Backport from mainline to arm-embedded branch r171096, r171251, r171379,
r171632, r171978, r172297, r174035.
Committed.
2011-09-19 chengbin bin.ch...@arm.com
Backport r174035 from mainline
2011-05-22 Tom de Vries t...@codesourcery.com
PR middle-end/48689
*
Hello,
I patched arm-arches.def and re-generated arm-tables.opt using command
./genopt.sh ../arm arm-tables.opt in directory gcc/config/arm. Now the
updated patch is as below. Is it OK to trunk?
BR,
Terry
2011-09-19 Terry Guo terry@arm.com
* config/arm/arm-arches.def
On Sat, Sep 3, 2011 at 11:11 PM, Uros Bizjak ubiz...@gmail.com wrote:
I've decided to not use four new bits from target_flags, and instead
created a new mask (recip_mask). Four bits would have fit in target
bits right now, but in the future we might want to add more
specialization, like
Committed
2011-09-06 chengbin bin.ch...@arm.com
Backport cortex-r5 from mainline based on r174526
* config/arm/arm-cores.def: Add cortex-r5. Add DIV flags to
Cortex-A15.
* config/arm/arm-tune.md: Regenerate.
* config/arm/arm.c (FL_DIV): Rename...
Backport from mainline to ARM/embedded-4_6-branch.
Committed.
2011-09-19 Jiangning Liu jiangning@arm.com
Backport r175427 from mainline
2011-06-27 Richard Guenther rguent...@suse.de
PR tree-optimization/49169
* fold-const.c
In Alfa mode, consider the underlying entity renamed instead of the renaming
for the reference, because the latter is needed to compute a valid set of
effects for the enclosing subprogram. If no such enclosing object, then it
could be a reference to any location not tracked individually, like
The body that is created for an expression function does not freeze. When a
new declaration and body are created for an expression function, the new body
must be inserted at the end of the current declarative part or package spec,
to prevent premature freezing of related types.
The following must
Get the enclosing entity through possible multiple renamings, not only once,
which may come from source or from the translation of generic instantiations.
Only for the Alfa mode for formal verification.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-09-19 Yannick Moy m...@adacore.com
Ada 2012 AI-241. Restriction No_Implementation_Aspect_Specifications
disallows any implementation defined aspect specifications. It is
included in the profile No_Implementation_Extensions. Restriction
No_Implementation_Of_Aspects = Aspect_Identifier disallows a
specified aspect specification. The
Tagged type expansion should not be performed in the formal verification mode.
It was previously performed, to avoid some errors, which are now corrected.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-09-19 Yannick Moy m...@adacore.com
* gnat1drv.adb (Adjust_Global_Switches):
On 19 Sep 2011, at 07:50, Terry Guo terry@arm.com wrote:
Hello,
I patched arm-arches.def and re-generated arm-tables.opt using command
./genopt.sh ../arm arm-tables.opt in directory gcc/config/arm. Now the
updated patch is as below. Is it OK to trunk?
BR,
Terry
2011-09-19
Hello,
The problem here is the case where the selective scheduler doesn't expect
having a block with zero successors. The patch is near obvious which is
why it was forgotten for nearly half a year, but still it applies cleanly
and fixes the testcase, full testing is in progress. OK for
Backport following mainline revs to ARM/embedded-4_6-branch:
177890, 177891.
Committed.
2011-09-19 Joey Ye joey...@arm.com
Backport r177891 from mainline
2011-08-19 Matthew Gretton-Dann matthew.gretton-d...@arm.com
PR target/49437
* config/arm/arm.c
The GNAT tasking runtimes using a signal (SIGADAABRT) for asynchronous
task abortion. On most platforms, SIGADAABRT is SIGABRT, but Darwin used
to use SIGTERM for that purpose. This change makes Darwin consistent with
the majority of platforms. This allows tasking programs to install their
own
Expansion for set membership in normal mode is also useful in Alfa mode for
formal verification, hence apply it too in that case.
Tested on x86_64-pc-linux-gnu, committed on trunk
2011-09-19 Yannick Moy m...@adacore.com
* exp_alfa.adb, exp_alfa.ads (Expand_Alfa_N_In): New function
This change causes the compiler to copy components from runtime record type
System.Partition_Interface.RACW_Stub_Type instead of hard-coding the
component list in Build_Stub_Type. This allows the structure of that type
to be changed in the PCS without having to change the compiler.
