Re: [PATCH PING] c++-specific bits of tree-slimming patches

On Apr 22, 2011, Jason Merrill wrote: > On 04/22/2011 02:13 AM, Mike Stump wrote: >> http://gcc.gnu.org/ml/gcc/2005-04/msg00161.html >> >> has the details of why the code was put in. > Right. At the time, we were sorting the goto queue based on pointer > values, which caused the problem. We n

Re: Disable tracer by default for profile use (issue4428074)

2011/4/29 Sharad Singhai (शरद सिंघई) : > On Thu, Apr 28, 2011 at 4:38 PM, Richard Guenther > wrote: >> >> On Fri, Apr 29, 2011 at 1:34 AM, Xinliang David Li >> wrote: >> > Sharad can provide some some performance data -- we have seen up to 2% >> > degradation to with tracer turned on for one of g

Re: [pph] Save/restore PARM_DECL DECL_ARG_TYPE (issue4441079)

On Fri, Apr 29, 2011 at 3:10 AM, Lawrence Crowl wrote: > This patch saves and restores the PARM_DECL DECL_ARG_TYPE in the PPH file. Should be already streamed via lto_output_ts_decl_common_tree_pointers as it is aliased to DECL_INITIAL. Richard. > Index: gcc/cp/ChangeLog.pph > > 2011-04-28  Law

Re: [libffi] Provide unwind info for Tru64 UNIX in osf.S

Hello! > On 04/28/2011 10:55 AM, Rainer Orth wrote: > > I cannot test the Alpha/Linux side, though. > Neither can I, at the moment. The bootstrap fails due to link error [1] on alphaev68-pc-linux-gnu for some time ... [1] http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47230 Uros.

Re: [PR 48093] document -mtls-dialect opt (and GCC_COMPARE_DEBUG env)

Alexandre, > Index: gcc/doc/invoke.texi > === > --- gcc/doc/invoke.texi.orig 2011-04-27 15:59:46.0 -0300 > +++ gcc/doc/invoke.texi 2011-04-29 01:31:58.603523692 -0300 > @@ -12581,6 +12581,13 @@ SYSV ABI. You can contro

Re: [libffi] Provide unwind info for Tru64 UNIX in osf.S

Richard Henderson writes: >> 2011-04-28 Rainer Orth >> >> * src/alpha/osf.S (UA_SI, FDE_ENCODING, FDE_ENCODE, FDE_ARANGE): >> Define. >> Use them to handle ELF vs. ECOFF differences. >> [__osf__] (_GLOBAL__F_ffi_call_osf): Define. > > Looks good. If there are any hidden L

Re: [google] Patch to support calling multi-versioned functions via new GCC builtin. (issue4440078)

On Fri, Apr 29, 2011 at 4:52 AM, Sriraman Tallam wrote: > I want this patch to be considered for google/main now. This is intended to > be submitted to trunk for review soon. > This patch has been tested with crosstool bootstrap using buildit and by > running all tests. > > > Patch Description :

Re: [google]Add support for sampled profile collection (issue4438083)

On Fri, Apr 29, 2011 at 1:42 AM, Easwaran Raman wrote: > This patch from Silvius Rus  adds support for sampled edge profile collection > to reduce instrumentation run overhead. Bootstraps and no test regressions. > Ok for google/main? > > 2011-04-28  Silvius Rus   > >        * doc/invoke.texi: D

Re: libgo patch committed: Inherit environment in http/cgi

Ian, > This libgo patch brings over a patch to the master Go library to inherit > environment variables in http/cgi. This should fix PR go/48503. not really :-) It needs the following supplement to handle Solaris and IRIX. I'm not only including LD_LIBRARY_PATH (although this would suffice for

[PATCH][6/n] Alias housekeeping

This makes PTA handle OBJ_TYPE_REF (not sure why I thought we need to give up). Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk. Richard. 2011-04-29 Richard Guenther * tree-ssa-structalias.c (get_fi_for_callee): Restructure. Handle OBJ_TYPE_REF.

