The moxie port had CC_REG referring to a real hard register ($r9) by
mistake instead of the virtual CC register. This never resulted in
incorrect code, but we would often see $r9 marked as used in a function
when it actually wasn't. I'm checking this in.
2015-01-09 Anthony Green
I'm committing the following patch, which cleans up the assembly output
by using tabs between opcodes and operands.
Thanks,
AG
2015-01-09 Anthony Green gr...@moxielogic.com
* config/moxie/moxie.md: Tabify assembly output.
Index: gcc/config/moxie/moxie.md
On Fri, 09 Jan 2015, Kyrill Tkachov wrote:
On 09/01/15 16:11, Christophe Lyon wrote:
On 9 January 2015 at 11:26, Martin Liška mli...@suse.cz wrote:
On 01/09/2015 06:21 AM, Jeff Law wrote:
On 01/07/15 04:38, Martin Liška wrote:
Hello.
Following patch adds support for target and
On Wed, 19 Nov 2014, David Malcolm wrote:
On Wed, 2014-11-19 at 10:09 -0700, Jeff Law wrote:
On 11/19/14 04:47, Richard Biener wrote:
On Wed, Nov 19, 2014 at 11:46 AM, David Malcolm dmalc...@redhat.com
wrote:
Valgrind complains about uninitialized data within sparseset_bit_p.
This patch from Peter Collingbourne backports a patch from the master
repository to pass CGO_LDFLAGS to the linker when using cgo with
gccgo. Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu.
Committed to mainline.
Ian
diff -r b0a83aacb539 libgo/go/cmd/go/build.go
---
On Fri, Jan 9, 2015 at 12:12 PM, Magnus Granberg zo...@gentoo.org wrote:
fredag 09 januari 2015 13.00.14 skrev Daniel Micay:
On 09/01/15 12:49 PM, Joseph Myers wrote:
On Fri, 9 Jan 2015, Daniel Micay wrote:
--with-specs=%{pie|fpic|fPIC|fpie|fPIE|fno-pic|fno-PIC|fno-pie|fno-PIE|
On 01/07/2015 12:39 AM, Eric Botcazou wrote:
Some ports are missing (lm32, moxie, nios2, nvptx, rl78, rx) so the relevant
maintainers are CCed (see 6.3.9 Anatomy of a Target Back End in the doc).
The page is directly browsable at https://gcc.gnu.org/backends.html
For the moxie, nvptx, rl178
When not bootstrapping, the newly built Go compiler is not passed down
to the Go tools as GOC. This patch changes the gotools Makefile to
use GOC_FOR_TARGET for a native build. I also set MOSTLYCLEANFILES.
Bootstrapped on x86_64-unknown-linux-gnu. Committed to mainline.
Ian
2015-01-09 Ian
On Jan 9, 2015, at 4:20 AM, Matthew Fortune matthew.fort...@imgtec.com
wrote:
Robert Suchanek robert.sucha...@imgtec.com writes:
gcc/
* simplify-rtx.c (simplify_replace_fn_rtx): Simplify (lo_sum (high x)
(const (plus x offset))) to (const (plus x offset)).
The fix appears
On January 9, 2015 4:10:44 PM CET, Bernd Edlinger bernd.edlin...@hotmail.de
wrote:
On Fri, 9 Jan 2015 14:04:27, Richard Biener wrote:
FYI: the VIEW_CONVERT_EXPR did not fail in the
gcc_checking_assert (is_gimple_addressable (base))
but much later, somewhere in tree-cfg.c it dropped out.
Jakub,
For the oacc kernels patch series I need a fortran builtin with fn spec
attribute (as mentioned here: https://gcc.gnu.org/ml/gcc/2014-12/msg1.html ).
Attached patch adds a function gfc_define_builtin_with_spec that allows me to
define such a builtin.
At this point there's no
On Fri, Jan 9, 2015 at 5:52 AM, Segher Boessenkool
seg...@kernel.crashing.org wrote:
This fixes 88 testsuite FAILs.
