Re: libgo patch committed: Change build procedure to use build tags

On Fri, Aug 12, 2016 at 6:56 AM, Rainer Orth wrote: > Hi Ian, > this patch broke i386-pc-solaris2.12 and sparc-sun-solaris2.12 bootstrap, however: in both cases, the 64-bit build of os.lo fails like this:

Re: libgo patch committed: Change build procedure to use build tags

On Fri, Aug 12, 2016 at 4:19 AM, Andreas Krebbel wrote: > On 08/06/2016 02:36 AM, Ian Lance Taylor wrote: >> Go packages use build tags (see the section on Build Constraints at >> https://golang.org/pkg/go/build/) to select which files to build on >> specific systems.

Re: libgo patch committed: Change build procedure to use build tags

On Fri, Aug 12, 2016 at 2:15 AM, Rainer Orth wrote: > >>> this patch broke i386-pc-solaris2.12 and sparc-sun-solaris2.12 >>> bootstrap, however: in both cases, the 64-bit build of os.lo fails like >>> this: >>> >>>

Go patch committed to GCC 6 branch

I've backported a change to the go tool to the GCC 6 branch. This change came in on mainline as part of the update to Go 1.7 library, and update that will not be made on GCC 6 branch. The original change was https://golang.org/cl/21692. Backport bootstrapped and tested on x86_64-pc-linux-gnu,

RE: [ARM][PATCH] Add support for overflow add, sub, and neg operations

Christophe, I have resolved the issue you identified. I had not update my upstream patch with a previous fix I had made to the usubv pattern. Tested and bootstrapped on arm hardware. This patch improves code generations for builtin arithmetic overflow operations for the arm backend. As an

Re: [PATCH] Add mark_spam.py script

Next observation on this script: it dies if a bug number in the given range doesn't exist, with an error like: Marking as spam: PR75336 Traceback (most recent call last): File "./mark_spam.py", line 98, in mark_as_spam(id, args.api_key, args.verbose) File "./mark_spam.py", line 38, in

Re: [PATCH] Add mark_spam.py script

On 08/12/2016 06:10 PM, Joseph Myers wrote: > On Wed, 10 Aug 2016, Martin Liška wrote: > >> On 08/10/2016 10:49 PM, Joseph Myers wrote: >>> The latest spam bugs have spam attachments as well. I'm not sure if the >>> API can delete attachments, but it would be helpful for the script to do >>> as

Ping! Re: [PATCH, Fortran] New flag -finit-derived to initialize components of derived types

https://gcc.gnu.org/ml/fortran/2016-08/msg00014.html On Thu, Aug 4, 2016 at 2:07 PM, Fritz Reese wrote: > ... > To increase usability and compatibility I have thus a patch which > introduces a new flag -finit-derived into GNU Fortran, allowing > initialization of automatic

Re: [PATCH] Fix ubsan EH updating (PR c/71512)

On August 12, 2016 5:58:06 PM GMT+02:00, Jakub Jelinek wrote: >Hi! > >With -ftrapv -fnon-call-exceptions -fexceptions +/-/* arithmetics can >have >EH edges, and ubsan code didn't take it into account. >Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok >for

Re: [gomp4] Extend libgomp's fortran test coverage of host_data

On 08/11/2016 03:38 PM, Cesar Philippidis wrote: > This patch ports libgomp.oacc-c-c++-common/host_data-1.c to fortran. > Fortunately, the existing fortran host_data infrastructure was already > in place, so I had to do was port over the calls to Nvidia's CUDA BLAS > library. > > There are a

Re: [PATCH] Extend -falign-FOO=N to N[,M]: the second number is max padding

On 08/12/2016 05:20 PM, Denys Vlasenko wrote: Yes, I know all that. Fetching is one thing. Loop cache is for instance another (more important) thing. Not aligning the loop head increases chance of the whole loop being split over more cache lines than necessary. Jump predictors also don't