[PATCH][1/n] Fix build_int_cst callers with NULL type

First batch, obvious in tail of grep. Bootstrapped and tested on x86_64-unknown-linux-gnu, applied. Richard. 2011-04-29 Richard Guenther * tree-nested.c (get_trampoline_type): Use size_int. (get_nl_goto_field): Likewise. * tree-eh.c (lower_try_finally_switch): Use in

Re: better wpa [1/n]: merge types during read-in

Hi, On Thu, 21 Apr 2011, Richard Guenther wrote: > >> > It would have been nice to have the top-level tree merging as a > >> > separate patch, as I am not convinced it is correct, but see below ... > >> > >> I'll split it out. > > > > Like so (also including the other remarks). > > > > Regstrappi

Re: [patch, 4.5] Fix PR middle-end/43085 (make profiledbootstrap crashes due to dataflow bug)

Eric Botcazou wrote: > > I guess that patch does indeed address both correctness and performance > > aspects, but I wasn't sure attempting to pull those apart was a safer > > option, given that at least in its current form it has seen testing in > > mainline ... > > Reasonable enough indeed. > >

Re: [pph] Save/restore PARM_DECL DECL_ARG_TYPE (issue4441079)

On Fri, Apr 29, 2011 at 04:29, Richard Guenther wrote: > On Fri, Apr 29, 2011 at 3:10 AM, Lawrence Crowl wrote: >> This patch saves and restores the PARM_DECL DECL_ARG_TYPE in the PPH file. > > Should be already streamed via lto_output_ts_decl_common_tree_pointers > as it is aliased to DECL_INITI

[PATCH][2/n] Fix build_int_cst callers with NULL type

Some more obvious cases in the middle-end. Next is expand_shift and callers. Bootstrapped and tested on x86_64-unknown-linux-gnu, applied. Richard. 2011-04-29 Richard Guenther * tree-inline.c (remap_eh_region_tree_nr): Use integer_type_node for the remapped region number.

[PATCH][3/n] Fix built_int_cst callers with NULL type - expand_shift reorg

Noting that there is a single caller that calls expand_shift with a possible non-constant shift value this patch renames expand_shift to expand_variable_shift and changes expand_shift to take a constant integer shift amout, updating all callers to drop various variants of building an integer cons

[PATCH] convert nonlocal_goto_handler_labels to a VEC

As $SUBJECT suggests. The memory savings from this conversion is negligible; the real benefit, IMHO, is use of a proper container instead of EXPR_LIST. remove_node_from_expr_list is unused after this patch; I will delete it as an obvious followon patch if this patch is approved. Tested on x86_64

[PATCH] convert forced_labels to a VEC

As $SUBJECT suggests. Just like the nonlocal_goto_handler_labels, the real benefit is a proper container instead of an EXPR_LIST. in_expr_list_p is unused after this patch; I will delete it as obvious in a followon patch if this patch is approved. Tested on x86_64-unknown-linux-gnu. OK to commi

[PATCH][4/n] Fix build_int_cst callers with NULL type

Another round. Missing is now tree-data-ref.c which better should use double_ints. Frontends also need a look, so do backends. Bootstrap & regtest running on x86_64-unknown-linux-gnu, will apply after that succeeded. Richard. 2011-04-29 Richard Guenther * builtins.c (fold_builtin_

[google] Do not emit NULL warnings for implicit template args (issue4436067)

This patch from Le-Chun Wu distinguishes explicit from implicit template arguments to avoid emitting NULL warnings for implicit arguments. Additionally, Le-Chun extended the pointer to boolean conversion warnings to also warn when assigning a boolean value to a pointer. Le-Chun, if I'm missing a

[google] Add two new -Wshadow warnings (issue4452058)

This patch from Le-Chun Wu adds two two new shadow warning flags for C and C++: -Wshadow-local which warns if a local variable shadows another local variable or parameter, -Wshadow-compatible-local which warns if a local variable shadows another local variable or parameter whose t

[google] Add new warning -Wreal-conversion (issue4436068)

This patch from Le-Chun Wu adds a new warning flag "-Wreal-conversion" that warns about implicit type conversions from real (double or float) values to integral values. OK for trunk? Applied to google/main. 2011-04-27 Le-Chun Wu Google ref 39133 * c.opt (Wreal-conversion): N

[PATCH,c++] delete TREE_NEGATED_INT

As $SUBJECT suggests. It is write-only; I'm not sure what it was ever used for. Tested on x86_64-unknonw-linux-gnu. OK to commit? -Nathan gcc/cp/ * cp-tree.h (TREE_NEGATED_INT): Delete. * semantics.c (finish_unary_op_expr): Don't try to set it. diff --git a/gcc/cp/cp-tree.h b/

Re: [google] Add new warning -Wreal-conversion (issue4436068)

