Re: [patch, libgfortran] PR48906 Wrong rounding results with -m32

On Fri, Jun 10, 2011 at 20:27, jerry DeLisle wrote: > On 06/03/2011 05:51 AM, jerry DeLisle wrote: >> >> Hi, >> >> The attached patch, which includes test cases, fixes this bug by >> eliminating the >> code which used floating point instructions to determine the 'r' value as >> outlined in the For

Go patch committed: Use backend interface for map descriptors

This patch to the Go frontend uses the backend interface for map descriptors. In order to do this I had to change the types of the map descriptor fields from size_t to uintptr_t, which led to a few changes in libgo. Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu. Committed to main

Go patch committed: Use backend interface for type descriptors

This patch to the Go frontend uses the backend interface for type descriptors. Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu. Committed to mainline. Ian 2011-06-10 Ian Lance Taylor * go-gcc.cc: Include "toplev.h". (Gcc_backend::immutable_struct): New functio

[google] Enhance Annotalysis to support cloned functions/methods (issue4591066)

Just identified a bug in my previous patch after running the compiler on google code base. Basically the difference from the previous patch is for the compiler to handle the case where the parameters of a cloned method are optimized away. Bootstrapped OK. Testing is still on-going. OK for google/m

Re: [google/gcc-4_6] merge google/main r174890 to google/gcc-4_6 branch (issue4576055)

On Fri, Jun 10, 2011 at 16:48, Chris Demetriou wrote: > testing with a native bootstrap, not quite done yet but since there were > no conflicts (except for the ChangeLog) it'll be fine. > > OK for google/gcc-4_6 assuming tests pass? OK. > (Note that the properties changes were generated by svnme

[google/gcc-4_6] merge google/main r174890 to google/gcc-4_6 branch (issue4576055)

testing with a native bootstrap, not quite done yet but since there were no conflicts (except for the ChangeLog) it'll be fine. OK for google/gcc-4_6 assuming tests pass? (Note that the properties changes were generated by svnmerge.py, and TBH I have *no idea* what some of them are about.) tha

[pph contrib] Add support for multiple spawn patterns in repro_fail (issue4571061)

So I had to go back to grep|sed instead of just sed because I want to use extended regexps supported by grep -E. In pph this is useful so that you can specify patterns that match both the writing and the reading of the PPH files. I'm thinking that this script is better written in python, but tha

Re: PING^3 [PATCH] Support for AMD64 targets running GNU/kFreeBSD

2011/6/10 Richard Henderson : > On 06/10/2011 01:59 PM, Robert Millan wrote: >> 2011-06-02  Robert Millan   >> >>       * config/i386/kfreebsd-gnu.h: Resync with `config/i386/linux.h'. >>       * config/kfreebsd-gnu.h (GNU_USER_DYNAMIC_LINKER): Resync with >>       `config/linux.h'. >> >>       * c

Re: PING^3 [PATCH] Support for AMD64 targets running GNU/kFreeBSD

On 06/10/2011 01:59 PM, Robert Millan wrote: > 2011-06-02 Robert Millan > > * config/i386/kfreebsd-gnu.h: Resync with `config/i386/linux.h'. > * config/kfreebsd-gnu.h (GNU_USER_DYNAMIC_LINKER): Resync with > `config/linux.h'. > > * config/i386/kfreebsd-gnu64.h: New file

PING^3 [PATCH] Support for AMD64 targets running GNU/kFreeBSD

Please someone look at this if you have time. The patch is very small now. 2011/6/3 Robert Millan : > Actually, please consider this patch instead.  It's the same but fixes a > mistake in ld.so pathname. > > 2011/6/2 Robert Millan : >> Hi, >> >> 2011/5/21 Joseph S. Myers : >>> Please send a patch

Cgraph alias reorg 14/14 (ipa-pure-const, ipa-reference and tree-profile fixes)

Hi, the patch series will be a bit longer as I found some problems exposed by the breakup. This patch fixes latent bug in tree-profile that needs to call execute_fixup_cfg. This is because local pure const might find some edges nothrow and cfg needs to be updated (that usually happens only after

