This patch fixes the problem with substituting expressions into IF_THEN_ELSE
during combining. Without this patch combining of conditionals inside
IF_THEN_ELSE is seriously inhibited.
The problem this patch fixes is that combine_simplify_rtx() prefers to return
an expression (say, xor (a)
Janne Blomqvist wrote:
On Thu, Mar 24, 2011 at 18:51, Jim Meyering j...@meyering.net wrote:
Janne Blomqvist wrote:
On Tue, Mar 8, 2011 at 19:53, Jim Meyering j...@meyering.net wrote:
Relative to v2, I've added libgo/ to the list of exempt directories and
added
this recently discussed
2011-04-14 Eric Botcazou ebotca...@adacore.com
* cfgexpand.c (expand_call_stmt): Rematerialize the original function
type if this is not a builtin function.
Hum, using fold_convert seems to be more appropriate here.
Bootstrapped/regtested on i586-suse-linux, applied as obvious.
Signed-off-by: Jim Meyering meyer...@redhat.com
---
While most of these are in comments, the corrections
in gcc/tree-cfg.c and gcc/config/sh/constraints.md are in strings.
The former at least is marked for translation, and hence appears
in every .po file.
gcc/config/alpha/vms-unwind.h |
On Fri, Apr 15, 2011 at 10:54, Jim Meyering j...@meyering.net wrote:
Janne Blomqvist wrote:
On Thu, Mar 24, 2011 at 18:51, Jim Meyering j...@meyering.net wrote:
Janne Blomqvist wrote:
On Tue, Mar 8, 2011 at 19:53, Jim Meyering j...@meyering.net wrote:
Relative to v2, I've added libgo/ to the
Janne Blomqvist wrote:
On Fri, Apr 15, 2011 at 10:54, Jim Meyering j...@meyering.net wrote:
Janne Blomqvist wrote:
On Thu, Mar 24, 2011 at 18:51, Jim Meyering j...@meyering.net wrote:
Janne Blomqvist wrote:
On Tue, Mar 8, 2011 at 19:53, Jim Meyering j...@meyering.net wrote:
Relative to v2,
Long lines, superfluous 's' and missing head comment. No functional changes.
There is one function left without head comment: gimplify_adjust_omp_clauses.
Jakub, when you have a few minutes, would you mind adding one? TIA.
Tested on i586-suse-linux, applied on the mainline as obvious.
On Thu, 14 Apr 2011, Diego Novillo wrote:
On Thu, Apr 14, 2011 at 15:28, Jakub Jelinek ja...@redhat.com wrote:
If bitpack_word_t has BITS_PER_BITPACK_WORD bits, then for
nbits = BITS_PER_BITPACK_WORD this will be undefined.
Use say
mask = ((bitpack_word_t) 2 (nbits - 1)) - 1;
or
The patch was successfully tested on {i686, arm, mips}-linux, both GCC
testsuites and SPEC2000 runs. For all targets there was no observable code
difference in SPEC2000 benchmarks, so the example does not trigger very
often. Still, it speeds up CoreMark by about 1%.
OK for trunk?
Yes,
On Thu, 14 Apr 2011, Eric Botcazou wrote:
I have the following, the gimple_call_chain check is to prevent
Ada bootstrap from failing.
On x86?
Yes, though maybe because I had get_callee_fndecl patched to not
strip conversions.
I suppose your patch is ok, maybe with simply not wrapping
On Thu, Apr 14, 2011 at 8:14 PM, Denis Chertykov cherty...@gmail.com wrote:
2011/4/14 Georg-Johann Lay a...@gjlay.de:
This patchlet adds -finline-limit=0 to dg-options in
testsuite/gcc.target/avr/torture/pr41885.c
because otherwise optimizers will fold all tests and actually no test
I fixed this on the and added sanity check that the fields are initialized.
This has shown problem with early inliner iteration fixed thusly and fact
that
early inliner is attempting to compute overall growth at a time the inline
parameters are not computed for functions not visited
2011/4/15 Jan Hubicka hubi...@ucw.cz:
I fixed this on the and added sanity check that the fields are initialized.