On Fri, Apr 29, 2011 at 10:59:31AM -0400, Diego Novillo wrote: > * g++.dg/warn/Wreal-conversion-1.C: New. > * gcc.dg/Wreal-conversion-1.c: New. Could a single copy of the test be placed in c-c++-common, instead? -Nathan

[google] Enable -Wnonnull for C++ (issue4431076)

This patch from Le-Chun Wu enables -Wnonnull in C++ as well as C. OK for trunk? Applied to google/main. 2011-04-27 Le-Chun Wu * c.opt (Wnonnull): Enable for C++. ChangeLog.google-main 2011-04-27 Le-Chun Wu * doc/extend.texi (Wnonnull): Add documentation for C++.

Re: [google] Enable -Wnonnull for C++ (issue4431076)

On Fri, Apr 29, 2011 at 11:04, Diego Novillo wrote: > This patch from Le-Chun Wu enables -Wnonnull in C++ as well as C. > > OK for trunk?  Applied to google/main. > > 2011-04-27  Le-Chun Wu   > >        * c.opt (Wnonnull): Enable for C++. Google ref 42718

[google] Check if the nonnull attribute is applied to 'this' (issue4446070)

This patch from Le-Chun Wu adds support to check whether a nonnull attribute is applied to 'this' pointer for non-static methods. OK for trunk? Applied to google/main 2011-04-27 Le-Chun Wu Google ref 45339. * c-common.c (handle_nonnull_attribute): Check whether the nonnull

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On Fri, Apr 29, 2011 at 5:08 PM, Diego Novillo wrote: > This patch from Le-Chun Wu adds support to check whether a nonnull > attribute is applied to 'this' pointer for non-static methods. > > OK for trunk?  Applied to google/main > > 2011-04-27  Le-Chun Wu   > >        Google ref 45339. > >      

[Patch, libfortran] Some path handling fixes

Hello all, since about a decade or so ago, POSIX specifies that PATH_MAX includes the trailing null byte (previously it was undefined). However, libgfortran has incorrectly assumed the opposite, and has thus created temporary buffers of length PATH_MAX + 1 for holding paths when converting from Fo

Re: [C++ Patch] PR 48606

OK. Jason

Re: [Patch, Fortran] Fix regressions PRs 48810 and 48800: wrong access flag and missing deferred-shape diagnostics

On 04/28/2011 04:18 PM, Tobias Burnus wrote: The attached patch fixes two regressions: a) PR 48810: For type-bound procedures, the access flags should be checked only for the generic function, not for the specific function the generic resolves to. (4.6/4.7 rejects-valid regression.) b) PR 48800

Re: [PING][PATCH] Change rclass argument type in memory_move_cost function from enum reg_class to reg_class_t.

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/27/11 12:17, Anatoly Sokolov wrote: > Hello. > > Patch ping. > >> This patch change memory_move_cost function to stop using back end >> specific >> type 'enum reg_class' in favor to reg_class_t. Also this allow do small >> cleanup in ia64_

Re: [google] Do not emit NULL warnings for implicit template args (issue4436067)

OK. Jason

Re: [Patch, libfortran] Some path handling fixes

On Fri, Apr 29, 2011 at 06:20:17PM +0300, Janne Blomqvist wrote: > > since about a decade or so ago, POSIX specifies that PATH_MAX includes > the trailing null byte (previously it was undefined). However, > libgfortran has incorrectly assumed the opposite, and has thus created > temporary buffers

Re: [Patch, libfortran] Some path handling fixes

On 04/29/2011 08:20 AM, Janne Blomqvist wrote: Hello all, since about a decade or so ago, POSIX specifies that PATH_MAX includes the trailing null byte (previously it was undefined). However, libgfortran has incorrectly assumed the opposite, and has thus created temporary buffers of length PATH_

Re: [PATCH] Canonicalize compares in combine [2/3] Modifications to try_combine()

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/26/11 05:44, Chung-Lin Tang wrote: > > Hi Jeff, thanks for reviewing this quite convoluted patch :) FWIW, I don't think it's necessarily your patch that is convoluted, but instead the original code. It's also the case that I haven't spent nearl

[Patch, libfortran] PR 48488 Fix comments

Hi all, now that we fixed real output, I update the comments per the patch below and committed. Index: io/write.c === --- io/write.c (revision 173168) +++ io/write.c (working copy) @@ -1456,10 +1456,18 @@ set_fnode_default (st_para