Re: Cgraph alias reorg 13/14 (disable inlining functions called once at -O0

> On Fri, Jun 10, 2011 at 8:42 PM, Jan Hubicka wrote: > > Hi, > > by some mistake we enable functions called once at -O0 and it actually > > happens from > > time to time. > > Why do it for -O1? It definitely makes debugging less reliable. I'd say do > it > for -O[23s] only. Well, that is wh

Re: Cgraph alias reorg 13/14 (disable inlining functions called once at -O0

On Fri, Jun 10, 2011 at 8:42 PM, Jan Hubicka wrote: > Hi, > by some mistake we enable functions called once at -O0 and it actually > happens from > time to time. Why do it for -O1? It definitely makes debugging less reliable. I'd say do it for -O[23s] only. Richard. > Bootstrapped/regtested

Re: [PATCH] PR debug/49348 (DW_TAG_template_* missing from template specializations)

Hmm, I'm not sure about this; it seems to me that we want to know about the actual parameters that a particular specialization has, which in the case of an explicit specialization is none. Or for a partial specialization, template struct A; template struct A { } A a; here A has a type para

Cgraph alias reorg 13/14 (disable inlining functions called once at -O0

Hi, by some mistake we enable functions called once at -O0 and it actually happens from time to time. Bootstrapped/regtested x86_64-linux, will commit it shortly. Honza * opts.c (default_options): Enlist OPT_finline_functions_called_once. * common.opt (flag_inline_functions_call

Re: [C++ Patch] PR 41769

OK. Jason

Re: [PATCH, PR43864] Gimple level duplicate block cleanup.

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 06/08/11 03:42, Tom de Vries wrote: > Hi Richard, > > I have a patch for PR43864. The patch adds a gimple level duplicate block > cleanup. The patch has been bootstrapped and reg-tested on x86_64, and > reg-tested on ARM. The size impact on ARM for

[C++ Patch] PR 41769

Hi, as it happens, a couple of years ago Johannes filed a PR quite similar to 29003 but for function parameters, and we have a similar issue for catch clauses too. Patch tested x86_64-linux. Is it Ok? Thanks, Paolo. / /cp 2011-06-10 Paolo Carlini PR c++/29003

[PATCH] PR debug/49348 (DW_TAG_template_* missing from template specializations)

Hello, For a given template instantiation, the dwarf backend emits debug info that describes its template parameters and arguments if generic_type_p returns TRUE on the the instantiation. For that, primary_template_instantiation_p must be also return TRUE. The problem in this PR is that primary_

Cgraph alias reorg 12/14 (ipa.c update)

Hi, the usual update to walk alias nodes, this time for cgraph_local_node_p and logic deicing whether address is taken. Bootstrapped/regtested x86_64-linux, comitted. Honza * ipa.c (cgraph_non_local_node_p_1): Break out from ... (cgraph_local_node_p): ... here. (has_addr_

Re: [Patch : H8300] Bug fix for bit insn and minor tweaks to insns

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 06/09/11 01:20, Kaushik Phatak wrote: > Hi, > The following patch fixes an ICE that is generated when the compiler tries > to perform bit manipulation for logical operations when the source and > destination address does not match. The testcase is

Cgraph alias reorg 10/14 (C++ fe tweek)

Hi, this patch makes C++ FE to use cgraph_for_node_and_aliases instead of doing it on its own. It helps the transition to new aliases since I don't have to update the code in question. Bootstrapped/regtested x86_64-linux, comitted as obvoius. Honza * decl2.c (clear_decl_external): New

[google] Enhance Annotalysis to support cloned functions/methods (issue4591066)

Enhance Annotalysis to support cloned functions/methods (especially created by IPA-SRA). The patch basically does the following: 1. For a FUNCTION_DECL, check whether it's a clone, and if so, grab its original DECL. 2. Deal with the situation where a reference/pointer parameter is converted to a v

Re: [google] limit excessive load/store motions (issue4563044)

Ok for google/main after the minor cleanups. Incorporate comments from maintainers when available. David http://codereview.appspot.com/4563044/diff/1/gcc/gcse.c File gcc/gcse.c (right): http://codereview.appspot.com/4563044/diff/1/gcc/gcse.c#newcode5050 gcc/gcse.c:5050: if (ld_motion_count >=