This has shown problem with early inliner iteration fixed thusly and fact
that
early inliner is attempting to compute overall growth at a time the inline
parameters are
Hi!
This testcase doesn't use avx-check.h and is dg-do run testcase,
as it is compiled with -mavx whenever it emits any AVX instructions,
it won't be able to run on any older CPUs.
Fixed thusly, committed to trunk/4.6.
2011-04-15 Jakub Jelinek ja...@redhat.com
PR target/48614
And make IN_COND a bool instead of an int?
I had the same initial reaction but then came to the same conclusion as Maxim.
--
Eric Botcazou
Ping.
On 25/03/11 16:19, Andrew Stubbs wrote:
On 28/01/11 15:20, Richard Earnshaw wrote:
On Fri, 2011-01-28 at 15:13 +, Andrew Stubbs wrote:
On 28/01/11 14:12, Richard Earnshaw wrote:
So what happens to a variation of your testcase:
long long foolong (long long x, short *a, short *b)
{
Fix annoying gcov filename handling:
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01380.html
(rs6000) Fix thinko in output_profile_hook:
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01624.html
Introduce -Wstack-usage:
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01992.html
Thanks in
Richard Guenther schrieb:
On Thu, Apr 14, 2011 at 8:14 PM, Denis Chertykov cherty...@gmail.com wrote:
2011/4/14 Georg-Johann Lay a...@gjlay.de:
This patchlet adds -finline-limit=0 to dg-options in
testsuite/gcc.target/avr/torture/pr41885.c
because otherwise optimizers will fold all tests
Hi,
On Thu, 14 Apr 2011, Dominique Dhumieres wrote:
I have forgotten to mentionned that I have a variant of fatigue
in which I have done the inlining manually along with few other
optimizations and the timing for it is
[macbook] lin/test% gfc -Ofast fatigue_v8.f90
[macbook] lin/test% time
On 04/14/2011 11:53 PM, Janne Blomqvist wrote:
Hi,
as is well known, sprintf() is prone to buffer overflow, hence
snprintf(). libgfortran uses snprintf() in some places, but not
everywhere. Rather than analyzing every sprintf() call for a potential
overflow, the attached patch takes the
On Apr 15, 2011, at 3:34 PM, Eric Botcazou wrote:
The problem this patch fixes is that combine_simplify_rtx() prefers to
return an expression (say, xor (a) (b)) even when a comparison is
prefered (say, neq (xor (a) (b)) 0). Expressions are not recognized as
valid conditions of if_then_else
On Tue, Apr 12, 2011 at 4:14 PM, Richard Sandiford
richard.sandif...@linaro.org wrote:
We have separate vect_extract_even_odd and vect_extract_even_odd_wide
target selectors, and separate vect_strided and vect_strided_wide
selectors. The comment suggests that wide is for 32+ bits,
but we
On Tue, Apr 12, 2011 at 4:19 PM, Richard Sandiford
richard.sandif...@linaro.org wrote:
The next patch introduces separate vect_stridedN target selectors
for each tested stride factor N. At the moment, some tests contain
several independent loops that have different stride factors.
It's easier
On Tue, Apr 12, 2011 at 4:34 PM, Richard Sandiford
richard.sandif...@linaro.org wrote:
This patch adds a test for stride-3 accesses. I didn't add any
particularly complicated cases because I think the testsuite already
covers the interaction between the strided loads stores and other
On Fri, Apr 15, 2011 at 04:56, Richard Guenther rguent...@suse.de wrote:
@@ -518,7 +518,8 @@ pack_ts_type_value_fields (struct bitpack_d *bp, tree
expr)
bp_pack_value (bp, TYPE_USER_ALIGN (expr), 1);
bp_pack_value (bp, TYPE_READONLY (expr), 1);
bp_pack_value (bp, TYPE_ALIGN (expr),
On Fri, 15 Apr 2011, Diego Novillo wrote:
On Fri, Apr 15, 2011 at 04:56, Richard Guenther rguent...@suse.de wrote:
@@ -518,7 +518,8 @@ pack_ts_type_value_fields (struct bitpack_d *bp, tree
expr)
bp_pack_value (bp, TYPE_USER_ALIGN (expr), 1);
bp_pack_value (bp, TYPE_READONLY
Hi,
ipa_analyze_virtual_call_uses contains code that was meant to deal
with situation where OBJ_TYPE_REF_OBJECT is a (number of)
COMPONENT_REFs on top of a dereferenced default definition SSA_NAME of
a parameter.