[PATCH] Fix suggestions for non-trivial Wformat type cases (PR c/72858)

In r239260 I attempted to add fix-it hints for -Wformat type warnings. Unfortunately, my implementation was too simplistic, and only worked correctly for the most simple format strings: the fix-it hint would suggest replacement of an entire conversion specification, but the replacement would only

Re: [PATCH] RFC: -fasm-show-source

On Fri, 2016-08-12 at 11:00 +0530, Prathamesh Kulkarni wrote: > On 12 August 2016 at 02:04, David Malcolm > wrote: > > I sometimes find myself scouring assembler output from the compiler > > and trying to figure out which instructions correspond to which > > lines of source

[PATCH] Add source information to -fverbose-asm

On Thu, 2016-08-11 at 20:00 -0600, Sandra Loosemore wrote: > On 08/11/2016 02:34 PM, David Malcolm wrote: > > I sometimes find myself scouring assembler output from the compiler > > and trying to figure out which instructions correspond to which > > lines of source code; I believe this is a common

[PATCH] lra: A multiple_sets is not a simple_move_p (PR73650)

Hi! In the PR we have a PARALLEL of a move and a compare (a "mr." instruction). The compare is dead, so single_set on it returns just the move. Then, simple_move_p returns true; but the instruction does need reloads in this case. This patch solves this by making simple_move_p return false for

Re: [PATCH] Fix early debug regression with DW_AT_string_length (PR debug/71906)

On Fri, Aug 12, 2016 at 01:47:14PM -0400, Jason Merrill wrote: > On 07/21/2016 12:53 PM, Jakub Jelinek wrote: > > size = int_size_in_bytes (TREE_TYPE (szdecl)); > ... > >+ if (size != DWARF2_ADDR_SIZE) > >+ add_AT_unsigned (array_die, DW_AT_byte_size,

Re: [PATCH] Fix early debug regression with DW_AT_string_length (PR debug/71906)

On 07/21/2016 12:53 PM, Jakub Jelinek wrote: size = int_size_in_bytes (TREE_TYPE (szdecl)); ... + if (size != DWARF2_ADDR_SIZE) + add_AT_unsigned (array_die, DW_AT_byte_size, size); For DWARF5, where DW_AT_byte_size is always the size of the array

[PATCH v2] Extend -falign-FOO=N to N[,M]: the second number is max padding

falign-functions=N is too simplistic. Ingo Molnar ran some tests and it seems that on latest x86 CPUs, 64-byte alignment runs fastest (he tried many other possibilites). However, developers are less than thrilled by the idea of a slam-dunk 64-byte aligning everything. Too much waste: On

[PATCH] Remove "at least 8 byte alignment" code from x86 targets

This change drops forced alignment to 8 if requested alignment is higher than 8: before the patch, -falign-functions=9 was generating .p2align 4,,8 .p2align 3 which means: "align to 16 if the skip is 8 bytes or less; else align to 8". After this change, ".p2align 3" is not

Re: [PATCH 1/4][Ada,DJGPP] Ada support for DJGPP

> 2016-07-30 Andris Pavenis > > * ada/adaint.c (__gnat_is_djgpp): define (1 for DJGPP host, 0 > otherwise). * ada/s-os_lib.ads (Is_Djgpp): import __gnat_is_djgpp as > constant. * ada/s-os_lib.adb (Normalize_Pathname): support DJGPP special > paths (/dev/*) for DJGPP

[PR fortran/72741] Handle intrinsic functions specified in !$ACC ROUTINE ( NAME )

Hi! Let me actually break this out of the other pending patches; this should be uncontroversial. Originally by Cesar, extended by me. OK for trunk? commit a0fee96c0f204814e87ddf6635f9cbec2afc6887 Author: Thomas Schwinge Date: Fri Aug 12 17:19:05 2016 +0200 [PR