-mpowerpc64 does not change the ABI, but it does change the value of
UNITS_PER_WORD. We could use POINTER_SIZE_UNITS instead of 4 here,
but that does not seem quite right.
On 01/08/2015 10:10 PM, Jeff Law wrote:
On 01/08/15 15:08, Sandra Loosemore wrote:
This patch cleans up the documentation of -fvtable-verify, -fvtv-debug,
and -fvtv-counts. The substantive change is to correct the location of
the debug log files per discussion here:
On Fri, Jan 9, 2015 at 1:38 AM, Segher Boessenkool
seg...@kernel.crashing.org wrote:
This fixes gcc.target/powerpc/recip-[67].c with -m32 -mpowerpc64.
Tested etc.; okay for mainline?
Segher
2015-01-08 Segher Boessenkool seg...@kernel.crashing.org
gcc/testsuite/
*
Some hooks return word_mode by default, which is incorrect for -m32
-mpowerpc64. This patch creates a new function rs6000_abi_word_mode
to implement these hooks, and does so.
This fixes 163 testuite FAILs.
Tested as usual; okay for mainline?
2015-01-09 Segher Boessenkool
Hi,
On Thu, 8 Jan 2015 22:27:26, Jakub Jelinek wrote:
Any objections to approving it now?
LGTM.
Jakub
would it be OK to apply this patch also to the 4.9 testsuite,
except for c-c++-common/tsan/bitfield_race.c and
g++.dg/tsan/aligned_vs_unaligned_race.C of course?
Bernd.
On Fri, 9 Jan 2015 14:04:27, Richard Biener wrote:
FYI: the VIEW_CONVERT_EXPR did not fail in the
gcc_checking_assert (is_gimple_addressable (base))
but much later, somewhere in tree-cfg.c it dropped out.
How did it fail there? It doesn't look like VIEW_CONVERT_EXPR
is forbidden.
without
On 12/23/2014 07:50 PM, Thomas Schwinge wrote:
GCC needs this, if nvptx-none-ar and nvptx-none-ranlib aren't found in $PATH.
I've pushed the three patches you sent to my github repository.
Bernd
On 9 January 2015 at 11:26, Martin Liška mli...@suse.cz wrote:
On 01/09/2015 06:21 AM, Jeff Law wrote:
On 01/07/15 04:38, Martin Liška wrote:
Hello.
Following patch adds support for target and optimization nodes
comparison, which is
based on Honza's newly added infrastructure.
Apart
On 09/01/2015 15:24, Ian Lance Taylor wrote:
This should work automatically, the only difference is that you must
omit $(LIBGODEP) from the dependencies.
What will happen if there is no installed gccgo at the right version?
Compilation fails.
What should happen?
Compilation fails. :)
On Fri, Jan 9, 2015 at 10:26 AM, Segher Boessenkool
seg...@kernel.crashing.org wrote:
Some hooks return word_mode by default, which is incorrect for -m32
-mpowerpc64. This patch creates a new function rs6000_abi_word_mode
to implement these hooks, and does so.
This fixes 163 testuite FAILs.
On Fri, Jan 09, 2015 at 04:32:47PM +0100, Bernd Edlinger wrote:
Hi,
On Thu, 8 Jan 2015 22:27:26, Jakub Jelinek wrote:
Any objections to approving it now?
LGTM.
Jakub
would it be OK to apply this patch also to the 4.9 testsuite,
except for c-c++-common/tsan/bitfield_race.c and
On Fri, Jan 09, 2015 at 09:54:12AM +0100, Tom de Vries wrote:
0001-Fix-gcc_assert-in-expand_omp_for_static_chunk.patch
2014-11-12 Tom de Vriest...@codesourcery.com
Two spaces before .
* omp-low.c (expand_omp_for_static_chunk): Fix assert.
Ok, thanks.
gcc/omp-low.c | 2 +-
1
On Fri, 9 Jan 2015, Richard Biener wrote:
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
Hi!
