Il 10/04/2013 21:33, Steven Bosscher ha scritto:
Hello,
df_find_def and df_find_use do not work properly for hard registers
because rtx_equal_p returns false for the case where
REGNO(x)==REGNO(y) but the modes are different. This happened as
follows in my case:
Breakpoint 9, df_reg_used
On Wed, Apr 10, 2013 at 06:55:10PM +0530, Senthil Kumar Selvaraj wrote:
--- gcc/opts.c
+++ gcc/opts.c
@@ -1699,6 +1699,18 @@ common_handle_option (struct gcc_options *opts,
set_debug_level (SDB_DEBUG, false, arg, opts, opts_set, loc);
break;
+case OPT_gdwarf:
+ if
This is yet another bug just like PR target/52610, we need to pass
-Av8 to the assembler for every cpu type that will use MASK_V8 in
sparc.c:sparc_option_override().
What on Earth are you doing with -mcpu=supersparc and -mcpu=hypersparc? :-)
I found this while testing GMP builds.
I'd like
Ping ^ 2
-Original Message-
From: Joey Ye
Sent: Saturday, January 05, 2013 3:41 PM
To: Ramana Radhakrishnan
Cc: Joey Ye; gcc-patches@gcc.gnu.org
Subject: RE: [PATCH][ARM][thumb1] Reduce lr save for leaf function with
non-far jump
Ping
-Original Message-
From: Joey
One leftover problem: The patch currently fails on the auto_char_len_4
test case, which is not being rejected any more. Actually I'm not
fully convinced that the dg-errors there are correct: If the EXTERNAL
statements in auto_char_len_{1,2} do not trigger an explicit
interface required
On Thu, Apr 11, 2013 at 3:54 AM, Gabriel Dos Reis
g...@integrable-solutions.net wrote:
On Wed, Apr 10, 2013 at 1:42 PM, Manuel López-Ibáñez
lopeziba...@gmail.com wrote:
On 9 April 2013 15:21, Jakub Jelinek ja...@redhat.com wrote:
white). The default is still -fdiagnostics-color=never, can be
2013/4/11 Tobias Burnus bur...@net-b.de:
http://gcc.gnu.org/gcc-4.9/changes.html was so empty :-)
I have committed the attached patch - comments and suggestions are welcome.
Tobias
Hi, Tobias,
On behalf of Andes Technology Co.,
I am going to propose a new target contribution, nds32 port,
Minor patch update:
- Changed FAILURE to false due to Janne's patch
- Removed a left-over #if 0 debug code
Tobias Burnus wrote:
Many compilers have some pragma or directive to disable the type, kind
and rank (TKR) checks. That feature matches C's void* pointer and
can be used in conjunction
Hello,
Thanks for the quick ok. Yes, I ran into this with out of tree patches
(my new delay slot scheduling pass, based on cfg, DF, and sched-deps).
The problem only happens with hard regs and I think that of the
existing users of these functions only the mips.c ones maybe are
buggy.
Ciao!
On 12/20/12 09:53, Joey Ye wrote:
Current GCC thumb1 has an annoying problem that always assuming far branch.
So it forces to save lr, even when unnecessarily. The most extreme case
complained by partner is:
// compiled with -mthumb -mcpu=cortex-m0 -Os.
void foo() { for (;;); }
=
foo:
On Thu, Apr 11, 2013 at 09:08:08AM +0200, Jakub Jelinek wrote:
On Wed, Apr 10, 2013 at 06:55:10PM +0530, Senthil Kumar Selvaraj wrote:
--- gcc/opts.c
+++ gcc/opts.c
@@ -1699,6 +1699,18 @@ common_handle_option (struct gcc_options *opts,
set_debug_level (SDB_DEBUG, false, arg, opts,
Hi Chung-Wu,
Chung-Ju Wu wrote:
On behalf of Andes Technology Co., I am going to propose a new target
contribution, nds32 port, to GCC on
June or July. If the contribution is adopted by the GCC Steering Committee and
community,
is following content OK for describing the new target 'nds32'? :)
On Thu, Apr 11, 2013 at 02:32:51PM +0530, Senthil Kumar Selvaraj wrote:
gcc/ChangeLog
2013-04-11 Senthil Kumar Selvaraj senthil_kumar.selva...@atmel.com
* gcc/opts.c : Fix formatting and add comment.
