On Fri, Mar 23, 2012 at 3:40 PM, Cary Coutant wrote:
>
> 2012-03-23 Cary Coutant
>
> * dwarf2out.c (resolve_addr): Remove address table entry for symbol
> when removing location list entry.
>
>
> Index: dwarf2out.c
> ===
2012-03-23 Cary Coutant
* dwarf2out.c (resolve_addr): Remove address table entry for symbol
when removing location list entry.
Index: dwarf2out.c
===
--- dwarf2out.c (revision 185594)
+++ dwarf2out.c (working cop
> Do you mean use the weights to decide the shape of the binary tree
Yes.
> (similar to COST_TABLE heuristic)?
COST_TABLE should die I hope.
> I am planning to send a separate patch for that.
Great.
-Andi
On Fri, Mar 23, 2012 at 3:29 PM, Andi Kleen wrote:
> Easwaran Raman writes:
>
>> Some more background on this patch: Right now, while the execution
>> counts of different case labels of a switch statement are obtained
>> during profile collection, they are not propagated to RTL. Instead,
>> count
Easwaran Raman writes:
> Some more background on this patch: Right now, while the execution
> counts of different case labels of a switch statement are obtained
> during profile collection, they are not propagated to RTL. Instead,
> counts are regenerated at the RTL level using static heuristics
Fix bindings for using declarations.
When an identifier is declared in a using declaration, we were
registering the USING_DECL object in the corresponding binding.
This was wrong, and at the time I hacked around it by simply ignoring
USING_DECLs when setting bindings. That was wrong too. What w
Some more background on this patch: Right now, while the execution
counts of different case labels of a switch statement are obtained
during profile collection, they are not propagated to RTL. Instead,
counts are regenerated at the RTL level using static heuristics that
tend to weigh branches equal
On Fri, Mar 23, 2012 at 11:47 AM, David Edelsohn wrote:
> Some early releases of AIX 6 include a malloc.h header with an
> incorrect use of "builtin" keyword:
>
> #ifdef __cplusplus
> extern "C" {
> extern "builtin" char *__alloca (size_t);
> # define alloca __alloca
> }
> #endif /* def __cplus
Some early releases of AIX 6 include a malloc.h header with an
incorrect use of "builtin" keyword:
#ifdef __cplusplus
extern "C" {
extern "builtin" char *__alloca (size_t);
# define alloca __alloca
}
#endif /* def __cplusplus */
GCC is not too happy about this. Now that GCC mainline and 4.7
This patch to libgo fixes the handling of errno in syscall.Syscall and
friends. Previously it would return a non-zero errno value even when
the system call succeeded. Bootstrapped and ran Go testsuite on
x86_64-unknown-linux-gnu. Committed to mainline and 4.7 branch.
Ian
diff -r b626c77ab40e l
This patch propagates execution count of thee case labels of a
switch-case statement after its expansion. Bootstraps and all
tests pass. OK for trunk?
2012-03-23 Easwaran Raman
* cfgbuild.c (non_zero_profile_counts): New function.
(compute_outgoing_frequencies): If at least on
> -Ursprüngliche Nachricht-
> Von: Andrew Stubbs [mailto:a...@codesourcery.com]
> Gesendet: Montag, 19. März 2012 17:12
> An: EXTERNAL Waechtler Peter (Fa. TCP, CM-AI/PJ-CF31)
> Cc: gcc-patches@gcc.gnu.org; libstd...@gcc.gnu.org;
> p...@codesourcery.com; pwaecht...@mac.com; d...@false.org
>
... here's the patch.2012-03-22 Benjamin Kosnik
* scripts/testsuite_flags.in (--cxxflags): Let CXXFLAGS set -g -O2.
* testsuite/lib/libstdc++.exp: Same.
diff --git a/libstdc++-v3/scripts/testsuite_flags.in b/libstdc++-v3/scripts/testsuite_flags.in
index 457adaf..f77784b 100755
--- a/libstd
Removes duplicates on command line when running test suite. CXXFLAG
info was being quoted twice.
tested x86/linux
-benjamin
This is a similar test to the previous acquire test. Here we are
incorrectly caching 'x' and failing to reload it after the __ATOMIC_ACQUIRE.
+ i = x + y;
+
+ if (__atomic_load_n (&flag, __ATOMIC_ACQUIRE))
+{
+ /* x here should not be reused from above. */
+ k = x;
+}
Note
After much pondering and talking with both you and Torvald, it has been
determined that the test at hand is technically allowed to hoist the
value of x/y because the standard guarantees that the code below is data
race free:
if (__atomic_load_n (&flag1, __ATOMIC_ACQUIRE))
On 03/23/2012 10:39 AM, Aldy Hernandez wrote:
On 03/21/12 12:54, Andrew MacLeod wrote:
On 03/21/2012 01:35 PM, Aldy Hernandez wrote:
In the test below, we cannot cache either [x] or [y] neither before
the load of flag1 nor the load of flag2. This is because the
corresponding store/release can f
On 03/21/12 12:54, Andrew MacLeod wrote:
On 03/21/2012 01:35 PM, Aldy Hernandez wrote:
In the test below, we cannot cache either [x] or [y] neither before
the load of flag1 nor the load of flag2. This is because the
corresponding store/release can flush a different value of x or y:
+ if (__atom
The build process of libquadmath sometimes encounters problems on AIX
due to multilib and LD_LIBRARY_PATH interfering with GCC's own library
dependencies. Libquadmath is not used on AIX, so this patch adds it
to noconfigdirs.