The following testcase is miscompiled on s390x. The problem is that there
is massive cross-jumping going on, and after that post_order_compute
decides to call tidy_fallthru_edges, including on an
On Mon, Jan 5, 2015 at 2:12 PM, Martin Liška mli...@suse.cz wrote:
On 12/19/2014 12:04 PM, Richard Biener wrote:
On Thu, Dec 18, 2014 at 6:38 PM, Martin Liška mli...@suse.cz wrote:
On 12/17/2014 04:23 PM, Richard Biener wrote:
On Wed, Dec 17, 2014 at 12:17 PM, Martin Liška mli...@suse.cz
On Fri, Jan 09, 2015 at 10:36:09AM +0100, Richard Biener wrote:
I wonder why post_order_compute calls tidy_fallthru_edges at all - won't
that break the just computed postorder?
Dunno, but I think it shouldn't break anything, the function doesn't remove
any blocks, just in the typical case of an
On Mon, Jan 5, 2015 at 9:26 PM, Jeff Law l...@redhat.com wrote:
On 12/29/14 06:30, Yuri Rumyantsev wrote:
Hi All,
Here is a patch which fixed several performance degradation after
operand canonicalization (r216728). Very simple approach is used - if
operation is commutative and its second
Hi,
On Fri, 9 Jan 2015 10:57:14, Richard Biener wrote:
On Mon, Jan 5, 2015 at 9:00 PM, Jeff Law l...@redhat.com wrote:
On 01/03/15 06:49, Bernd Edlinger wrote:
Hi,
I was experimenting with enabling TSAN for Ada recently.
I think this gives rather interesting results.
The Instrumentation
On 08/01/2015 22:35, Ian Lance Taylor wrote:
+if NATIVE
+
+# For a native build we build the programs using the newly built libgo
+# and install them as regular programs.
+
+bin_PROGRAMS = go$(EXEEXT) gofmt$(EXEEXT)
+libexecsub_PROGRAMS = cgo$(EXEEXT)
+
+go$(EXEEXT):
This patch flattens tree.h and tree-core.h. This is a revised patch that
does not include tree-core.h as a result of flattening.
Version 3 of the patch adds the header files removed from tree-core.h to
gcc-plugin.h in order to allow ggc-common.c to compile. This is a recent
issue seen on
On Mon, Jan 5, 2015 at 9:00 PM, Jeff Law l...@redhat.com wrote:
On 01/03/15 06:49, Bernd Edlinger wrote:
Hi,
I was experimenting with enabling TSAN for Ada recently.
I think this gives rather interesting results.
The Instrumentation worked almost out of the box, we just have
the problem
Hi!
On Fri, 9 Jan 2015 03:40:29 +0300, Ilya Verbin iver...@gmail.com wrote:
Currently check_effective_target_lto properly works only in gcc/testsuite/
dir,
since it checks for ENABLE_LTO, which is defined in gcc/configure.ac.
But it doesn't work in other subdirectories, e.g. in
On 01/09/2015 06:21 AM, Jeff Law wrote:
On 01/07/15 04:38, Martin Liška wrote:
Hello.
Following patch adds support for target and optimization nodes
comparison, which is
based on Honza's newly added infrastructure.
Apart from that, there's a small hunk that corrects formatting and
removes
On 26-11-14 14:25, Tom de Vries wrote:
On 12-11-14 11:00, Tom de Vries wrote:
Jakub,
this patch fixes a gcc_assert in expand_omp_for_static_chunk.
The assert follows a loop with composite loop condition:
...
vecedge_var_map *head = redirect_edge_var_map_vector (re);
ene =
On 08/01/15 20:28, Andreas Tobler wrote:
On 08.01.15 17:22, Richard Earnshaw wrote:
On 27/11/14 20:56, Andreas Tobler wrote:
Hi all,
this is the second attempt.
I reworked the issues Richard mentioned in the previous review.
I also found one issue which will break build/bootstrap if I pass
Hello.
