Hi,
this patch adds support for the 64x64-128 bit unsigned multiplication
to the s390 backend. We always the instruction but for some reason
forgot to add a pattern :(
Thanks to Torbjorn Granlund for the pointer!
Committed to mainline after testing on s390 and s390x.
Bye,
-Andreas-
On Thu, Oct 6, 2011 at 11:33 PM, H.J. Lu hjl.to...@gmail.com wrote:
OTOH, x86_64 and i686 targets can also benefit from this change. If
combine can't create more complex address (covered by lea), then it
will simply propagate memory operand back into the add insn. It looks
to me that we can't
Dear Steve,
Thanks - I just noticed that the { dg-do run } is missing a space. It
seemed to run correctly during regtesting but I will set it right
anyway.
Cheers
Paul
On Fri, Oct 7, 2011 at 1:29 AM, Steve Kargl
s...@troutmask.apl.washington.edu wrote:
On Thu, Oct 06, 2011 at 10:22:12PM
On 10/07/2011 12:15 AM, Richard Henderson wrote:
diff --git a/gcc/recog.c b/gcc/recog.c
index d3ecb73..e0557c5 100644
--- a/gcc/recog.c
+++ b/gcc/recog.c
@@ -3769,20 +3769,33 @@ struct rtl_opt_pass pass_split_all_insns =
static unsigned int
rest_of_handle_split_after_reload (void)
{
- /*
On Fri, Oct 7, 2011 at 6:22 AM, Artem Shinkarov
artyom.shinkar...@gmail.com wrote:
On Wed, Oct 5, 2011 at 12:35 PM, Richard Guenther
richard.guent...@gmail.com wrote:
On Wed, Oct 5, 2011 at 1:28 PM, Artem Shinkarov
artyom.shinkar...@gmail.com wrote:
On Wed, Oct 5, 2011 at 9:40 AM, Richard
Instead of adding the parameter, let's have the C++ front end call a
different function that doesn't try to pull out the value from aggregate
variables, like C has decl_constant_value_for_optimization.
Jason
On Thu, 6 Oct 2011, William J. Schmidt wrote:
On Thu, 2011-10-06 at 16:16 +0200, Richard Guenther wrote:
snip
Doh, I thought you were matching gimple stmts that do the address
computation. But now I see you are matching the tree returned from
get_inner_reference. So no need to
On Thu, 6 Oct 2011, Jeff Law wrote:
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 10/06/11 04:13, Richard Guenther wrote:
People have already commented on pieces, so I'm looking only at the
tree-ssa-reassoc.c pieces (did you consider piggy-backing on IVOPTs
instead? The idea
Committed.
Richard.
2011-10-07 Richard Guenther rguent...@suse.de
PR testsuite/50637
* gcc.dg/vect/vect-align-2.c: Increase array size.
Index: gcc/testsuite/gcc.dg/vect/vect-align-2.c
===
---
On Thu, Oct 6, 2011 at 5:57 PM, Rainer Orth r...@cebitec.uni-bielefeld.de
wrote:
After almost two months, two tests are still XPASSing everywhere:
XPASS: gcc.dg/uninit-B.c uninit i warning (test for warnings, line 12)
XPASS: gcc.dg/uninit-pr19430.c (test for warnings, line 32)
XPASS:
On Fri, Oct 7, 2011 at 12:49 AM, Michael Meissner
meiss...@linux.vnet.ibm.com wrote:
On Thu, Oct 06, 2011 at 10:44:03AM -0400, Michael Meissner wrote:
On Thu, Oct 06, 2011 at 03:23:07PM +0200, Tobias Burnus wrote:
On 10/06/2011 03:02 PM, Michael Meissner wrote:
On the x86 (with Fedora 13), I
On 10/07/2011 10:00 AM, Richard Guenther wrote:
It's a reasonable plan - you'd have to introduce a late reassoc
pass though. Can you separate out the RTL fwprop changes? So
we can iterate over the tree parts separately.
That's indeed better, also because they became CSE changes.
Paolo
On Oct 6, 2011, at 4:12 PM, Joseph S. Myers wrote:
On Thu, 6 Oct 2011, Tristan Gingold wrote:
So the consensus is for a dedicated option. Which one do you prefer ?
-funnamed-variadic-parameter
-fpointless-variadic-functions
-fallow-parameterless-variadic-functions
I prefer
Hi Janus,
I completely missed that this patch is still pending review as you wrote
committed. However, it only applied to the ICE and not to the
documentation and backporting.