Tested on
The patch below fixes gcc.c-torture/execute/builtin-bitops-1.c on ARM/Thumb-2
targets.
The story here is that ARM doesn't have a 64-bit shift operator, so these get
decomposed into something like:
low1 = (low n)
low1 |= (high (32 - n)) /**/
low2 = high (32 - n)
low_result = (n 32) ? low2 :
Backport following mainline revs to ARM/embedded-4_6-branch:
174115, 176760, 177594, 178628.
Committed.
2011-09-19 Terry Guo terry@arm.com
Backport r178628 from mainline
2011-09-07 Jiangning Liu jiangning@arm.com
PR tree-optimization/46021
*
Hi!
As mentioned in an earlier posting, for -msse4.1 and above
on the gcc.target/i386/{sse2,avx}-extract-1.c testcases we
generate inefficient code when trying to extract SFmode
from V4SFmode, unless it is the element 0.
The problem is that the sse4_1_extractps pattern matches, but
doesn't have a
Hi,
When we can't vectorize a certain statement in SLP we mark it as not
vectorizable and continue with the analysis. This is wrong when the
reason for the failure is that we can't analyze a data-ref, because
this way we may miss a data dependence. This patch fails SLP if the
data-refs analysis
Hi,
tested x86_64-linux multilib, committed to mainline.
Paolo.
2011-09-19 Paolo Carlini paolo.carl...@oracle.com
PR libstdc++/40856
* include/std/limits (numeric_limits__int128_t,
numeric_limits__uint128_t): Add.
*
On 09/19/11 02:16, Alan Modra wrote:
On Sat, Sep 17, 2011 at 03:26:21PM +0200, Bernd Schmidt wrote:
On 09/17/11 09:16, Alan Modra wrote:
This patch series adds shrink-wrap support for PowerPC. The patches
are on top of Bernd's Initial shrink-wrapping patch:
On Sep 16, 2011, Eric Botcazou ebotca...@adacore.com wrote:
Yes, modulo Jakub's remark and s/NULL/NULL_TREE for zeroing in.base.from.
Thanks, here's what I've just checked in.
for gcc/ada/ChangeLog
from Alexandre Oliva aol...@redhat.com
* gcc-interface/decl.c (annotate_value): Look up
I caught this by chance: TREE_NOT_CHECK4 was not named properly,
deviating from the standard of other TREE_NOT_CHECK* macros, and
different from the equivalent macro in the without-tree-checking portion
of the macro setting.
I've checked this in as obvious, after bootstrapping it along with other
Hi,
On Fri, Sep 16, 2011 at 04:29:44PM +0200, Jan Hubicka wrote:
* ipa-prop.h (jump_func_type): Updated comments.
(ipa_known_type_data): New type.
(ipa_jump_func): Use it to describe known type jump functions.
* ipa-prop.c (ipa_print_node_jump_functions_for_edge):
On 19/09/11 10:12, Paul Brook wrote:
In Thumb-2 mode we try and match an M constraint (const_int between 0 and
31) with no other alternatives. This fails, and with nowhere else to go we
ICE.
The fix is to restrict the predicate for these insns to only accept valid
const_int shift counts. This
Jason Merrill ja...@redhat.com writes:
On 09/16/2011 03:55 AM, Dodji Seketeli wrote:
+test.c: In function ‘g’:
+test.c:5:14: error: invalid operands to binary (have ‘double’ and
‘int’)
+test.c:2:9: note: in expansion of macro 'OPERATE'
+test.c:5:3: note: expanded from
On Mon, Sep 19, 2011 at 4:46 AM, Ira Rosen ira.ro...@linaro.org wrote:
Hi,
When we can't vectorize a certain statement in SLP we mark it as not
vectorizable and continue with the analysis. This is wrong when the
reason for the failure is that we can't analyze a data-ref, because
this way we
This patch improves an existing test case for memcpy. The test case
memcpy-2.c checks that memory outside the destination is not corrupted. The
check fails if SEQUENCE_LENGTH MAX_OFFSET, although it does not happen
with the default values of these macros. This patch fixes the check to work
for
Richard Earnshaw wrote:
This patch fixes a couple of cases where reload incorrectly thinks it
should be messing around with the result of a match_operator operand.
