers to tbegin, tend, etc.

On Fri, 2015-10-09 at 16:41 +0200, Torvald Riegel wrote: > On Thu, 2015-09-03 at 16:58 -0500, Peter Bergner wrote: >> +Note that the semantics of the above HTM builtins are required to mimic the >> +locking semantics used for critical sections. Builtins that are used to >> +create a new transactio

[PATCH, i386]: Use ROUND_DOWN some more

Hello! No functional changes. 2015-10-09 Uros Bizjak * config/i386/i386.c (expand_set_or_movmem_prologue_epilogue_by_misaligned_moves): Use ROUND_DOWN where applicable. Bootstrapped and regression tested on x86_64-linux-gnu {,-m32}, committed to mainline SVN: Uros. Index: config

[gomp4] Backport from trunk

Hi, The attached patch was backported from trunk. commit 13d478244d380d1564eefc204201b7f681218ce3 Author: jnorris Date: Fri Oct 9 12:35:58 2015 + * config/rs6000/rs6000.c (rs6000_offload_options): New. (TARGET_OFFLOAD_OPTIONS): New. Jim diff --git a/gcc/config/rs6000/rs

Re: [gomp4.1] OpenMP 4.1 is dead, long live OpenMP 4.5

On Fri, Oct 09, 2015 at 09:55:07 +0200, Jakub Jelinek wrote: > -GOMP_4.1 { > +GOMP_4.5 { >global: > GOMP_target_41; > GOMP_target_data_41; Should we rename it to GOMP_target*_45, or do you know some more mnemonic name? -- Ilya

Re: [PATCH] New attribute to create target clones

On 10/08/2015 03:36 PM, Jan Hubicka wrote: Yes, here you have different names for different variants of the function body. Basically this pass takes ctarget attribute and creates bunch of verisons of the functions and assigns them the proper target attributes, right? Right. Given a single func

Re: Do not use TYPE_CANONICAL in useless_type_conversion

> > The type has BLKmode and size 32. DECL_SIZE of the FIELD_DECL is however 24 > > (see it printed as Ada size). > > Yes, no wonder since it's a bitfield, i.e. DECL_BIT_FIELD is set. > > > The DECL_MODE of the FIELD_DECL is VOIDmode (not printed), while the > > TYPE_MODE of type contained is BLK

Re: [PATCH] New attribute to create target clones

> >Of course it also depends what you inline into function. You can have > > > >bar() target(-mavx) {fancy avx code} > >foobar() { .. if (avx) bar();} > >foo() ctarget(-mavx,-mno-avx) {foobar();} > > > >Now if you compile with -mavx and because ctarget takes effect only after > >inlini

Re: Do not use TYPE_CANONICAL in useless_type_conversion

> Jan Hubicka writes: > > > Does the patch in https://gcc.gnu.org/ml/gcc-patches/2015-10/msg00902.html > > help? > > No, it doesn't. > > #0 fancy_abort (file=0x43f1ce48 "../../gcc/expr.c", line=282, > function=0x43f1ec38 int)::__FUNCTION__> "convert_move") at ../../gcc/d

Re: [PATCH] Fix another bootstrap-ubsan failure due to -Werror=maybe-uninitialized.

Le 08/10/2015 11:06, Maxim Ostapenko a écrit : Hi, running UBSan bootstrap on trunk, I've run to the such issue: /home/max/workspace/downloads/svn/trunk/gcc/fortran/parse.c: In function ‘gfc_statement decode_statement()’: /home/max/workspace/downloads/svn/trunk/gcc/fortran/parse.c:368:51: error

Re: [PATCH] Unswitching outer loops.

On Wed, Oct 7, 2015 at 2:53 AM, Yuri Rumyantsev wrote: > Richard, > > I've fixed adding virtual phi argument and add check on irreducible basic > block. > New patch is attached. > > I checked it for bootstrap and regression testing, no new failures. > > ChangeLog: > 2015-10-07 Yuri Rumyantsev

Re: [PATCH] New attribute to create target clones

On Fri, Oct 9, 2015 at 9:27 PM, Jan Hubicka wrote: >> >Of course it also depends what you inline into function. You can have >> > >> >bar() target(-mavx) {fancy avx code} >> >foobar() { .. if (avx) bar();} >> >foo() ctarget(-mavx,-mno-avx) {foobar();} "no-" targets are not supported

Re: [PATCH] New attribute to create target clones

> On Fri, Oct 9, 2015 at 9:27 PM, Jan Hubicka wrote: > >> >Of course it also depends what you inline into function. You can have > >> > > >> >bar() target(-mavx) {fancy avx code} > >> >foobar() { .. if (avx) bar();} > >> >foo() ctarget(-mavx,-mno-avx) {foobar();} > > "no-" targets are

Re: [PATCH] PR66870 PowerPC64 Enable gold linker with split stack

Here's a new one. Tried all the variations, verified that split stack is still enabled and uses gold linker for 64 bit targets when using a 64 bit default compiler, and does not give the split stack error for 32 bit default compilers. On 10/08/2015 03:46 PM, Matthias Klose wrote: On 08.10.20

Re: [PATCH, rs6000] Add memory barriers to tbegin, tend, etc.

On Fri, 2015-10-09 at 11:52 -0500, Peter Bergner wrote: > On Fri, 2015-10-09 at 16:41 +0200, Torvald Riegel wrote: > > On Thu, 2015-09-03 at 16:58 -0500, Peter Bergner wrote: > >> +Note that the semantics of the above HTM builtins are required to mimic > >> the > >> +locking semantics used for cri

Re: [Boolean Vector, patch 1/5] Introduce boolean vector to be used as a vector comparison type

On 10/02/2015 07:59 AM, Ilya Enkovich wrote: 2015-10-02 Ilya Enkovich * doc/tm.texi: Regenerated. * doc/tm.texi.in (TARGET_VECTORIZE_GET_MASK_MODE): New. * stor-layout.c (layout_type): Use mode to get vector mask size. * target.def (get_mask_mode): New.

  1   2   >