Gerald Pfeifer writes:
> Like some other domains, redhat.com now defaults to http, so this
> adjusts the link.
Are you sure it's not something on your side?
Andreas.
--
Andreas Schwab, sch...@linux-m68k.org
GPG Key fingerprint = 58CA 54C7 6D53 942B 1756 01D3 44D5 214B 8276 4ED5
"And now for
Hi!
As discussed in the PR, *bswapdi2_doubleword splitter would generate wrong
code if there is overlap in between the destination register and the memory
address. Without bswapdi2 pattern for !TARGET_64BIT, we generate sometimes
better, usually same quality and sometimes slightly worse code, so
On Thu, Nov 01, 2012 at 10:27:57AM +0100, Jakub Jelinek wrote:
> The patch unfortunately doesn't speed test-tgmath2.i compilation
> significantly, but decreases number of debug stmts from ~ 36000 to ~ 16000,
> the 2+ were clearly redundant.
Testing found a buglet in the patch (doing is_gimple_
> Index: gcc/ChangeLog
>
> 2012-10-31 Lawrence Crowl
>
> * is-a.h: New.
> (is_a (U*)): New. Test for is-a relationship.
> (as_a (U*)): New. Treat as a derived type.
> (dyn_cast (U*)): New. Conditionally cast based on is_a.
> * cgraph.h (varpool_node): Rename
Hi,
I posted a patch improving register pressure directed hoist at
http://gcc.gnu.org/ml/gcc-patches/2012-10/msg02552.html
Turns out it has a mistake in update_bb_reg_pressure resulting in changing
register pressure incorrectly. Here comes the 2nd version patch for review.
Unfortunately and strange
Applied.
2012-11-02 Gerald Pfeifer
* doc/xml/manual/allocator.xml: Update reference to Hoard.
Update reference to Berger's OOPSLA 2002 paper.
Index: doc/xml/manual/allocator.xml
===
--- doc/xml/manual/allocator.xm
"Steve Ellcey " writes:
> I checked in this patch as obvious. It fixes PR 55103 which is a compiler
> abort on MIPS with the -mips16 flag. After committing I realized I forgot
> to put the bug number in the ChangeLog, I will update PR 55103 by hand.
Thanks
Hello,
With this patch a few more cases are handled where the T bit is stored
and then re-tested again before conditional branches.
Tested on rev 193061 with
make -k check RUNTESTFLAGS="--target_board=sh-sim
\{-m2/-ml,-m2/-mb,-m2a/-mb,-m4/-ml,-m4/-mb,-m4a/-ml,-m4a/-mb}"
and no new failures.
OK?
Hello,
This mainly reformats some of the asm output code in sh.c to use
multi-line strings.
Tested on rev 193061 with
make -k check RUNTESTFLAGS="--target_board=sh-sim
\{-m2/-ml,-m2/-mb,-m2a/-mb,-m4/-ml,-m4/-mb,-m4a/-ml,-m4a/-mb}"
and no new failures.
OK?
Cheers,
Oleg
gcc/ChangeLog:
*
... and this is what I'm actually going to commit, a shorter version in
4_7-branch too. The issue I mentioned yesterday was actually trivial, we
were just missing an overload. Note that I'm not convinced we couldn't
clean up things even more, but certainly the issue as reported is
completely fi
On 2012.10.27 at 09:02 +0200, Markus Trippelsdorf wrote:
> The problem here is that we end up with an INDIRECT_REF TREE_CODE with a
> null TREE_TYPE in lvalue_kind. Is this possible at that point, or does
> it point to a deeper underlying problem?
It looks like there is indeed a deeper problem. Se
On 10/30/2012 05:30 PM, Florian Weimer wrote:
On 10/30/2012 05:17 PM, Paolo Carlini wrote:
Sorry, I don't know the code well enough to review your patch, but
since I'm in CC, I still don't understand why, instead of adding a
full libstdc++ testcase you are extending a C++ testcase, in old-deja
cfg-flags.def reads:
/* Edge out of a basic block that ends with a CALL_INSN with abnormal
exit, like an exception, or a sibcall.
