-Original Message-
From: Thomas Schwinge [mailto:tho...@codesourcery.com]
Sent: Friday, April 20, 2012 01:46
To: Bernd Schmidt; Richard Earnshaw
Cc: Richard Guenther; Joey Ye; d...@redhat.com; GCC Patches; Mitchell,
Mark
Subject: Re: Continue strict-volatile-bitfields fixes
That
Fu, Chao-Ying f...@mips.com writes:
Index: gcc/config/mips/gnu-user.h
===
--- gcc/config/mips/gnu-user.h(revision 186580)
+++ gcc/config/mips/gnu-user.h(working copy)
@@ -45,8 +45,10 @@
/* A standard GNU/Linux
On Fri, Apr 13, 2012 at 02:24:09PM +0200, Olivier Hainque wrote:
On typical VxWorks environments, WindRiver integrated tools are used as
much if not more than gdb for debugging purposes.
These evolve at an industrial pace, traditionally not as fast as GCC
regarding the support of latest
On Thu, 19 Apr 2012, William J. Schmidt wrote:
This enhances constant folding for division by complex and vector
constants. When -freciprocal-math is present, such divisions are
converted into multiplies by the constant reciprocal. When an exact
reciprocal is available, this is done for
Hi!
Sorry for the delay.
On Tue, 27 Mar 2012 05:34:30 -0700, Ian Lance Taylor i...@airs.com wrote:
Thomas Schwinge tho...@codesourcery.com writes:
libgcc/
* config/alpha/linux-unwind.h (alpha_fallback_frame_state): Use
siginfo_t instead of struct siginfo.
*
*ping*
Tobias
PS: I know I should be also faster in reviewing patches of others...
On 04/12/2012 05:23 PM, Tobias Burnus wrote:
This patch is a kind of follow up to the other one for the same PR -
though this one is for a separate test case, it is not a regression
and it's about
On Thu, Apr 19, 2012 at 5:16 PM, Steven Bosscher stevenb@gmail.com wrote:
Hello,
If we want to use profiling to expand switches in GIMPLE, we'll have
to run switch-conversion after profiling.
Bootstrapped and tested on x86_64-unknown-linux-gnu. OK?
That's too early still. Profile data
On Thu, Apr 19, 2012 at 7:56 PM, Thomas Schwinge
tho...@codesourcery.com wrote:
Hi!
On Thu, 19 Apr 2012 19:46:17 +0200, I wrote:
diff -ru fnsvb/20030922-1.c.003t.original fsvb/20030922-1.c.003t.original
--- fnsvb/20030922-1.c.003t.original 2012-04-19 16:51:18.322150866 +0200
+++
On Thu, Apr 19, 2012 at 8:51 PM, Xinliang David Li davi...@google.com wrote:
Compiling the attached test case with -O2 using the trunk compiler,
the compiler will ICE. The proposed patch is also attached. The test
is under going, but I like to have discussion on the right fix first.
The patch
On Thu, Apr 19, 2012 at 10:35 PM, Jim Meyering j...@meyering.net wrote:
Richard Guenther wrote:
Sure, my point was that the
if (strlen (m-name) = sizeof buf)
{
error (%s:%d:mode name \%s\ is too long,
m-file, m-line, m-name);
continue;
On Thu, Apr 19, 2012 at 5:16 PM, Steven Bosscher stevenb@gmail.com
wrote:
Hello,
If we want to use profiling to expand switches in GIMPLE, we'll have
to run switch-conversion after profiling.
Bootstrapped and tested on x86_64-unknown-linux-gnu. OK?
That's too early still.
On Fri, Apr 20, 2012 at 10:52 AM, Jan Hubicka hubi...@ucw.cz wrote:
On Thu, Apr 19, 2012 at 5:16 PM, Steven Bosscher stevenb@gmail.com
wrote:
Hello,
If we want to use profiling to expand switches in GIMPLE, we'll have
to run switch-conversion after profiling.