Following patch is motivated by seen utilization of
cgraph_edge::redirect_callee in Chromium
compiled with LTO. Thus, I decided to move the function to header file as well
as few small
functions it calls. Changes are just of mechanical nature.
Patch has been tested on x86_64-linux-pc
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
Hi!
The following testcase is miscompiled on s390x. The problem is that there
is massive cross-jumping going on, and after that post_order_compute
decides to call tidy_fallthru_edges, including on an edge from a bb ending
with a table jump to a bb
On 2015-01-09 19:03, Jonathan Wakely wrote:
The attached patch should be correct.
Tested x86_64-linux, committed to trunk and 4.9.
Awesome, thanks!
2015-01-09 13:39 GMT+00:00 Jiong Wang jiong.w...@arm.com:
the following code in store_bit_field_using_insv haven't consider above
MEM-REG situation,
it always assume bitnum + bitsize is within unit which is wrong.
if (BITS_BIG_ENDIAN != BYTES_BIG_ENDIAN)
bitnum = unit - bitsize -
On 09/01/15 16:11, Christophe Lyon wrote:
On 9 January 2015 at 11:26, Martin Liška mli...@suse.cz wrote:
On 01/09/2015 06:21 AM, Jeff Law wrote:
On 01/07/15 04:38, Martin Liška wrote:
Hello.
Following patch adds support for target and optimization nodes
comparison, which is
based on Honza's
On Jan 8, 2015, at 9:35 PM, Jeff Law l...@redhat.com wrote:
Do you want to be a reviewer for libobjc?
I think things are fine as is. If things were pinged and there were no
response, or if someone wanted to do major updated on the library to bring it
up a decade, then we might want to change
On 09-01-15 11:48, Jakub Jelinek wrote:
On Fri, Jan 09, 2015 at 11:35:41AM +0100, Tom de Vries wrote:
2015-01-09 Tom de Vries t...@codesourcery.com
PR rtl-optimization/64539
* regcprop.c (copyprop_hardreg_forward_1): Handle clobbers in
CALL_INSN_FUNCTION_USAGE.
To
On Fri, 9 Jan 2015, Daniel Micay wrote:
--with-specs=%{pie|fpic|fPIC|fpie|fPIE|fno-pic|fno-PIC|fno-pie|fno-PIE|shared|static|nostdlib|nodefaultlibs|nostartfiles:;:-fPIE
-pie}
at configure time (using CONFIGURE_SPECS).
I have no idea if the above is really the proper spec to use -
On 09/01/15 12:49 PM, Joseph Myers wrote:
On Fri, 9 Jan 2015, Daniel Micay wrote:
--with-specs=%{pie|fpic|fPIC|fpie|fPIE|fno-pic|fno-PIC|fno-pie|fno-PIE|shared|static|nostdlib|nodefaultlibs|nostartfiles:;:-fPIE
-pie}
at configure time (using CONFIGURE_SPECS).
I have no idea if the above
Hi,
This change:
+2014-12-05 Martin Jambor mjam...@suse.cz
+
PR ipa/64192
ipa-prop.c (ipa_compute_jump_functions_for_edge): Convert alignment
from bits to bytes after checking they are byte-aligned.
+
causes this regression on AArch64.
FAIL: gcc.target/aarch64/vect-movi.c
Committed to trunk as r219395.
gcc/jit/ChangeLog:
PR jit/64206
* docs/internals/test-hello-world.exe.log.txt: Update, the log now
shows tempdir creation/cleanup.
* docs/_build/texinfo/libgccjit.texi: Regenerate.
* jit-logging.h (class gcc::jit::log_user):
On Thu, Jan 8, 2015 at 8:10 PM, Segher Boessenkool
seg...@kernel.crashing.org wrote:
As the existing comment explains, we should always promote function
arguments and return values. However, notwithstanding its name,
default_promote_function_mode_always_promote does not always promote.
If unused on trunk, why would we commit it there?