This flag is only used for the RTL CFG. */
DEF_EDGE_FLAG(ABNORMAL_CALL, 2)
That's wrong, ABNORMAL_CALL isn't set for a sibcall, only ABNORMAL is.
That's eve
On 11/02/2012 01:09 PM, Florian Weimer wrote:
I looked at this again and made a new copy of the test case instead.
It has been successfully tested on x86_64-redhat-linux-gnu.
Is this okay for trunk?
Looks very nice to me, and after all the issue seems rather simple.
Let's say we wait another
On 12-11-01 7:15 PM, Steve Ellcey wrote:
I checked in this patch as obvious. It fixes PR 55103 which is a compiler
abort on MIPS with the -mips16 flag. After committing I realized I forgot
to put the bug number in the ChangeLog, I will update PR 55103 by hand.
Thanks, Steve.
On 07.08.2012 16:07, Ian Lance Taylor wrote:
On Tue, Aug 7, 2012 at 3:31 AM, Matthias Klose wrote:
ping?
On 08.07.2012 20:48, Matthias Klose wrote:
Please find attached v2 of the patch updated for trunk 20120706, x86 only,
tested on
x86-linux-gnu, KFreeBSD and the Hurd.
+[case "${withval}
Hi,
On Thu, Nov 01, 2012 at 11:11:13AM -0700, Xinliang David Li wrote:
>
...
> The TREE_ADDRESSABLE check seems redundant -- if the var_decl (instead
> of ssa name) appears in the assignment, why would it not be
> 'addressable'?
There are other reason beside being TREE_ADDRESSABLE that may ca
Hello!
Attached patch disable exceptions and rounding mode handling code in
soft-fp when _SOFT_FLOAT is enabled. We need real hardware (x87 or
SSE) to generate exceptions and real control register to know current
rounding mode.
RTEMS defines _SOFT_FLOAT, when -msoft-float is in effetct and this
m
Konstantin Serebryany writes:
>> [A cultural question I've kept asking myself is Why has address
>> sanitizer authors called these red zones (LEFT, MIDDLE, RIGHT)
>> instead of e.g, (BOTTOM, MIDDLE, TOP). Maybe they can step up and
>> educate me so that I get less confused in the future. :-)
The following patch fixes PR55130
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55130
The reason for the crash was in missed processing clobber insn with
pseudo changed on its equivalent memory. It resulted in presence of the
pseudo and in trying to remove equiv initialization insn the seco
On 12-11-02 10:45 AM, Vladimir Makarov wrote:
The following patch fixes PR55130
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=55130
The reason for the crash was in missed processing clobber insn with
pseudo changed on its equivalent memory. It resulted in presence of
the pseudo and in tryi
Xinliang David Li writes:
> Changing the option is part of the plan.
Indeed.
> Dodji, can you make the option change part of one the patches (e.g,
> the first one that introduces it) -- there seems no need for a
> separate patch for it.
Sure thing. I have done the change on my local tree. I'
Thanks, good to know that.
David
On Fri, Nov 2, 2012 at 6:31 AM, Martin Jambor wrote:
> Hi,
>
> On Thu, Nov 01, 2012 at 11:11:13AM -0700, Xinliang David Li wrote:
>>
>
> ...
>
>> The TREE_ADDRESSABLE check seems redundant -- if the var_decl (instead
>> of ssa name) appears in the assignment, why
On Fri, Nov 02, 2012 at 08:54:42AM -0700, Xinliang David Li wrote:
> Thanks, good to know that.
But note that such vars, even when accessed only partially, don't
need to be tracked by tsan just because of that.