The ChangeLog entry should have been
* opts.c (common_handle_option): Fix
Hi,
a rather straightforward SFINAE issue. Isn't a regression but should be
safe for the branch too.
Tested x86_64-linux. Ok?
Thanks,
Paolo.
/cp
2013-04-11 Paolo Carlini paolo.carl...@oracle.com
PR c++/56913
* typeck2.c (build_m_component_ref):
This patch adds a parameter ? to the -gnatW message for better
consistency with other messages
Compiling:
1. procedure DummyW is
2. begin
3.null;
4. end;
with -gnath yields an output file contaning the line:
-gnatW? Wide character encoding method (?=h/u/s/e/8/b)
This patch improves the handling of an aggregate like (others = 'A')
where the bounds are known, and the effect is almost like a string
literal except that it is not static.
The following test:
1. package NonSOthers2 is
2.B : constant String (1 .. 6) := (others = 'A');
3.
This patch improves the messages given to explain why expressions
are not static (when they are required to be). There are two changes.
First such messages are now continuations, which means they work
nicely with -gnatj. Second if a string literal comes from aggregates
that are never static, the
Hi,
The same motivation as for:
http://gcc.gnu.org/ml/gcc-cvs/2013-03/msg00786.html
Since -fpic option is turned on by default in Android we have certain test
fails. The reason for that is that those tests rely on the
availability of functions, defined in them
and with -fpic compiler
When invoked with an existing project file, gnatname now creates a backup
copy of the project file before modifying it, unless it is invoked with
the new switch --no-backup.
Tested on x86_64-pc-linux-gnu, committed on trunk
2013-04-11 Vincent Celier cel...@adacore.com
* gnatname.adb
Another important change was done in this commit, I've updated the changelog
accordingly:
(Ada_Version_Default): Switch to Ada 2012 by default.
Since Ada 2012 is the new Ada standard, GNAT is switched to Ada 2012 instead of
Ada 2005 by default.
Index: opt.ads
A subtype is created from the renamed object in an object renaming declaration
with an unconstrained nominal subtype. That subtype must be frozen ahead of
the renaming declaration, to prevent order of elaboration issues when the
nominal subtype of the object has unknown discriminants. The renamed
2013/4/11 Tobias Burnus bur...@net-b.de:
Hi Chung-Wu,
Chung-Ju Wu wrote:
On behalf of Andes Technology Co., I am going to propose a new target
contribution, nds32 port, to GCC on
June or July. If the contribution is adopted by the GCC Steering Committee
and community,
is following
On 03/19/2013 04:37 PM, Richard Biener wrote:
On Tue, 19 Mar 2013, Richard Biener wrote:
Which shows that I need to merge the IRA and reload/lra passes.
Honza tells me that they are considered separate has historical
reasons only. Given that reload pushes TV_IRA and that the boundary
isn't
On Thu, 11 Apr 2013, Bernd Schmidt wrote:
On 03/19/2013 04:37 PM, Richard Biener wrote:
On Tue, 19 Mar 2013, Richard Biener wrote:
Which shows that I need to merge the IRA and reload/lra passes.
Honza tells me that they are considered separate has historical
reasons only. Given that
Hi,
This patch adds the new -gdwarf option to dg-options for testcases in
gcc.dg/debug/dwarf2 that don't already explicitly request DWARF.
Not asking gcc to generate DWARF causes these tests to fail if gcc is
built/configured with support for multiple debugging formats and DWARF
is not the
On 04/11/2013 12:05 PM, Richard Biener wrote:
Any particular suggestions? The easiest split is to emit
a ' reload/lra =' marker (or similar).
It's about getting a dump of the insns, in two separate files so that
one can do diffs (between sched1/ira or ira/reload).
This patch fixes a number of problems that arose from the handling
of assertions (more generally checks from pragma Checks). In
particular if a check was explicitly disabled with Check_Policy
then the argument was not analyzed leading to junk warnings.