Bootstrapped and
Dear all,
I seriously struggle with the following patch. It replaces a library
call by inline code for
C_F_POINTER(C-pointer, Fortran-array-pointer, shape-of-the-array)
(Purpose: (a) The current library version fails for SHAPE with strides.
(b) For the new array descriptor (fortran-dev
Dear all,
some compilers support using q to indicate quad precision, e.g.
4.0q0. Since GCC 4.7, gfortran supports this vendor extension in the
source code. However, READing the floating-point number 4.0q0 was not
supported.
The attached patch adds this support, which some users expect (cf.
The original motivation to do switch conversion early was to get function
bodies smaller (i.e. when inlining the static var don't need duplication,
the
switch code does)
This was motivated by real world examples, i.e. mesa that inlines function
converting
error codes into
Thanks!
K
On Thu, Apr 19, 2012 at 9:18 PM, Uros Bizjak ubiz...@gmail.com wrote:
On Thu, Apr 19, 2012 at 5:21 PM, Kirill Yukhin kirill.yuk...@gmail.com
wrote:
Folks,
Thanks a lot for prompts!
I've updated my patch, so cmparing to previous it is:
- have dedicated hook var, to define target
On Fri, Apr 20, 2012 at 11:24 AM, Jan Hubicka hubi...@ucw.cz wrote:
The original motivation to do switch conversion early was to get function
bodies smaller (i.e. when inlining the static var don't need duplication,
the
switch code does)
This was motivated by real world examples,
I'm applying this to mainline as obvious. If we save fprs inline for
any reason, we must also restore them inline. (The same goes for
gprs, and that is handled later in this function.)
Bootstrapped and regresion tested powerpc-linux.
PR target/53040
* config/rs6000/rs6000.c
Hi,
Please consider this mail as a reminder to review the patch posted
at the following link:-
http://gcc.gnu.org/ml/gcc-patches/2012-04/msg00255.html
Please review the patch and let us know if there should be any
modifications in it.
Thanks Regards,
Naveen
Small follow up.
On 04/20/2012 11:18 AM, Tobias Burnus wrote:
Testing other compilers, the result is:
- 'q' not supported: g95, NAG f95, PGI, PathScale, Crayftn
- 'q' supported: g77, ifort, sunf95
It seems to be also supported by IBM's xlf - at least it supports such
literals in the source
Hi,
Please find attached the patch v850_indent.patch that adapts V850
machine descriptions to the braced string notation. The changes makes
the Machine Descriptor file more easier to read.
It is similar to the following changes in SH target.
Hi,
tested x86_64-linux, committed.
Thanks,
Paolo.
//
2012-04-20 Paolo Carlini paolo.carl...@oracle.com
PR libstdc++/53052
* include/std/type_traits (is_explicitly_convertible): Remove.
* testsuite/20_util/is_explicitly_convertible: Likewise.
My only other comment is that I think the MEMMODEL_MASK ought to be defined
globally where the MEMMODEL enum is defined instead of defining it
everywhere it is needed.
Agree. Done (along with Uros's input).
New define added to gcc/coretypes.h
Updated patch attached.
updated ChangeLog entry:
Il 20/04/2012 02:27, Joel Brobecker ha scritto:
Hello everyone,
I wasn't sure if I needed approval for this patch or not, but better
be safe than sorry. I'll apply to both GCC and then src when I receive
confirmation that it's OK to apply.
I would like to update the config.sub script to
On Fri, Apr 20, 2012 at 10:41 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Apr 19, 2012 at 8:51 PM, Xinliang David Li davi...@google.com wrote:
Compiling the attached test case with -O2 using the trunk compiler,
the compiler will ICE. The proposed patch is also attached. The
On Fri, 2012-04-20 at 10:04 +0200, Richard Guenther wrote:
On Thu, 19 Apr 2012, William J. Schmidt wrote:
This enhances constant folding for division by complex and vector
constants. When -freciprocal-math is present, such divisions are
converted into multiplies by the constant
This fixes PR53050.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied.