When your branch is merged, you'll merge it along. Otherwise that defeats the
purpose of working on a branch, unless I misunderstand something...
FX
Le 9 janv. 2015 à 16:37, Tom de Vries tom_devr...@mentor.com a écrit :
Jakub,
For the
On 01/09/15 04:32, Robert Suchanek wrote:
Hi Steven/Vladimir,
It's hard to say what the correct fix should be, but it sounds like
the address you get after the substitutions should be simplified
(folded).
Coming back to the original testcase and re-analyzing the problem, it appears
that
On 09-01-15 18:11, FX wrote:
If unused on trunk, why would we commit it there?
When your branch is merged, you'll merge it along. Otherwise that defeats the
purpose of working on a branch, unless I misunderstand something...
This patch is not branch-specific.
Thanks,
- Tom
FX
Le 9
On Jan 8, 2015, at 9:38 PM, Andrew Pinski pins...@gmail.com wrote:
2014-12-24 Dimitris Papavasiliou dpapa...@gmail.com
PR libobjc/51891
* libobjc/ivars.c: Add a check for classes without instance
variables, which have a NULL ivar list pointer.
*
Previously it was only possible to create integer constants via
gcc_jit_context_new_rvalue_from_int [1], which takes a host int and
a gcc_jit_type representing a target type
Hence it wasn't possible to create e.g. the constant (long)LONG_MAX
if int != long.
(strictly speaking, one might be able
On Sat, 2015-01-10 at 01:50 +0530, Prathamesh Kulkarni wrote:
On 9 January 2015 at 16:21, Richard Biener richard.guent...@gmail.com wrote:
On Fri, Jan 9, 2015 at 10:39 AM, Michael Collison
michael.colli...@linaro.org wrote:
This patch flattens tree.h and tree-core.h. This is a revised patch
2015-01-09 21:29 GMT+00:00 Andrew Pinski pins...@gmail.com:
On Fri, Jan 9, 2015 at 12:40 PM, Jeff Law l...@redhat.com wrote:
On 01/09/15 06:39, Jiong Wang wrote:
as reported at https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64304
given the following test:
unsigned char byte = 0;
void
jit.dg/test-combination.c was spelling out all of the passing
test cases, twice, when test-threads.c already had this as metadata.
Move the metadata from test-threads.c into all-non-failing-tests.h,
and use it from test-combination.c to avoid this repetition.
Before/after test-combination.c both
On 28/12/14 13:47 +0100, Eelis wrote:
On 2014-12-28 00:18, Eelis wrote:
Trivial fix attached.
Please don't commit this patch.
I just noticed that the assignability test is wrong in an additional way: it
should look at assignability of /output/ elements, not /input/ elements.
As a result,
On 9 January 2015 at 16:21, Richard Biener richard.guent...@gmail.com wrote:
On Fri, Jan 9, 2015 at 10:39 AM, Michael Collison
michael.colli...@linaro.org wrote:
This patch flattens tree.h and tree-core.h. This is a revised patch that
does not include tree-core.h as a result of flattening.
On 01/09/15 06:39, Jiong Wang wrote:
as reported at https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64304
given the following test:
unsigned char byte = 0;
void set_bit(unsigned int bit, unsigned char value) {
unsigned char mask = (unsigned char)(1 (bit 7));
if (!value) {
fredag 09 januari 2015 13.00.14 skrev Daniel Micay:
On 09/01/15 12:49 PM, Joseph Myers wrote:
On Fri, 9 Jan 2015, Daniel Micay wrote:
--with-specs=%{pie|fpic|fPIC|fpie|fPIE|fno-pic|fno-PIC|fno-pie|fno-PIE|
shared|static|nostdlib|nodefaultlibs|nostartfiles:;:-fPIE -pie}
at configure
Hi,
The tree.h and tree-core.h flattening patch:
(https://gcc.gnu.org/ml/gcc-patches/2015-01/msg00467.html
broke visium build. The attached patch fixes that.