The code is useless because that never happens in the IL, if an
ancestor object of
Hi,
in order to speed up astar, I had to persuade the function that
decides whether a statement potentially modifies the dynamic type of
an object by storing a new value to the VMT pointer to consider the
following statement harmless (all types are integers of some sort):
MEM[(i32 *)b2arp_3(D) +
Hi,
this is the patch that actually speeds up astar (together with the
previous one). It implements devirtualization based on global
objects. In the past we have refrained from doing this because in
general it is difficult to say whether the object is currently being
constructed and so it might
Hi,
early inlining can create virtual calls based on the part of an object
that represents an ancestor. This patch makes ipa-prop analysis able
to recognize such calls and store the required offset along with such
calls (the field is already there for similar purposes of indirect
inlining). The
On Fri, Apr 15, 2011 at 08:49, Richard Guenther rguent...@suse.de wrote:
On Fri, 15 Apr 2011, Diego Novillo wrote:
On Fri, Apr 15, 2011 at 04:56, Richard Guenther rguent...@suse.de wrote:
@@ -518,7 +518,8 @@ pack_ts_type_value_fields (struct bitpack_d *bp, tree
expr)
bp_pack_value
On Tue, 2011-04-12 at 15:01 +0100, Richard Sandiford wrote:
This patch adds vec_load_lanes and vec_store_lanes patterns for NEON.
They feed directly into the corresponding intrinsic patterns.
Tested on x86_64-linux-gnu and arm-linux-gnueabi. OK to install?
Richard
gcc/
*
Janne == Janne Blomqvist blomqvist.ja...@gmail.com writes:
Jim Can someone add me to the gcc group? That would help.
Jim I already have ssh access to sourceware.org.
Janne I'm not sure if I'm considered to be well-established
Janne enough, so could someone help Jim out here, please?
I added
There are a couple places that check GIMPLE_ASMs for clobbering memory;
this patch centralizes the logic in gimple.c.
Tested on x86_64-unknown-linux-gnu. OK to commit?
-Nathan
* gimple.h (gimple_asm_clobbers_memory_p): Declare.
* gimple.c (gimple_asm_clobbers_memory_p): Define.
On Thu, 2011-04-14 at 21:19 +0800, Carrot Wei wrote:
On Fri, Apr 8, 2011 at 6:51 PM, Ramana Radhakrishnan
ramana.radhakrish...@linaro.org wrote:
On 08/04/11 10:57, Carrot Wei wrote:
Hi
This is the second part of the fixing for
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=47855
Michael,
Yes, this is due to the DECL_EXPR statement which is rendered by the
dumper just the same as a normal decl. The testcase looks for exactly one
such decl, but with -fstack-arrays there are exactly two for each such
array.
The testsuite is run without -fstack-arrays, so I dont'
This makes us avoid 323.
Tested on x86_64-unknown-linux-gnu/-m32, committed.
Richard.
2011-04-15 Richard Guenther rguent...@suse.de
PR testsuite/48286
* gfortran.dg/cray_pointers_8.f90: Use -ffloat-store.
gcc/testsuite/gfortran.dg/cray_pointers_8.f90
Index:
On Fri, Apr 15, 2011 at 3:19 PM, Nathan Froyd froy...@codesourcery.com wrote:
There are several cut-and-pasted loops in expand_asm_stmt that could be
parameterized by the functions used to access the particular operands.
The patch below does that.
