Hello!
It looks that GP relative relocations do not fit anymore into GPREL16
reloc, so bootstrap on alpha hosts fail in stage2 with relocation
truncated to fit: GPREL16 against I found no other solution but
to pass --no-relax to linker in order to finish the bootstrap.
2011-05-02 Uros
On 04/27/2011 03:22 PM, Ulrich Weigand wrote:
* haifa-sched.c (sched_emit_insn): Emit insn before first
non-scheduled insn. Inform back-end about new insn. Add
new insn to scheduled_insns list.
Seems like this function is only used from spu.c. OK.
Bernd
On Sun, 1 May 2011, Tom G. Christensen wrote:
Latest results for 4.6.x
Thanks, online too, now.
Gerald
David,
2011-05-01 David Li davi...@google.com
* testsuite/lib/profopt.exp (proc): Multiple source file support.
^
This is wrong, the proc is called profopt-execute.
* testsuite/gcc.dg/tree-prof/lipo/lipo.exp (load_lib):
On Fri, Apr 29, 2011 at 6:23 PM, Xinliang David Li davi...@google.com wrote:
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
On Fri, Apr 29, 2011 at 6:43 PM, Jakub Jelinek ja...@redhat.com 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 array,
which is wrong,
On Sun, May 1, 2011 at 9:31 AM, Ira Rosen i...@il.ibm.com wrote:
gcc-patches-ow...@gcc.gnu.org wrote on 20/04/2011 02:24:55 PM:
Hi,
In gcc.dg/vect/slp-3.c and gcc.dg/vect/no-vfa-pr29145.c vectorization is
expected to fail on targets vect_no_align. But no realignment is
necessary
here
Hi,
we have this simple issue where we ICE because we don't check the return
value of build_expr_type_conversion for NULL_TREE, at variance with all
its other uses elsewhere.
The fix seems also simple, we have already logic for emitting an error
message (or just returning error_mark_node
On Mon, 2 May 2011, Simon Martin wrote:
2011-05-01 Simon Martin simar...@users.sourceforge.net
PR c/35445
* c-decl.c (finish_decl): Only create a composite if the types are
compatible.
2011-05-01 Simon Martin simar...@users.sourceforge.net
PR c/35445
*
On Tue, 26 Apr 2011, Rainer Orth wrote:
* doc/sourcebuild.texi (Ada Tests): Adjust reference to ACATS
testsuite and make it version agnostic.
I doubt this is a good idea: while you can find out what version of
ACATS is included in GCC, the information is somewhat hard to find
This patch makes us count individual tree codes instead of globbing
them only.
Bootstrapped on x86_64-unknown-linux-gnu, installed.
Richard.
2011-05-02 Richard Guenther rguent...@suse.de
* tree.c (tree_code_counts): New global array.
(record_node_allocation_statistics):
Hi,
lto_input_1_unsigned and lto_output_1_stream are the most frequently called
functions
during the WPA stage. Obviously these are simple accesors that should be
inline.
Note that even with LTO bootstrap these won't get inlined unless we declare them
inline or use -O3 or FDO.
* doc/sourcebuild.texi (Ada Tests): Adjust reference to ACATS
testsuite and make it version agnostic.
I doubt this is a good idea: while you can find out what version of
ACATS is included in GCC, the information is somewhat hard to find
(gcc/testsuite/ada/acats/support/acats25.lst)
Arnaud Charlet char...@adacore.com writes:
I'll be happy to add the concrete version number if Arnaud prefers.