Richard.
2012-04-20 Richard Guenther rguent...@suse.de
PR tree-optimization/53050
* tree-ssa-forwprop.c (ssa_forward_propagate_and_combine):
Do only one transform on COND_EXPRs at the
On Fri, 20 Apr 2012, William J. Schmidt wrote:
On Fri, 2012-04-20 at 10:04 +0200, Richard Guenther wrote:
On Thu, 19 Apr 2012, William J. Schmidt wrote:
This enhances constant folding for division by complex and vector
constants. When -freciprocal-math is present, such divisions are
Hi,
insn_invalid_p might add clobbers in order to match an instruction.
Unfortunately that change is not recorded in the 'changes' array. So
if it is decided to rollback the changes the added clobbers will stay.
This is a problem if the back-end provides two (mostly) identical insn
patterns, one
On Tue, Apr 17, 2012 at 09:11:37PM +0100, Richard Sandiford wrote:
* rtl.def (ADDRESS): Turn operand into a HOST_WIDE_INT.
* alias.c (reg_base_value): Expand and update comment.
(arg_base_value): New variable.
(unique_id): Move up file.
On Fri, Apr 20, 2012 at 12:07 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 20, 2012 at 10:41 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Apr 19, 2012 at 8:51 PM, Xinliang David Li davi...@google.com
wrote:
Compiling the attached test case with -O2
.
* g++.dg/torture/20120420-1.C: New testcase.
Index: gcc/tree-ssa-copy.c
===
--- gcc/tree-ssa-copy.c (revision 186619)
+++ gcc/tree-ssa-copy.c (working copy)
@@ -257,13 +257,11 @@ propagate_tree_value_into_stmt (gimple_s
else
On Fri, 20 Apr 2012, Jakub Jelinek wrote:
On Tue, Apr 17, 2012 at 09:11:37PM +0100, Richard Sandiford wrote:
* rtl.def (ADDRESS): Turn operand into a HOST_WIDE_INT.
* alias.c (reg_base_value): Expand and update comment.
(arg_base_value): New variable.
On Tue, Apr 17, 2012 at 12:27 AM, Igor Zamyatin izamya...@gmail.com wrote:
On Fri, Apr 13, 2012 at 4:20 PM, Andrey Belevantsev a...@ispras.ru wrote:
On 13.04.2012 14:18, Igor Zamyatin wrote:
On Thu, Apr 12, 2012 at 5:01 PM, Andrey Belevantseva...@ispras.ru
wrote:
On 12.04.2012 16:38,
Hi,
This patch handles duplicating of the last iteration correctly.
The current code always duplicates the complete static iteration from
the entry to the latch,
and then sets the number of iterations according to the pattern of the
loop (according to whether the cond before the body, or the
On Fri, Apr 20, 2012 at 2:07 PM, Razya Ladelsky ra...@il.ibm.com wrote:
Hi,
This patch handles duplicating of the last iteration correctly.
The current code always duplicates the complete static iteration from
the entry to the latch,
and then sets the number of iterations according to the
Jakub Jelinek ja...@redhat.com writes:
On Tue, Apr 17, 2012 at 09:11:37PM +0100, Richard Sandiford wrote:
* rtl.def (ADDRESS): Turn operand into a HOST_WIDE_INT.
* alias.c (reg_base_value): Expand and update comment.
(arg_base_value): New variable.
(unique_id):
Hi,
Here is the patch that adds support for divide 64-bit by constant for
32-bit target machines, this patch was tested on arm-7a with no new
regressions, also I am not sure on how to avoid for example i686
targets since div operation there is fast compared to over targets and
it showed better
Hello,
here is a patch that requires parentheses around the argument of
sizeof I am not sure what variadic76.C tests, so I put 2 lines, with
parentheses in different places.
bootstrapped and regression tested (make -k check, compared to an
unpatched version).
Note that I can't commit.