On 10/01/2011 02:00 PM, Janus Weil wrote:
I have just committed as obvious a one-line patch to fix a regression
Hi Richard, Hi Paulo,
Following Paulo's suggestion, here is a patch to remove the calls to
df_analyze from the compare-elimination pass, and to use
df_get_live_[in/out]. I can confirm that with this patch applied PR
49801 is fixed.
OK to apply ? For mainline and the 4.6 branch ?
Hi all,
here is a small patch for an OOP regression. The title of the PR is a
bit misleading, since it's actually not the ALLOCATABLE attribute
which gets lost, but the 'class_ok' attribute, which we use for
checking if a class declaration is valid. When reading class variables
from a module, we
On 10/07/2011 12:06 PM, Nick Clifton wrote:
- df_analyze ();
-
gcc_checking_assert (all_compares == NULL);
Ok except that this particular df_analyze should stay. If it doesn't
fix the bug, please try the parts of rth's patch that add the gate and
the TODO_df_finish flag.
Paolo
On 10/07/2011 12:18 PM, Janus Weil wrote:
Regtested successfully on x86_64-unknown-linux-gnu (except for failure
on entry_4 and select_type_12, which are known middle-end
regressions). Ok for trunk and 4.6?
OK. Thanks for the patch - and for telling me that select_type_12 is a
middle end
Hi,
this is another LTO bootstrap failure with Ada enabled. There are 2 problems:
1. dwarf2out.c thinks that it is processing pure C++ code for gnat1 so it is
invoking some C++-specific code paths that aren't compatible with Ada. Fixed
by building TRANSLATION_UNIT_DECL nodes in Ada as well.
On Fri, Oct 7, 2011 at 12:37 PM, Eric Botcazou ebotca...@adacore.com wrote:
Hi,
this is another LTO bootstrap failure with Ada enabled. There are 2 problems:
1. dwarf2out.c thinks that it is processing pure C++ code for gnat1 so it is
invoking some C++-specific code paths that aren't
On Wed, Oct 05, 2011 at 09:44:34PM -0400, Jason Merrill wrote:
On 10/05/2011 05:15 PM, Jakub Jelinek wrote:
If optimize_specialization_lookup_p (tmpl) doesn't change return value in
between the two calls, then you are right. But perhaps in that case
you could avoid the second call and use
This implements a form of Named Return Value optimization in gigi. This helps
a lot for small functions returning array types as well as for vectorization.
Tested on i586-suse-linux, applied on the mainline.
2011-10-07 Eric Botcazou ebotca...@adacore.com
* gcc-interface/gigi.h
Hi,
Instead of adding the parameter, let's have the C++ front end call a
different function that doesn't try to pull out the value from
aggregate variables, like C has decl_constant_value_for_optimization.
Ok. The below lives entirely inside the C++ front-end (actually I had
already
Hi!
Calling vect_is_simple_cond from the pattern recognized was a bad idea,
as it calls vect_is_simple_use_1 rather than vect_is_simple_use and the
former looks at STMT_VINFO_VECTYPE which by that time is only set
for the data ref stmts (which is why in my test it worked well, that
one did (a[i]
On 06/10/11 18:17, Paul Brook wrote:
I believe this patch to be nothing but an improvement over the current
state, and that a fix to the constraint problem should be a separate patch.
In that basis, am I OK to commit?
One minor nit:
(define_special_predicate shift_operator
...
+ (ior
On Fri, 7 Oct 2011, Jakub Jelinek wrote:
Hi!
Calling vect_is_simple_cond from the pattern recognized was a bad idea,
as it calls vect_is_simple_use_1 rather than vect_is_simple_use and the
former looks at STMT_VINFO_VECTYPE which by that time is only set
for the data ref stmts (which is
On 06/10/11 16:01, Andrew Stubbs wrote:
(define_special_predicate shift_operator
(and (ior (ior (and (match_code mult)
(match_test power_of_two_operand (XEXP (op, 1), mode)))
(and (match_code rotate)
(match_test GET_CODE (XEXP
Hm, but canonicalize_cond_expr_cond is supposed to produce a
tree that is suitable for a condition in a GIMPLE_COND, or a
COND_EXPR. So the issue is that it is used for a tcc_comparison
on a assignment RHS?
It is called on (Ada_Boolean_Type) iftmp.xxx, which comes from the RHS of
gcc.c-torture/execute/vector-compare-2.c was FAILing on Solaris 8 and
9/x86 with ABI warnings
output is:
/vol/gcc/src/hg/trunk/local/gcc/testsuite/gcc.c-torture/execute/vector-compare-2.c:
In function 'foo':
/vol/gcc/src/hg/trunk/local/gcc/testsuite/gcc.c-torture/execute/vector-compare-2.c:7:1:
On Fri, Oct 7, 2011 at 1:37 PM, Eric Botcazou ebotca...@adacore.com wrote:
Hm, but canonicalize_cond_expr_cond is supposed to produce a
tree that is suitable for a condition in a GIMPLE_COND, or a
COND_EXPR. So the issue is that it is used for a tcc_comparison
on a assignment RHS?