Re: [google] limit excessive load/store motions (issue4563044)

This is for google/main for now. This is a port from my work in 4.4.3. I may need to rework the register pressure estimation (try to see if I can reuse the function in trunk) before submitting to trunk. -Rong On Fri, Jun 10, 2011 at 10:52 AM, Diego Novillo wrote: > On Fri, Jun 10, 2011 at 10:45,

Re: [google] limit excessive load/store motions (issue4563044)

On Fri, Jun 10, 2011 at 10:45, Rong Xu wrote: > Use a counter to avoid excessive load/store motions in tree and RTL level. > This recovers some of the performance loss in FDO mode for openssl. Rong, is this for trunk or google/main? If it's for trunk, please make sure you mark it as such. OK fo

[google] limit excessive load/store motions (issue4563044)

Use a counter to avoid excessive load/store motions in tree and RTL level. This recovers some of the performance loss in FDO mode for openssl. 2011-06-10 Rong Xu * gcc/tree-ssa-loop-im.c (maxmimu_lsm): New define. (find_refs_for_sm): Limit excessive lsm. * gcc/gcse.c (c

Re: [patch, fortran] Some more TRIM optimizations

On 06/10/2011 10:08 AM, Thomas Koenig wrote: I wrote: Hello world, the attached patch extends removing trailing TRIMs in assignments for cases like a // trim(b). Regression-tested. OK for trunk? Thomas This time with the test case corrected (cleanup of the *.original file) and a more mean

Re: [Patch, Fortran] PR 49324: Deep copy array constr of DT with allocatable components

On 06/10/2011 05:12 AM, Tobias Burnus wrote: This patch fixes parts of the PR 49324: There was a deep copy missing for assigning an array constructor of a DT with allocatable components. Whether a deep copy is done, depends on a flag. I think the flag has been added to avoid a deep copy and mult

Re: [patch, libgfortran] PR48906 Wrong rounding results with -m32

On 06/03/2011 05:51 AM, jerry DeLisle wrote: Hi, The attached patch, which includes test cases, fixes this bug by eliminating the code which used floating point instructions to determine the 'r' value as outlined in the Fortran standard under G formatting. Essentially, the code now examines the

[v3] Use noexcept in the allocators

Hi, tested x86_64-linux, committed. Paolo. /// 2011-06-10 Paolo Carlini * include/ext/throw_allocator.h: Use noexcept. * include/ext/pool_allocator.h: Likewise. * include/ext/bitmap_allocator.h: Likewise. * include/ext/new_allocator.h: Lik

Re: [patch, fortran] Some more TRIM optimizations

I wrote: Hello world, the attached patch extends removing trailing TRIMs in assignments for cases like a // trim(b). Regression-tested. OK for trunk? Thomas This time with the test case corrected (cleanup of the *.original file) and a more meaningful Subject line. OK? Ping ** 0.5714 Fu

Re: [PATCH, PR43864] Gimple level duplicate block cleanup.

Hi Richard, thanks for the review. On 06/08/2011 11:55 AM, Richard Guenther wrote: > On Wed, Jun 8, 2011 at 11:42 AM, Tom de Vries wrote: >> Hi Richard, >> >> I have a patch for PR43864. The patch adds a gimple level duplicate block >> cleanup. The patch has been bootstrapped and reg-tested on x

[patch, testsuite] Fix g++.dg/opt/devirt2.C scanning, PR 48727

I will check in this testsuite fix as obvious if there are no objections. On IA64 and HPPA the test generates global declarations for the function as well as calls so the scan-assembler-times fails because it finds the declarations as well as the two calls. The fix is to make the scans more specif

Re: Dump before flag

On Fri, Jun 10, 2011 at 1:52 AM, Richard Guenther wrote: > On Thu, Jun 9, 2011 at 5:47 PM, Xinliang David Li wrote: >> See attached. > > Hmm.  I don't like how you still wire dumping in the TODO routines. > Doesn't it work to just dump the body from pass_fini_dump_file ()? > Or if that doesn't so