The following should compile quietly with
On Thu, 11 Apr 2013, Bernd Schmidt wrote:
On 04/11/2013 12:05 PM, Richard Biener wrote:
Any particular suggestions? The easiest split is to emit
a ' reload/lra =' marker (or similar).
It's about getting a dump of the insns, in two separate files so that
one
On 04/11/2013 12:26 PM, Richard Biener wrote:
The point was not to remove TODO_ggc_collect, the point was to
both enforce the GC boundary at all passes (that's what the pass
manager assumes anyway) and to collect when necessary at such
well-defined boundaries. Another approach would have
On Thu, Apr 11, 2013 at 12:26:00PM +0200, Richard Biener wrote:
So, the option I didn't mention is to add a way to disable the
GC (clearing/setting a property for example).
Yeah, TODO_ggc_avoid_collect or similar would be easiest, plus
perhaps some way of expressing the requirement that this
This patch implements Ada 2012 AI (AI12-0022) which intorduces a new
expression type raiae expression. This can appear within an expression
and results in raising the given exception. Most importantly, if this
form is used in an assertion such as a precondition, it can be used to
change the
On Thu, 11 Apr 2013, Bernd Schmidt wrote:
On 04/11/2013 12:26 PM, Richard Biener wrote:
The point was not to remove TODO_ggc_collect, the point was to
both enforce the GC boundary at all passes (that's what the pass
manager assumes anyway) and to collect when necessary at such
This patch implements Ada 2012 AI-0022, which specifies that a raise
expression that is executed in a predicate that is tested during the
execution of a membership test causes the test to fail (or succeed
for NOT IN), rather than raising an exception.
The following, compiled with -gnata:
1.
On Thu, 11 Apr 2013, Jakub Jelinek wrote:
On Thu, Apr 11, 2013 at 12:26:00PM +0200, Richard Biener wrote:
So, the option I didn't mention is to add a way to disable the
GC (clearing/setting a property for example).
Yeah, TODO_ggc_avoid_collect or similar would be easiest, plus
perhaps
Hi Yvan, My understanding of the 'Free for all' section of
http://gcc.gnu.org/svnwrite.html#policies is that since this is a
backport from an upstream project you do not need to seek further
approval to commit this change.
Cheers
/Marcus
On 2 April 2013 11:50, Yvan Roux yvan.r...@linaro.org
The testcase in PR56878 and my previous changes to the vectorizer
expose that there is amongst the DRs with unknown misalignment
with the same number of related accesses a better candidate if
the DRs base address is invariant in some loops. This is because
then you can hoist the misalign
Hi Marcus,
thanks for pointing this, but unfortunately I don't have the svn write
access for the moment, thus
if somebody can commit it for me, i would be very thankful.
Cheers,
Yvan
On 11 April 2013 13:02, Marcus Shawcroft marcus.shawcr...@gmail.com wrote:
Hi Yvan, My understanding of the
There's some interest inside Oracle to support compressed debug sections
inside their toolchain, both on Solaris and Linux. So far, there's the
GNU style supported by gas, gld, gold, and gdb, which mangles section
names (.debug_* - .zdebug_*), but consultation with the Solaris linker
engineers
Rainer Orth r...@cebitec.uni-bielefeld.de writes:
There's some interest inside Oracle to support compressed debug sections
inside their toolchain, both on Solaris and Linux. So far, there's the
GNU style supported by gas, gld, gold, and gdb, which mangles section
names (.debug_* -
Hi,
Usually stdlib.h does not include string.h but on bionic it is
historically included. memcmp() reacts on a volatile argument
differently, depending on whether string.h is included or not. If it
is included, then the compiler will generate a warning:
warning: passing argument 2 of 'memcmp'
On 5 April 2013 11:48, Bernhard Reutner-Fischer rep.dot@gmail.com wrote:
On 5 April 2013 11:23, Gabriel Dos Reis g...@integrable-solutions.net wrote:
On Fri, Apr 5, 2013 at 4:13 AM, Rainer Orth r...@cebitec.uni-bielefeld.de
wrote:
Gabriel Dos Reis g...@integrable-solutions.net writes:
On 04/11/13 00:32, Jakub Jelinek wrote:
On Wed, Apr 10, 2013 at 05:16:17PM -0700, Andrew Pinski wrote:
On Wed, Apr 10, 2013 at 3:24 PM, Aldy Hernandez al...@redhat.com wrote:
Hi Tom. Hi folks.