Re: [google]Add support for sampled profile collection (issue4438083)

> How is code-size affected with this patch, non-instrumented vs. > regular-instrumented vs. sample-instrumented? I don't have the numbers, but the increase in code size from regular-instrumented to sample-instrumented is larger than that from non-instrumented to sample-instrumented. Easwaran, co

Re: [google] Patch to support calling multi-versioned functions via new GCC builtin. (issue4440078)

Here is the background for this feature: 1) People relies on function multi-version to explore hw features and squeeze performance, but there is no standard ways of doing so, either a) using indirect function calls with function pointers set at program initialization; b) using manual dispatch at e

Re: Toplevel cleanup: disable Java when libffi not supported

> "Joseph" == Joseph S Myers writes: Joseph> This patch, relative to a tree with Joseph> applied, Joseph> continues the cleanup of toplevel cases relating to disabling Java or Joseph> Java libraries by arranging for Java to be disabled

Re: Toplevel cleanup: disable Java when libffi not supported

On 04/29/2011 06:26 PM, Tom Tromey wrote: "Joseph" == Joseph S Myers writes: Joseph> This patch, relative to a tree with Joseph> applied, Joseph> continues the cleanup of toplevel cases relating to disabling Java or Joseph> Java li

[PATCH] Fix switch conversion (PR tree-optimization/48809)

Hi! The following patch fixes a bug in tree-switch-conversion.c with signed index_expr's. build_arrays would compute index_expr - range_min in index_expr's type and use that as index into CSWTCH.N array, which is wrong, because in this case index_expr 98 - (-62) computed in signed char type resul

[Patch, fortran] PR48462 - [4.6/4.7 Regression] realloc on assignment: matmul Segmentation Fault with Allocatable Array + PR48746

Dear All, These are both quite trivial fixes and can be understood from ChangeLogs and comments in the patch. Bootstrapped and regtested on FC9/x86_64 - OK for trunk and 4.6? Cheers Paul 2011-04-29 Paul Thomas PR fortran/48462 * trans-expr.c (arrayfunc_assign_needs_temporar

Re: [Patch, fortran] PR48462 - [4.6/4.7 Regression] realloc on assignment: matmul Segmentation Fault with Allocatable Array + PR48746

On Fri, Apr 29, 2011 at 06:55:11PM +0200, Paul Richard Thomas wrote: > Dear All, > > These are both quite trivial fixes and can be understood from > ChangeLogs and comments in the patch. > > Bootstrapped and regtested on FC9/x86_64 - OK for trunk and 4.6? OK for both. -- Steve

[google] Add missing reference numbers to ChangeLogs

I forgot to update the ChangeLog entries with the reference numbers in my previous patches. Committed to google/main. Diego. diff --git a/gcc/ChangeLog.google-main b/gcc/ChangeLog.google-main index 771435b..dfebff2 100644 --- a/gcc/ChangeLog.google-main +++ b/gcc/ChangeLog.google-main @@ -19,6

Re: [google]Add support for sampled profile collection (issue4438083)

On Fri, Apr 29, 2011 at 9:12 AM, Silvius Rus wrote: >> How is code-size affected with this patch, non-instrumented vs. >> regular-instrumented vs. sample-instrumented? > > I don't have the numbers, but the increase in code size from > regular-instrumented to sample-instrumented is larger than that

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On 2011/04/29 15:12:52, richard.guenther_gmail.com wrote: > > + spurious white-space change. Thanks. Fixed. Diego. http://codereview.appspot.com/4446070/

Re: [google] Patch to support calling multi-versioned functions via new GCC builtin. (issue4440078)

Hi Richard, Thanks for the comments. Please find inline responses. On Fri, Apr 29, 2011 at 1:56 AM, Richard Guenther wrote: > > On Fri, Apr 29, 2011 at 4:52 AM, Sriraman Tallam wrote: > > I want this patch to be considered for google/main now. This is intended to > > be submitted to trunk fo

Re: libgo patch committed: Inherit environment in http/cgi

Rainer Orth writes: >> This libgo patch brings over a patch to the master Go library to inherit >> environment variables in http/cgi. This should fix PR go/48503. > > not really :-) It needs the following supplement to handle Solaris and > IRIX. I'm not only including LD_LIBRARY_PATH (although

Re: [google] Add new warning -Wreal-conversion (issue4436068)