Re: -fdump-passes -fenable-xxx=func_name_list

On Fri, Jun 10, 2011 at 2:03 AM, Richard Guenther wrote: > On Fri, Jun 10, 2011 at 12:51 AM, Xinliang David Li > wrote: >> Patch is temporally rolled back. >> >> Richard, looks like deeper pass manager cleanup is needed -- I would >> like to delay that. For now, this leaves us two choices -- 1)

Re: [testsuite] skip ARM tests with conflicting options

On 06/10/2011 03:00 AM, Richard Earnshaw wrote: > On 10/06/11 01:04, Janis Johnson wrote: >> On 06/08/2011 03:39 AM, Richard Earnshaw wrote: >>> On 08/06/11 03:14, Janis Johnson wrote: On 06/07/2011 06:25 PM, Mike Stump wrote: > On Jun 7, 2011, at 4:24 PM, Janis Johnson wrote: >> On 06

Re: [PATCH] c-pragma: adding a data field to pragma_handler

I guess this is better now. Changelog (gcc/c-family): 2011-06-10 Pierre Vittet * c-pragma.h (pragma_handler_1arg, pragma_handler_2arg): New handler. (gen_pragma_handler): New union. (internal_pragma_handler): New type. (c_register_pragma_with_data, c_register_pragma_with_e

Re: [pph] Do not emit PCH if generating PPH (issue4591061)

On Thu, Jun 9, 2011 at 11:49, Diego Novillo wrote: > > We were trying to generate PCH and PPH information at the same time. > We never noticed because PPH is generated after PCH, so we were just > clobbering over the previous dump. > > Found it by accident while debugging a GC ICE.  This should ma

Cgraph alias reorg 11/14 (stream indirect call jump functions)

Hi, this patch fixes omission I noticed while debugging unrelated problem. While we compute jump functions for indirect calls (so they can possibly be used after the indirect calls become direct) we forget to stream them. Fixed thus. Regtested/bootstrapped x86_64-linux, will commit it shortly. H

Cgraph alias reorg 9/14 (skip aliases in ipa-cp)

Hi. this patch updates ipa-cp to skip aliases. Since it includes some reformating, I decided to leave it for separate patch. It also prevents ipa-cp from an attempt to clone or propagate through thunks. We still don't do any jump functions on those. Regtested/boostrapped x86_64-linux, will com

[PATCH] PR48454 Fix length of vec_pack_trunc

Hi, It appears as though the length of the vec_pack_trunc was incorrectly not set to anything. This has been fixed now at 8 bytes. I have also taken the oppurtunity to remove the \n\t and replace it with a \; instead. I'll commit this to both trunk and 4.6 branch after a round of sanity checking.

Re: [Design notes, RFC] Address-lowering prototype design (PR46556)

On Tue, 2011-06-07 at 16:49 +0200, Richard Guenther wrote: > On Tue, Jun 7, 2011 at 4:14 PM, William J. Schmidt > wrote: > >> > Loss of aliasing information > >> > > >> > The most serious problem I've run into is degraded performance due to > >> > poorer > >> > ins

Cgraph alias reorg 8/14 (ipa-cp and ipa-prop update)

Hi, this patch updated ipa-cp and ipa-prop for aliases. It is basically easy - we don't analyze nodes represneting aliases and when propagating we skip them, like everywhere else. There are two problems I noticed. First we should not propagate through calls that are overwritable. When non-ov

Re: [testsuite] Run TLS torture tests with -fpic etc.

Rainer Orth writes: > As suggested by Joseph, the gcc.dg/torture/tls tests should be run with > -fpic/-fPIC and -fpie/-fPIE if supported. > > The patch below implements part of this. Unlike > gcc.dg/torture/stackalign/stackalign.exp, I'm using the code below to > add -fpic etc. to the torture op

Ping^2 Re: Create common hooks structure shared between driver and cc1

Ping^2. This patch , and the followup , are pending review after two weeks. Together these patches create and populate a common structure of target hooks shared between the driver

Re: [testsuite] Run TLS torture tests with -fpic etc.