Built on visium-elf.
OK to commit ?
Thank you,
Prathamesh
2015-01-09 Prathamesh Kulkarni prathamesh.kulka...@linaro.org
*
On Fri, Jan 9, 2015 at 7:40 AM, Paolo Bonzini bonz...@gnu.org wrote:
On 09/01/2015 15:24, Ian Lance Taylor wrote:
This should work automatically, the only difference is that you must
omit $(LIBGODEP) from the dependencies.
What will happen if there is no installed gccgo at the right
Committed initial gotools patch with following ChangeLog entries:
./:
2015-01-09 Ian Lance Taylor i...@google.com
* configure.ac (host_tools): Add gotools.
* Makefile.def (host_modules): Add gotools.
(dependencies): Add dependency of all-gotools on all-target-libgo.
On Sat, Jan 10, 2015 at 01:50:42AM +0530, Prathamesh Kulkarni wrote:
I bootstrapped on x86 with all languages. I also bootstrapped on all
targets
listed in contrib/config-list.mk with c and c++ enabled.
Is this okay for trunk?
Ok.
Committed as r219402 on behalf of Michael.
Please
This patch changes the way that cmd/go finds gccgo to match the
upstream sources (which had changed since the 1.3 sources currently in
libgo). Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Committed to mainline.
Ian
diff -r 882d8b02b84b libgo/go/cmd/go/build.go
---
On Fri, Jan 9, 2015 at 12:40 PM, Jeff Law l...@redhat.com wrote:
On 01/09/15 06:39, Jiong Wang wrote:
as reported at https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64304
given the following test:
unsigned char byte = 0;
void set_bit(unsigned int bit, unsigned char value) {
unsigned
On 10 January 2015 at 02:58, Jakub Jelinek ja...@redhat.com wrote:
On Sat, Jan 10, 2015 at 01:50:42AM +0530, Prathamesh Kulkarni wrote:
I bootstrapped on x86 with all languages. I also bootstrapped on all
targets
listed in contrib/config-list.mk with c and c++ enabled.
Is this okay for
Hi!
The following testcase is miscompiled on s390x. The problem is that there
is massive cross-jumping going on, and after that post_order_compute
decides to call tidy_fallthru_edges, including on an edge from a bb ending
with a table jump to a bb with now a single successor where all the
On 12/19/2014 05:10 AM, Sandra Loosemore wrote:
On 11/19/2014 09:34 PM, Jan Hubicka wrote:
[snip]
As for timeline, I have a workshop next week and need to prepare draft for it.
So ideally I would like to work on this only after the workshop (ending
November 28). I would be also happy to help
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
On Fri, Jan 09, 2015 at 10:36:09AM +0100, Richard Biener wrote:
I wonder why post_order_compute calls tidy_fallthru_edges at all - won't
that break the just computed postorder?
Dunno, but I think it shouldn't break anything, the function doesn't
On 01/07/2015 12:38 PM, Martin Liška wrote:
Hello.
Following patch adds support for target and optimization nodes comparison,
which is
based on Honza's newly added infrastructure.
Apart from that, there's a small hunk that corrects formatting and removes
unnecessary
call to a comparison
Jakub,
Attached patch handles CALL_INSN_FUNCTION_USAGE clobbers in
copyprop_hardreg_forward_1.
Terry reported a cprop_hardreg misbehaviour here (
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64154#c2 ), in the context of
trying out -fipa-ra for thumb1. The -fipa-ra flag is currently
Hi,
can this be added for GCC 5?
It would be interesting for SUSE too.
Ciao, Marcus
On Mon, Nov 10, 2014 at 09:26:39PM +0100, Magnus Granberg wrote:
fredag 01 augusti 2014 10.52.27 skrev Rainer Orth:
Hi Magnus,
a couple of comments, mostly nits.
2014-07-31 Magnus Granberg
On Fri, Jan 9, 2015 at 10:39 AM, Michael Collison
michael.colli...@linaro.org wrote:
This patch flattens tree.h and tree-core.h. This is a revised patch that
does not include tree-core.h as a result of flattening.