Tested on x86_64-unknown-linux-gnu. OK to
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg01060.html
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg02247.html
-BEGIN PGP SIGNATURE-
Version: GnuPG v1.4.11 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/
Hi,
On Fri, 15 Apr 2011, Dominique Dhumieres wrote:
Michael,
Yes, this is due to the DECL_EXPR statement which is rendered by the
dumper just the same as a normal decl. The testcase looks for exactly one
such decl, but with -fstack-arrays there are exactly two for each such
array.
On Fri, 15 Apr 2011, Martin Jambor wrote:
Hi,
ipa_analyze_virtual_call_uses contains code that was meant to deal
with situation where OBJ_TYPE_REF_OBJECT is a (number of)
COMPONENT_REFs on top of a dereferenced default definition SSA_NAME of
a parameter.
The code is useless because that
On 04/15/2011 07:28 AM, Michael Matz wrote:
--- snip ---
I'll make the DECL_EXPR conditional on the size being variable. As Tobias
already okayed the patch I'm planning to check in the slightly modified
variant as below, after a new round of testing.
Thats A-OK
Thanks,
Jerry
I have forgotten to say that the 125 and 516 thresholds are for
x86_64-apple-darwin10. On powerpc-apple-darwin9 they are
repsectively 172 and 638.
Dominique
Hello,
This patch fixes the scenario where SMS fails to
schedule a loop and continue to the next one without
freeing data structures allocated while scheduling
the first loop.
Bootstrap and regtested on ppc64-redhat-linux.
OK for mainline?
Thanks,
Revital
Changelog:
* modulo-sched.c
Hello,
The attached patch adds missing free operation for storage
allocated while calculating SCCs.
Bootstrap and regtested on ppc64-redhat-linux.
OK for mainline?
Thanks,
Revital
Changelog:
* ddg.c (free_ddg_all_sccs): Free sccs field in struct ddg_all_sccs.
Index: ddg.c
On Fri, 15 Apr 2011, Martin Jambor wrote:
Hi,
early inlining can create virtual calls based on the part of an object
that represents an ancestor. This patch makes ipa-prop analysis able
to recognize such calls and store the required offset along with such
calls (the field is already there
On Fri, 15 Apr 2011, Martin Jambor wrote:
Hi,
in order to speed up astar, I had to persuade the function that
decides whether a statement potentially modifies the dynamic type of
an object by storing a new value to the VMT pointer to consider the
following statement harmless (all types are
On Fri, 15 Apr 2011, Martin Jambor wrote:
Hi,
this is the patch that actually speeds up astar (together with the
previous one). It implements devirtualization based on global
objects. In the past we have refrained from doing this because in
general it is difficult to say whether the
On Fri, Apr 15, 2011 at 5:26 PM, Revital Eres revital.e...@linaro.org wrote:
Hello,
This patch fixes the scenario where SMS fails to
schedule a loop and continue to the next one without
freeing data structures allocated while scheduling
the first loop.
Bootstrap and regtested on
On Fri, Apr 15, 2011 at 5:27 PM, Revital Eres revital.e...@linaro.org wrote:
Hello,
The attached patch adds missing free operation for storage
allocated while calculating SCCs.
Bootstrap and regtested on ppc64-redhat-linux.
OK for mainline?
Ok.
Thanks,
Richard.
Thanks,
Revital
On 04/13/2011 02:38 PM, Bernd Schmidt wrote:
This still requires the i386 output_set_got which I think I can cope
with [...]
Patch below, to be applied on top of all the others. Only lightly tested
so far beyond standard (fairly useless) regression tests, by comparing
generated assembly
Hi,
I plan to commit the following simple and obvious patch.
Regression tested on x86-64-linux-gnu.
I will apply the test case in the PR to the testsuite.
Regards,
Jerry
2011-04-15 Jerry DeLisle jvdeli...@gcc.gnu.org
PR libgfortran/48589
* io/write_float.def
On 04/15/2011 04:18 PM, Jeff Law wrote:
http://gcc.gnu.org/ml/gcc-patches/2011-03/msg02247.html
I don't know. I sympathize with the goal, but I'm not too happy about
the structure of this patch. Doesn't this do the scan once for every
reload in an insn? It seems to me like the loop (or rather,
Resent in plain text mode ..