"Joseph S. Myers" writes: > On Fri, 10 Jun 2011, Rainer Orth wrote: > >> Jakub Jelinek writes: >> >> > On Tue, Jun 07, 2011 at 06:48:08PM +0200, Rainer Orth wrote: >> >> any word on this patch? I think I only need approval for the gcc.c >> > >> > I'm not a maintainer of gcc.c, and I think it

Re: RFA (fold): PATCH for c++/49290 (folding *(T*)(ar+10))

On 06/10/2011 10:20 AM, Richard Guenther wrote: no, a VIEW_CONVERT_EXPR is generally not an lvalue (fold for example would turn the above to (volatile int) a[1]). Really? Seems like it would be a lot more useful if it were an lvalue. I guess I'll just copy the whole function into the front en

Re: RFA (fold): PATCH for c++/49290 (folding *(T*)(ar+10))

On Fri, 10 Jun 2011, Jason Merrill wrote: > On 06/10/2011 10:03 AM, Richard Guenther wrote: > > > > *((volatile int *)&a[0] + 1) > > > > > > It would be correct to fold it to > > > > > > VIEW_CONVERT_EXPR > > > > No, it wouldn't be correct. It isn't correct to fold it to an array-ref > > that

Re: RFA (fold): PATCH for c++/49290 (folding *(T*)(ar+10))

On 06/10/2011 10:03 AM, Richard Guenther wrote: *((volatile int *)&a[0] + 1) It would be correct to fold it to VIEW_CONVERT_EXPR No, it wouldn't be correct. It isn't correct to fold it to an array-ref that isn't volatile. Hmm? The C expression produces a volatile int lvalue referring to

Re: RFA (fold): PATCH for c++/49290 (folding *(T*)(ar+10))

On Fri, 10 Jun 2011, Jason Merrill wrote: > On 06/10/2011 04:35 AM, Richard Guenther wrote: > > I'm out of good suggestions ;) You can do the same-qualifier matching > > and simply have a mismatched array element vs. array-ref type. > > But I need to allow different qualifiers, too. > > > We co

Re: [testsuite] Run TLS torture tests with -fpic etc.

On Fri, 10 Jun 2011, Rainer Orth wrote: > Jakub Jelinek writes: > > > On Tue, Jun 07, 2011 at 06:48:08PM +0200, Rainer Orth wrote: > >> any word on this patch? I think I only need approval for the gcc.c > > > > I'm not a maintainer of gcc.c, and I think it is a bad idea. > > PIE is just a (sli

Re: RFA (fold): PATCH for c++/49290 (folding *(T*)(ar+10))

On 06/10/2011 04:35 AM, Richard Guenther wrote: I'm out of good suggestions ;) You can do the same-qualifier matching and simply have a mismatched array element vs. array-ref type. But I need to allow different qualifiers, too. We could also argue that whoever calls fold_indirect_ref_1 with

Re: FORBIDDEN_INC_DEC_CLASSES in ira-costs.c

On 11-06-10 8:35 AM, Hans-Peter Nilsson wrote: On Thu, 9 Jun 2011, Hans-Peter Nilsson wrote: * ira-costs.c: Remove #ifdefs on dead FORBIDDEN_INC_DEC_CLASSES. Adjust comments. * system.h (FORBIDDEN_INC_DEC_CLASSES): Poison. Index: ira-costs.c =

Re: [PATCH] c-pragma: adding a data field to pragma_handler

Please make sure that with each revision you include *both* the patch *and* the ChangeLog entries so they can be reviewed together. The last version of the ChangeLog entries that I saw still needed more work to follow the normal style for ChangeLog entries. -- Joseph S. Myers jos...@codesourc

Re: [RFA] Cgraph alias reorg 7/14 (no longer try to diable devirtualization on thunks)

On Fri, 10 Jun 2011, Jan Hubicka wrote: > Hi, > code to refuse thunks in gimple_get_virt_method_for_binfo is no-op since I > comitted the thunk rewrite: thunks no longer have same_bodyy_alias flag set. > Only case where it chould diable the devirtualization is when NODE is NULL, > but that won't r

Re: Cgraph alias reorg 2/14 (introduction of alias walkers)

> Hi, > > On Fri, 10 Jun 2011, Jan Hubicka wrote: > > > +static inline struct cgraph_node * > > +cgraph_function_or_thunk_node (struct cgraph_node *node, enum availability > > *availability) > > +{ > > + if (availability) > > +*availability = cgraph_function_body_availability (node); > > +