Version 3 of the patch adds the header files removed from tree-core.h to
On Fri, Jan 09, 2015 at 12:07:26PM +0100, Thomas Schwinge wrote:
On Thu, 8 Jan 2015 15:11:49 +0100, Jakub Jelinek ja...@redhat.com wrote:
On Thu, Nov 20, 2014 at 01:27:08PM +0100, Bernd Schmidt wrote:
On 11/13/2014 05:06 AM, Jan Hubicka wrote:
this patch adds infrastructure for proper
On Fri, 9 Jan 2015 12:04:26, Richard Biener wrote:
There may be multiple VIEW_CONVERT_EXPRs in a reference chain
so simply stripping the outermost only doesn't work (the assert).
Hmm, that did not happen in any of the Ada tests in ada/acats nor in gnat.dg,
but with Ada anything may be
Robert Suchanek robert.sucha...@imgtec.com writes:
gcc/
* simplify-rtx.c (simplify_replace_fn_rtx): Simplify (lo_sum (high x)
(const (plus x offset))) to (const (plus x offset)).
The fix appears valid to me. Just some comments on the test case.
On Mon, Dec 22, 2014 at 3:39 PM, Yuri Rumyantsev ysrum...@gmail.com wrote:
Richard,
I changed algorithm for bool pattern repair.
It turned out that ifcvt_local_dce phaase is required since for
test-case I sent you in previous mail vectorization is not performed
without dead code elimination:
On 09/01/15 07:58 AM, Richard Biener wrote:
Looking at the actual implementation I wonder why it's not similar
to how darwin gets at it default (not sure how it does). Also
looking at how DRIVER_SELF_SPECS is used I wonder if the
functionality can be enabled with a simple
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
On Fri, Jan 09, 2015 at 11:59:44AM +0100, Richard Biener wrote:
If you want, I can try instead of disabling it for tablejumps
just move the label.
Yeah, I'd prefer that - it can't be too difficult, no?
So like this (tested just on the
Checked in as
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219384
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219389
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail
Checked in as
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219385
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219390
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail
Checked in slightly modified as
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219387
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219391
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89
On 09.01.15 10:27, Richard Earnshaw wrote:
On 08/01/15 20:28, Andreas Tobler wrote:
On 08.01.15 17:22, Richard Earnshaw wrote:
On 27/11/14 20:56, Andreas Tobler wrote:
Hi all,
this is the second attempt.
I reworked the issues Richard mentioned in the previous review.
I also found one issue
On Fri, Jan 9, 2015 at 12:54 AM, Paolo Bonzini bonz...@gnu.org wrote:
+
+# For a non-native build we have to build the programs using a
+# previously built host (or build - host) Go compiler. We should
+# only do this if such a compiler is available. Figure this out
+# later.
+
+endif
On Fri, Jan 09, 2015 at 03:10:16PM +0100, Richard Biener wrote:
Well, you have until the end of next week ;) For GIMPLE this is
a switch with all cases going to the same basic-block, right?
I think we optimize that in cleanup_control_expr_graph via the
single_succ_p case?
No, it is a switch
On 09/01/15 05:46, Jeff Law wrote:
On 12/30/14 03:21, Jiong Wang wrote:
PR64011 is actually a general problem on all target support bit insertion
instructions.
we overflow check at the start of store_bit_field_1, but that only check
the
situation where the field lies completely outside the
as reported at https://gcc.gnu.org/bugzilla/show_bug.cgi?id=64304
given the following test:
unsigned char byte = 0;
void set_bit(unsigned int bit, unsigned char value) {
unsigned char mask = (unsigned char)(1 (bit 7));
if (!value) {
byte = (unsigned char)~mask;
} else {
Checked in as
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219382
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : sebastian.hu...@embedded-brains.de
PGP : Public key
Checked in as
https://gcc.gnu.org/viewcvs/gcc?view=revisionrevision=219383
--
Sebastian Huber, embedded brains GmbH
Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone : +49 89 189 47 41-16
Fax : +49 89 189 47 41-09
E-Mail : sebastian.hu...@embedded-brains.de
PGP : Public key
On Fri, Jan 09, 2015 at 11:59:44AM +0100, Richard Biener wrote:
If you want, I can try instead of disabling it for tablejumps
just move the label.