On Fri, Apr 29, 2011 at 11:02, Nathan Froyd wrote: > On Fri, Apr 29, 2011 at 10:59:31AM -0400, Diego Novillo wrote: >>       * g++.dg/warn/Wreal-conversion-1.C: New. >>       * gcc.dg/Wreal-conversion-1.c: New. > > Could a single copy of the test be placed in c-c++-common, instead? Good point. W

Re: [google]Add support for sampled profile collection (issue4438083)

On Apr 29, 2011, at 9:12 AM, Silvius Rus wrote: > When you build with > -fprofile-generate=./fdoprof, the output gets dumped under > ./fdoprof/..., but there seems to be no easy way to know this path > within the profiling process. For Google servers this makes > collection fragile. The user gen

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On Fri, 29 Apr 2011 11:08:24 -0400 (EDT) dnovi...@google.com (Diego Novillo) wrote: > This patch from Le-Chun Wu adds support to check whether a nonnull > attribute is applied to 'this' pointer for non-static methods. This bring me to a question. Does the C++ standard imply that this is never a n

Re: [PATCH] Fix switch conversion (PR tree-optimization/48809)

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 On 04/29/11 10:43, Jakub Jelinek wrote: > Hi! > > The following patch fixes a bug in tree-switch-conversion.c with > signed index_expr's. build_arrays would compute index_expr - range_min > in index_expr's type and use that as index into CSWTCH.N arr

Re: branch optimizations

Someone pointed out a patch by Zdenek Dvorak which implements TSP-based basic block reordering: http://gcc.gnu.org/ml/gcc-patches/2004-03/msg02335.html to me. From the spec numbers presented, it looks enticing. Zdenek, do you think this is still a reasonable direction to go? If so, I'd li

Re: [google] Use different peeling parameters with available profile (issue4438079)

On 2011/04/29 00:02:40, singhai wrote: 2011-04-28 Sharad Singhai gcc/ChangeLog.google-main * params.def: Add new parameters to control peeling. * tree-ssa-loop-ivcanon.c (try_unroll_loop_completely): Use different peeling parameter

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On Fri, Apr 29, 2011 at 14:46, Basile Starynkevitch wrote: > On Fri, 29 Apr 2011 11:08:24 -0400 (EDT) > dnovi...@google.com (Diego Novillo) wrote: > >> This patch from Le-Chun Wu adds support to check whether a nonnull >> attribute is applied to 'this' pointer for non-static methods. > > This brin

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On Apr 29, 2011, at 11:46 AM, Basile Starynkevitch wrote: > On Fri, 29 Apr 2011 11:08:24 -0400 (EDT) > dnovi...@google.com (Diego Novillo) wrote: > >> This patch from Le-Chun Wu adds support to check whether a nonnull >> attribute is applied to 'this' pointer for non-static methods. > > This brin

Fix 64-bit libjava breakage on Solaris 2/SPARC (PR middle-end/48819)

This patch fixes 350 64-bit libjava execution failures on Solaris 2/SPARC. It was provided and approved by Richard in the PR. Tested by rebuilding libjava with the patched jc1 on sparc-sun-solaris2.11. All testsuite failures are now gone. Installed on mainline. Rainer 2011-04-29 Ric

Re: (build) Patch to fix cp/cfns.gperf building issues

Alexandre thanks for the review :-) > -pedantic review: how about outputting to a temporary file (say > cp/cfns.hT) and only renaming to the intended name on success, so that, > if gperf crashes or we reboot part-way through it, we don't end up with > a partially-generated file that will seem to

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On Fri, 29 Apr 2011 12:32:24 -0700 Mike Stump wrote: > On Apr 29, 2011, at 11:46 AM, Basile Starynkevitch wrote: > > > > This bring me to a question. Does the C++ standard imply that this is > > never a null pointer? > > Does this: > > 4 Certain other operations are described in this Internati

Re: [google] Add new warning -Wreal-conversion (issue4436068)

On 04/29/2011 11:02 AM, Nathan Froyd wrote: On Fri, Apr 29, 2011 at 10:59:31AM -0400, Diego Novillo wrote: * g++.dg/warn/Wreal-conversion-1.C: New. * gcc.dg/Wreal-conversion-1.c: New. Could a single copy of the test be placed in c-c++-common, instead? -Nathan Committed to go

Re: [Patch, fortran] PR48462 - [4.6/4.7 Regression] realloc on assignment: matmul Segmentation Fault with Allocatable Array + PR48746