Re: Cgraph alias reorg 2/14 (introduction of alias walkers)

Hi, On Fri, 10 Jun 2011, Jan Hubicka wrote: > +static inline struct cgraph_node * > +cgraph_function_or_thunk_node (struct cgraph_node *node, enum availability > *availability) > +{ > + if (availability) > +*availability = cgraph_function_body_availability (node); > + return node; > + ret

[RFA] Cgraph alias reorg 7/14 (no longer try to diable devirtualization on thunks)

Hi, code to refuse thunks in gimple_get_virt_method_for_binfo is no-op since I comitted the thunk rewrite: thunks no longer have same_bodyy_alias flag set. Only case where it chould diable the devirtualization is when NODE is NULL, but that won't really happen for thunks anyway. Consequentely this

Re: FORBIDDEN_INC_DEC_CLASSES in ira-costs.c

On Thu, 9 Jun 2011, Hans-Peter Nilsson wrote: > * ira-costs.c: Remove #ifdefs on dead FORBIDDEN_INC_DEC_CLASSES. > Adjust comments. > * system.h (FORBIDDEN_INC_DEC_CLASSES): Poison. > Index: ira-costs.c > === > ---

[PATCH] Fix PR49361

This fixes a fallout of pow(x,2.0) expander removal. As discussed several times we shouldn't really fold x*x to pow(x,2) - at least not when we're in gimple form already as we'll never see followup folding opportunities that folding was designed to catch. Bootstrapped and tested on x86_64-unknow

[PATCH] Fix PR49344

This allegedly fixes PR49344. Bootstrapped and tested on x86_64-unknown-linux-gnu, applied. Richard. 2011-06-10 Richard Guenther PR bootstrap/49344 * tree-ssa-math-opts.c (convert_mult_to_fma): Use FOR_EACH_PHI_OR_STMT_USE. Index: gcc/tree-ssa-math-opts.c ==

[Patch, Fortran] PR 49324: Deep copy array constr of DT with allocatable components

This patch fixes parts of the PR 49324: There was a deep copy missing for assigning an array constructor of a DT with allocatable components. Whether a deep copy is done, depends on a flag. I think the flag has been added to avoid a deep copy and multiple evaluation for functions, which return

Cgraph alias reorg 6/14 (ipa-sra update)

Hi, this patch updates ipa-sra. Again it is quite easy: function now can be local even if they have aliases that are local as well. Consequentely the analysis and transform code needs to be updated to walk the aliases of the function being updated. Bootstrapped/regtested x86_64-linux, will commit

Re: v3 testsuite PATCH to avoid dg-excess-errors

On 06/10/2011 10:36 AM, Jonathan Wakely wrote: I'll look at the broken tests asap, but that will be after work in about 8 hours. I'm ok if you want to go ahead and commit the patch, but if you do let's warn HJ that the regression testers will see the FAILs. Thanks Jon. The good news is that the

Cgraph alias reorg 5/14 (ipa-inline update)

Hi, this patch updates inliner. Inliner also mostly walks from callers to callees and it needs to get into the real function, not alias node. There is quite a lot of walking however, so the patch is quite busy. There are few bits I had to leave out of this patch since the infrastructure is not i

[PATCH] Back to forward-scan for combine in forwprop

Dunno why I thought a backward scan was preferable (it would scan inserted stmts), but it clearly is better to first visit the defs of the uses of stmts we are trying to combine. The following patch uses some trick to make sure stmts inserted by the combining are processed as well. Bootstrapped

Re: [PATCH, Fortran] (Coarray) Change declaration of CAF sync functions.