On 20/04/12 01:27, Joel Brobecker wrote:
Hello everyone,
I wasn't sure if I needed approval for this patch or not, but better
be safe than sorry. I'll apply to both GCC and then src when I receive
confirmation that it's OK to apply.
I would like to update the config.sub script to the
Hi,
this patch is first of bigger cleanups I have in queue after symtab changes.
The main purpose of the patch is to remove needed flag from cgraph nodes.
Long time ago, when cgraph code was written, it was simple datastructure to
compute reachablity and inlinng decisions on.
We still did
Hi,
On Thu, 19 Apr 2012, Michael Matz wrote:
I've audited all other calls of va_end in the compiler and these were the
only problematic ones.
Well, I did, but forgot the fact that I found one in cp/error.c, so the
generated diff and the commit forgot that file too. It was in the
Thanks very much Paolo.
I'll apply this patch on my side for a while.
I ll tell you if i see anything strange.
Regards,
Laurent
On 04/19/12 17:52, Paolo Carlini wrote:
On 04/19/2012 05:02 PM, Laurent Alfonsi wrote:
Well, I don't know mt_allocator enough to know if this is a fix for
real or a
Here is a patch to fix the failing test gcc.dg/pr52283.c.
Adding the missing dg-warning and dg-options.
OK?
gcc/testsuite/ChangeLog
2012-04-20 Greta Yorsh greta.yo...@arm.com
* gcc.dg/pr52283.c: Add missing dg-warning and dg-options.
diff --git a/gcc/testsuite/gcc.dg/pr52283.c
Hi,
the new symtab verifier introduced by me two days ago fires during bootstrap on
AIX (and probably some other platforms not having COMDAT group support). The
problem is that same_comdat_group list is set for symbols that are not COMDAT.
This happens on variables only, but not for functions.
Richard Guenther wrote:
On Thu, Apr 19, 2012 at 10:35 PM, Jim Meyering j...@meyering.net wrote:
Richard Guenther wrote:
Sure, my point was that the
if (strlen (m-name) = sizeof buf)
{
error (%s:%d:mode name \%s\ is too long,
m-file, m-line, m-name);
Hi,
two days ago I talked to Richi on IRC about the functions to determine
the expected alignment of objects and pointers we have and he
suggested that get_object_alignment_1 and get_pointer_alignment_1
should return whether the alignment is actually known and return the
actual alignment in a
Hi!
This patch fixes ICE in adjust_bool_pattern, trueval in that case has
the vector type's element type, which is full precision of mode,
so when we see a smaller precision, we need to use build_nonstandard_type.
Bootstrapped/regtested on x86_64-linux and i686-linux, ok
for trunk and 4.7 (where
Hi,
there are only two places left where frontend is fiddling with
TREE_SYMBOL_REFERENCED.
The first should not be needed because alias code now properly analyze the
aliases
in callgraph.
The second can be replaced by TREE_USED (and I think should be). It is only
needed to the following
On Fri, Apr 20, 2012 at 4:50 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 20, 2012 at 12:07 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 20, 2012 at 10:41 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Apr 19, 2012 at 8:51 PM,
On Fri, Apr 20, 2012 at 3:07 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Fri, Apr 20, 2012 at 10:41 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Apr 19, 2012 at 8:51 PM, Xinliang David Li davi...@google.com
wrote:
Compiling the attached test case with -O2
Hello!
So, a patch of mine runs into the following testsuite issues. Fixed
with the following patch, tested on x86_64-unknown-linux-gnu and
committed.
2012-04-18 Richard Guenther rguent...@suse.de
* gcc.target/x86_64/abi/test_passing_unions.c: Avoid undefined
array access.
On Thu, Apr 19, 2012 at 6:58 PM, William J. Schmidt
wschm...@linux.vnet.ibm.com wrote:
This enhances constant folding for division by complex and vector
constants. When -freciprocal-math is present, such divisions are
converted into multiplies by the constant reciprocal. When an exact
OK.