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

On 08/12/2016 09:48 AM, Joseph Myers wrote: On Thu, 11 Aug 2016, Martin Sebor wrote: * New target hooks remove hardcoding target-specific assumptions about libc implementation-specific details (%p format and printf floating point rounding mode). But the rounding mode may vary at

Re: [PATCH] Add mark_spam.py script

On Wed, 10 Aug 2016, Martin Liška wrote: > On 08/10/2016 10:49 PM, Joseph Myers wrote: > > The latest spam bugs have spam attachments as well. I'm not sure if the > > API can delete attachments, but it would be helpful for the script to do > > as much as possible with them (change filenames,

Re: [C PATCH] Fix set_nonincremental_init_from_string (PR c/67410)

On Fri, Aug 12, 2016 at 05:54:55PM +0200, Jakub Jelinek wrote: > Hi! > > The wide-int conversion of set_nonincremental_init_from_string broke this > function. > Before we used val[bitpos < HOST_BITS_PER_WIDE_INT], but val[1] and val[0] > has been swapped, so either it should be bitpos >=

Re: [C PATCH] Fix set_nonincremental_init_from_string (PR c/67410)

On Fri, 12 Aug 2016, Jakub Jelinek wrote: > Hi! > > The wide-int conversion of set_nonincremental_init_from_string broke this > function. > Before we used val[bitpos < HOST_BITS_PER_WIDE_INT], but val[1] and val[0] > has been swapped, so either it should be bitpos >= HOST_BITS_PER_WIDE_INT, > or

Re: Implement -Wimplicit-fallthrough: core

On Fri, 12 Aug 2016, Marek Polacek wrote: > Not sure if error is appropriate here, or whether I should downgrade the > error to a warning and ignore the attribute. I'd say the semantics for uses of attributes that currently parse should remain unchanged: they should not be reinterpreted (for

[PATCH] Fix ubsan EH updating (PR c/71512)

Hi! With -ftrapv -fnon-call-exceptions -fexceptions +/-/* arithmetics can have EH edges, and ubsan code didn't take it into account. Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk/6.2? 2016-08-12 Jakub Jelinek PR c/71512 *

[C PATCH] Fix set_nonincremental_init_from_string (PR c/67410)

Hi! The wide-int conversion of set_nonincremental_init_from_string broke this function. Before we used val[bitpos < HOST_BITS_PER_WIDE_INT], but val[1] and val[0] has been swapped, so either it should be bitpos >= HOST_BITS_PER_WIDE_INT, or bitpos / HOST_BITS_PER_WIDE_INT. Bootstrapped/regtested

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

On Thu, 11 Aug 2016, Martin Sebor wrote: > * New target hooks remove hardcoding target-specific assumptions > about libc implementation-specific details (%p format and printf > floating point rounding mode). But the rounding mode may vary at runtime; optimally the conversions should

Re: [PATCH GCC]Resolve compilation time known alias checks in vectorizer

On Tue, Aug 9, 2016 at 4:43 PM, Bin.Cheng wrote: > On Sat, Aug 6, 2016 at 9:20 PM, Andreas Schwab wrote: >> On Mi, Jul 13 2016, "Bin.Cheng" wrote: >> >>> Patch re-tested/applied on trunk as r238301. >> >> This breaks

Re: [PATCH] Extend -falign-FOO=N to N[,M]: the second number is max padding

On 08/12/2016 04:28 PM, Michael Matz wrote: Hi, On Fri, 12 Aug 2016, Denys Vlasenko wrote: Have you tested the performance impact of your patch? Note that the macro you changed is used for function and code label alignment. So, unless I misunderstand something that means that if the large

Re: backward threading heuristics tweek

On Thu, Aug 11, 2016 at 01:35:16PM +0200, Jan Hubicka wrote: > > On Mon, Jun 6, 2016 at 3:19 AM, Jan Hubicka wrote: > > > Hi, > > > while looking into profile mismatches introduced by the backward > > > threading pass > > > I noticed that the heuristics seems quite simplistics.