I'd even go further and clarify that the tests represent a modified subset of
executable tests from the ACATS 2.5 testsuite, to make it clear that these
have no official status and
../../gcc-4.7/gcc/tree-inline.c: In function 'tree_node*
maybe_inline_call_in_expr(tree)':
../../gcc-4.7/gcc/tree-inline.c:5229:40: error: converting 'false' to
pointer type 'void (*)(tree)' [-Werror=conversion-null]
2011-05-02 Dmitry Gorbachev d.g.gorbac...@gmail.com
* tree-inline.c
On Mon, May 2, 2011 at 08:52, Dmitry Gorbachev d.g.gorbac...@gmail.com wrote:
../../gcc-4.7/gcc/tree-inline.c: In function 'tree_node*
maybe_inline_call_in_expr(tree)':
../../gcc-4.7/gcc/tree-inline.c:5229:40: error: converting 'false' to
pointer type 'void (*)(tree)' [-Werror=conversion-null]
This fixes PR48822, an issue we ran into the past as well. The problem
is that due to various reasons (mostly simplification of expressions)
we might be tempted to go up the lattice (from VARYING to some constant)
during value-numbering. This is of course a recipie to oscillation
as can be seen
2011-05-02 Dmitry Gorbachev d.g.gorbac...@gmail.com
* tree-inline.c (maybe_inline_call_in_expr): Replace false by NULL.
Thanks for spotting this. I'd apply it to all relevant branches (4.5/4.6).
--
Eric Botcazou
gfortran.dg/fmt_g0_6.f08 execution seems to fail on all 32-bit x86
targets at -O1 and above. Running it under gdb, I find
Program received signal SIGABRT, Aborted.
[Switching to Thread 1 (LWP 1)]
0xfed0ff85 in _lwp_kill () from /lib/libc.so.1
(gdb) where
#0 0xfed0ff85 in _lwp_kill () from
Uh, there is another one:
../../gcc-4.7/gcc/cp/parser.c: In function 'tree_node*
cp_parser_init_declarator(cp_parser*, cp_decl_specifier_seq*,
VEC_deferred_access_check_gc*, bool, bool, int, bool*, tree_node**)':
../../gcc-4.7/gcc/cp/parser.c:14612:19: error: converting 'false' to
pointer type
Ping? Ping? Ping? Ping?
This is getting ridiculous. Would someone please accept the patch,
tell me what to fix in it to make it acceptable, or explain why it is
a bad idea?
Thanks!
On Mon, Apr 25, 2011 at 9:08 AM, Paul Pluzhnikov ppluzhni...@google.com wrote:
Ping? Ping? Ping?
On Mon, Apr
This is a fix for an optimization flaw when a long value is composed
from byte values.
For -fsplit-wide-types (which is still default for avr) the code is
worse than with -fno-split-wide-types. The code for the test case is
better in either situations, i.e. compared to code without the patch,
but
On Mon, May 2, 2011 at 4:13 PM, Rainer Orth r...@cebitec.uni-bielefeld.de
wrote:
Given that g++.dg/tree-ssa/pr43411.C seems to XPASS everywhere, it seems
time to remove the xfail.
Tested on i386-pc-solaris2.11 with the appropriate runtest invocation,
ok for mainline?
I think the testcase
commit 99835abc5d2a7fd3ae0950c8a16fd6d223d408c9
Author: Ollie Wild a...@google.com
Date: Fri Apr 29 13:03:57 2011 -0400
Remove unqualified lookups into dependent template base classes from
STL headers. These break clang.
See
Prompted by Joseph's message
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg01796.html
I've done some cleanup on mips-tfile.c and mips-tdump.c that are now
only used by Tru64 UNIX.
Apart from the issues mentioned by Joseph, I'm removing Saber C support,
use of private Size_t/Ptrdiff_t,
On Mon, May 2, 2011 at 4:16 PM, Jan Hubicka hubi...@ucw.cz wrote:
Hi,
according to oprofile, libiberty hashing takes about 30% of streaming in time
and according to callgrind, the most busy cache is node_map cache in the
streamer.
This patch turns it into pointer-map that also saves about
On Mon, May 2, 2011 at 4:38 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Mon, May 2, 2011 at 4:16 PM, Jan Hubicka hubi...@ucw.cz wrote:
Hi,
according to oprofile, libiberty hashing takes about 30% of streaming in time
and according to callgrind, the most busy cache is node_map
2011-05-02 Dmitry Gorbachev d.g.gorbac...@gmail.com
* parser.c (cp_parser_init_declarator): Replace false by NULL.