Dear Paul, first, thanks for the patch. There is one thing it does not appear to do correctly: It should also set the dtype on the variable itself: ig25@linux-fd1f:~/Krempel/H> cat mm.f90 program main implicit none integer, parameter :: m=10, n=12, count=4 double precision :: a(m, count),

[google][patch] Add new -gmlt option for min. debug info with line tables (issue4440072)

I'd like to put this patch into google/main while it's being reviewed for trunk. Original message: http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02075.html Rietveld: http://codereview.appspot.com/4440072 Bootstraps on x86_64, with testsuite output identical to a vanilla build. OK for google/main

Re: [patch] Add new -gmlt option for min. debug info with line tables (issue4440072)

On Wed, Apr 27, 2011 at 16:24, Cary Coutant wrote: >>> set_debug_level should not use global state; this needs to check >>> opts->x_debug_info_level (not the global debug_info_level) and set >>> opts->x_generate_debug_line_table. >> >> Oops, missed that. Thanks! > > I've uploaded the revised patch

Re: [google][patch] Add new -gmlt option for min. debug info with line tables (issue4440072)

On Fri, Apr 29, 2011 at 16:36, Cary Coutant wrote: > I'd like to put this patch into google/main while it's being reviewed for > trunk. > > Original message: http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02075.html > > Rietveld: http://codereview.appspot.com/4440072 > > Bootstraps on x86_64, with

Re: [google] Use R_ARM_GOT_PREL to simplify global address loading from GOT (issue4433079)

On Fri, Apr 29, 2011 at 01:24, Carrot Wei wrote: >>> http://codereview.appspot.com/4433079/diff/1/gcc/hooks.c#newcode287 >>> gcc/hooks.c:287: return NULL; >>> +hook_rtx_void_null (void) >>> +{ >>> +  return NULL; >>> >>> s/NULL/NULL_RTX/ >>> >> done. >> > Oops. File hooks.c doesn't include rtl.h,

Re: [google]Add support for sampled profile collection (issue4438083)

Hi, > + Honza > > This patch may be a candidate for trunk as well. This feature not only > allows profile collection with much less overhead (for multi-thread > programs with hot regions, the slow down can be significant due to > cache ping-pong effect of counter update) without sacrificing too mu

Re: [PATCH] Typed DWARF stack

On 04/16/2011 04:11 AM, Jakub Jelinek wrote: + case dw_val_class_const_double: + { + unsigned HOST_WIDE_INT first, second; + l = 2 * HOST_BITS_PER_WIDE_INT / HOST_BITS_PER_CHAR; + + dw2_asm_output_data (1, l, NULL); + if (WORDS_BIG

Re: Disable tracer by default for profile use (issue4428074)

On 2011/04/28 18:53:24, Diego Novillo wrote: OK. Committed rev 173186. Diego. http://codereview.appspot.com/4428074/

Re: [google] Use different peeling parameters with available profile (issue4438079)

On 2011/04/29 19:21:00, Diego Novillo wrote: On 2011/04/29 00:02:40, singhai wrote: > 2011-04-28 Sharad Singhai > >gcc/ChangeLog.google-main >* params.def: Add new parameters to control peeling. >* tree-ssa-loop-ivcanon.c (try_unroll_loop_completely):

Re: [Patch, fortran] PR48462 - [4.6/4.7 Regression] realloc on assignment: matmul Segmentation Fault with Allocatable Array + PR48746

Hello Paul, there's another point: The sizes are also not set correctly. ig25@linux-fd1f:~/Krempel/H> cat mm.f90 program main implicit none integer, parameter :: m=10, n=12, count=4 double precision :: a(m, count), b(count, n), c(m, n) double precision, dimension(:,:), allocatable :: tmp

[pph] Relativize pph test commands. (issue4445076)

Copy the pph map file from the source area to the run area to avoid absolute paths to the map file. These were causing tests to not match up in the test comparison scripts. Index: gcc/testsuite/ChangeLog.pph 2011-04-29 Lawrence Crowl * g++.dg/pph/pph.exp: Copy pph map file to avoid a

Re: [google] Use different peeling parameters with available profile (issue4438079)

Thanks a bunch! Sharad On Fri, Apr 29, 2011 at 1:58 PM, wrote: > > On 2011/04/29 19:21:00, Diego Novillo wrote: >> >> On 2011/04/29 00:02:40, singhai wrote: > >> > 2011-04-28  Sharad Singhai   >> > >> >       gcc/ChangeLog.google-main >> >       * params.def: Add new p

Re: [pph] Relativize pph test commands. (issue4445076)

On Fri, Apr 29, 2011 at 17:00, Lawrence Crowl wrote: > Copy the pph map file from the source area to the run area to avoid > absolute paths to the map file.  These were causing tests to not match > up in the test comparison scripts. > > Index: gcc/testsuite/ChangeLog.pph > > 2011-04-29  Lawrence C

[pph] Added missing ChangeLog file in gcc/fortran

When I modified the fortran #include callback to return bool, I forgot to add the ChangeLog file for it. Fixed with rev. 173188. Diego.

