> >
> > No, we shouldn't arrive with a NULL use_stmt here.
>
> I think a proper fix will be to fail if there are no uses.
> I'll prepare a patch on Sunday.
>
Here is the patch. It bails out if LHS has no uses.
Bootstrapped and tested on powerpc64-suse-linux.
Committed.
Ira
ChangeLog:
Due to the inliner change (in google/main), a couple of test cases under
gcc.dg/matrix starts to fail (ICE in matrix reorg). The problem can be
reproduced in gcc46 with increases inline limit (but not in trunk -- the
problem is either fixed or inline behavior is still different). The following
work
> Fixed thusly, bootstrapped/regtested on x86_64-linux and i686-linux
> and tested with a cross to arm-linux on the testcase, ok for trunk/4.6?
>
> 2011-05-18 Jakub Jelinek
>
> PR middle-end/49029
> * expmed.c (extract_fixed_bit_field): Test whether target can be used
> only af
On Sat, May 21, 2011 at 6:33 PM, Joseph S. Myers
wrote:
> On Sat, 21 May 2011, Nicola Pero wrote:
>
>> As a consequence, GTFILES is different depending on whether
>> ${srcdir}/gcc/cp/ exists or not. If it exists, usually the cp gtfiles
>> come before the objc ones (due to alphabetical ordering by
Hi Steve,
* frontend-passes.c (cfe_register_funcs): Also register
character functions if their charlens are known and constant.
Also register allocatable functions.
2011-05-18 Thomas Koenig
* gfortran.dg/function_optimize_8.f90: New test case.
OK.
Wai
On Fri, 20 May 2011, Basile Starynkevitch wrote:
> By the way, I am quite happy of Pierre patches to the MELT branch. Is
> this enough to get him a write access to GCC SVN (all legalese is done)?
> I fear that to really get that write access, Pierre would also need to
> have accepted patches to
On Fri, 13 May 2011, Bernd Schmidt wrote:
> Patch appended.
Just raised on the steering committee.
Gerald
Hi Guys,
I am checking in the patch below to fix a couple of bugs in the
xstormy16_init_builtins function. The first was that the code was
initialising the args[2] array element, which does not exist. The
second was that the arguments are constructed in the "args" array, but
the build_
On Sun, 22 May 2011 15:24:30 +0200 (CEST)
Gerald Pfeifer wrote:
> On Fri, 20 May 2011, Basile Starynkevitch wrote:
> > By the way, I am quite happy of Pierre patches to the MELT branch. Is
> > this enough to get him a write access to GCC SVN (all legalese is done)?
> > I fear that to really get
Hello world,
the attached patch fixes PR 48955, a wrong-code regression for 4.6 and
4.7, including the test case from
http://gcc.gnu.org/ml/fortran/2011-05/msg00093.html . It follows the
outline from http://gcc.gnu.org/bugzilla/show_bug.cgi?id=48955#c7 .
Regression-tested. OK for trunk?
The following applies to -fcoarray=lib; for -fcoarray=single there
should be no change.
The coarray communication library needs to know about the coarrays even
before the function containing them has been invoked. Thus, the coarrays
(of all translation units) need to be registered at start up.
On 05/21/2011 09:40 AM, Eric Botcazou wrote:
>> 2011-05-05 Tom de Vries
>>
>> PR target/45098
>> * tree-ssa-loop-ivopts.c: Include expmed.h.
>> (get_shiftadd_cost): New function.
>> (force_expr_to_var_cost): Declare forward. Use get_shiftadd_cost.
>
> This breaks the Ada co
Hi,
I also ran into this assert:
...
:0:0: internal compiler error: tree check: expected tree that
contains 'common' structure, have 'integer_cst' in fold_checksum_tree, at
fold-const.c:13798
...
With a recent change in r172386, using a TREE_CHAIN on an INTEGER_CST has become
illegal:
...
This patch adds the rank intrinsic of the TR 29113 draft. It is a bit
boring until gfortran supports DIMENSION(..) ;-)
Build and regtested on x86-64-linux.
OK for the trunk?
Tobias
PS: The current draft of TR 29113 is available at
http://j3-fortran.org/pipermail/interop-tr/2011-May/000710.htm
On Sun, May 22, 2011 at 6:26 PM, Tom de Vries wrote:
> Hi,
>
> I also ran into this assert:
> ...
