Let's CC Jason.
---BeginMessage---
Hello,
I've been investigating a bug in gcc I came across recently and after some
difficulties, I have produced a patch that actually fixes that behavior.
However, I don't think the patch is very good and I would really
appreciate your help in making it
PING^2.
Dodji Seketeli do...@redhat.com writes:
Hello,
Consider this short test snippet:
-8---
#define STRINGIFY(x) #x
#define TEST(x) \
_Pragma(STRINGIFY(GCC diagnostic ignored -Wunused-local-typedefs)) \
typedef int myint;
On Sun, 26 Aug 2012, Ulrich Drepper wrote:
On Sun, Aug 26, 2012 at 1:36 PM, Jonathan Wakely jwakely@gmail.com wrote:
But I've just seen Paolo's suggestion for a __generate_impl that both
use, and I prefer that idea.
How about the following. I introduced a __generate_impl() member
On Mon, Aug 27, 2012 at 09:52:04AM +0200, Dodji Seketeli wrote:
PING^2.
This is ok for trunk. Thanks.
Bootstrapped and tested on x86_64-unknown-linux-gnu against trunk.
libcpp/
PR preprocessor/53469
* directives.c (do_pragma): Use the virtual location for the
pragma
Hello,
This patch intends to provide a chance to prune common warning messages for
tests defined in an exp file. It can avoid adding dg-prune to such cases one
by one. The story behind this patch is that when torture cases in arm.exp
with various arch/cpu options, there are always warnings on
On 08/27/2012 02:44 AM, Ulrich Drepper wrote:
On Sun, Aug 26, 2012 at 1:36 PM, Jonathan Wakely jwakely@gmail.com wrote:
But I've just seen Paolo's suggestion for a __generate_impl that both
use, and I prefer that idea.
How about the following. I introduced a __generate_impl() member
On 08/27/2012 11:46 AM, Paolo Carlini wrote:
(the advance patch is also pending, I guess).
I meant discard of course.
Paolo.
Hi Oleg,
Your movhi_index_disp and movsi_index_disp split patterns match several
times with the same bodies and names. I suppose it is a merge glitch
during your commit ?
Cheers
Christian
On 08/26/2012 08:12 PM, Tobias Burnus wrote:
This patch fixes one ICE and several memory leaks. But there are more.
I have now committed the patch with the following additional patch
* module.c (mio_symbol): Don't increase sym-refs for its
use in sym-formal_ns-proc_name.
The
On 27 Aug 2012, at 12:55, Christian Bruel christian.br...@st.com
wrote:
Hi Oleg,
Your movhi_index_disp and movsi_index_disp split patterns match
several
times with the same bodies and names. I suppose it is a merge glitch
during your commit ?
Hm, it seems there is a typo in the name of
On Mon, Aug 27, 2012 at 5:46 AM, Paolo Carlini paolo.carl...@oracle.com wrote:
One last observation from me: I think we are being a little inconsistent in
terms of inlining. I see some __generate_impl with a non-trivial body inline
whereas other, with a tad smaller body, in random.tcc.
For the
On 27/08/2012 14:14, Tobias Burnus wrote:
On 08/26/2012 08:12 PM, Tobias Burnus wrote:
This patch fixes one ICE and several memory leaks. But there are more.
I have now committed the patch with the following additional patch
* module.c (mio_symbol): Don't increase sym-refs for its
[ Richard, sorry for the duplicate message where I omitted the mailing-list]
On 24 August 2012 10:40, Richard Earnshaw rearn...@arm.com wrote:
Has this been tested for big-endian?
Hi,
While improving my tests and trying to turn them into execution tests,
I realized that vector initialization
On 08/27/2012 03:19 PM, Mikael Morin wrote:
as you seem to be very much into memory issues,
Well, it started as middle-end project by Steven Bosscher and others
which looked at memory issues due to http://gcc.gnu.org/PR54146 (a
program using attribute((flatten)), which essentially causes
On 27/08/2012 17:11, Tobias Burnus wrote:
could you add comments
in gfortran.h telling which pointers account for reference counting?
As far as I remember for symbols, there are:
gfc_symtree::n::sym;
gfc_namespace::proc_name;
Well, I have still not gained a full overview about refs,
On 08/23/2012 06:08 PM, Maxim Kuvyrkov wrote:
This patch fixes an ICE in sched-deps.c due to a dereference of an empty list.
The problem is in handling of DEBUG_INSNs and its peculiarity of disregarding
requests to free dependence lists in certain cases. This patch fixes up an
instance of
Hi,
On Sun, Aug 19, 2012 at 07:43:45AM +0200, Jan Hubicka wrote:
* gcc.dg/ipa/iinline-1.c: Update testcase to test inline hints.
* ipa-inline.c (want_inline_small_function_p): Bypass
inline limits for hinted functions.
