On Tue, Aug 28, 2012 at 11:54 PM, Eric Botcazou wrote:
The CLEANUP_CFG_CHANGED path looks unnecessary, it looks like this is
mostly for repairing loops but I don't see a reason for this in
postreload (loops have been freed at that point). I could do
cleanup_cfg(0) but there shouldn't be much
This is a regression present on the mainline and 4.7 branch. The compiler
ICEs on the attached Ada testcase at -O -fstack-check in ix86_expand_prologue:
gcc_assert (m-fs.sp_offset == frame.stack_pointer_offset);
The problem is that release_scratch_register_on_entry uses a POP instruction
Hi,
Since x32 runs in 64-bit mode, for address -0x4300(%rax), hardware
sign-extends displacement from 32-bits to 64-bits and adds it to %rax.
But x32 wants 32-bit -0x4300, not 64-bit -0x4300. This patch
uses 32-bit registers instead of 64-bit registers when displacement
Updated patch attached, tested on powerpc64-unknown-linux-gnu.
This is exactly the same patch as before, except that it now uses
cleanup_cfg (0) instead of delete_unreachable_blocks.
OK for trunk?
OK if you adjust the ChangeLog as well, thanks.
--
Eric Botcazou
Hi,
this is a regression present on the mainline and 4.7 branch. -fvar-tracking
is automatically enabled on Darwin when -g is passed, including at -O0. This
results in bloated and incomplete debug info.
This comes from a latent issue in the handling of flag_var_tracking_uninit in
the Darwin
Hi Kaz,
Now that the cross-jumping problem is fixed since rev #193350, I'd like
to remove this restriction and close PR/54546.
Checked with default sh-sim target_board and
--target_board=sh-sim/-freorder-blocks-and-partition.
Thanks
Christian
2012-11-09 Christian Bruel
-- Forwarded message --
From: Vladimir Yakovlev vbyakov...@gmail.com
Date: 2012/11/9
Subject: Re: [off-list] Re: [PATCH] Vzeroupper placement/47440
To: Uros Bizjak ubiz...@gmail.com
Копия: H.J. Lu hjl.to...@gmail.com, Igor Zamyatin izamya...@gmail.com
I did changes that moves
This fixes a couple of cases where gnat_to_gnu would recurse on an empty node
and removes all traces of SCIL references from gigi.
Tested on x86_64-suse-linux, applied on mainline.
2012-11-09 Eric Botcazou ebotca...@adacore.com
* gcc-interface/trans.c (gnat_to_gnu)
On 11/08/2012 08:43 PM, Lubos Lunak wrote:
The patch implements an attribute for marking types for which gcc cannot on
its own issue warnings about unused variables (e.g. because the ctor is
external), but for which such a warning might be useful anyway (e.g.
std::string).
I'm not sure if
Hi,
when writing find_aggregate_values_for_callers_subset, I omitted a
test that is present in propagate_aggs_accross_jump_function and that
lead to triggering an assert in the former.
Fixed by moving the test into a separate predicate function in the
following patch.
Bootstrapped and
These assert should tell you what is wrong with the control flow.
Please look at control_flow_insn_p, which condition returns true.
There is a note after call insn.
(call_insn:TI 908 35558 50534 1681 (call (mem:QI (symbol_ref:DI
(_gfortran_stop_string) [flags 0x41] function_decl 0x77eb6200
On Fri, Nov 9, 2012 at 1:18 PM, Vladimir Yakovlev vbyakov...@gmail.com wrote:
These assert should tell you what is wrong with the control flow.
Please look at control_flow_insn_p, which condition returns true.
There is a note after call insn.
(call_insn:TI 908 35558 50534 1681 (call (mem:QI
On 30.10.2012 12:09, Andrey Belevantsev wrote:
Hello,
This PR is due to the selective scheduling missing the dependencies with
implicit_sets. From the sched-deps.c code it looks like implicit sets
generate anti dependencies with either sets, uses or clobbers, so that's
that I've done with the
On Fri, Nov 09, 2012 at 01:29:18PM +0100, Uros Bizjak wrote:
On Fri, Nov 9, 2012 at 1:18 PM, Vladimir Yakovlev vbyakov...@gmail.com
wrote:
These assert should tell you what is wrong with the control flow.
Please look at control_flow_insn_p, which condition returns true.