Propagate BB predicates in ipa-inline-analysis

Hi, this more or less complettes the infrastructure for predicates by adding logic propagating predicates across CFG. I also added switch statement handling and __builtin_constant_p construct, so we "understand" functions using those as one in the new testcase. It also turned out that the predica

[pph] DECL_INITIAL instead of DECL_ARG_TYPE (issue4442102)

In the pph file, save and restore DECL_INITIAL instead of just PARM_DECL DECL_ARG_TYPE. Index: gcc/cp/ChangeLog.pph 2011-04-29 Lawrence Crowl * pph-streamer-out.c (pph_stream_write_tree): Write DECL_INITIAL instead of PARM_DECL DECL_ARG_TYPE. * pph-streamer-in.c (pph_st

Re: [pph] Save/restore PARM_DECL DECL_ARG_TYPE (issue4441079)

On 4/29/11, Diego Novillo wrote: > On Apr 29, 2011 Richard Guenther wrote: > > On Apr 29, 2011 Lawrence Crowl wrote: > > > This patch saves and restores the PARM_DECL DECL_ARG_TYPE in the > > > PPH file. > > > > Should be already streamed via lto_output_ts_decl_common_tree_pointers > > as it is

Re: [pph] DECL_INITIAL instead of DECL_ARG_TYPE (issue4442102)

On Fri, Apr 29, 2011 at 17:28, Lawrence Crowl wrote: > In the pph file, save and restore DECL_INITIAL instead of just > PARM_DECL DECL_ARG_TYPE. > > Index: gcc/cp/ChangeLog.pph > > 2011-04-29  Lawrence Crowl > >        * pph-streamer-out.c (pph_stream_write_tree): Write DECL_INITIAL >        inst

[Patch, Fortran] PR 48800 - fix "IMPORT :: symbol"

Nearly obvious patch. Build and regtested on x86-64-linux. OK for the trunk? Tobias 2011-04-30 Tobias Burnus PR fortran/48800 * decl.c (gfc_match_import): Don't try to find the symbol if already found. 2011-04-30 Tobias Burnus PR fortran/48800 * gfortran.dg/interface_36.f90: New.

[Fortran] RFC patch for gfc_trans_deferred_vars (PR 48786)

Dear all, gfc_trans_deferred_vars is a bit of a mess; there is first a block which handles function results of the type proc_sym->result == proc_sym. Afterwards, deferred variables - local, dummys, and proc_sym->result (!= proc_sym) are handled. The problem is that for allocatable results (e

Re: [google] Check if the nonnull attribute is applied to 'this' (issue4446070)

On Apr 29, 2011, at 1:08 PM, Basile Starynkevitch wrote: > Not really. How's this then: http://stackoverflow.com/questions/2474018/when-does-invoking-a-member-function-on-a-null-instance-result-in-undefined-behav ? :-)

Re: [pph] Enable nested namespaces (issue4431071)

Looks OK. Some comments below. http://codereview.appspot.com/4431071/diff/1/gcc/c-family/c.opt File gcc/c-family/c.opt (right): http://codereview.appspot.com/4431071/diff/1/gcc/c-family/c.opt#newcode943 gcc/c-family/c.opt:943: +fpph-dump-tree +C++ Var(flag_pph_dump_tree) +-fpph-dump-tree

Re: Toplevel cleanup: disable Java when libffi not supported

On Fri, 29 Apr 2011, Tom Tromey wrote: > > "Joseph" == Joseph S Myers writes: > > Joseph> This patch, relative to a tree with > Joseph> applied, > Joseph> continues the cleanup of toplevel cases relating to disabling Java or > Joseph>

[google]: Add new test case on integer literal address

This test case was extracted from kernel source exposed in some older version of gcc. Works fine in trunk, but need to add a test for it. Ok for trunk? Thanks, David Index: testsuite/gcc.dg/tree-ssa/integer-addr.c === --- testsuite/