Jason
This libgo patch disables memory profiling in one of the tests of the
garbage collector. This test measures the amount of memory allocated
after a loop that allocates and then forgets about a large number of
large slices. The test was failing on PPC64 GNU/Linux because that
system uses a much
* config/s390/s390.h (LINK_SPEC): Remove, no longer needed.
(LIBSTDCXX): Change to CPP2.
Can one S390 maintainers approve this please?
http://gcc.gnu.org/ml/gcc-patches/2012-03/msg01545.html
On 20 April 2012 16:54, Martin Jambor mjam...@suse.cz wrote:
two days ago I talked to Richi on IRC about the functions to determine
the expected alignment of objects and pointers we have and he
suggested that get_object_alignment_1 and get_pointer_alignment_1
should return whether the
This patch to libgo fixes the DWARF line lookup when one function has
different types of lines. This happens when the line info includes,
e.g., discriminators. Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu and power64-unknown-linux-gnu. Committed to
mainline and 4.7 branch.
Ian
The libgo test of using sendfile on GNU/Linux runs strace. The version
of strace I have installed on my powerpc64-unknown-linux-gnu system
crashes when tracing the sigaltstack system call. This patch works
around this problem by telling strace to not trace sigaltstack. This
does not affect the
Hi!
This patch fixes -Woverride-init. IMHO we don't want to ever warn
from within set_nonincremental_init, which moves over values from
the constructor_elements vector to constructor_pending_elts AVL.
The reason for that is that when switching from the non-incremental
mode where values only live
This patch to the Go compiler and runtime adds explicit checks for
division by zero and for division overflow (INT_MIN / -1).
In Go, a division by zero is supposed to produce a panic; that did not
happen on processors like PPC that do not issue a SIGFPE signal on
division by zero. This patch
On Fri, 2012-04-20 at 11:32 -0700, H.J. Lu wrote:
On Thu, Apr 19, 2012 at 6:58 PM, William J. Schmidt
wschm...@linux.vnet.ibm.com wrote:
This enhances constant folding for division by complex and vector
constants. When -freciprocal-math is present, such divisions are
converted into
Hi,
cxx_callgraph_analyze_expr contains logic that should be obsolette now.
PTRMEM_CST/BASELINK should be obsoletted by gimplification and VAR_DECL handling
seems weird from very start and I think it is remainder from time we was not
outputting function local vars from varpool but they needed a
This patch to libgo changes the way that GNU/Linux netlink code is
handled on big-endian systems. Rather than test whether the system is
big- or little-endian, this code simply uses the native endianness,
whatever it is, via the unsafe package. Bootstrapped and ran Go
testsuite on
We're still hitting the overzealous assert identified in PR49246 that is
looking for empty LTO partitions, even after the change to a checking assert.
How about we backport Honza's mainline patch that stops the creation of empty
partitions in the first place? Speaking with honza on IRC, he
This patch to the Go testsuite driver recognizes PowerPC as a Go
architecture, and disables the nilptr.go test on PowerPC. The nilptr.go
test is only valid if values are allocated in relatively low memory,
which does not happen on PowerPC. Ran Go testsuite on
x86_64-unknown-linux-gnu and
Hi,
sparc has
Mask(LONG_DOUBLE_128)
Target Report RejectNegative Mask(LONG_DOUBLE_128)
Mask(LONG_DOUBLE_128) is defined as extra_masks, which leads to
MASK_LONG_DOUBLE_128 be defined twice. This patch checks if
MASK_ and TARGET_ macros are defined for extra_masks. Tested
on Linux/x86 and
Hi,
This patch is for google-4_6 branch only.
It disables the localization of hidden and internal symbols in streaming
LIPO. Otherwise, we may have undefines in link time because of the reference in
other module that is not include the define module into module group.
Tested with google
ok.
thanks,
David
On Fri, Apr 20, 2012 at 2:13 PM, Rong Xu x...@google.com wrote:
Hi,
This patch is for google-4_6 branch only.