TREE_ADDRESSABLE (decl) || is_global_var (decl) is really a conservative
check for what
Konstantin Serebryany writes:
> http://research.google.com/pubs/archive/37752.pdf
> The horizontal drawing is given in section 3.3 and hence the redzones there
> are called left/right.
> The stack poisoning is only explained using an example in C.
Great, thanks. This makes it easier to understa
Yes, agree. For now using the !(TREE_ADDRESSABLE (decl) ||
is_global_var (decl)) can be used to skip instrumentation if the
points-to interface is not available for use.
David
On Fri, Nov 2, 2012 at 8:58 AM, Jakub Jelinek wrote:
> On Fri, Nov 02, 2012 at 08:54:42AM -0700, Xinliang David Li wro
Currently on targets like x86_64 darwin which don't support -mx32
-maddress-mode=short,
tests that pass those on dg-options fail with excessive errors at -m64. The
attached patch
resolves this by adding a check_effective_target_maybe_x32 proc to test if the
-mx32
-maddress-mode=short flags are
The attached patch ports Joseph's GLIBC fmal patch from today;
additionally, I made a small adjustment to lgamma's signgam handling.
Build and tested on x86-64-gnu-linux and applied as Rev. 193099
Otherwise, it still applies what I wrote at
http://gcc.gnu.org/ml/gcc-patches/2012-11/msg00042.h
On 11/02/2012 02:08 AM, Jakub Jelinek wrote:
On Thu, Nov 01, 2012 at 10:27:57AM +0100, Jakub Jelinek wrote:
The patch unfortunately doesn't speed test-tgmath2.i compilation
significantly, but decreases number of debug stmts from ~ 36000 to ~ 16000,
the 2+ were clearly redundant.
Testing fo
Hi!
This is a RFC patch which implements masked (conditional) vector
loads/stores roughly as discussed with richi (i.e. internal function
calls, added during if-conversion pass, and if not vectorized
removed after vectorization again).
The if-unconversion pass is just very basic, I guess it could
On Fri, Nov 2, 2012 at 6:26 AM, Matthias Klose wrote:
>
>> +ifeq ($(enable_multiarch),yes)
>> + if_multiarch = $(1)
>> +else ifeq ($(enable_multiarch),auto-detect)
>> + if_multiarch = $(if $(wildcard $(shell echo
>> $(SYSTEM_HEADER_DIR))/../../usr/lib/*/crti.o),$(1))
>> +else
>> + if_multiarch
On Nov 1, 2012, Jakub Jelinek wrote:
> Even for stmt frontiers it is IMHO undesirable to duplicate
> (perhaps many times) the whole sequence, as one would then reply the var
> changing sequence in the debugger perhaps once in the original bb, then
> once or many times again in the successor bb.
On Fri, Nov 02, 2012 at 12:22:07AM +0100, Gerald Pfeifer wrote:
> On Thu, 1 Nov 2012, Michael Meissner wrote:
> > * doc/install.texi (--with-extra-prefix=): Document new configure
> > switches.
> > (--with-extra-exec-prefix=): Likewise.
> > (--with-extra-startfile-prefix=): Likewise
On 11/02/2012 05:59 PM, Tobias Burnus wrote:
The attached patch ports Joseph's GLIBC fmal patch from today;
additionally, I made a small adjustment to lgamma's signgam handling.
Build and tested on x86-64-gnu-linux and applied as Rev. 193099
My builds are now failing in libquadmath with IEEE854
Paolo Carlini wrote:
On 11/02/2012 05:59 PM, Tobias Burnus wrote:
The attached patch ports Joseph's GLIBC fmal patch from today;
additionally, I made a small adjustment to lgamma's signgam handling.
Build and tested on x86-64-gnu-linux and applied as Rev. 193099
My builds are now failing in li
Tobias Burnus:
Sorry, I seemingly applied an old premature version of the patch
instead of the one I had build and tested. Fixed by the attached patch
(Rev. 193100).
Now with a nonzero file as attachment, it helps diffing -r193099:193100
instead of -r193109:193100.