We've asked Balaji to rewrite the #pragma simd handling for cilkplus as we
currently do for OMP,
Hi,
On 04/11/2013 02:04 PM, Bernhard Reutner-Fischer wrote:
I would have expected that somebody would tell me that omitting
::tmpnam violates 27.9.2 cstdio from the spec but noone yelled at me
yet?
Frankly, I didn't because the targets I really care about aren't
affected. The actual
Hi,
On Tue, Sep 4, 2012 at 12:33 AM, Andrew Pinski
andrew.pin...@caviumnetworks.com wrote:
Hi,
On MIPS it is better sometimes not to use the word mode (DImode) but
rather SImode. The main reason is for 32bits, the value has to be
sign extended to 64bits so using the 32bit instruction for
Andi Kleen a...@firstfloor.org writes:
Rainer Orth r...@cebitec.uni-bielefeld.de writes:
There's some interest inside Oracle to support compressed debug sections
inside their toolchain, both on Solaris and Linux. So far, there's the
GNU style supported by gas, gld, gold, and gdb, which
The routine Safe_To_Capture_Values performs simple data-flow to determine
whether some characteristic of a variable (its value after assignment, or the
fact that an access value is non-null after a test) can be used downstream.
The value cannot be saved if the expression that determines that value
128 is about upper bound you can expand with sse moves.
Tuning did not take into account code size and measured only when code
is in tigth loop.
For GPR-moves limit is around 64.
Thanks for the data - I've not performed measurements with this
implementation yet, but we surely should adjust
This patch provides the initial implementation of aspect Depends. This
construct is intended for formal verification proofs.
The syntax of the aspect is as follows:
DEPENDENCY_RELATION ::= null
| DEPENDENCY_CLAUSE {, DEPENDENCY_CLAUSE}
DEPENDENCY_CLAUSE ::=
On Thu, Apr 11, 2013 at 07:10:46AM -0500, Aldy Hernandez wrote:
Yeah, the above is definitely wrong. Just
if (flag_cilkplus)
cpp_register_deferred_pragma (parse_in, NULL, simd, PRAGMA_SIMD,
true, false);
Well, the thing is that you can't just use NULL for #pragma simd
as a
The Ada 2012 RM introduces the notion of an incomplete view of an ancestor
type: in a child unit, a derived type is within the derivation class of an
ancestor declared in a parent unit, even if there is an intermediate derivation
that does not see the full view of that ancestor. This makes some
This change replaces a call to a thick socket operation binding that uses
an exception to report an expected condition (return with EINPROGRESS
error) with a thin call reporting the same information through a return
value and errno setting, to avoid an unnecessary exception raise.
Tested on
OK.
Jason
On Thu, Apr 11, 2013 at 1:58 PM, Alexander Ivchenko aivch...@gmail.com wrote:
Hi,
Usually stdlib.h does not include string.h but on bionic it is
historically included. memcmp() reacts on a volatile argument
differently, depending on whether string.h is included or not. If it
is included,
In Ada 2012, aspect specifications can appear on subprogram body stubs, as long
as there is no previous corresponding subprogram declaration. This patch
handles properly aspects on such stubs, and rejects aspects when a previous
subprogram declaration exists.
executing the following:
gnatmake
Instead of creating then deleting multiple times the same temporary file,
gnatname now creates then deletes temporary files with distinct names.
In addition, if environment variable TMPDIR exists, the temporary files
are created in this directory.
No change in gnatname behavior, so no test.