> :0:0: internal compiler error: tree check: expected tree that
> contains 'common' structure, have 'integer_cst' in fold_checksum_tree, at
> fold-const.c:13798
> ...
>
> With a recent change in r1723
On Sun, May 22, 2011 at 6:06 PM, Tom de Vries wrote:
> On 05/21/2011 09:40 AM, Eric Botcazou wrote:
>>> 2011-05-05 Tom de Vries
>>>
>>> PR target/45098
>>> * tree-ssa-loop-ivopts.c: Include expmed.h.
>>> (get_shiftadd_cost): New function.
>>> (force_expr_to_var_cost): Declar
Hi,
Now that struct tree_type does not exist anymore, 'sizeof (struct tree_type)'
generates an error in the following assert in fold_checksum_tree:
...
gcc_assert ((sizeof (struct tree_exp) + 5 * sizeof (tree)
<= sizeof (struct tree_function_decl))
&& sizeof (struct
The SFINAE code in composite_pointer_type_r correctly avoids giving an
error in SFINAE context, but then it just uses the permissive fallback
without indicating to the caller that there was an error, so we get
wrong results. This patch fixes it to return error_mark_node when we
see an error.
Dear Thomas,
I am sorry for the long silence on this PR. I have been up to my
eyeballs in daytime work.
I have posted a simpler alternative on the PR that uses your
suggestion that forward and backward dependences need to to be
recorded to get this right.
I believe that it's OK but have only no
> I didn't manage to reproduce the breakage, but I think this patch will fix
> it.
You need a 32-bit host.
> The patch makes sure cst_and_fits_in_hwi is tested before using
> int_cst_value.
This should be sufficient indeed, thanks.
> 2011-05-22 Tom de Vries
>
> PR target/45098
>
Hello!
The problem is in optimized search_line_mmx function in libcpp/lex.c.
This function is compiled for SSE target, so cmove instructions are
automatically enabled. In addition, "pmovmskb" instruction is part of
"AMD extensions to MMX" instruction set, marked by bit 22 in register
%edx from cp
On Thu, Mar 18, 2010 at 01:50:54PM -0700, Janis Johnson wrote:
> The ISO C++ standard says, in Section 3.3.2 sentence 4, that a name
> declared in the for-init-statement or in the condition of an if, for
> while, or switch statement can't be redeclared in the outermost block
> of the controlled sta
Committed as Rev. 174039
Tobias
Index: libgfortran/runtime/stop.c
===
--- libgfortran/runtime/stop.c (Revision 174038)
+++ libgfortran/runtime/stop.c (Arbeitskopie)
@@ -67,8 +67,7 @@ stop_string (const char *string, GFC_INTEGER_4 len
This re-enables the compilation of crti.o crtn.o for sparc-elf and sparc-rtems
that was disabled in 4.6.0 but turned out to be disturbing. This also enables
it for sparc64-elf and sparc64-rtems and removes obsolete EXTRA_MULTILIB_PARTS
settings in a couple of files.
Bootstrapped/regtested on S
In preparation for another change.
Bootstrapped/regtested on SPARC/Solaris and SPARC64/Solaris, applied on the
mainline.
2011-05-22 Eric Botcazou
* config/sparc/sparc.c (sparc_delegitimize_address): Handle
UNSPEC_MOVE_PIC pattern.
--
Eric Botcazou
Index: config/sparc/spar
Dear Janne,
Sorry I'm a bit late on this, but since async-signal-safe code is so hard to
get right (at least for me), I wanted to help review your new code. It's very
nice to have this instead of my initial ugly implementation, and I have only
spotted on issue: AFAICT, execvp() is not safe to u
Hi,
after http://gcc.gnu.org/ml/gcc-patches/2011-04/msg02139.html profiling uses
TLS support where available. This breaks on SPARC/Solaris 8 and 9 (at least)
because the associated libraries aren't linked in.
Tested on SPARC/Solaris 8 and SPARC64/Solaris 9 (the Solaris 10 machine isn't
in ver
var-tracking doesn't fully work with dynamic stack realignment on x86 when a
DRAP register is used for a function: the value of the incoming parameters is
very quickly wrong (as soon as the DRAP register is clobbered), in particular
in backtraces. The reason is that var-tracking doesn't track pa
On Sat, 14 May 2011, John David Anglin wrote:
> > OK, thanks. Note that another delay slot bug was fixed around the same
> > time:
> >
> > 2010-09-20 Eric Botcazou
> >
> > PR rtl-optimization/42775
> > * cfgrtl.c (rest_of_pass_free_cfg): Recompute notes if delay slot
> > schedul
I didn't get this reviewed previously so I'll try again for 4.7 ...