Replace by NULL_TREE instead and install under the obvious rule.
--
Eric Botcazou
Hi,
On Mon, 2 May 2011, Richard Guenther wrote:
/* The mapping between tree nodes and slots into the nodes array. */
! struct pointer_map_t GTY((skip)) *node_map;
If you skip node_map you can end up with false entries for re-used
trees. So I don't think that's a good idea.
Or
Submitted to google/main.
Adding libstd...@gcc.gnu.org for trunk approval.
Ollie
On Mon, May 2, 2011 at 9:34 AM, Ollie Wild a...@google.com wrote:
commit 99835abc5d2a7fd3ae0950c8a16fd6d223d408c9
Author: Ollie Wild a...@google.com
Date: Fri Apr 29 13:03:57 2011 -0400
Remove
If you skip node_map you can end up with false entries for re-used
trees. So I don't think that's a good idea.
The GTY marker is bogus there. I believed that the cache is GTY annotated
and then the skip would be safe, since whether is in the map is also in the
following array.
But it is not
On 05/02/2011 04:52 PM, Ollie Wild wrote:
Submitted to google/main.
Adding libstd...@gcc.gnu.org for trunk approval.
This is Ok for mainline and 4_6-branch.
Why nobody noticed so far? Is there a latent C++ front-end bug? Please
double check (the last time, I took care of filing the
In my last --enable-threads cleanup patch
[build, doc] Cleanup --enable-threads support
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00735.html
I missed another opportunity, handled by this patch.
BTW, feel free to also remove any trace of --enable-threads=gnat.
This is a
On Mon, 2 May 2011, Rainer Orth wrote:
Apart from the issues mentioned by Joseph, I'm removing Saber C support,
use of private Size_t/Ptrdiff_t, and STATIC. Those were the low-hanging
fruit; I currently don't plan to go for other stuff (like removal of the
remaining warnings) since I may
On Mon, May 2, 2011 at 4:35 PM, Michael Matz m...@suse.de wrote:
Hi,
On Mon, 2 May 2011, Jan Hubicka wrote:
! d_entry.base.from = t;
! slot = htab_find_slot (cache-node_map, d_entry, INSERT);
! if (*slot == NULL)
{
/* Determine the next slot to use in the cache. */
Hi,
On Mon, 2 May 2011, Richard Guenther wrote:
--- 348,367
bool insert_at_next_slot_p)
{
void **slot;
unsigned ix;
bool existed_p;
gcc_assert (t);
! slot = pointer_map_insert (cache-node_map, t);
! if (!*slot)
ix
This changes the code that deals with too large array sizes to
use int_fits_type_p instead of relying on the TREE_OVERFLOW setting
of the tree folder. The latter will break once we don't treat
sizetypes specially (and they keep being unsigned).
Bootstrapped and tested on
2011/5/2 Georg-Johann Lay a...@gjlay.de:
This is a fix for an optimization flaw when a long value is composed
from byte values.
For -fsplit-wide-types (which is still default for avr) the code is
worse than with -fno-split-wide-types. The code for the test case is
better in either
On 05/02/2011 04:58 PM, Paolo Carlini wrote:
Why nobody noticed so far? Is there a latent C++ front-end bug? Please
double check (the last time, I took care of filing the corresponding
C++ front-end issue)
-- Jason,
is it known that we accept the below?
Paolo.
//
This is another tiny step towards making TYPE_IS_SIZETYPE not special
wrt sign-extension. In output_constructor_regular_field we call
tree_low_cst without first assuring it won't ICE. Instead of
doing that simply compute the zero-based index in double-ints
and assert that that result is
Ping^4. Parts of this patch
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00130.html are still
pending review, for targets fr30, m32r, mn10300, rx, v850 (target
maintainers CC:ed). This version applies cleanly to current trunk.