On 04/11/13 07:36, Jakub Jelinek wrote:
On Thu, Apr 11, 2013 at 07:10:46AM -0500, Aldy Hernandez wrote:
Yeah, the above is definitely wrong. Just
if (flag_cilkplus)
cpp_register_deferred_pragma (parse_in, NULL, simd, PRAGMA_SIMD, true,
false);
Well, the thing is that you can't just
Ok, here is an updated patch, which does the discussed checking for
procedure pointer assignments. For this I have introduced a new
function 'gfc_explicit_interface_required', which checks all the items
in F08:12.4.2.2 and is loosely based on the present checks in
'resolve_global_procedure'
Currently for initial values for inductions the vectorizer generates
stmp_var_.3_2 = 0 + 1;
stmp_var_.3_19 = stmp_var_.3_2 + 1;
stmp_var_.3_20 = stmp_var_.3_19 + 1;
vect_cst_.4_21 = {0, stmp_var_.3_2, stmp_var_.3_19, stmp_var_.3_20};
instead of properly folding elements and eventually
Hi Paolo,
About 3 years ago I've sent a patch which was submitted by Kenneth
Zadeck on revision 153924 (See
http://gcc.gnu.org/ml/gcc-patches/2009-11/msg00232.html)
Recently we tried to update our gcc port from gcc-4.4 to gcc-4.8 and
discovered that the same lines of code were changed, thus
The flag Do_Discriminant_Check was set during semantic analysis only when
expansion was also performed. Now set it unconditionally when needed. Also
clarify in the associated documentation that it is set also for Unchecked_Union
but not expanded into an actual check in that case.
Tested on
This change fixes an oversight in the predicate that determines whether two
composite types have the same representation (difference of scalar storage
order was ignored, causing the compiler to incorrectly omit a per-component
assignment in this case).
The following test case must compile and
Il 11/04/2013 14:57, Amir Gonnen ha scritto:
Hi Paolo,
About 3 years ago I've sent a patch which was submitted by Kenneth
Zadeck on revision 153924 (See
http://gcc.gnu.org/ml/gcc-patches/2009-11/msg00232.html)
Recently we tried to update our gcc port from gcc-4.4 to gcc-4.8 and
This patch corrects the implementation of static predicates in the context of
case expressions and/or statements. The maching of individual case alternatives
is now done against the sets of legal values defined by the static predicate.
-- Source --
--
This bug has accumulated a lot of duplicates over the years. The issue
is that the standard says that
Implicit conversions (Clause 4) will be performed on a function
argument to convert it to the type of the corresponding function
parameter if the parameter type contains no
Here it is (also copied back the comment from the original patch):
@@ -3207,9 +3207,14 @@
DF_REF_REG_USE, bb, insn_info, flags);
df_uses_record (collection_rec, XEXP (dst, 2),
DF_REF_REG_USE, bb, insn_info, flags);
+
+ /* Handle the case of zero_extract(mem(...)) in the set dest.
+ This
This patch allows classwide postconditions on null procedures when the
aspect expression includes calls to abstract functions that must be rewrittne
as dispatching calls.
The following must compile and execute quietly:
gnatmake -gnat12 -gnata -q test_class
test_class
--
with objects; use
This patch modifies the analysis of aspects Depends and Global. The machinery
can now process renamings of entire objects. Legal renamings are replaced by
the object they rename.
-- Source --
-- renamings.ads
package Renamings
with Abstract_State = (Input_State with
-Original Message-
From: Zhenqiang Chen [mailto:zhenqiang.c...@linaro.org]
Sent: 11 April 2013 02:33
To: Marcus Shawcroft
Cc: James Greenhalgh; gcc-patches@gcc.gnu.org; Marcus Shawcroft;
pins...@gmail.com; Richard Earnshaw
Subject: Re: [PATCH, AARCH64] Fix unrecognizable insn issue
A new environment variable GPR_PROJECT_PATH_FILE is used by the Project
Manager to set up the project path. When defined, GPR_PROJECT_PATH_FILE
indicates the path of a text file that contains directories to be added
to the project path. GPR_PROJECT_PATH_FILE is taken into account before
This patch fixes a missing check on the priority of a protected type, when
it is set by a pragma or an aspect. A static value for priority that is out
of the bounds of the corresponding System type is an error. A dynamic value
that is out of bounds must raise a runtime exception.
Compiling t.adb
Hi,
the GCC 4.8.0 release unfortunately does not bootstrap on S/390. It
fails with a bootstrap comparison failure.
The problem is an RTL sharing bug created by ifcvt. I never noticed it
because the problem does *not* occur with --enable-checking which I'm
usually using on my daily build system.