There is a note
On Fri, Nov 9, 2012 at 1:36 PM, Jakub Jelinek ja...@redhat.com wrote:
On Fri, Nov 09, 2012 at 01:29:18PM +0100, Uros Bizjak wrote:
On Fri, Nov 9, 2012 at 1:18 PM, Vladimir Yakovlev vbyakov...@gmail.com
wrote:
These assert should tell you what is wrong with the control flow.
Please look at
Hi,
r193098 introduced a change to the loop unroll behaviour, where exits
beyond nb_iterations_upper_bound were removed as being redundant. This
assumption is not true for an undefined behaviour, which is when a loop
causes access beyond bounds of an array. In such a case, all exits are
removed
Dodji Seketeli wrote:
This patch imports the initial state of asan as it was in the
Google branch.
It provides basic infrastructure for asan to instrument memory
accesses on the heap, at -O3. Note that it supports neither stack nor
global variable protection.
I tried the 01/10 to 10/10 patch
This patch is an updated version of my patch 3 with all of richard
sandiford's comments resolved.
Richi had approved it before he went on vacation.
Committed as revision 193360.
Kenny
On 10/08/2012 03:42 PM, Richard Sandiford wrote:
Kenneth Zadeck zad...@naturalbridge.com writes:
diff
On Fri, Nov 9, 2012 at 1:47 PM, Uros Bizjak ubiz...@gmail.com wrote:
These assert should tell you what is wrong with the control flow.
Please look at control_flow_insn_p, which condition returns true.
There is a note after call insn.
(call_insn:TI 908 35558 50534 1681 (call (mem:QI
Christian Bruel christian.br...@st.com wrote:
Now that the cross-jumping problem is fixed since rev #193350, I'd like
to remove this restriction and close PR/54546.
Checked with default sh-sim target_board and
--target_board=sh-sim/-freorder-blocks-and-partition.
OK.
Regards,
kaz
On Fri, Nov 09, 2012 at 10:42:45AM +0100, Eric Botcazou wrote:
Hi,
this is a regression present on the mainline and 4.7 branch. -fvar-tracking
is automatically enabled on Darwin when -g is passed, including at -O0. This
results in bloated and incomplete debug info.
This comes from a
On 11/07/2012 02:05 PM, Bin Cheng wrote:
2012-11-08 Bin Cheng bin.ch...@arm.com
* gcse.c: (struct bb_data): Add new fields, old_pressure, live_in
and backup.
(calculate_bb_reg_pressure): Initialize live_in and backup.
(update_bb_reg_pressure): New.
On Fri, Nov 09, 2012 at 02:14:04PM +0100, Tobias Burnus wrote:
Dodji Seketeli wrote:
This patch imports the initial state of asan as it was in the
Google branch.
It provides basic infrastructure for asan to instrument memory
accesses on the heap, at -O3. Note that it supports neither stack
On 12-11-08 11:59 AM, Hans-Peter Nilsson wrote:
On Wed, 7 Nov 2012, Vladimir Makarov wrote:
On 12-11-07 5:27 PM, H.J. Lu wrote:
You should check !ia32 target:
/* { dg-do compile { target { ! { ia32 } } } } */
Thanks, H.J. I've just fixed it.
Index: testsuite/ChangeLog
On Wed, Nov 07, 2012 at 07:02:04AM -0800, Kostya Serebryany wrote:
1. LLVM unittests (Text file with LLVM IR and the expected result of
the transformations).
Example:
From: Andi Kleen a...@linux.intel.com
get_atomic_generic_size would error out for
__atomic_store(...,__ATOMIC_HLE_RELEASE)
Just mask it out. All the memory orders are checked completely
in builtins.c anyways.
I'm not sure what that check is for, it could be removed in theory.
Passed bootstrap
From: Andi Kleen a...@linux.intel.com
Make a first stab at the undocumented HLE_ACQUIRE/RELEASE
memory model modifiers in the documentation
gcc/:
2012-11-09 Andi Kleen a...@linux.intel.com
* doc/extend.texi: Document __ATOMIC_HLE_ACQUIRE and RELEASE.
---
gcc/doc/extend.texi | 10
On 2012-11-09 07:03, Andi Kleen wrote:
PR55139
* c-common.c (get_atomic_generic_size): Mask with
MEMMODEL_MASK
Ok.
r~
Hello Jakub,
with your patch and 4.8.0 20121109 the problem in PR55137 vanishes and I am
able to run the RTEMS testsuite on PowerPC again. Thanks.
--
Sebastian Huber, embedded brains GmbH
Address : Obere Lagerstr. 30, D-82178 Puchheim, Germany
Phone : +49 89 18 90 80 79-6
Fax : +49 89
Hi!