I have seen no comments from the mn10300 maintainers. For the
PR45099 is an extension that gives an error when a fixed register is
needed to pass a parameter to a function.
Because the program will show malfunction when such code is generated,
anyway, I think an error is more appropriate than a warning (as
proposed in the PR).
Johann
2011-05-02
On Mon, May 2, 2011 at 5:15 PM, Michael Matz m...@suse.de wrote:
Hi,
On Mon, 2 May 2011, Richard Guenther wrote:
--- 348,367
bool insert_at_next_slot_p)
{
void **slot;
unsigned ix;
bool existed_p;
gcc_assert (t);
! slot =
2011/5/2 Jan Hubicka hubi...@ucw.cz:
Hi,
On Mon, 2 May 2011, Richard Guenther wrote:
--- 348,367
bool insert_at_next_slot_p)
{
void **slot;
unsigned ix;
bool existed_p;
gcc_assert (t);
! slot = pointer_map_insert
On Mon, 2 May 2011, Joseph S. Myers wrote:
On Mon, 2 May 2011, Richard Guenther wrote:
This changes the code that deals with too large array sizes to
use int_fits_type_p instead of relying on the TREE_OVERFLOW setting
of the tree folder. The latter will break once we don't treat
thanks. Will correct.
David
On Mon, May 2, 2011 at 2:03 AM, Rainer Orth r...@cebitec.uni-bielefeld.de
wrote:
David,
2011-05-01 David Li davi...@google.com
* testsuite/lib/profopt.exp (proc): Multiple source file support.
^
This is wrong, the
On Mon, May 2, 2011 at 2:11 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 29, 2011 at 6:23 PM, Xinliang David Li davi...@google.com wrote:
Here is the background for this feature:
1) People relies on function multi-version to explore hw features and
squeeze performance,
It has long been on my agenda to perform/complete the toplevel libgcc
migration for my targets. Although Solaris is the most complicated of
the three, it lays some infrastructure for the others, so I'm starting
with this patch. The ones for IRIX 6 and Tru64 UNIX will follow once
the dust on this
Checked into trunk.
Andreas.
2011-05-02 Andreas Schwab sch...@linux-m68k.org
PR target/47955
* config/m68k/m68k.c (m68k_expand_prologue): Set
current_function_static_stack_size.
Index: gcc/config/m68k/m68k.c
2011-05-02 Dmitry Gorbachev d.g.gorbac...@gmail.com
* parser.c (cp_parser_init_declarator): Replace false by NULL.
I've installed the attached patch as obvious.
2011-05-02 Dmitry Gorbachev d.g.gorbac...@gmail.com
Eric Botcazou ebotca...@adacore.com
* parser.c
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
This is merely an infrastructure patch in preparation for another
improvement in the jump threading code. It shouldn't affect the code we
generate at all.
Let's consider this CFG (from a routine in cfgexpand.c)
A
/ \
On Mon, 2 May 2011, Rainer Orth wrote:
* Another wart was that gcc/ada/gcc-interface/Makefile.in requires
TARGET_LIBGCC2_CFLAGS, which is otherwise only used in libgcc. It
often contains -fPIC, and if the variable is removed, libgnat.so won't
link (text relocations remain).
This
Quoting Joern Rennecke joern.renne...@embecosm.com:
2010-04-14 Joern Rennecke joern.renne...@embecosm.com
* config-list.mk: New file.
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg01044.html
On 05/02/2011 05:27 PM, Richard Guenther wrote:
On Mon, May 2, 2011 at 5:21 PM, Joseph S. Myers jos...@codesourcery.com
wrote:
Ping^4. Parts of this patch
http://gcc.gnu.org/ml/gcc-patches/2011-04/msg00130.html are still
pending review, for targets fr30, m32r, mn10300, rx, v850 (target
On 04/14/2011 10:52 AM, Jason Merrill wrote:
This patch avoids this issue by saving bounds values which have side
effects into a local automatic variable (since VLAs can only appear in
automatic variables). We stick with the SAVE_EXPR approach for bounds
without side-effects to avoid breaking
On Sat, Apr 30, 2011 at 2:14 PM, Chris Demetriou c...@google.com wrote:
Makes -S output more easily preprocessable -- otherwise, the __i686 in
__i686.get_pc_think.reg chokes things.
IIRC the reason why it uses __i686 is because 586 and before does not
need to worry about the return stack
Hello!