(edge_badness): Dump hints; decrease
This fixes a segfault seen in the fix-vr4130* cases when run with
-O3 -mabi=64. The problem was caused by vr4130_align_insns simulating
zero-length ghost gp instructions, even though the idea is that the
scheduler should ignore them. (These instructions postdate the vr4130
stuff by quite some
mips_pic_call_symbol_from_set has code to get the symbol associated with
a %call16 reloc, and can also get the symbol from a REG_EQUAL or REG_EQUIV
note. But in call-3.c, which tests a call to hidden symbol, we have a plain:
(set (reg ...) (symbol_ref ...))
This patch handles that case too.
Let's do away with ba_quiet entirely and instead pass tf_none in cases
where we would have set ba_quiet.
#define DERIVED_FROM_P(PARENT, TYPE) \
- (lookup_base ((TYPE), (PARENT), ba_any, NULL) != NULL_TREE)
+ (lookup_base ((TYPE), (PARENT), ba_any, NULL, tf_warning_or_error) \
+ !=
As an experiment, I tried removing all the explicit -O flags from
gcc.target/mips and making it cycle through the usual torture options
instead. This showed up a fair few bugs, which I've tried to fix with
today's and yesterday's patches.
Even though it does add to the testing time, I think it's
The patch to make gcc.target/mips use gcc-dg triggered some false
failures for -O3 -g, where the file's pathname was included in the
output and happened to match some of the scan-assembler-not regexps.
This patch tightens the regexps by:
- making sure that register names are prefixed by $.
-
A NOMIPS16 janitorial.
Tested on mipsisa64-elf, mips64-elf and mips64-linux-gnu. Applied.
Richard
gcc/testsuite/
* gcc.target/mips/ext_ins.c, gcc.target/mips/octeon-pop-2.c,
gcc.target/mips/pr54240.c, gcc.target/mips/stack-1.c,
gcc.target/mips/unaligned-1.c: Add
Diego Novillo dnovi...@google.com writes:
+# Skip all inline functions in tree.h.
+# These are used in accessor macros.
+# Note that this is added at the end because older gdb versions
+# do not understand the 'skip' command.
+skip tree.h
(gdb) skip tree.h
No function found named tree.h.
Sandra Loosemore san...@codesourcery.com writes:
On 08/19/2012 11:22 AM, Richard Sandiford wrote:
Not sure whether a peephole is the right choice here. In practice,
I'd imagine these opportunities would only come from a DImode move of
$0 into a doubleword register, so we could simply emit
Unless anyone has objections I'm going to commit this to trunk,
implementing Sebastian's idea to disable the verbose terminate handler
and the pure virtual function called message, which write to stderr
when a process terminates. This allows embedded systems to avoid
pulling in the demangler and
On Aug 27, 2012, at 1:14 AM, Terry Guo wrote:
This patch intends to provide a chance to prune common warning messages for
tests defined in an exp file.
Is it OK to trunk?
Ok.
If you can find where to document this... :-) That'd be nice.
On 08/27/2012 06:28 PM, Jason Merrill wrote:
Let's do away with ba_quiet entirely and instead pass tf_none in cases
where we would have set ba_quiet.
Thanks for the feedback. Indeed, as I tried to explain in an older
message, that was also my understanding. The practical difficulty with
the
This patch properly sets the latency of atomic ops to the approximate
latency of a remote memory operation.
* config/tilegx/sync.md (atomic_compare_and_swap_baremode,
atomic_exchange_baremode,
atomic_fetch_fetchop_name_baremode): Set type to X1_remote.
*
On 2012/08/24 23:20:21, davidxl wrote:
On Fri, Aug 24, 2012 at 3:56 PM, mailto:cm...@google.com wrote:
http://codereview.appspot.com/6427063/diff/11002/gcc/gcov-io.h
File gcc/gcov-io.h (right):
http://codereview.appspot.com/6427063/diff/11002/gcc/gcov-io.h#newcode688
gcc/gcov-io.h:688:
On Mon, Aug 27, 2012 at 10:55 AM, cm...@google.com wrote:
On 2012/08/24 23:20:21, davidxl wrote:
On Fri, Aug 24, 2012 at 3:56 PM, mailto:cm...@google.com wrote:
http://codereview.appspot.com/6427063/diff/11002/gcc/gcov-io.h
File gcc/gcov-io.h (right):
On 08/27/2012 01:36 PM, Paolo Carlini wrote:
Thanks for the feedback. Indeed, as I tried to explain in an older
message, that was also my understanding. The practical difficulty with
the mapping (which you can also guess from the current structure of
get_delta_difference_1, the if (kind ==
This patch stops trying to represent the bfins instruction with the
insv pattern, because it causes an extra copy when bfins reuses one of
its inputs.
* config/tilegx/tilegx.md (*bfins): Rename to insn_bfins.
(insn_bfins): Delete.
Index: gcc/config/tilegx/tilegx.md
Hi Mikael,
Is this patch approved? I realize it's not the final step (no pun
intended), but I will be very excited to see this hit the trunk.
Supporting FINAL will have broad impact on my work and the work of others
writing modern Fortran libraries and applications.
Damian
On 8/25/12 1:42 PM,
This patch fixes a typo in the mcount function_profiler.