Just nits.
On Fri, Nov 09, 2012 at 04:03:47PM +0100, Andi Kleen wrote:
From: Andi Kleen a...@linux.intel.com
get_atomic_generic_size would error out for
__atomic_store(...,__ATOMIC_HLE_RELEASE)
Just mask it out. All the memory orders are checked completely
in builtins.c anyways.
On Fri, Nov 09, 2012 at 04:04:15PM +0100, Andi Kleen wrote:
+Additional target specific flags that can be ored with the memory models.
+@table @code
+@item __ATOMIC_HLE_ACQUIRE
+(on i386/x86_64). Start a Hardware Lock Elision transaction with the atomic
+operation. The memory model must be
On Fri, Nov 9, 2012 at 5:11 AM, Siddhesh Poyarekar siddh...@redhat.com wrote:
I understand that the behaviour is undefined, but this is easily
avoidable by skipping removal of the exits if it results in an infinite
loop. Attached patch does exactly that.
I don't mind saying that GCC should
The following patch fixes
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55154
The insn in question has 2 alternatives. The first alternative is
rejected because operand constraint is class containing non-allocated
hard reg for given options. The second alternative was rejected by LRA
as
On Fri, Nov 9, 2012 at 5:08 AM, Florian Weimer fwei...@redhat.com wrote:
On 11/08/2012 08:43 PM, Lubos Lunak wrote:
The patch implements an attribute for marking types for which gcc cannot
on
its own issue warnings about unused variables (e.g. because the ctor is
external), but for which
On Fri, 9 Nov 2012, Terry Guo wrote:
You are right that we should make script POSIX compliant. This v3 patch
removed function and local which don't belong to POSIX standard. I also
verified that CONFIG_SHELL is passed to this script with value /bin/sh.
Suppose /bin/sh is not a POSIX shell but
On 2012-11-09 07:18, Jakub Jelinek wrote:
I think it actually should be removed, given that e.g.
int i, j;
void
foo (void)
{
__atomic_store (i, j, 12);
}
warns twice for the same thing, when it should only warn once.
Richard?
Makes sense.
r~
It seems that my one line fix in asan branch (r192605) is not included
in Dodji's patch set.
David
On Fri, Nov 9, 2012 at 5:58 AM, Jakub Jelinek ja...@redhat.com wrote:
On Fri, Nov 09, 2012 at 02:14:04PM +0100, Tobias Burnus wrote:
Dodji Seketeli wrote:
This patch imports the initial state
Richard Henderson r...@redhat.com writes:
On 2012-11-09 07:18, Jakub Jelinek wrote:
I think it actually should be removed, given that e.g.
int i, j;
void
foo (void)
{
__atomic_store (i, j, 12);
}
warns twice for the same thing, when it should only warn once.
Richard?
Makes sense.
On 2012-11-08 17:23, Richard Henderson wrote:
I believe this is the sort of patch that Torvald was talking about
for handling abortTransaction with RTM.
FYI, I realized that this patch doesn't handle aborts on a
nested transaction properly.
r~
On Tue, 2012-11-06 at 08:11 -0800, Richard Henderson wrote:
On 2012-11-05 17:09, Torvald Riegel wrote:
+ if (likely(htm_fastpath (prop pr_hasNoAbort)))
For reference, could the NoAbort clause be relaxed with an htm check
in abortTransaction, and the use of an xabort insn with an
Jakub Jelinek wrote:
On Fri, Nov 09, 2012 at 02:14:04PM +0100, Tobias Burnus wrote:
I tried the 01/10 to 10/10 patch series but it doesn't trigger for
the following test case:
[...]
Can't reproduce that (admittedly with asan SVN branch rather than the patchset):
I can reproduce both;
Other issues:
* libasan does not seem to be a multilib, at least I only find the 64bit
version on x86-64-gnu-linux such that -m32 compilation fails.
That is because originally configure file is shared between asan and
tsan (tsan doesn't support 32 bit). Diego has suggested me to split
the
Until and unless we need to do something special in the hook,
use a generic function. Change this before other targets end
up duplicating the mistake.
Committed.
r~
* config/i386/i386.c (ix86_lra_p): Remove.