Thanks, David
* configure.ac: Add libquadmath to noconfigdirs
Hi,
in a few places in libgfortran we use malloc (or rather, our own
xmalloc-like wrapper, get_mem) followed by a memset to zero the
memory. The attached patch replaces this usage with calloc().
Regtested on x86_64-unknown-linux-gnu, unless somebody objects I'll
commit it as obvious in a few days
On Wed, 21 Mar 2012, Richard Sandiford wrote:
> Richard Guenther writes:
> > This patch makes us preserve loop structures from the start of tree
> > loop optimizers to the end of RTL loop optimizers. It uses a new
> > property, PROP_loops to indicate we want to preserve them and
> > massages loo
Hi,
since we now should be able to expand misaligned MEM_REFs properly and
both SRA and IPA-SRA now tag the memory accesses with the appropriate
alignment information, we should now be able to get rid off the SRA
disabling in the face of potential strict-alignment expansion
problems. The patch be
On 03/23/2012 02:02 PM, Paolo Bonzini wrote:
It was approved by Mike Stump who obviously confused the *.in with *.am,
or libstdc++-v3 with gcc. It happens. :)
Good, good, thanks Paolo. The messages on the mailing list aren't fully
threaded and I didn't see immediately the approval.
And of cou
Il 23/03/2012 13:40, Paolo Carlini ha scritto:
> Hi,
>
> On 03/07/2012 06:22 AM, Terry Guo wrote:
>> Hello,
>>
>> Can anybody please review and approve the following simple patch? Thanks
>> very much.
>>
>> http://gcc.gnu.org/ml/libstdc++/2011-08/msg00063.html
> apparently somebody, somewhere, app
Hi,
On 03/07/2012 06:22 AM, Terry Guo wrote:
Hello,
Can anybody please review and approve the following simple patch? Thanks
very much.
http://gcc.gnu.org/ml/libstdc++/2011-08/msg00063.html
apparently somebody, somewhere, approved this patch, because I see:
2012-03-13 Terry Guo
* test
On Wed, Mar 21, 2012 at 08:46:49AM +0100, Richard Guenther wrote:
> On Tue, 20 Mar 2012, Martin Jambor wrote:
>
> > Hi,
> >
> > On Tue, Mar 20, 2012 at 04:08:31PM +0100, Richard Guenther wrote:
> > > On Tue, 20 Mar 2012, Martin Jambor wrote:
> > >
> > > > Hi,
> > > >
> > > > this patch which re
This patch fixes PR52678 - re-computing the evolution of a loop
PHI node after updating it during prologue loop peeling is fragile
(and I remember several cases in the past where we ICEd here and
adjusted the code). So, instead of re-computing the evolution
this patch saves the evolution part at
Joern Rennecke writes:
> 2012-03-23 Joern Rennecke
>
> * htdocs/gcc-4.7/changes.html: Document addition of Epiphany support.
>
> Index: htdocs/gcc-4.7/changes.html
> ===
> RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.7/changes.ht
2012-03-23 Joern Rennecke
* htdocs/gcc-4.7/changes.html: Document addition of Epiphany support.
Index: htdocs/gcc-4.7/changes.html
===
RCS file: /cvs/gcc/wwwdocs/htdocs/gcc-4.7/changes.html,v
retrieving revision 1.102
di
Mike Stump writes:
>> Sorry, with this bit, I meant that the current svn code is correct
>> for GET_MODE_BITSIZE (op_mode) == HOST_BITS_PER_WIDE_INT * 2.
>> In that case, hv < 0 can just mean that we have a uint128_t
>> (or whatever) whose high bit happens to be set.
(To be clear, I was using uin
On Wed, 21 Mar 2012 11:20:13 +
Richard Earnshaw wrote:
> Semantically the neon intrinsic vgetq_lane_[su]64 returns a 64 bit
> sub-object of a 128-bit vector; there's no real need for the intrinsic
> to map onto a specific machine instruction.
>
> Indeed, if force a particular instruction tha
On 23 March 2012 07:47, David Miller wrote:
>
> I verified that this matches what other Linux platforms have added
> recently, and I was careful not to accidently include the TLS symbols.
>
> In fact, this baseline file now matches powerpc-linux-gnu's precisely.
>
> Ok for mainline and the 4.7 bran
On Thu, 22 Mar 2012, Richard Guenther wrote:
> On Thu, 22 Mar 2012, Eric Botcazou wrote:
>
> > > bitregion_start == 11 looks bogus. The representative is starting at
> > >
> > > DECL_FIELD_BIT_OFFSET (repr)
> > > = size_binop (BIT_AND_EXPR,
> > > DECL_FIELD_BIT_OFFSET (fi
Hi,
the maximum line length is nowadays decided dynamically at runtime as
specified by the chosen standard version, or overridden by the
-fline-length option. The patch below removes the GFC_MAX_LINE macro
which is no longer used. Committed to trunk as obvious.
Index: gfortran.h
Committed.
Richard.
Committed as obvious.
Richard.
2012-03-23 William Bader
c-family/
* c-lex.c (c_lex_with_flags): Avoid declarations after stmts.
Index: gcc/c-family/c-lex.c
===
*** gcc/c-family/c-lex.c(revision 185724)
-
I verified that this matches what other Linux platforms have added
recently, and I was careful not to accidently include the TLS symbols.
In fact, this baseline file now matches powerpc-linux-gnu's precisely.
Ok for mainline and the 4.7 branch?
Thanks.
2012-03-23 David S. Miller
*
37 matches
Mail list logo