Yeah, I'd prefer that - it can't be too difficult, no?
So like this (tested just on the testcase, fully bootstrap/regtest
will follow)?
On Fri, 9 Jan 2015 13:30:45, Jakub Jelinke wrote:
On Fri, Jan 09, 2015 at 01:12:09PM +0100, Bernd Edlinger wrote:
should be equivalent to
if (DECL_P (base) !may_be_aliased (base))
return false;
is that right?
Yes, well, not exactly, but I wonder if its worth doing the extra check
if
On Fri, Jan 09, 2015 at 01:12:09PM +0100, Bernd Edlinger wrote:
should be equivalent to
if (DECL_P (base) !may_be_aliased (base))
return false;
is that right?
Yes, well, not exactly, but I wonder if its worth doing the extra check
if you only check decl accesses anyway and not
On Tue, Dec 30, 2014 at 10:23 PM, Magnus Granberg zo...@gentoo.org wrote:
fredag 14 november 2014 23.31.48 skrev Magnus Granberg:
måndag 10 november 2014 21.26.39 skrev Magnus Granberg:
Rainer
Thanks Rainer for the nits and comments.
Have updated the patches and Changelogs.
But i
On Fri, Jan 9, 2015 at 1:44 PM, Bernd Edlinger
bernd.edlin...@hotmail.de wrote:
On Fri, 9 Jan 2015 13:30:45, Jakub Jelinke wrote:
On Fri, Jan 09, 2015 at 01:12:09PM +0100, Bernd Edlinger wrote:
should be equivalent to
if (DECL_P (base) !may_be_aliased (base))
return false;
is that
Hi all, hi Paul,
I started to implement the changes requested below, but I stumbled over an
oddity:
For a deferred length kind4 char array, the length of the string is stored
without multiplication by 4 in the length variable attached. So when we now
decide to store the length of the string in
This fixes 88 testsuite FAILs.
-mpowerpc64 does not change the ABI, but it does change the value of
UNITS_PER_WORD. We could use POINTER_SIZE_UNITS instead of 4 here,
but that does not seem quite right. This code is for SVR4 only, so
a literal 4 isn't so bad I think. Better suggestions welcome
On Fri, Jan 09, 2015 at 11:15:14AM +0100, Richard Biener wrote:
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
On Fri, Jan 09, 2015 at 10:36:09AM +0100, Richard Biener wrote:
I wonder why post_order_compute calls tidy_fallthru_edges at all - won't
that break the just computed postorder?
Hi!
On Thu, 8 Jan 2015 15:11:49 +0100, Jakub Jelinek ja...@redhat.com wrote:
On Thu, Nov 20, 2014 at 01:27:08PM +0100, Bernd Schmidt wrote:
On 11/13/2014 05:06 AM, Jan Hubicka wrote:
this patch adds infrastructure for proper streaming and merging of
TREE_TARGET_OPTION.
This breaks the
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
On Fri, Jan 09, 2015 at 11:15:14AM +0100, Richard Biener wrote:
On Fri, 9 Jan 2015, Jakub Jelinek wrote:
On Fri, Jan 09, 2015 at 10:36:09AM +0100, Richard Biener wrote:
I wonder why post_order_compute calls tidy_fallthru_edges at all - won't
This fixes the specific case of complex arithmetic vectorization
in the PR which is caused by loads/stores of complex types which
the vectorizer does not like.
The patch implements two things, first a late variant of
gimplify_modify_expr_complex_part in update-address-taken
when we can write the
1 - 100 of 115 matches
Mail list logo