Given that it's never even tried to validate the contents of such an
operand the changes it tries to make are bogus.
I'm
On 09/19/2011 04:33 AM, Jakub Jelinek wrote:
* config/i386/sse.md (*sse4_1_extractps): Change into
define_insn_and_split, add =x 0 n and =x x n alternatives
and split them after reload.
Ok.
r~
Hello,
Ping!
I would like to get a return on this patch. I don't know quite well the
status of libiberty in GNU, please if I must this patch on another
mailing list, please say me on which.
Thanks!
Pierre Vittet
On Mon, Sep 19, 2011 at 1:33 PM, Jakub Jelinek ja...@redhat.com wrote:
As mentioned in an earlier posting, for -msse4.1 and above
on the gcc.target/i386/{sse2,avx}-extract-1.c testcases we
generate inefficient code when trying to extract SFmode
from V4SFmode, unless it is the element 0.
The
On 19/09/11 15:14, Ulrich Weigand wrote:
Richard Earnshaw wrote:
This patch fixes a couple of cases where reload incorrectly thinks it
should be messing around with the result of a match_operator operand.
Given that it's never even tried to validate the contents of such an
operand the
Richard Earnshaw wrote:
On 19/09/11 15:14, Ulrich Weigand wrote:
So it seems to me that for match_operator operands, the
goal_alternative_win flag should always be true ...
Can you find out why this isn't true in your case?
Hmm, I think it must be because of this:
/* If the
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 09/19/11 02:51, Carrot Wei wrote:
Hi
We should not combine insns cross volatile insns. Combine.c already
does this check, postreload should also do so.
Bootstrapped and regtested on x86_64-unknown-linux-gnu. Regtested
on arm qemu.
On 19/09/11 16:53, Ulrich Weigand wrote:
Richard Earnshaw wrote:
On 19/09/11 15:14, Ulrich Weigand wrote:
So it seems to me that for match_operator operands, the
goal_alternative_win flag should always be true ...
Can you find out why this isn't true in your case?
Hmm, I think it must be
Hello!
2011-09-19 Uros Bizjak ubiz...@gmail.com
* config/i386/i386.md (maxmin): New code iterator.
* config/i386/sse.md (maxmin:codemode3): Macroize expander
from umaxmin:codemode3 and smaxmin:codemode3 using maxmin
code iterator.
Hello
Here is a patch allowing to compute correctly the md5 of a file sequence
even with the bug rapported here:
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg00963.html.
The function will work with both gcc 4.6 without the patch and a gcc
with the patch applied.
The problem cames from the fact
Hello!
Hello!
No functional change.
2011-09-19 Uros Bizjak ubiz...@gmail.com
* config/i386/mmx.md (*movmode_internal_rex64): Use if_then_else RTX
to calculate unit, prefix_rep and prefix_data16 attributes.
(*movmode_internal): Ditto for unit attribute.
On 09/19/2011 09:58 AM, Dodji Seketeli wrote:
+ The part that goes from the third to the heighth line of this
An extra 'h' snuck in there. :)
Inside the definition of a macro M, you can have another macro M'.
And M' is going to be eventually expanded into a token FOO.
So, to paraphrase
On Mon, 19 Sep 2011 20:02:16 +0200
Pierre Vittet pier...@pvittet.com wrote:
Hello
Here is a patch allowing to compute correctly the md5 of a file sequence
even with the bug rapported here:
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg00963.html.
The function will work with both gcc 4.6
Hi!
AVX2 prefers v*i128 instructions for vector integer
modes, while AVX only supports v*f128.
In one of my recent patches I've changed vextract* so that
it emits v*i128 instead of v*f128 for AVX2 on integer modes,
but I wonder what is the prefered way to change the rest
of the insn.
E.g. we
On 09/15/2011 06:02 PM, Andrew MacLeod wrote:
On 09/15/2011 05:49 PM, Jason Merrill wrote:
Do we want to start exporting these from libstdc++?
I dont know the ins and outs of the library... the typedefs and
specializations are already there, I can write:
atomic_char c;
c = 'a'
in my
Hello world,
this patch implements the library side of PR 36313, minloc and friends
on character arguments.
I haven't started tackling the inline versions yet. I suspect that
these are easy for somebody who knows the scalarizer (not me :-) but
I'll start looking into how to do it. Unless, of
2011-09-19 Wei Guozhi car...@google.com
PR 49452
* postreload.c (reload_combine): Invalidate use information when
across volatile insn.