(TARGET_LRA_P): Use hook_bool_void_true.
diff --git
On Fri, 9 Nov 2012 17:34:26 +0100, Jan wrote:
I don't mind saying that GCC should define cases that the language
standards leave undefined. But that does not seem to be what this
patch does. I don't see why this is a good idea. It seems to
produce a program that is unpredictable in a
On Fri, Nov 9, 2012 at 10:17 AM, H.J. Lu hjl.to...@gmail.com wrote:
Since x32 runs in 64-bit mode, for address -0x4300(%rax), hardware
sign-extends displacement from 32-bits to 64-bits and adds it to %rax.
But x32 wants 32-bit -0x4300, not 64-bit -0x4300. This patch
uses 32-bit
Is this a hack around the underlying problem on darwin or the
implementation of the same approach used on other targets? If the former,
shouldn't we have a PR opened to properly fix this in the long term?
Neither, it's a fix and other targets don't care about -fvar-tracking-uninit.
--
On Thu, 2012-11-08 at 17:23 -0800, Richard Henderson wrote:
+ // Honor an abort from abortTransaction.
+ else if (htm_abort_is_cancel(ret))
+ return a_abortTransaction | a_restoreLiveVariables;
The problem is that we cannot reliably detect whether an abort with a
On Thu, 2012-11-08 at 20:34 -0800, Andi Kleen wrote:
Richard Henderson r...@redhat.com writes:
static inline void
-htm_abort ()
+htm_abort_retry ()
{
// ??? According to a yet unpublished ABI rule, 0xff is reserved and
// supposed to signal a busy lock. Source:
Torvald Riegel trie...@redhat.com writes:
I'm not sure this is quite true. If a libitm-executed transaction is
started from within some other transactional region (e.g., managed
explicitly by the user), and those two disagree about what is an abort
that should be retried or not, then this
Torvald Riegel trie...@redhat.com writes:
On Thu, 2012-11-08 at 17:23 -0800, Richard Henderson wrote:
+ // Honor an abort from abortTransaction.
+ else if (htm_abort_is_cancel(ret))
+ return a_abortTransaction | a_restoreLiveVariables;
The problem is that we cannot
On Fri, 2012-11-09 at 10:24 -0800, Andi Kleen wrote:
Torvald Riegel trie...@redhat.com writes:
On Thu, 2012-11-08 at 17:23 -0800, Richard Henderson wrote:
+ // Honor an abort from abortTransaction.
+ else if (htm_abort_is_cancel(ret))
+ return
Now that we have two code paths, the testsuite needs to be adjusted.
Committed as obvious.
commit 1cfee58f7559592cc21e8c68754a1c0093c5876d
Author: Aldy Hernandez al...@redhat.com
Date: Fri Nov 9 12:42:00 2012 -0600
+ * g++.dg/tm/pr47530-2.C: Adjust for uninstrumented code path.
On 11/08/2012 11:55 PM, Ed Smith-Rowland wrote:
+literal number suffixes as Gnu extensions.
GNU is all-caps. OK with that change.
Jason
On 11/09/2012 12:12 AM, Siddhesh Poyarekar wrote:
Hi,
I found that some test cases access arrays beyond their bounds. I
looked up their originating bugzillas and found that the test
cases for pr22506 and pr34005 were likely to be typos since the
original test cases in the report do not have
2. Large Gtest-based unittest. This is a set of c++ files that should
be built with the asan switch, depends on gtest
(http://code.google.com/p/googletest/).
http://llvm.org/viewvc/llvm-project/compiler-rt/trunk/lib/asan/tests/asan_test.cc?revision=166104view=markup
This should be
On Fri, Nov 09, 2012 at 11:05:37AM -0800, Wei Mi wrote:
gtest integrate multiple tests into the same file with each test being
a single line check. I cannot think out a method to migrate it to
dejagnu without using gtest, except splitting a single gtest file to
multiple files with each file
On Fri, 09 Nov 2012 11:57:44 -0700, Jeff wrote:
The off-by-one aspects of 31227 ought to be corrected.
Ok for the trunk,
I don't have write access (not sure if I qualify for it yet with just a
couple of trivial patches so far) so I need someone to commit for me.
Thanks,
Siddhesh
On Fri, Nov 09, 2012 at 08:13:12PM +0100, Jakub Jelinek wrote:
On Fri, Nov 09, 2012 at 11:05:37AM -0800, Wei Mi wrote:
gtest integrate multiple tests into the same file with each test being
a single line check. I cannot think out a method to migrate it to
dejagnu without using gtest,
On 11/09/2012 12:14 PM, Siddhesh Poyarekar wrote:
On Fri, 09 Nov 2012 11:57:44 -0700, Jeff wrote:
The off-by-one aspects of 31227 ought to be corrected.