It disables the localization of hidden and internal symbols in streaming
LIPO. Otherwise, we may have undefines in link time because of the reference
in
other
Honza removed the function cgraph_mark_needed_node from mainline. The
Go frontend called it, but that call was left over from days long ago
when package initialization was run as a global constructor. Package
initialization is now done by a direct call from the main function
(found in
On 04/20/2012 03:37 AM, Naveen H. S wrote:
Hi,
Please find attached the patch h8300_indent.patch that adapts H8300
machine descriptions to the braced string notation. The changes makes
the Machine Descriptor file more easier to read.
It is similar to the following changes in SH target.
On 04/20/2012 03:39 AM, Naveen H. S wrote:
Hi,
Please find attached the patch v850_indent.patch that adapts V850
machine descriptions to the braced string notation. The changes makes
the Machine Descriptor file more easier to read.
It is similar to the following changes in SH target.
Honza removed the function cgraph_mark_needed_node from mainline. The
Go frontend called it, but that call was left over from days long ago
when package initialization was run as a global constructor. Package
initialization is now done by a direct call from the main function
(found in
Hello,
This patch for google 4.6 stops gcc from emitting end-of-sibling
markers for skeleton debug info sections, which don't have siblings.
This fixes an errors reported by readelf. OK for google 4.6?
Sterling
2012-04-20 Sterling Augustine saugust...@google.com
* gcc/dwarf2out.c
This time with the patch included.
Sterling
On Fri, Apr 20, 2012 at 3:51 PM, Sterling Augustine
saugust...@google.com wrote:
Hello,
This patch for google 4.6 stops gcc from emitting end-of-sibling
markers for skeleton debug info sections, which don't have siblings.
This fixes an errors
Oops, meant to CC gcc-patches ...
On 21 April 2012 01:01, Jonathan Wakely jwakely@gmail.com wrote:
On 21 April 2012 00:37, Todd Edwards wrote:
In Section New Languages and Language specific improvements In subsection
C Family Objective-C is repeated twice. :
A new experimental
2012-04-20 Sterling Augustine saugust...@google.com
* gcc/dwarf2out.c (output_skeleton_debug_sections): Fix off-by-one
error.
OK for google/gcc-4_6 branch.
-cary
On 20/04/2012, at 1:34 PM, Fu, Chao-Ying wrote:
Hi Maxim, Richard,
I built cross-toolchains for 3 different targets as follows.
1. mips-linux-gnu
2. mips-linux-gnu --enable-targets=all
3. mips64-linux-gnu
These targets are affected by this MIPS Android patch.
Then, I checked the
For abiv4, mtcr/mtcrf emitted in the epilogue can be scheduled past
the stack deallocation, where the cfa_restore notes are emitted. The
stack tie doesn't stop movement of this insn because it doesn't touch
memory. So emit the cfa_restore on the mtcr/mtcrf itself. The PR was
about cr, but
Hi,
I noticed a missed simple optimization on the tree level where the
and expression could be commoned out.
This patch implements the optimization in tree-ssa-forwprop.c. I
thought it would be good to get it in even before my tree combiner
work gets in (which I am still working on but
On Thu, Jan 19, 2012 at 3:13 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Thu, Jan 19, 2012 at 10:00 AM, Andrew Pinski
andrew.pin...@caviumnetworks.com wrote:
On Tue, Jan 17, 2012 at 1:38 AM, Richard Guenther
richard.guent...@gmail.com wrote:
On Tue, Jan 17, 2012 at 8:06 AM,
This time with the patch and describing what the bug was. The problem
was defcodefor_name does not always set arg1 and arg2. This fixes it
so it is always set to NULL if they don't exist.
Thanks,
Andrew
On Fri, Apr 20, 2012 at 9:05 PM, Andrew Pinski
andrew.pin...@caviumnetworks.com wrote:
On
82 matches
Mail list logo