Hi,
enable-checking sometimes hides RTL checking bugs instead of detecting
them.
The problem is that the `used' flag in the RTXs used in
verify_rtl_sharing is not reset after doing the checks. But the flag
is also used by other passes in order to fix RTX sharing (e.g. ifcvt).
These passes then
Thanks!
Paolo.
On 11/04/13 14:33, James Greenhalgh wrote:
-Original Message-
From: Zhenqiang Chen [mailto:zhenqiang.c...@linaro.org]
Sent: 11 April 2013 02:33
To: Marcus Shawcroft
Cc: James Greenhalgh; gcc-patches@gcc.gnu.org; Marcus Shawcroft;
pins...@gmail.com; Richard Earnshaw
Subject: Re: [PATCH,
On 11 April 2013 03:54, Gabriel Dos Reis g...@integrable-solutions.net wrote:
As I stated before, our pursuit of enabling everything new thing by default
may have made C++ diagnostics more terrifying.
Do you have examples of C++ diagnostics that are more terrifying now
than in, say, GCC 4.2?
-Original Message-
From: gcc-patches-ow...@gcc.gnu.org [mailto:gcc-patches-
ow...@gcc.gnu.org] On Behalf Of Aldy Hernandez
Sent: Thursday, April 11, 2013 8:55 AM
To: Jakub Jelinek
Cc: Andrew Pinski; Tom Tromey; Iyer, Balaji V; gcc-patches
Subject: Re: libcpp: registering both a
[Btw, I also thought about doing a full gfc_compare_interfaces in
resolve_global_procedure, but that would probably be too strict.]
Comment to self: It's certainly more strict, but I think this is a
good thing. I just tried it and in fact it catches a number of cases
in the testsuite, which
Il 11/04/2013 15:23, Amir Gonnen ha scritto:
Here it is (also copied back the comment from the original patch):
@@ -3207,9 +3207,14 @@
DF_REF_REG_USE, bb, insn_info, flags);
df_uses_record (collection_rec, XEXP (dst, 2),
DF_REF_REG_USE, bb, insn_info, flags);
+
+ /* Handle the case
On Thu, Apr 11, 2013 at 03:44:56PM +0200, Andreas Krebbel wrote:
2013-04-11 Andreas Krebbel andreas.kreb...@de.ibm.com
* ifcvt.c (end_ifcvt_sequence): Mark a and b for unsharing as
well.
Ok, thanks.
*** gcc/ifcvt.c.orig
--- gcc/ifcvt.c
*** end_ifcvt_sequence
Hi,
while working on the last patch I noticed that we have got quite a few
unconditional error calls in functions taking a tusbst_flags_t
parameter. The below mostly adjusts those.
I'm also finding a bit misleading the name warn_ref_binding for a the
function emitting a permerror, thus I'm
On Thu, Apr 11, 2013 at 12:55 AM, Jakub Jelinek ja...@redhat.com wrote:
On Wed, Apr 10, 2013 at 09:04:06PM -0500, Gabriel Dos Reis wrote:
We might be saying the same thing using different languages.
I was the %r/%R markers are ways of implementing the IL language
I suggested in that message.
As noted in PR56900, the new vrp87 test does not work on some targets as
the code ultimately fed to vrp is dependent on BRANCH_COST.
I scanned all the backends for those which use the default BRANCH_COST,
which is known to perturb the code in a way that VRP can't see the
optimization
OK.
Jason
Naveen, I've committed your original REVERSIBLE_CC_MODE patch.
Cheers
/Marcus
On 8 January 2013 05:09, Hurugalawadi, Naveen
naveen.hurugalaw...@caviumnetworks.com wrote:
Hi,
AARCH64 back-end defines REVERSE_CONDITION but not REVERSIBLE_CC_MODE.
REVERSE_CONDITION is used only in case
On Thu, Apr 11, 2013 at 10:20:18AM -0500, Gabriel Dos Reis wrote:
On Thu, Apr 11, 2013 at 12:55 AM, Jakub Jelinek ja...@redhat.com wrote:
On Wed, Apr 10, 2013 at 09:04:06PM -0500, Gabriel Dos Reis wrote:
We might be saying the same thing using different languages.