Attached patch merges MMX move patterns in mmx.md and fixes a couple
of problems in i386.md file.
2011-05-02 Uros Bizjak ubiz...@gmail.com
* config/i386/mmx.md (*movmode_internal_rex64): Use %vmovq for
reg-xmm moves.
(*movmode_internal): Merge with
Hi,
I want to submit this patch to google/main to make sure it is
available for our internal use at Google in order to materialize some
optimization opportunities. Let us continue this dicussion as I make
changes and submit this for review for trunk.
Thanks,
-Sri.
On Mon, May 2, 2011 at 9:41
Ok. There may be more correctness related comments -- but those can be
addressed when available. For trunk, you need to address issues such
as multi-way dispatch.
Thanks,
David
On Mon, May 2, 2011 at 12:11 PM, Sriraman Tallam tmsri...@google.com wrote:
Hi,
I want to submit this patch to
Since google/gcc-4_6 follows the 4.6 branch, changes in minor
revisions cause unnecessary churn in directory names.
Fixed with this. OK for google/gcc-4_6?
Google ref 4335466.
* BASE-VER: Change to 4.6.x-google.
diff --git a/gcc/BASE-VER b/gcc/BASE-VER
index 4110f74..33d4edd
On Mon, May 2, 2011 at 2:53 PM, Diego Novillo dnovi...@google.com wrote:
Since google/gcc-4_6 follows the 4.6 branch, changes in minor
revisions cause unnecessary churn in directory names.
Fixed with this. OK for google/gcc-4_6?
Yes, okay for google/gcc-4_6.
Thanks,
Ollie
On Mon, May 2, 2011 at 11:49, Andrew Pinski pins...@gmail.com wrote:
On Sat, Apr 30, 2011 at 2:14 PM, Chris Demetriou c...@google.com wrote:
Makes -S output more easily preprocessable -- otherwise, the __i686 in
__i686.get_pc_think.reg chokes things.
IIRC the reason why it uses __i686 is
On Mon, May 2, 2011 at 7:07 PM, Jeff Law l...@redhat.com wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
This is merely an infrastructure patch in preparation for another
improvement in the jump threading code. It shouldn't affect the code we
generate at all.
Let's consider this CFG
With this patch, now all coarray index intrinsics are implemented for
num_images() = 1 - except for bugs.
Built and regtesting on x86-64-linux. The testcase was also tested with
-fcoarray=single and with -fcoarray=lib -lcaf_mpi with mpirun -n 8.
OK for the trunk?
Tobias
PS: I would like to
On Mon, May 2, 2011 at 6:41 PM, Xinliang David Li davi...@google.com wrote:
On Mon, May 2, 2011 at 2:11 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 29, 2011 at 6:23 PM, Xinliang David Li davi...@google.com
wrote:
Here is the background for this feature:
1) People
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 05/02/11 15:18, Richard Guenther wrote:
For (edge *)xmalloc (sizeof (edge) * 1) use XNEWVEC and friends.
Yup. Good catch. Will fix. Old habits die hard.
I wonder if it makes sense to use a VEC instead (will the vector
be of variable
On Mon, May 2, 2011 at 11:34 PM, Jeff Law l...@redhat.com wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 05/02/11 15:18, Richard Guenther wrote:
For (edge *)xmalloc (sizeof (edge) * 1) use XNEWVEC and friends.
Yup. Good catch. Will fix. Old habits die hard.
I wonder if it
This patch improves unix 2003 conformance on HP-UX 11.31 systems
by linking in the unix2003.o object instead of the unix98.o object.