David
On Fri, Apr 15, 2011 at 9:28 AM, Xinliang David Li davi...@google.com wrote:
Honza, do you have a chance to take a look at it?
Thanks,
David
On Wed, Apr 13, 2011 at 3:25 PM, Xinliang David Li davi...@google.com wrote:
Hi, in current FDO
On Apr 15, 2011, at 3:44 AM, Eric Botcazou ebotca...@adacore.com wrote:
And make IN_COND a bool instead of an int?
I had the same initial reaction but then came to the same conclusion as Maxim.
If it can be bool, it should be bool.
Hello,
On 15 April 2011 18:53, Nathan Froyd froy...@codesourcery.com wrote:
On Fri, Apr 15, 2011 at 06:27:05PM +0300, Revital Eres wrote:
+ if (all_sccs-sccs)
+ free (all_sccs-sccs);
No need to check for non-NULL prior to free'ing.
OK, I'll commit the patch without the check then.
Nicola,
Ok for mainline if both pass?
Yes.
[and by the way, I think you're fixing PR libobjc/32037 in the process. :-)]
indeed, thanks for checking. Testing revealed that I'd been lazy with
quoting. I'm including the patch I've installed, which encloses
OBJC_BOEHM_GC in configure.ac in
While testing the close-to-final version of my COMDAT-group-with-Sun as
patch
[build, c++, lto] Support COMDAT group with Sun as (PR target/40483)
http://gcc.gnu.org/ml/gcc-patches/2010-05/msg01365.html
http://gcc.gnu.org/ml/gcc-patches/2010-06/msg00600.html
which only
This patch rewrites the check for duplicate instance variables
done by the Objective-C (and Objective-C++) compiler.
The new code has many advantages:
* it avoids copying all the instance variables into a temporary
tree chain in the way that the all code was doing (according
to some notes
Hi,
lower_emutls_function_body in emultls.c should use
cgraph_get_create_node to create call graph nodes if need be because
it is introducing TLS builtin decls into IL.
I have bootstrapped and tested this on x86_64-linux (where the bug
does not happen) and in bugzilla it has been confirmed it
If it can be bool, it should be bool.
Rather basic principle IMO. Consistency is of much greater value.
--
Eric Botcazou
Denis Chertykov schrieb:
2011/4/14 Georg-Johann Lay a...@gjlay.de:
Denis Chertykov schrieb:
2011/4/14 Georg-Johann Lay a...@gjlay.de:
The rotlmode3 expanders (mode \in {HI,SI,DI}) violates synopsis by
using 4 operands instead of 3. This runs in ICE in top of
optabs.c:maybe_gen_insn.
The
This concludes this mini project. :-)
Committed and the pages on gcc.gnu.org regenerated; www.gnu.org will
follow over night.
Gerald
Index: style.mhtml
===
RCS file: /cvs/gcc/wwwdocs/htdocs/style.mhtml,v
retrieving revision 1.115
This patch cleans up positive tests. First, stop on first failure so
as to avoid littering the logfile. Second, change the extensions of
the generated assembly files to .s-pph and .s+pph to be clear on the
provenence of each file.
Index: gcc/testsuite/ChangeLog.pph
2011-04-15 Lawrence Crowl
Vladimir and Dmitry,
The new test you added, gcc.dg/pr48235.c, is failing on IA64 because
it gets this excess message:
cc1: note: -freorder-blocks-and-partition does not work on this architecture
Could you modify the test to either filter out this message or not
use the
This patch for the Objective-C compiler gets rid of another case where
we'd create a temporary tree chain with a copy of all the instance variables
of a class just to do a simple check on them.
The check is the one to check the scope of the instance variable that is
being accessed; without this
This is a trivial patch to deal with bad profile data (from
multi-threaded programs) that leads to divide by zero error.
Ok for trunk?
Thanks,
David
2011-04-15 Xinliang David Li davi...@google.com
* ipa-inline.c (cgraph_decide_recursive_inlining): Fix
div by zero error
On Fri, Apr 15, 2011 at 15:20, Xinliang David Li davi...@google.com wrote:
This is a trivial patch to deal with bad profile data (from
multi-threaded programs) that leads to divide by zero error.