Re: Implement -Wswitch-fallthrough

On Wed, Jul 13, 2016 at 09:05:26PM +0200, Marc Glisse wrote: > > > Unrelated question: are there cases where __builtin_fallthrough() has any > > > impact on code generation? > > > > It should not -- my implementation gets rid of all __builtin_fallthrough() > > during gimple-low, before most of

Re: [PATCH] Extend -falign-FOO=N to N[,M]: the second number is max padding

Hi, On Fri, 12 Aug 2016, Denys Vlasenko wrote: > > Have you tested the performance impact of your patch? Note that the > > macro you changed is used for function and code label alignment. So, > > unless I misunderstand something that means that if the large > > alignment can't be achieved

Re: Implement -Wimplicit-fallthrough: core

On Fri, Jul 29, 2016 at 10:58:28PM +, Joseph Myers wrote: > On Fri, 22 Jul 2016, Marek Polacek wrote: > > > * I noticed that we don't diagnose e.g. '__attribute__((noreturn));' in > > the C FE in certain contexts. Shouldn't be a showstopper though. > > Bug number? shadow_tag_warned is

Re: [PATCH] Extend -falign-FOO=N to N[,M]: the second number is max padding

On 08/12/2016 04:00 PM, Michael Matz wrote: On Thu, 11 Aug 2016, Denys Vlasenko wrote: This change makes it possible to align function to 64-byte boundaries *IF* this does not introduce huge amount of padding. Patch drops forced alignment to 8 if requested alignment is higher than 8: before

Re: [PATCH, RFC] gcov: dump in a static dtor instead of in an atexit handler

On 08/10/2016 02:53 PM, Nathan Sidwell wrote: > On 08/10/16 06:43, Martin Liška wrote: >> Hello. >> >> There are multiple PRs (mentioned in ChangeLog) which suffer from missing >> capability of gcov >> to save counters for functions with constructor/destructor attributes. I >> done that by

Re: [PATCH] Fix 4 avx512* mask reg constraints

On Fri, Aug 12, 2016 at 12:51 PM, Jakub Jelinek wrote: > Hi! > > For the VEXTRACT* insns, I can't find anything that would make them > special wrt. masking, so I believe they can't accept {%k0}, just > no masking or {%k1} through {%k7}. > > I've tried to reproduce it with: >

Re: [RFC] ipa bitwise constant propagation

> On 11 August 2016 at 18:25, Jan Hubicka wrote: > >> @@ -266,6 +267,38 @@ private: > >>bool meet_with_1 (unsigned new_align, unsigned new_misalign); > >> }; > >> > >> +/* Lattice of known bits, only capable of holding one value. > >> + Similar to ccp_lattice_t, mask

Re: [PATCH] Extend -falign-FOO=N to N[,M]: the second number is max padding

Hi, On Thu, 11 Aug 2016, Denys Vlasenko wrote: > This change makes it possible to align function to 64-byte boundaries > *IF* this does not introduce huge amount of padding. > > Patch drops forced alignment to 8 if requested alignment is higher than > 8: before the patch, -falign-functions=9

Re: libgo patch committed: Change build procedure to use build tags

Hi Ian, >>> this patch broke i386-pc-solaris2.12 and sparc-sun-solaris2.12 >>> bootstrap, however: in both cases, the 64-bit build of os.lo fails like >>> this: >>> >>> /vol/gcc/src/hg/trunk/local/libgo/go/os/dir.go:82:8: error: reference to >>> undefined name 'libc_readdir_r' >>>i :=

[PATCH, testsuite]: Fix PR71008, FAIL: gcc.target/i386/avx512f-vprord-1.c scan-assembler-times vprord

As shown in the PR, during the combine pass, the compiler can convert ROTATERT to an equivalent ROTATE pattern. Attached patch adjusts a couple of scan-asm patterns to also check for vprol insns in addition to vpror. 2016-08-12 Uros Bizjak PR testsuite/71008 *