I think I need a break…
On 11/02/2012 07:10 PM, Tobias Burnus wrote:
Paolo Carlini wrote:
On 11/02/2012 05:59 PM, Tobias Burnus wrote:
The attached patch ports Joseph's GLIBC fmal patch from today;
additionally, I made a small adjustment to lgamma's signgam handling.
Build and tested on x86-64-gnu-linux and applied
During development, I often times have powerpc in the name of the compiler.
This trips up several tests that want to make sure the various pow
optimizations are done and the pow function is not called, since "powerpc" is
in the compiler identification string.
I committed the following patch as obv
David pointed out to me that pr48258-2 was failing on AIX, and it was failing
under Linux as well. In terms of the Linux failures, this was because the test
was failing if the xxsldwi was generated. I decided the test conditions were
too specific, and loosened them to just check that the compiler
On 11/1/2012 10:48 AM, Bruce Korb wrote:
Hi Robert,
On Thu, Nov 1, 2012 at 6:35 AM, rbmj wrote:
and now my patches will build on top of
trunk. Bruce, can you give steps on how to reproduce the error you
reported?
rm -rf gcc-bld gcc-ins
cp -l gcc-svn gcc-bld
pfx=$PWD/gcc-ins
cd gcc-bld
hi,
this patch makes predict_loop to predict even loop with exits not
dominating the latch when their iteration bound is known.
Honza
Index: ChangeLog
===
--- ChangeLog (revision 193101)
+++ ChangeLog (working copy)
@@ -1,5 +1,10
Lawrence, this is the change I was referring to yesterday. See
if it helps in comparing your results against clean builds.
Add a new option --clean_build to validate_failures.py
This is useful when you have two builds of the same compiler. One with
your changes. The other one, a clean build a
On Fri, Nov 2, 2012 at 3:29 PM, Michael Meissner
wrote:
> David pointed out to me that pr48258-2 was failing on AIX, and it was failing
> under Linux as well. In terms of the Linux failures, this was because the
> test
> was failing if the xxsldwi was generated. I decided the test conditions we
This is covered in the installation documentation and really not
specific to GCJ (and probably predates the merge of GCJ into GCC).
Also, it's not really an FAQ in this context.
Better focus on what is really relevant..
Applied.
Gerald
2012-11-02 Gerald Pfeifer
* faq.html: Remove q
On 11/2/12, Eric Botcazou wrote:
>> Index: gcc/ChangeLog
>>
>> 2012-10-31 Lawrence Crowl
>>
>> * is-a.h: New.
>> (is_a (U*)): New. Test for is-a relationship.
>> (as_a (U*)): New. Treat as a derived type.
>> (dyn_cast (U*)): New. Conditionally cast based on is_a.
>>
On Fri, Nov 2, 2012 at 3:24 PM, Michael Meissner
wrote:
> During development, I often times have powerpc in the name of the compiler.
> This trips up several tests that want to make sure the various pow
> optimizations are done and the pow function is not called, since "powerpc" is
> in the compil
Just backporting the fix for 54985 from the trunk to gcc-4_7-branch.
Bootstrapped and regression tested on x86_64-unknown-linux-gnu.
2012-11-02 Jeff Law
PR tree-optimization/54985
* tree-ssa-threadedge.c (cond_arg_set_in_bb): New function extracted
from thread_across_
Rainer (or others),
the FAQ entry below seems obsolete to me (dates back more than a
decade). Shall we remove it, or is there something else we still
should document (in addition to gcc/doc/install.texi)?
Gerald
Index: faq.html
===
Ping.