* config/tilegx/tilegx.c (tilegx_function_profiler): Fix typo.
config/tilepro/tilepro.c (tilepro_function_profiler): Ditto.
Index: gcc/config/tilegx/tilegx.c
Rather simple fix. The first one was leaking expr-value.integer, the
second one sym-formal_ns.
Build, tested and committed as Rev. 190728.
Tobias
Index: gcc/fortran/ChangeLog
===
--- gcc/fortran/ChangeLog (Revision 190714)
+++
On 27/08/2012 20:20, Rouson, Damian wrote:
Hi Mikael,
Is this patch approved?
There are a few overlooks to be fixed and the components walking code
that I would like to see shared.
Then I think it can go in. But there is no big stopper.
Mikael
On 08/27/2012 08:02 AM, Christophe Lyon wrote:
[ Richard, sorry for the duplicate message where I omitted the mailing-list]
On 24 August 2012 10:40, Richard Earnshaw rearn...@arm.com wrote:
Has this been tested for big-endian?
Hi,
While improving my tests and trying to turn them into
Hello,
With my patch to visit insns in topological order in
init_alias_analysis, an issue in postreload-cse got exposed: Blocks
may become unreachable between calls of reload_cse_regs_1 but no CFG
cleanups are scheduled between them, resulting in an ICE in
pre_and_rev_post_order_compute at an
On Sun, 26 Aug 2012, Richard Sandiford wrote:
I'm preparing a patch to turn gcc.target/mips into a torture-like
testsuite.
While on the subject of gcc.target/mips and its extensions, it
also doesn't handle a build configured with --with-synci=yes.
(Well, not on the 4.7 branch at least.)
brgds,
This patch fixes the problem by deleting unreachable basic blocks if
edges were purged during a reload-cse transformation on an insn. I
didn't add a test case because reliably triggering problems this late
in the pass pipeline is so hard.
A pervasive idiom in the RTL land is to call
Another obvious leakage fix: For if( e1 == NULL || e2 == NULL) one has
to free the non-NULL expression (if there is one). Similarly, vector of
type tree * has to be freed - we only need the elements of that vector.
Build and tested on x86-64-linux.
Committed as Rev. 190731
Tobias
Index:
On 08/27/2012 08:04 PM, Jason Merrill wrote:
On 08/27/2012 01:36 PM, Paolo Carlini wrote:
Thanks for the feedback. Indeed, as I tried to explain in an older
message, that was also my understanding. The practical difficulty with
the mapping (which you can also guess from the current structure of
On Mon, Aug 27, 2012 at 10:32 PM, Eric Botcazou ebotca...@adacore.com wrote:
This patch fixes the problem by deleting unreachable basic blocks if
edges were purged during a reload-cse transformation on an insn. I
didn't add a test case because reliably triggering problems this late
in the pass
Hello!
Attached patch fixes PR 46254. The problem was, that PIC code models,
different than small PIC still consume %rbx, even in 64bit mode. The
patch fixes this by enhancing atomic_compare_and_swapdwi_doubleword,
so the same macroized pattern now handles PIC and non-PIC compilations
for 32bit
Hi,
A truncate of a memory location can be simplified to a memory load
just like what is done for a subreg of a memory load.
This patch adds that simplification to simplify-rtx.c.
OK? Bootstrapped and tested on mips64-linux-gnu with no regressions.
Thanks,
Andrew Pinski
ChangeLog:
*
Moved pmu writing utilities to global header.
The patch should be applied to google/main
Tested with crosstools.
2012-08-27 Chris Manghane cm...@google.com
* libgcc/pmu-profile.c
(gcov_write_load_latency_infos): Removed unused function.
Ok for google branches.
David
http://codereview.appspot.com/6427063/
ping
Thanks,
Dehao
On Sat, Aug 18, 2012 at 6:02 AM, Dehao Chen de...@google.com wrote:
Hi, Richard,
Thanks for the review. I've addressed most of the issues except the
java unittest (see comments below). The new patch is attached in the
end of this email.
Thanks,
Dehao
On Fri, Aug
Google ref b/7041390.
2012-08-25 Cary Coutant ccout...@google.com
gcc/
* dwarf2out.c (is_template_instantiation): New function.
(should_move_die_to_comdat): Reject types that are template
instantiations.
This patch is OK for google 4.7
Sterling
OK.
Jason
Especially after Carlo's comment from earlier that attention is paid
to not inlining unnecessarily it is surprising that the random_device
code is inlined even though it is not a template class.
How about not doing this and moving the definition into the library?
This is done in the attached
This patch adds support for the -mcmodel=MODEL flag on TILE-Gx. The
models supported are small and large. In the small model (default),
distance for direct calls is limited to 500M in either direction.
PC-relative addresses are 32 bits. Absolute addresses support the
full address range. In
That's probably the last patch of this kind for a while.
By using formal_ns, the namespace is freed when the main symbol is freed.
Committed as Rev.
Tobias
Index: gcc/fortran/ChangeLog
===
--- gcc/fortran/ChangeLog (Revision
54 matches
Mail list logo