Ok for trunk?
Thanks,
David
2011-04-15 Xinliang David Li davi...@google.com
*
Hi!
On IRC richi mentioned yesterday that when gcc.dg/guality/asm-1.c
testcase is compiled with g++ instead of gcc, it fails.
Apparently that is because of slightly different ARRAY_REF
in the code, which for g++ leads into
(sign_extend:DI (zero_extend:SI (something:HI)))
in DEBUG_INSN (as opposed
There is no way to get a test case that can produce the problem in a
deterministic way. There is really not much for Honza to review
though :)
David
On Fri, Apr 15, 2011 at 12:52 PM, Diego Novillo dnovi...@google.com wrote:
On Fri, Apr 15, 2011 at 15:20, Xinliang David Li davi...@google.com
On Apr 15, 2011, at 11:52 AM, Nicola Pero wrote:
This patch for the Objective-C compiler gets rid of another case where
we'd create a temporary tree chain with a copy of all the instance variables
of a class just to do a simple check on them.
Ok to commit ?
Ok.
On Fri, Apr 15, 2011 at 16:21, Xinliang David Li davi...@google.com wrote:
There is no way to get a test case that can produce the problem in a
deterministic way. There is really not much for Honza to review
though :)
The check may be needed somewhere else, you may be papering over the
real
On 04/12/2011 08:22 PM, Alan Modra wrote:
On Tue, Apr 12, 2011 at 04:00:45PM -0500, Pat Haugen wrote:
--- gcc/config/rs6000/rs6000.c (revision 172327)
+++ gcc/config/rs6000/rs6000.c (working copy)
@@ -7976,7 +7976,7 @@ call_ABI_of_interest (tree fndecl)
return true;
This patch to the Go frontends adds a convenience function
compound_statement to build a statement list from two statements.
Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu.
Committed to mainline.
Ian
2011-04-15 Ian Lance Taylor i...@google.com
* go-gcc.cc
Hi all,
the following patch fixes an issue with a multiple decl for procedures
with procedure dummy arguments. Without the patch, multiple declarations
are generated, which causes link failures (euler is optimized away)
with -fwhole-program. (Thanks goes to Richard for spotting the problem
I was curious if anyone was still looking at this problem? I see this
on IA64 HP-UX in 32 bit mode where ptr_mode(SImode) != Pmode(DImode).
As H.J. points out, expand_expr_real_2 is calling simplify_gen_subreg
(expr.c, line 7366) and at that point op0 is (label_ref/v:DI 32) and
innermode is
On Sat, Apr 16, 2011 at 12:30:01AM +0200, Tobias Burnus wrote:
I have no idea why there should be a problem with inlining (cf. deleted
comment), but removing that part did not give any test suite failure
(check-gfortran + libgomp's check). Additionally, I have build and run
the Polyhedron
Applied (with a few formatting tweaks).
Thanks,
Jason
Hi,
tested x86_64-linux, committed mainline and 4_6-branch.
Paolo.
2011-04-15 Daniel Krugler daniel.krueg...@googlemail.com
Paolo Carlini paolo.carl...@oracle.com
PR libstdc++/48631
* include/bits/unique_ptr.h (default_delete_Tp[]): Add
Hello,
If it's for debugging, can you use a --parm instead (like
modulo-sched-min-sc or similar)?
I think I can use --param for debugging purposes in this case.
(I might add modulo-sched-max-sc as well)
Thanks,
Revital
Thanks,
Richard.
Thanks,
Revital
Changelog:
* common.opt
Hi Richard
Thank you for the detailed explanation. It sounds like an inherent
difficulty of rtl passes. Then the only opportunity is ldrb/strb
instructions because they never affect cc registers.
thanks
Carrot
On Fri, Apr 15, 2011 at 9:34 PM, Richard Earnshaw rearn...@arm.com wrote:
On Thu,
81 matches
Mail list logo