On 06/10/2011 12:23 PM, Tobias Burnus wrote: + /* SYNC ALL => stat == null_pointer_node + SYNC ALL(stat=s) => stat has an integer type + + If "stat" has the wrong integer type, use a temp variable of + the right type and later cast the result back into "stat". */ Intending: The last two lines w

Re: [patch] Fix PR tree-optimization/49318

On 10 June 2011 12:14, Richard Guenther wrote: > On Fri, Jun 10, 2011 at 9:19 AM, Ira Rosen wrote: >> Hi, >> >> The test in PR 49318 fails because the vectorizer recognizes address >> computation sequence as a widening-multiplication pattern, while such >> sequence is not relevant to vectorizatio

Re: [PATCH] Move error_operand_p one level up

On 06/10/2011 12:14 PM, Richard Guenther wrote: On Fri, Jun 10, 2011 at 11:58 AM, Paolo Carlini wrote: Hi, yesterday I noticed that, as I vaguely suspected for quite some time, error_operand_p can be useful also outside the c++ front-end. The below, which I bootstrapped c, c++, go, on x86_64-

Re: unwinding fallback for mips-irix6 n32

Hello Olivier, > We have been using an unwinding fallback on mips-irix6/n32 for a while > (number of years) across different versions of GCC (3.4, 4.1, 4.3, and > 4.5 lately). > > We haven't submitted this for inclusion so far because of the pretty > limited range of supported configurations (irix

Re: Reintroduce -mflat option on SPARC

> Yes, I have changed a few things, patch attached. The basic stuff works, > but not yet the fancy features (non-local gotos, setjmp/longjmp and > exceptions). I'm also working on the GDB side. I've now installed the attached version. It adds support for the above fancy features plus sibling ca

Re: [PATCH, Fortran] (Coarray) Change declaration of CAF sync functions.

On 06/10/2011 11:12 AM, Daniel Carrera wrote: This is the patch recently discussed in the GFortran list, now ready for official submission. [...] 2011-06-07 Daniel Carrera * gfortran.dg/coarray/sync_1.f90: New Test "SYNC ALL", "SYNC MEMORY" and "SYNC IMAGES". "Test" -> "test for".

Re: [PATCH] Move error_operand_p one level up

On Fri, Jun 10, 2011 at 11:58 AM, Paolo Carlini wrote: > Hi, > > yesterday I noticed that, as I vaguely suspected for quite some time, > error_operand_p can be useful also outside the c++ front-end. The below, > which I bootstrapped c, c++, go, on x86_64-linux (and I'm finishing testing) > tries t

Re: [Patch, AVR]: Fix PR46779

Denis Chertykov schrieb: > 2011/6/9 Georg-Johann Lay : >> Denis Chertykov schrieb: >>> 2011/6/9 Georg-Johann Lay : >> I agree with you. However, I think that the risk of spill failure >> should be minimized. I have no idea how to fix a splill failure like >> the following that occurs in testsuite (

Cgraph alias reorg 4/14 (ipa-pure-const and ipa-reference update)

Hi, this patch updates ipa-pure-const and ipa-reference to be ready for aliases. There is nothing difficult - the passes propagate from caller to callee and thus all they need is to skip the aliases and be aware of the fact that body visibility may change from the edge to edge. Bootstrapped/regtes

Re: [testsuite] Run TLS torture tests with -fpic etc.

Jakub Jelinek writes: > On Tue, Jun 07, 2011 at 06:48:08PM +0200, Rainer Orth wrote: >> any word on this patch? I think I only need approval for the gcc.c > > I'm not a maintainer of gcc.c, and I think it is a bad idea. > PIE is just a (slightly) more secure form of an executable, therefore > i

Re: [testsuite] skip ARM tests with conflicting options

On 10/06/11 01:04, Janis Johnson wrote: > On 06/08/2011 03:39 AM, Richard Earnshaw wrote: >> On 08/06/11 03:14, Janis Johnson wrote: >>> On 06/07/2011 06:25 PM, Mike Stump wrote: On Jun 7, 2011, at 4:24 PM, Janis Johnson wrote: > On 06/07/2011 02:07 PM, Joseph S. Myers wrote: >> On Tue

[PATCH] Move error_operand_p one level up

Hi, yesterday I noticed that, as I vaguely suspected for quite some time, error_operand_p can be useful also outside the c++ front-end. The below, which I bootstrapped c, c++, go, on x86_64-linux (and I'm finishing testing) tries to do that, moving the macro one level up and using it in all t

[PATCH] [Patch ARM] Fix PR target/49335 .