Ok for the trunk,
I don't have write access (not sure if I qualify for it yet with just a
couple of trivial patches so far) so I need
Hello Teresa,
It seems to me that it's better if you commit it along with your set
of fixes. My patch doesn't fix any bugs, it just exposes them :-)
Ciao!
Steven
On Fri, Nov 9, 2012 at 9:09 PM, Teresa Johnson tejohn...@google.com wrote:
Hi Steven,
I've spent this week trying to clean up
Ok, thanks. Will do, with appropriate credit. =)
Teresa
On Fri, Nov 9, 2012 at 12:12 PM, Steven Bosscher stevenb@gmail.com wrote:
Hello Teresa,
It seems to me that it's better if you commit it along with your set
of fixes. My patch doesn't fix any bugs, it just exposes them :-)
Ciao!
The attached test case ICEs (segfault) both on the asan branch and on
the trunk with Dodji's patches:
fail31.ii: In static member function 'static std::size_t
std::char_traitschar::length(const char_type*)':
fail31.ii:13:19: internal compiler error: Segmentation fault
static size_t
I'm not sure this is quite true. If a libitm-executed transaction is
It's just a convention. You don't have to use it. Not doing it will
just make abort profiling harder.
-Andi
My patch for 54575 started using instantiate_template for alias
templates, but instantiate_template wasn't prepared to instantiate with
dependent template arguments. Fixed thus.
Tested x86_64-pc-linux-gnu, applied to trunk.
commit 11ab97f0fe9a7262362f4ef997562f32028e18e4
Author: Jason Merrill
I've merged latest changes made on the trunk (up to rev. 193365) into
lra branch.
This removes s-taspri-lynxos.ads, which is unused, and converts the 3 last
files under GPL V2+ (plus exception) to GPL V3+ (plus exception), as all the
other similar files. Applied on mainline and 4.7 branch.
2012-11-09 Eric Botcazou ebotca...@adacore.com
PR other/52438
*
Tobias Burnus wrote:
The attached test case ICEs (segfault) both on the asan branch and on
the trunk with Dodji's patches:
I found another ICE - this time without a patch.
[That's with the patch, which I posted in this thread. Without, one
seems to run into the problem I tried to fix with
On 11/09/2012 05:09 PM, Jason Merrill wrote:
Now that G++ uses the value of __cplusplus specified by the standard,
we don't need the other macro anymore.
OK for trunk, or should I save it for the next stage 1?
Jason
This looks like a non-user-facing mechanical change that makes sense.
FWIW
Hi Jason,
Made all the changes and attached patch. Ok to commit?
Thanks,
-Sri.
On Tue, Nov 6, 2012 at 7:52 AM, Jason Merrill ja...@redhat.com wrote:
On 11/05/2012 09:38 PM, Sriraman Tallam wrote:
+ /* For multi-versioned functions, more than one match is just fine.
+Call
[re-sending so that it gets to the list]
asan's design choice is to hard crash on the first error.
Doing many asan checks in the same process might become too messy too soon.
I wouldn't try to implement it myself.
Playing with longjmp in asan is not the best idea because asan
intercepts longjmp
On Mon, 2012-11-05 at 11:20 -0800, Andrew Pinski wrote:
On Mon, 2012-11-05 at 19:19 +, Richard Sandiford wrote:
Hi Andrew,
Andrew Pinski andrew.pin...@caviumnetworks.com writes:
On Fri, Oct 5, 2012 at 8:43 PM, Andrew Pinski
andrew.pin...@caviumnetworks.com wrote:
On Sun, Aug
Hi,
While I was working on using zero_extract more on the left hand side
in combine, I found a bug in my code but there was no testcase in the
testsuite that tested it. The code originally comes from uboot which
was failing when doing a ping or tftp.
Committed the testcase as obvious after a
Hi,
The problem here is that set PLUGIN_LD_SUFFIX to ld-new which is not
the final installed binary name. This patch fixes the problem by
changing if we got ld-new to just ld.
Note this issue has been around since 4.6 but not many people test
installed binary and it does not show up while doing
Hi,
In
(insn 19 17 20 2 (set (reg:TI 85 [ *_15 ])
(mem:TI (zero_extend:DI (reg:SI 82)) [0 *_15+0 S16 A32])) x.i:29 61
{*movti_internal_rex64}
(expr_list:REG_DEAD (reg:SI 82)
(expr_list:REG_EQUIV (mem/c:TI (plus:DI (reg/f:DI 20 frame)
(const_int -16
74 matches
Mail list logo