I was the %r/%R markers
Am 11.04.2013 16:23, schrieb Janus Weil:
[Btw, I also thought about doing a full gfc_compare_interfaces in
resolve_global_procedure, but that would probably be too strict.]
Comment to self: It's certainly more strict, but I think this is a
good thing. I just tried it and in fact it catches a
56901 turns out to be related to an issue that has been bugging me for a
while; we would hand off instantiation of a use of a variable to
tsubst_decl, which might end up producing a whole new instantiation of
the variable rather than look up the one that we (presumably) already
have. 56901 is
This fixes a link failure for an array object declared with an array of array
type under very specific circumstances, as the compiler doesn't consistently
consider that the size of the array type is constant, depending on the unit
which is being compiled.
I need to export the
Hello,
here is a patch to handle constant folding of mixed vector-integer
operations. I could have shared the loop with the vector-vector case, but
that would have meant re-testing if arg2 was a vector at every iteration
(I can go back to that version if you prefer).
bootstrap+testsuite on
My earlier patch to have lambda_capture_field_type and lambda_proxy_type
just return the captured variable's type in more cases caused this
regression; we were failing to recognize the case of an auto variable
captured by reference. Fixed by stripping references.
Tested x86_64-pc-linux-gnu,
On Wed, Apr 10, 2013 at 10:15 PM, Vladimir Makarov vmaka...@redhat.com wrote:
The following patch fixes
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56903
In this test case reload pass gets correct value HARD_REGNO_MODE_OK because
it can not create pseudos (can_create_pseudo) and this was
On Thu, Apr 11, 2013 at 06:24:02PM +0200, Marc Glisse wrote:
here is a patch to handle constant folding of mixed vector-integer
operations. I could have shared the loop with the vector-vector
case, but that would have meant re-testing if arg2 was a vector at
every iteration (I can go back to
On Thu, 11 Apr 2013, Jakub Jelinek wrote:
On Thu, Apr 11, 2013 at 06:24:02PM +0200, Marc Glisse wrote:
here is a patch to handle constant folding of mixed vector-integer
operations. I could have shared the loop with the vector-vector
case, but that would have meant re-testing if arg2 was a
Here is a patch to enable LRA for rs6000. The patch includes code
changes in rs6000 machine-dependent parts and in LRA parts.
I've added a new switch -mlra for rs6000 to make debugging LRA for
rs6000 easier but not documented it as it will be gone at the end of
stage1 (may be with
Hi!
Hopefully last problematic spot, other places where maybe_constant_value
calls have been introduced are either guarded with
!processing_template_decl, or I haven't succeeded in triggering ICEs (say in
build_new_1).
Bootstrapped/regtested on x86_64-linux and i686-linux, ok for trunk?
Hi!
The op1 - op2 line instead of multiplication is something that has been
introduced in C++ double_int changes, but without the 2 * TYPE_PRECISION
prec it would always return 0 or -1 anyway. And, with wide_mul_with_sign
we can also handle 2 * HOST_BITS_PER_WIDE_INT precisions.
On Thu, Apr 11, 2013 at 1:30 PM, Vladimir Makarov vmaka...@redhat.com wrote:
Here is a patch to enable LRA for rs6000. The patch includes code changes
in rs6000 machine-dependent parts and in LRA parts.
I've added a new switch -mlra for rs6000 to make debugging LRA for rs6000
easier but
Hi,
*mmintrin headers does not work with function specific opts.
Example 1:
#include smmintrin.h
__attribute__((target(sse4.1)))
__m128i foo(__m128i *V)
{
return _mm_stream_load_si128(V);
}
$ g++ test.cc
smmintrin.h:31:3: error: #error SSE4.1 instruction set not enabled
# error
On Thu, 11 Apr 2013, Jakub Jelinek wrote:
The op1 - op2 line instead of multiplication is something that has been
introduced in C++ double_int changes, but without the 2 * TYPE_PRECISION
prec it would always return 0 or -1 anyway.
The comments on PREC in the rshift function really confused
OK.
Jason
1 - 100 of 121 matches
Mail list logo