- Easwaran
On Wed, Oct 31, 2012 at 12:16 PM, Easwaran Raman wrote:
> On Wed, Oct 31, 2012 at 4:36 AM, Richard Biener
> wrote:
>> On Wed, Oct 24, 2012 at 4:02 AM, Easwaran Raman wrote:
>>> On Tue, Oct 23, 2012 at 2:52 AM, Richard Biener
>>> wrote:
On Mon, Oct 22, 2012 at 8:31 PM, Ea
Hi,
As Richard suggested, if a location is UNKNOWN_LOCATION, instead of
change_scope to DECL_INITIAL (cfun->decl), we should let it inherit
location from its previous instructions. This is implemented in the
attached patch.
Bootstrapped and passed gcc regression tests and gdb regression tests.
I
Next installment in my "so that libstdc++ hackers do not feel
neglected" series. ;-)
Applied.
Any volunteer to regenerate the HTML docs after my patches now?
Gerald
2012-11-02 Gerald Pfeifer
* doc/xml/manual/configure.xml: Adjust GNU libc link.
* doc/xml/manual/debug.xml: A
On Fri, Nov 2, 2012 at 9:28 AM, Jack Howarth wrote:
> Currently on targets like x86_64 darwin which don't support -mx32
> -maddress-mode=short,
> tests that pass those on dg-options fail with excessive errors at -m64. The
> attached patch
> resolves this by adding a check_effective_target_mayb
"Joseph S. Myers" writes:
> On Thu, 1 Nov 2012, do...@redhat.com wrote:
>
>> From: dnovillo
>>
>> Following a discussion we had on this list, this patch renames the
>> file tree-asan.* into asan.*.
>>
>> * asan.c: Rename from tree-asan.c.
>> Update all users.
>> * asan.h: Rename
On Mon, 29 Oct 2012, Jakub Jelinek wrote:
> I'd like to close the stage 1 phase of GCC 4.8 development
Documented via the patch below. I also changed "Active Development"
to "Development" to reduce text density and improve formatting on a
wider range of window/text sizes.
Gerald
Index: index.ht
do...@redhat.com writes:
> The first patch is the initial import of the asan state from the
> Google branch into the [asan] branch. Subsequent patches clean the
> code up, add features like protection of stack and global variables,
> instrumentation of memory access through built-in functions, an
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.
The rest of the patches of the set is intended to further
This patch defines a new asan_shadow_offset target macro, instead of
having a mere macro in the asan.c file. It becomes thus cleaner to
define the target macro for targets that supports asan, namely x86 for
now. The ASAN_SHADOW_SHIFT (which, along with the asan_shadow_offset
constant, is used to
This patch cleanups the instrumentation code generation by emitting
GIMPLE directly, as opposed to emitting GENERIC tree and then
gimplifying them. It also does some cleanups here and there
* Makefile.in (GTFILES): Add $(srcdir)/asan.c.
(asan.o): Update the dependencies of asan.o.
This patch defines a new asan pass gate that is activated at -O0, in
addition to the pass that was initially activated at -O3 level The
patch also does some comment cleanups here and there.
* asan.c (build_check_stmt): Rename join_bb variable to else_bb.
(gate_asan_O0): New functio
Oleg Endo wrote:
> With this patch a few more cases are handled where the T bit is stored
> and then re-tested again before conditional branches.
> Tested on rev 193061 with
> make -k check RUNTESTFLAGS="--target_board=sh-sim
> \{-m2/-ml,-m2/-mb,-m2a/-mb,-m4/-ml,-m4/-mb,-m4a/-ml,-m4a/-mb}"
>
> an
Oleg Endo wrote:
> This mainly reformats some of the asm output code in sh.c to use
> multi-line strings.
> Tested on rev 193061 with
> make -k check RUNTESTFLAGS="--target_board=sh-sim
> \{-m2/-ml,-m2/-mb,-m2a/-mb,-m4/-ml,-m4/-mb,-m4a/-ml,-m4a/-mb}"
>
> and no new failures.
> OK?
OK.
Regards,
This patch implements the protection of stack variables.
To understand how this works, lets look at this example on x86_64
where the stack grows downward:
int
foo ()
{
char a[23] = {0};
int b[2] = {0};
a[5] = 1;
b[1] = 2;
return a[5] + b[1];
}
For this function, the stack pr
This patch implements the protection of global variables.