The only difference I am aware of between the two is that linking in
unix2003.o alters the behaviour of snprintf when the buffer is not
large enough to hold the
On Mon, May 2, 2011 at 17:34, Jeff Law l...@redhat.com wrote:
I've got no opinion on using a VEC if folks chime in with the request,
I'll make that change.
I have a slight preference for VEC by default.
Diego.
On 04/29/2011 04:08 PM, Basile Starynkevitch wrote:
[...]
which is also what I expect. I have no idea if it is conforming to
standards. But I notice that the test about this is remaining in the
code. So apparently GCC 4.6 does not make the hypothesis that this is
never null, otherwise it would
So, the problem in 48834 was that we had specified a particular target
for the vec initialization, but it was getting clobbered by ending up on
the rhs of an INIT_EXPR. So 48834.patch avoids that.
But Diego doesn't think there was any real reason to abort on trying to
copy a STATEMENT_LIST,
Is this one ok?
David
On Fri, Apr 29, 2011 at 4:38 PM, Xinliang David Li davi...@google.com wrote:
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
But Diego doesn't think there was any real reason to abort on trying to
copy a STATEMENT_LIST, so it seems to me that we could revert my earlier
patch for 40975 and just add support for copying STATEMENT_LIST. So
40975-2.patch adds that support.
FWIW this assertion caught an impressive
Nathan Froyd froy...@codesourcery.com wrote:
As $SUBJECT suggests. The only tricky bit is the initialization of
`args' to NULL_TREEs so that we can safely pass all of the relevant args
to build_function_type_list, regardless of whether the function type in
question has that many args.
On Mon, May 2, 2011 at 2:33 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Mon, May 2, 2011 at 6:41 PM, Xinliang David Li davi...@google.com wrote:
On Mon, May 2, 2011 at 2:11 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 29, 2011 at 6:23 PM, Xinliang David Li
2011-05-03 Jonathan Wakely jwakely@gmail.com
PR libstdc++/48848
* doc/xml/manual/status_cxx200x.xml: Update valarray status.
* doc/html/*: Regenerate.
A small tweak to the C++0x status table, as pointed out in PR 48848
Tested with 'make doc-xml-validate-docbook'
On 25 April 2011 19:11, Jonathan Wakely wrote:
Here's a patch which removes std::bind from the overload set when the
first argument is socketlike which I've defined as is_integral ||
is_enum. I considered using is_convertible to int to detect
socketlike types, but a callable class type could
On 05/02/2011 07:20 AM, Rainer Orth wrote:
gfortran.dg/fmt_g0_6.f08 execution seems to fail on all 32-bit x86
targets at -O1 and above. Running it under gdb, I find
Program received signal SIGABRT, Aborted.
[Switching to Thread 1 (LWP 1)]
0xfed0ff85 in _lwp_kill () from /lib/libc.so.1
(gdb)
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 05/02/11 15:40, Richard Guenther wrote:
I've got no opinion on using a VEC if folks chime in with the request,
I'll make that change.
If it's fixed size there's no need for a VEC, it would be just overhead.
VEC adds checking, which can be
On Sun, Apr 24, 2011 at 8:11 AM, H.J. Lu hjl.to...@gmail.com wrote:
On Fri, Apr 22, 2011 at 12:27 PM, Richard Guenther rguent...@suse.de wrote:
There is a scary comment before build_int_cst_type why build_int_cst
can't be sane. Fortunately it is not true. If it were there would
be other
On Fri, Apr 29, 2011 at 9:43 AM, Jakub Jelinek ja...@redhat.com 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 array,
which is wrong,
On 05/02/2011 06:23 PM, Eric Botcazou wrote:
I'm not sure you can copy statements if they have any side-effects; this looks
quite dangerous to me. Instead statement-expressions should be wrapped up in
a SAVE_EXPR/TARGET_EXPR to protect them and prevent copying.
It sounds like Ada and C++ are
82 matches
Mail list logo