[PR56974] output DWARF-5 markers for ref_qualifiers

When a method or a method or function type has a ref-qualifier, output DW_AT_reference or DW_AT_rvalue_reference in the subprogram or subroutine_type tag, as specified in DWARF version 5 drafts, see . Output pointer to member function types as

Re: [PATCH] Set -fprofile-update=atomic when -pthread is present

On 08/09/2016 09:03 PM, Andi Kleen wrote: > It could potentially make things a lot slower. I don't think it's a good > idea to do this by default. > > -Andi Ok, alternative can be a warning in the driver that would inform a user that combining -pthread and -fprofile-update=single can lead to

Re: [PATCH, COMMITTED] Add branch_changer.py script to maintainer-scripts

On 08/09/2016 09:28 AM, Gerald Pfeifer wrote: > On Wed, 3 Aug 2016, Martin Liška wrote: >> I've installed (r239066) the script which is used by maintainers to >> update PRs in a batch mode. > > I think it would be good to add a comment at the top that > describes what the scripts does and how to

Re: [PATCH] Fix PR57326

On Fri, 12 Aug 2016, Richard Biener wrote: > > The following fixes a missed optimization in PRE which was over-cautionous > with rejecting simplifications to SSA names during phi-translation. > > Bootstrapped on x86_64-unknown-linux-gnu, testing in progress. Requires some testsuite adjustments

Re: [PATCH, RFC] Introduce -fsanitize=use-after-scope

PING^1 > On 05/12/2016 12:41 PM, Jakub Jelinek wrote: >> On Wed, May 11, 2016 at 02:54:01PM +0200, Martin Liška wrote: >>> On 05/06/2016 02:22 PM, Jakub Jelinek wrote: On Fri, May 06, 2016 at 01:04:30PM +0200, Martin Liška wrote: > I've started working on the patch couple of month go,

[PATCH] Fix invalid memory access in gcc.c (driver/72765)

Hi. Simple patch corrects assumption about string length, however the hunk in save_string is kind of discussable as one can have a string with '\0' chars which is length enough? Thoughts? Patch can bootstrap on ppc64le-redhat-linux and survives regression tests. Ready to be installed? Martin

Re: [PATCH 0/4] Properly handle GCC target("march=") (PR71652)

PING^1 > Hello. > > Following small patch set targets $subject, where we ICE if someone > uses #pragma GCC target ("arch=generic"). My attempt is to not to > create a new target optimization node in case of a wrong value > of march string. Such approach does not generate multiple errors. > >

Re: divmod transform: add test-cases

On 12 August 2016 at 12:47, Richard Biener wrote: > On Thu, 11 Aug 2016, Jeff Law wrote: > >> On 08/09/2016 04:58 AM, Prathamesh Kulkarni wrote: >> > ping https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01869.html >> This seems to be dependent upon other patches, this is not OK

[PATCH] Fix PR73434

I am testing the following patch fixing a TBAA related miscompile. Bootstrap / regtest running on x86_64-unknown-linux-gnu. Richard. 2016-08-12 Richard Biener PR tree-optimization/73434 * tree-ssa-sccvn.c (vn_reference_maybe_forwprop_address): Preserve

Re: Early jump threading

> > * passes.def (pass_early_thread_jumps): Schedule after forwprop. > > * tree-pass.h (make_pass_early_thread_jumps): Declare. > > * tree-ssa-threadbackward.c (fsm_find_thread_path, > > fsm_find_thread_path, profitable_jump_thread_path, > >

Re: libgo patch committed: Change build procedure to use build tags

On 08/06/2016 02:36 AM, Ian Lance Taylor wrote: > Go packages use build tags (see the section on Build Constraints at > https://golang.org/pkg/go/build/) to select which files to build on > specific systems. > > Previously the libgo Makefile explicitly listed the set of files to > compile for