The basic idea is to insert a red zone between two global variables
and install a constructor function that calls the asan runtime to do
the populating of the relevant shadow memory regions at load time.
So the patch lays out the global va
This patch makes build_check_stmt accept its memory access parameter
to be an SSA name. This is useful for a subsequent patch that will
re-use.
Tested by running cc1 -fasan on the program below with and without the
patch and inspecting the gimple output to see that there is no change.
void
foo (
This patch splits a new create_cond_insert_point_before_iter function
out of build_check_stmt, to be used by a later patch.
Tested by running cc1 -fasan on the test program below with and
without the patch and by inspecting the gimple output to see that
there is no change.
void
foo ()
{
char fo
This patch instruments many memory access patterns through builtins.
Basically, for a call like:
__builtin_memset (from, 0, n_bytes);
the patch would only instrument the accesses at the beginning and at
the end of the memory region [from, from + n_bytes]. This is the
strategy used by the l
I believe the time has come (and probably was overdue) to switch the
web pages to UTF-8.
Which actually should not require any hanges to contents since we
have been generally using ä instead of ä and so forth.
Applied and htdocs/ on gcc.gnu.org regenerated; spot tests all were
fine (in fact, sile
On Nov 2, 2012, at 9:28 AM, Jack Howarth wrote:
> Currently on targets like x86_64 darwin which don't support -mx32
> -maddress-mode=short,
> tests that pass those on dg-options fail with excessive errors at -m64. The
> attached patch
> resolves this by adding a check_effective_target_maybe_x32
Part of my initial 64-bit support patch was bogus: I shadowed a variable
in a couple of functions used for reflect.Call. This patch fixes it.
Bootstrapped and ran Go testsuite on x86_64-unknown-linux-gnu.
Committed to mainline.
Ian
diff -r cecca11207b7 libgo/runtime/go-reflect-call.c
--- a/libgo
Hi,
For some reason cse produced:
(insn 27 43 28 2 (set (reg:SI 231 [+4 ])
(plus:SI (reg:SI 229 [+4 ])
(const_int 0 [0]))) t.c:24 10 {*addsi3}
(nil))
(insn 28 27 29 2 (set (reg:SI 212)
(ltu:SI (reg:SI 231 [+4 ])
(reg:SI 229 [+4 ]))) t.c:24 521 {*sltu_
This patch prepares the syscall package for the change to 64-bit int on
x86_64. This introduces a new type _C_int to the syscall package for
the C type int, and uses it for libc calls as needed. The Pipe and
Socketpair functions had to be adjusted. Bootstrapped and ran Go
testsuite on x86_64-unk
On 2 November 2012 22:00, Gerald Pfeifer wrote:
> Next installment in my "so that libstdc++ hackers do not feel
> neglected" series. ;-)
>
> Applied.
>
> Any volunteer to regenerate the HTML docs after my patches now?
Thanks, I've regenerated them.
Hi,
my patch on loop bound estimation had missing cast that shows up as a bootstrap
problem
on 32bit hosts. Fixed thus as obvious. I am sorry for the breakage.
Honza
Index: ChangeLog
===
--- ChangeLog (revision 193116)
+++ Chang
Hi,
Thanks for so many useful comments! I update the file according to the
comments. The major changes include adding sanitizer.def and
generating gimple directly. New patch file is attached.
> On Wed, Oct 31, 2012 at 11:34:10AM -0700, Wei Mi wrote:
>> gcc/ChangeLog:
>> 2012-10-31 Wei Mi
>
> I
Due to weird circumstances detailed in the PR, this test briefly
passed (it has always failed before), so technically I'm fixing
a regression. :)
The test checks that a certain label is mentioned twice; being
mentioned once infers that there are two identical initializer
vectors in the constant-po
75 matches
Mail list logo