PR rtl-optimization/49452, otherwise I'm not sure the commit will be xref'ed.
--
Eric Botcazou
Jan,
The following patch started as a one-liner for ipa-inline-analysis.c:
account_size_time() to merge predicates when we are adding data to entry[0]
(i.e., when space for 32 size_time entries is exhausted):
@@ -537,6 +592,9 @@ account_size_time (struct inline_summary
}
else
{
+
On 09/19/2011 02:29 PM, Jason Merrill wrote:
expansion location. So it seems like we ought to be able to get away
with only storing one location per token in a macro expansion map. Am I
missing something?
I notice that here:
+/* Resolve the location iter-where into the locus 1/ of
On 09/18/2011 06:47 AM, Dodji Seketeli wrote:
+cpp_hashnode *macro;
+
+ }c;
Extra blank line, missing space.
+/* Appends a token to the end of the token buffer BUFFER. Note that
+ this function doesn't enlarge BUFFER; it overwrite the last memory
+ location of BUFFER that holds a
On 09/19/2011 04:22 AM, Ed Smith-Rowland wrote:
+check_literal_operator_args( const_tree decl,
Space before the (, not after.
Done. I am now storing the numeric string along with the number and the suffix
ID for numeric operators. I'm keeping the numeric value because we need it too.
On 14/09/2011, at 7:40 AM, Maxim Kuvyrkov wrote:
OK then, attached is the trivial patch that removes
haifa-sched.c:check_cfg(). Please let me know if you have strong feelings
towards keeping check_cfg().
Tested on x86_64-linux-gnu. Absent any requests to the contrary, I will
check in
Jason Merrill ja...@redhat.com writes:
On 09/19/2011 09:58 AM, Dodji Seketeli wrote:
+ The part that goes from the third to the heighth line of this
An extra 'h' snuck in there. :)
Oops, fixed in my local copy, sorry.
Inside the definition of a macro M, you can have another macro
Hi,
Its been used in the automated toolchain builds for sometime, seems like
a good idea to enable it by default. It can be easily changed to match
for all mingw as well if needed.
OK for trunk?
Index: libstdc++-v3/configure.ac
===
The Go frontend was overloading Named_type::seen_ to mean too many
different things. It caused a compiler crash compiling mutually
recursive types as in:
type T18 *[10]T19
type T19 T18
This patch fixes the problem by splitting off the only complex use of
seen_ into a new field. I also changed
On 09/17/2011 09:44 AM, Fabien Chêne wrote:
I tried various things without success, and I ended up hacking
supplement_binding_1 to handle those ENUMERAL_TYPEs.
I am all ear for another solution ...
Your solution seems reasonable to me, but it needs a comment, along the
lines of
/* We allow
Hi.
This patch adds two functions, timeval_add and timeval_sub,
to libiberty. GDB has use for them in a few places and since
they're general purpose I wish to check them into libiberty.
Ok to check in?
2011-09-19 Doug Evans d...@google.com
include/
* timeval-utils.h: New
Hello,
This is an old case written in 1997. Perhaps it intends to cover an old gcc
bug. In fact the case itself passed on QEMU for cortex-m3. So I am proposing
a patch here to remove the test script to make this pass as a normal pass.
Otherwise the pass is regarded as XPASS. Is it ok to trunk?
Prepare for merging symbols. When writing a cp_binding_level, we emit
this_entity first, so that we can identify the context of the bindings
early. If this_entity is a namespace, we call alternate routines that
track the namespace. These ultimately call pph_write_namespace_tree,
which adds
On Mon, Sep 19, 2011 at 9:08 PM, Jakub Jelinek ja...@redhat.com wrote:
AVX2 prefers v*i128 instructions for vector integer
modes, while AVX only supports v*f128.
In one of my recent patches I've changed vextract* so that
it emits v*i128 instead of v*f128 for AVX2 on integer modes,
but I
2011/9/20 JonY jo...@users.sourceforge.net:
Hi,
Its been used in the automated toolchain builds for sometime, seems like
a good idea to enable it by default. It can be easily changed to match
for all mingw as well if needed.
OK for trunk?
Index: libstdc++-v3/configure.ac
61 matches
Mail list logo