Re: [google]: Add new test case on integer literal address

The old bug was that the second call to func got CSEed. David On Fri, Apr 29, 2011 at 4:08 PM, Xinliang David Li wrote: > This test case was extracted from kernel source exposed in some older > version of gcc. Works fine in trunk, but need to add a test for it. > > Ok for trunk? > > Thanks, > >

Re: [google] LIPO regression tests and bug fixes (issue4444076)

Hi, It seems that majority of testcases are independent of lipo. We could probably enjoy more of testing on mainline, so could you please take those working on mainline and make mainline patch and let me know what of the tests are not working there? > Index: testsuite/gcc.dg/tree-prof/lipo/inline

[PATCH] Cleanup of cgraph topological ordering functions

Hi, today while discussing the two functions that we have in gcc to topologically sort the call graph, Honza asked me to put them both into the ipa-utils file and rename them. So this patch moves cgraph_postorder there and renames it to ipa_reverse_postorder, and similarly, it renames ipa_utils_r

Re: Disable tracer by default for profile use (issue4428074)

> 2011/4/29 Sharad Singhai (? ???) : > > On Thu, Apr 28, 2011 at 4:38 PM, Richard Guenther > > wrote: > >> > >> On Fri, Apr 29, 2011 at 1:34 AM, Xinliang David Li > >> wrote: > >> > Sharad can provide some some performance data -- we have seen up to 2% > >> > degradation to wi

Re: [PATCH] Cleanup of cgraph topological ordering functions

> 2011-04-29 Martin Jambor > > * cgraph.h (cgraph_postorder): Remove declaration. > * ipa-utils.h (ipa_free_postorder_info): Declare. > (ipa_reverse_postorder): Likewise. > * cgraphunit.c: Include ipa-utils.h. > (cgraph_expand_all_functions): Update call to ipa_rev

Re: [google] LIPO regression tests and bug fixes (issue4444076)

On Fri, Apr 29, 2011 at 4:16 PM, Jan Hubicka wrote: > Hi, > It seems that majority of testcases are independent of lipo. We could probably > enjoy more of testing on mainline, so could you please take those working on > mainline and make mainline patch and let me know what of the tests are not >

[google]: initialize language field for clone function struct

During function cloning, the language field of the src func is not copied. This can lead to null dereference when gcc calls into langhook functions. Unfortunately, I lost track of the test case. Ok for trunk ? Thanks, David 2011-04-29 Xinliang David Li * tree-inline.c (ininitializ

Re: [google][patch] Track discriminators by instruction instead of by basic block (issue4441075)

>> 2011-04-28  Cary Coutant   >> >>        * basic-block.h (struct basic_block_def): Remove discriminator field. >>        * cfghooks.c (split_block): Remove discriminator field. >>        * cfglayout.c (insn_discriminator): New function. >>        * final.c (discriminator): Remove. >>        (over

[google]: add a wrong dce regression test case

The following test makes sure compiler does not wrongly eliminate store to ireg.ax. Ok for trunk? David 2011-04-29 Xinliang David Li * gcc.dg/tree-ssa/alias_bug.c: New test. Index: testsuite/gcc.dg/tree-ssa/alias_bug.c

Re: [google][patch] Add new -gmlt option for min. debug info with line tables (issue4440072)

>> OK for google/main? > > Yes. Thanks, committed to google/main. -cary

[patch, libgfortran] PR48767 Rounding Up followup patch

Hi, The attached patch does some cleanup and a check for trailing zeros to decide whether or not to round. I have added the additional test cases posted on the bugzilla to the existing test case round_3.f08. Regression tested on x86-64. OK for trunk and then I will back port the whole ench

Re: [patch] Split Parse Timevar (rev 2) (issue4433076)

OK. Jason

Re: [patch gimplifier]: Do folding on truth and/or trees

On Wed, 27 Apr 2011, Jakub Jelinek wrote: > You don't need to build cross binutils, all that is needed is > configure the cross and build just cc1, don't mind that the build > fails afterwards and just run it on your testcases by hand to see > what is in the dumps. FWIW "make all-gcc" doesn't fail

Improve Jump Threading 3 of N

-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Found by inspection while working on another threading patch. DOM tracks equivalences in a stack so that it can restore the state of the known equivalences as it completes processing a block in the dominator tree. The threading code also utilizes t