-- Forwarded message --
Here's my attempt to fix PR c++/18016 so that we get a warning from
struct S {
int i;
S() : i(i) { }
};
As I said in the audit trail, -Winit-self is broken for C++ (PR
c++/34772) so I
In PR 44826, we split a pre-reload call into another call followed
by a load of GP. The problem is that we're running the split pass after
var-tracking, and the original call had a NOTE_INSN_CALL_ARG_LOCATION
attached to it. We need to move the note to the new call, just like we
already move othe
"Joseph S. Myers" writes:
> The processing of MIPS_CPU_STRING_DEFAULT is replaced by much simpler
> use of strcmp to find a matching entry; the "from-abi" default
> definition of that macro is replaced by code in mips_default_arch if
> that macro is not defined. (Previously it would always have b
On Sun, 22 May 2011, Jack Howarth wrote:
>> 2010-03-18 Janis Johnson
>>
>> gcc/cp
>> PR c++/2288
>> PR c++/18770
>> * name-lookup.h (enum scope_kind): Add sk_cond.
>> * name-lookup.c (pushdecl_maybe_friend): Get scope of shadowed local.
>> Detect and report error for re
After processing each outgoing edge of a basic block, try_forward_edges tests
whether the block is a forwarder block. This doesn't make much sense since,
if there are still more than one outgoing edge, then this cannot be a forwarder
block. The attached patch arranges for the test to be done o
Following the discussion here:
http://gcc.gnu.org/ml/gcc/2011-05/msg00216.html
this patch makes -fno-delayed-branch imply -Wa,-O1. It also makes
sure that we explicitly assemble with -O0 by default. (Previously,
we'd assemble with -O0 if the user specified -O0, but assemble with
the default
Richard Sandiford writes:
> this patch makes -fno-delayed-branch imply -Wa,-O1. It also makes
> sure that we explicitly assemble with -O0 by default.
Um, I meant -O1. The point is that -Wa,-O0 vs -Wa,-O1 isn't really an
optimization decision; -O0 just inserts nops that weren't even implied
by t
On Sun, May 22, 2011 at 2:29 PM, Gerald Pfeifer wrote:
> On Sun, 22 May 2011, Jack Howarth wrote:
>>> 2010-03-18 Janis Johnson
>>>
>>> gcc/cp
>>> PR c++/2288
>>> PR c++/18770
>>> * name-lookup.h (enum scope_kind): Add sk_cond.
>>> * name-lookup.c (pushdecl_maybe_friend): Get
Hi, the following trial patch fixed PR 48988 which is a regression.
Bootstrap and tested on x86/linux. Verified the reported failure is fixed.
Ok for trunk?
David
2011-05-22 David Li
PR tree-optimization/48988
* tree-ssa-uninit.c (convert_control_dep_chain_into_preds):
Ping.
David
On Fri, May 20, 2011 at 9:06 AM, Xinliang David Li wrote:
> Ok to check in this one?
>
> Thanks,
>
> David
>
> On Wed, May 18, 2011 at 12:30 PM, Joseph S. Myers
> wrote:
>> On Wed, 18 May 2011, David Li wrote:
>>
>>> + error ("Unrecognized option %s", is_enable ? "-fenable" :
Hi,
see the audit trail for some interesting observations... Tested
x86_64-linux, committed mainline and 4_6-branch.
Thanks,
Paolo.
/
2011-05-22 Jonathan Wakely
* testsuite/20_util/bind/cv_quals_2.cc: New.
2011-05-22 Paolo Carlini
PR libstdc++/4905
On 05/22/2011 05:11 PM, Jonathan Wakely wrote:
As I said in the audit trail, -Winit-self is broken for C++ (PR
c++/34772)
Not anymore. :)
Even if -Winit-self wasn't broken I don't think it applies here,
there's no valid reason to initialize a member with itself, if you
really want to leave a
2011/5/20 Nicola Pero :
> I'd like to show you a testcase, but, as you discuss in your other email, we
> can't
> really write GTY testcases at this stage. So here is a walked tour of how to
> test
> the new atomic option --
[...]
> To test that the GC marking actually works, I mostly bootstrap t
42 matches
Mail list logo