Hi, This fixes an issue with PR target/49335 where we have a situation where the sp register is being used as a source operand in the rsb instruction in the shifted immediate form in Thumb2 i.e. rsb ip, sp, ip lsl #2 The only valid instructions in Thumb2 that can do this are the add and the

[PATCH, Fortran] (Coarray) Change declaration of CAF sync functions.

This is the patch recently discussed in the GFortran list, now ready for official submission. As my first submission as a GSoC student, this is a simple patch mainly intended to familiarize me with GFortran. I changed the signature of the functions _gfortran_caf_sync_all and _gfortran_caf_sync

Re: [Patch, AVR]: Fix PR46779

Denis Chertykov schrieb: > 2011/6/9 Georg-Johann Lay : >> This is a tentative patch to fix PR46779 and hopefully also related >> issues like PR45291. >> > - /* Disallow QImode in stack pointer regs. */ > - if ((regno == REG_SP || regno == (REG_SP + 1)) && mode == QImode) > + /* Don't allocate d

Re: [patch] Fix PR tree-optimization/49318

On Fri, Jun 10, 2011 at 9:19 AM, Ira Rosen wrote: > Hi, > > The test in PR 49318 fails because the vectorizer recognizes address > computation sequence as a widening-multiplication pattern, while such > sequence is not relevant to vectorization. The problem is that the > vectorizer doesn't check i

Re: [PATCH] c-pragma: adding a data field to pragma_handler

thanks! I formatted as you requested. I cannot commit myself as I haven't a "write after approval" status, maye you can do it, or I can wait my GSOC mentor, Basile Starynkevitch to do this (He mights be busy for a few days). Pierre Vittet Index: gcc/c-family/c-pragma.c

Re: [google] pessimize stack accounting during inlining

On Fri, Jun 10, 2011 at 1:45 AM, Mark Heffernan wrote: > On Wed, Jun 8, 2011 at 1:54 AM, Richard Guenther > wrote: >> Well, it's still not a hard limit as we can't tell how many spill slots >> or extra call argument or return value slots we need. > > Agreed.  It's not perfect.  But I've found thi

Re: -fdump-passes -fenable-xxx=func_name_list

On Fri, Jun 10, 2011 at 12:51 AM, Xinliang David Li wrote: > Patch is temporally rolled back. > > Richard, looks like deeper pass manager cleanup is needed -- I would > like to delay that. For now, this leaves us two choices -- 1) do cfunc > push/pop, or 2) do pass dump while executing. None of th

Re: [PATCH] Fix PR lto/49302 (corrected)

On Thu, Jun 9, 2011 at 7:34 PM, William J. Schmidt wrote: > Richard, thanks for the comments.  Here's the revised patch for > expanding cabs in gimple. > > Given the direction this pass is taking, should it eventually be renamed > from cse_sincos to something more general? Eventually yes - but na

Re: Dump before flag

On Thu, Jun 9, 2011 at 5:47 PM, Xinliang David Li wrote: > See attached. Hmm. I don't like how you still wire dumping in the TODO routines. Doesn't it work to just dump the body from pass_fini_dump_file ()? Or if that doesn't sound clean from (a subset of) places where it is called? (we might wa

Re: v3 testsuite PATCH to avoid dg-excess-errors

On 10 June 2011 04:45, Jason Merrill wrote: > > Doing this revealed that the bind/ref_neg.cc test was broken; the dg-error > tags were missing a space before the closing }, so they were being ignored. >  When I fix that, I find that two of the tests are failing. Doh. > I'm committing the first pa

Re: RFA (fold): PATCH for c++/49290 (folding *(T*)(ar+10))

On Thu, 9 Jun 2011, Jason Merrill wrote: > On 06/07/2011 10:24 AM, Jason Merrill wrote: > > On 06/07/2011 10:05 AM, Richard Guenther wrote: > > > In that case you could do what Jakub suggested - but only for rvalues > > > of course. > > > > Right, and I need to handle lvalues as well. > > > > >

[patch] Fix PR tree-optimization/49318

Hi, The test in PR 49318 fails because the vectorizer recognizes address computation sequence as a widening-multiplication pattern, while such sequence is not relevant to vectorization. The problem is that the vectorizer doesn't check if a statement is going to be vectorized before replacing it wi