[PATCH] Properly propagate into PHIs in forwprop

forwprop has a lattice now (mainly for match-and-simplify simplification) but it fails to substitute into PHIs or update the lattice for degenerate PHIs. Fixed thusly. Bootstrapped on x86_64-unknown-linux-gnu, testing in progress. Richard. 2016-08-12 Richard Biener

[PATCH] Fix PR57326

The following fixes a missed optimization in PRE which was over-cautionous with rejecting simplifications to SSA names during phi-translation. Bootstrapped on x86_64-unknown-linux-gnu, testing in progress. Richard. 2016-08-12 Richard Biener PR

[PATCH] Fix 4 avx512* mask reg constraints

Hi! For the VEXTRACT* insns, I can't find anything that would make them special wrt. masking, so I believe they can't accept {%k0}, just no masking or {%k1} through {%k7}. I've tried to reproduce it with: #include __m128i f1 (__m128i a, __m512i b, __mmask8 c) { register __mmask8 d __asm

Re: [PATCH] Fix PR middle-end/71654 (missed shortening of a compare)

On Tue, Aug 9, 2016 at 4:04 PM, Patrick Palka wrote: > Hi, this patch makes match.pd to shorten comparisons that involve a > sign-changing cast from a shorter unsigned type to a wider signed type. > This should be safe to do because a wider signed type can hold all the >

Re: [RFC][IPA-VRP] Early VRP Implementation

On Wed, Aug 3, 2016 at 3:17 AM, kugan wrote: > Hi Richard, > > Thanks for the review. > > On 28/07/16 21:34, Richard Biener wrote: >> >> On Thu, Jul 28, 2016 at 9:35 AM, kugan >> wrote: >>> >>> Hi Richard, >>> >>> Thanks for

Re: [Patch, libfortran] Multi-threaded random_number

On Fri, Aug 12, 2016 at 1:07 PM, Thomas Schwinge wrote: > Hi! > > On Thu, 11 Aug 2016 12:01:59 +0300, Janne Blomqvist > wrote: >> committed a slightly modified patch as r239356. [...] > > This breaks the build for nvptx Sorry about that! >

Re: [RFC] Use ccp mask and val to refine value_range

On Mon, Aug 8, 2016 at 3:41 AM, kugan wrote: > Hi All, > > I was experimenting with using ccp mask and value to refine value_ranges for > SSA_NAME. Attached patch does this. This might be particularly useful for > early_vrp. Bootstrap on x86_64-linux-gnu is OK.

Re: RFC: A few more fallthrus

On Fri, Aug 12, 2016 at 09:20:36AM +0200, Richard Biener wrote: > > --- gcc/gcc/tree-complex.c > > +++ gcc/gcc/tree-complex.c > > @@ -1336,6 +1336,8 @@ expand_complex_division (gimple_stmt_iterator *gsi, > > tree inner_type, > >rr = gimplify_build2 (gsi, code, inner_type, ai, bi); > >

Re: [Patch, libfortran] Multi-threaded random_number

Hi! On Thu, 11 Aug 2016 12:01:59 +0300, Janne Blomqvist wrote: > committed a slightly modified patch as r239356. [...] This breaks the build for nvptx (but it may actually be a problem specific to the nvptx-newlib); I filed "libgfortran:

Re: [RFC] ipa bitwise constant propagation

On 11 August 2016 at 18:25, Jan Hubicka wrote: >> @@ -266,6 +267,38 @@ private: >>bool meet_with_1 (unsigned new_align, unsigned new_misalign); >> }; >> >> +/* Lattice of known bits, only capable of holding one value. >> + Similar to ccp_lattice_t, mask represents which

Re: libgo patch committed: Change build procedure to use build tags

Hi Ian, >> this patch broke i386-pc-solaris2.12 and sparc-sun-solaris2.12 >> bootstrap, however: in both cases, the 64-bit build of os.lo fails like this: >> >> /vol/gcc/src/hg/trunk/local/libgo/go/os/dir.go:82:8: error: reference to >> undefined name 'libc_readdir_r' >>i :=

[PATCH GCC]Refactor doloop by moving cheaper checks earlier.

