This fixes PR56150 - mixed store/loads are not equal.
Bootstrapped and tested on x86_64-unknown-linux-gnu, applied to trunk.
Richard.
2013-01-30 Richard Biener
PR tree-optimization/56150
* tree-ssa-tail-merge.c (gimple_equal_p): Properly handle
mixed store non-store
This reverts the change to go into loop-closed SSA form for
virtual operands. Nothing relys on that and the verifier doesn't
verify it works. Furthermore SSA updating will destroy loop-closed
virtual SSA form very quickly. As the PR shows it can be quite
costly to go into loop-closed SSA form f
On Wed, 30 Jan 2013, Pat Haugen wrote:
> On 01/29/2013 04:53 AM, Richard Biener wrote:
> > I'm curious about the affect of -fno-aggressive-loop-optimizations
> > on SPEC CPU 2006 numbers (not curious enough to try for myself
> > though). Both on extra PASSes for official latest sources
> > (I hav
On Thu, 31 Jan 2013, Richard Biener wrote:
> On Wed, 30 Jan 2013, Pat Haugen wrote:
>
> > On 01/29/2013 04:53 AM, Richard Biener wrote:
> > > I'm curious about the affect of -fno-aggressive-loop-optimizations
> > > on SPEC CPU 2006 numbers (not curious enough to try for myself
> > > though). Bot
Kai Tietz writes:
> this patch fixes wrong handling of cases that bitness of size_t is
> wider as 32-bit.
>
> ChangeLog
>
> 2013-01-30 Kai Tietz
>
> PR other/543413
^ this is clearly wrong (6 digits)
> * md5.c (md5_process_block): Handle case that size_t is
>
On Wed, Jan 30, 2013 at 9:45 PM, Marc Glisse wrote:
> On Tue, 29 Jan 2013, Richard Biener wrote:
>
>> On Sun, Jan 27, 2013 at 4:28 PM, Marc Glisse wrote:
>>>
>>> Hello,
>>>
>>> this message is to check that I am not doing something absurd and ask for
>>> a
>>> bit of advice.
>>>
>>> In the attach
2013/1/31 Rainer Orth :
> Kai Tietz writes:
>
>> this patch fixes wrong handling of cases that bitness of size_t is
>> wider as 32-bit.
>>
>> ChangeLog
>>
>> 2013-01-30 Kai Tietz
>>
>> PR other/543413
>
> ^ this is clearly wrong (6 digits)
>
>> * md5.c (md5_process_
Hi!
As the pr56154-{1,2,3}.c testcases show, sometimes we start a location
list with an empty range (either for to PR49382, or because the first
insn in the function is inline asm (or many of them) which has zero
size and the compiler doesn't know it). If
!have_multiple_function_sections, we emit
Hi!
The Linux kernel started (recently?) using -mfentry -p on x86_64/i686?
for some instrumentation, I believe they normally overwrite the -mfentry
prologues with nops and only when some function needs to be instrumented,
overwrite it with some call or jump.
The problem in this PR is that var-trac
Hello,
Sorry for the long delay (ref http://patchwork.ozlabs.org/patch/199397/)
On 6 December 2012 20:26, Teresa Johnson wrote:
>
>
>
> On Wed, Nov 28, 2012 at 7:48 AM, Christophe Lyon
> wrote:
>>
>> I have updated my trunk checkout, and I can confirm that eval.c now
>> compiles with your pat
The attached patch fixes a race condition in libgomp. Based on the suggestions
by Dmitry, and verified that it fixes the corresponding sanitizer warnings.
2012-12-30 Dmitry Vyukov
PR libgomp/55561
* config/linux/wait.h (do_spin): Use atomic load for addr.
* config/lin
On Thu, Jan 31, 2013 at 02:50:51PM +, VandeVondele Joost wrote:
> The attached patch fixes a race condition in libgomp. Based on the
> suggestions by Dmitry, and verified that it fixes the corresponding sanitizer
> warnings.
>
> 2012-12-30 Dmitry Vyukov
Ok, but please use current date
In my 56104 patch I failed to notice that in the ARM (vbit in delta)
case, we need to adjust the delta even if we know we're doing a
non-virtual call.
Tested x86_64-pc-linux-gnu, applying to trunk.
commit 6a52d34876377448990d550cb763171180f4
Author: Jason Merrill
Date: Thu Jan 31 09:45:
We ended up inserting the type in the array of types to add template
parameters to twice, so we added the parameters twice. Fixed by only
adding it the first time; it doesn't matter whether we happen to be
creating a declaration or definition DIE at this point, since when we
get to EOF we'll a
This fixes PR56157 - when looking up operands for a vectorized
SLP stmt we have to match up SLP childs with operands. We do
so by looking for the operand in the DEF position of the SLP
child. But that get's complicated in face of either stmt being
a pattern stmt and its operands being DEFed by p
The libbacktrace interfaces says that the strings passed to the callback
routine may disappear. As it happens, in the current implementation
they will not. The Go backtrace routine needs to hang on to the
strings, and I've discovered that it can't reliably allocate memory,
since it might be calle
On Thu, Jan 31, 2013 at 05:31:12PM +0100, Richard Biener wrote:
> 2013-01-31 Richard Biener
>
> PR tree-optimization/
> * tree-vect-slp.c (vect_get_slp_defs): More thoroughly try to
> match up operand with SLP child.
>
> * gcc.dg/torture/pr56157.c: New testcase.
Ok, th
The updated changelog entry is below, but somebody with write access should do
the commit, please.
2013-01-31 Dmitry Vyukov
Joost VandeVondele
PR libgomp/55561
* config/linux/wait.h (do_spin): Use atomic load for addr.
* config/linux/ptrlock.c (gomp_
On 01/28/13 14:03, amol pise wrote:
Dear Ramana,
Thank You very much for the changelog and commit of my patch in gcc.
I will follow the steps mentioned by you.
There are no vector forms for the vfnma and vfnms instructions.
A co-worker (thanks Kyryll) just pointed out to me that I'd misread
This patch to libgo disables signals while creating a new thread.
Otherwise if a signal comes in between the time the thread starts and
the time the thread initializes its m and g TLS variables, the program
will crash with an error "signal received on thread not created by Go."
Signals are already
Using libbacktrace more extensively from Go has revealed a race
condition in libbacktrace. I cleverly save some memory by reusing a
vector in the DWARF data structure. However, that is fairly hard to do
if multiple threads are simultaneously getting a backtrace. This patch
changes the code to al
Hi Guys,
I am applying the patch below to add support for the E3V5 architecture
variant to the V850 backend. This patch was originally developed by
Renesas, but it is only recently that their copyright assignment was
completed so that it could be contributed.
I realize that the current
I've merged trunk revision 195620 to the gccgo branch.
Ian
None of these changes affect the code, but they provide some more information
that I've found useful when using the -mdebug=reg and -mdebug=addr options.
When GCC 4.9 opens up, can I install these patches in the source tree>
2013-01-31 Michael Meissner
* config/rs6000/rs6000.c (rs6000
In an earlier patch I tried to remove confusion about split-stack
functions in backtraces, but it still doesn't work in conjunction with
skipping functions. Both runtime.Callers and backtrace_full can skip
stack frames. backtrace_full will count split-stack frames in the skip
count, but runtime.C
Continuing my quest for some really stable gccgo sources, I've merged
trunk revision 195627 to the gccgo branch.
Ian
Hi Nick,
the web page update required a tweak: goes at the end of the
entry. Committed.
Gerald
Index: changes.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.8/changes.html,v
retrieving revision 1.91
diff -u -3 -p -r1.91 changes.
On Jan 30, 2013, at 12:18 AM, Jakub Jelinek wrote:
> I've noticed
> ERROR: (DejaGnu) proc "lreverse {{ASAN_OPTIONS 0}}" does not exist.
>
> errors when regtesting gcc 4.8 on RHEL 5, it seems lreverse has been added
>
When using an old assembler it's possible to get debug info that says
that a PC is in morestack.S but does not have the function name. This
patch changes the Go backtrace code to recognize this case.
Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu.
Committed to mainline.
Ian
diff -
On Wed, Jan 30, 2013 at 6:24 PM, Vladimir Makarov wrote:
> The following patch fixes
>
> http://gcc.gnu.org/bugzilla/show_bug.cgi?id=56144
>
> The patch was successfully bootstrapped and tested on x86/x86-64.
Hello Vlad,
Can you please put this patch on the lra-branch too, so that the
auto-tester
cmt...@google.com
On Wed, Jan 30, 2013 at 2:09 AM, Florian Weimer wrote:
> On 11/01/2012 09:07 PM, Caroline Tice wrote:
>>
>> We have been developing a new security hardening feature for GCC that
>> is designed to detect and handle (during program execution) when a
>> vtable pointer that is abou
When a source file has multiple identical unnamed structs, and those
structs have methods, gccgo would emit multiple copies of the methods.
That would be bad because each copy would have the same name, causing an
assembler error about multiply defined symbols. This patch fixes the
problem. Bootst
On Wed, Jan 30, 2013 at 9:26 AM, Jason Merrill wrote:
> I'm also touching on the middle-end and library changes, but would
> appreciate a more thorough review from others in those areas.
>
> On 01/23/2013 05:25 PM, Caroline Tice wrote:
>>
>> Index: gcc/cp/g++spec.c
>
>
> Changes to g++spec.c only
It's true, I've merged trunk to gccgo branch again, in this case
revision 195638.
Ian
I am an idiot. (Hope it is right this time.)
Ian
diff -r 686169e30cff libgo/runtime/go-callers.c
--- a/libgo/runtime/go-callers.c Thu Jan 31 16:22:11 2013 -0800
+++ b/libgo/runtime/go-callers.c Thu Jan 31 21:42:52 2013 -0800
@@ -49,7 +49,7 @@
p = strrchr (filename, '/');
if (p == N
I merged trunk revision 195640 to the gccgo branch.
Ian
36 matches
Mail list logo