Hi, This is a simple refactoring patch which can help long compile time issue reported as PR72855. Basically, it moves cheaper checks before expensive ones. By doing this, doloop -optimization fails on cheap checks, thus following expensive data-flow analysis is skipped. Bill helped

[committed] Fix up tsan on aarch64 (PR sanitizer/71042)

Hi! I've committed to trunk and 6.2 following backport from upstream, which ought to fix tsan on aarch64, after bootstrapping/regtesting it on aarch64 and verifying it no longer references any GLIBC_PRIVATE symbols. 2016-08-12 Jakub Jelinek PR sanitizer/71042

Re: [PATCH PR72817/PR73450]Fix wrong code caused by niter analyzer for NE_EXPR exit cond.

On Thu, Aug 11, 2016 at 6:35 PM, Bin Cheng wrote: > Hi, > I made a mistake when improving loop niter analysis for NE_EXPR exit > condition. > It can results in wrong code as reported in PR72817/PR73450. In previous > patch, > it simplifies below condition: > base

Re: Early jump threading

On Thu, 11 Aug 2016, Jan Hubicka wrote: > Hi, > this patch adds early jump threading pass. Jump threading is one of most > common cases where estimated profile becomes corrupted, because the branches > are predicted independently beforehand. This patch simply adds early mode to > jump threader

Re: [PATCH][RFC] Add the ability to record sub-timevars (-ftime-report-details)

On Wed, 10 Aug 2016, Richard Biener wrote: > On Wed, 10 Aug 2016, David Malcolm wrote: > > > On Wed, 2016-08-10 at 14:04 +0200, Richard Biener wrote: > > > The following patch adds the ability to record time spent in utility > > > to the pass using it. For things like CFG cleanup, alias stmt >

Re: [PATCH][RFC] Fix PR72851 by "randomizing" SSA propagator worklist extraction

On Thu, 11 Aug 2016, Richard Biener wrote: > On Wed, 10 Aug 2016, Richard Biener wrote: > > > > > The following randomizes SSA propagator worklist processing to make the > > processing order less likely hit the worst-case as in the PR. Ideally > > we'd process it in stmt order but that adds

Re: RFC: A few more fallthrus

On Thu, 11 Aug 2016, Marek Polacek wrote: > A few more cases where I'm unsure whether the fall through is intended. > Jason, can you please look at the cp/ part? > Richi, would you mind looking at the tree-complex.c bit? > What 'bout the pch.c? > > Thanks, > > 2016-08-11 Marek Polacek

Re: divmod transform: add test-cases

On Thu, 11 Aug 2016, Jeff Law wrote: > On 08/09/2016 04:58 AM, Prathamesh Kulkarni wrote: > > ping https://gcc.gnu.org/ml/gcc-patches/2016-07/msg01869.html > This seems to be dependent upon other patches, this is not OK until all > prereqs are resolved. > > You're using SI/DI in the

Re: [PATCH] Increase alignment for bit-field access when predictive commoning (PR 71083)

On Thu, 11 Aug 2016, Bernd Edlinger wrote: > On 08/11/16 09:07, Richard Biener wrote: > > The patch looks mostly ok, but > > > > + else > > + { > > + boff >>= LOG2_BITS_PER_UNIT; > > + boff += tree_to_uhwi (component_ref_field_offset (ref)); > > + coff =

Re: Early jump threading

On August 11, 2016 8:41:53 PM GMT+02:00, Jeff Law wrote: >On 08/11/2016 08:27 AM, Jan Hubicka wrote: >> >> On tramp3d all VRP passes threads together 730 branches, all DOM >passes 393, so >> FSM threading (with 1957 branches) is the most effective one. Perhaps >eventually >>