It is
On Fri, 2011-10-07 at 11:17 +0200, Paolo Bonzini wrote:
On 10/07/2011 10:00 AM, Richard Guenther wrote:
It's a reasonable plan - you'd have to introduce a late reassoc
pass though. Can you separate out the RTL fwprop changes? So
we can iterate over the tree parts separately.
That's
Oh, I forgot to say, I don't understand why the rotate operator is
special cased?
If I understand it correctly, the effect of the (existing) rotate is
both to check the constant range, AND to disallow registers as the shift
amount. This difference has no effect on Thumb, but might cause ARM
Done, and attached.
Ok.
Two changes to the testcase that I'll pre-approve:
- Add a comment along the lines of
/* ARM has shift-and-alu insns. Depending on the ALU op GCC represents some
of these as a left shift, others as a multiply. Check that we match the
right one. */
- Also test
Hi,
On Wed, 5 Oct 2011, Uros Bizjak wrote:
Like so. Currently regstrapping on x86_64-linux. Okay if that
succeeds?
OK, with a nit - I'd introduce RECIP_MASK_NONE and use it in place of 0
in a couple of places.
Okay. r179608 it is.
Ciao,
Michael.
Fix of wrong data_section_start for ATmega164A.
Ok?
Johann
PR target/50652
* config/avr/avr-mcus.def (AVR_MCU): Set .data_section_start of
atmega164a to 0x100.
Index: config/avr/avr-mcus.def
===
---
Jeff == Jeff Law l...@redhat.com writes:
Jeff Presumably it hasn't been included because not all gdb's understand
Jeff those bits and we typically don't build with -g3.
GCC is pretty much the perfect candidate for a -g3 build. All those
macros...
The needed gdb changes have been in since
On Fri, Oct 07, 2011 at 07:42:44AM -0600, Tom Tromey wrote:
Jeff == Jeff Law l...@redhat.com writes:
Jeff Presumably it hasn't been included because not all gdb's understand
Jeff those bits and we typically don't build with -g3.
GCC is pretty much the perfect candidate for a -g3 build.
Hi,
this fixes the tls testsuite fails on s390. Calls to __tls_get_offset
require the GOT pointer to be loaded what so far only happened when
building pic code. With the attached patch the GOT pointer is loaded
on demand when calling __tls_get_offset.
Committed to mainline.
Bye,
-Andreas-
Hello,
the following patches propose to fix bug fortran/50420:
gfortran has been rejecting coarrays as argument to the coarray intrinsics
({l,u}cobound, image_index) if they had a subreference.
The standard, however, has:
A subobject of a coarray is a coarray if it does not have any
dim_corank_check uses gfc_find_array_ref to get coarray's array ref.
It won't work for obvious reasons for the case coarray(1,1) which is a
non-arrray (but still a coarray).
As the whole point of this is to get the corank, patch 10 replaces the call
to gfc_find_array_ref with a call to
The walk_coarray function introduced at
http://gcc.gnu.org/ml/fortran/2011-09/msg00074.html doesn't (like the code it
replaces) support subreferences after the coarray.
These patches are going to change that.
- The first step is, while looking for the coarray reference, to really look for
it,
This fixes a failing assertion in gfc_build_array_ref.
While array(x,:) still refers to an array, array(x,y) is a single element,
and thus coarray(x,y) is a descriptor whose data component points to something
that is not an array.
gfc_conv_expr_descriptor, to update data pointer with
These patches change the descriptor initialization code in
gfc_conv_expr_descriptor introduced by the recent scalarizer patchset at
http://gcc.gnu.org/ml/fortran/2011-09/msg00056.html.
It was supposing (just like the code it was replacing) that coarrays were
always full, but a subobject of a
simplify_cobound, when it looks for the coarray reference, in the AR_ELEMENT
case, first checks that it is the last reference in the chain.
As it is what we are trying to avoid, this patch removes that and uses the
corank field directly.
OK?
2011-10-06 Mikael Morin mikael.mo...@sfr.fr
gfc_build_array_type was forcing the lower bound to zero in cases where it was
unspecified (NULL). This is wrong for deferred (allocatable) coarrays at least,
as their cobound is specified at allocation time.
This patch fixes that.
I don't think other cases beyond AS_DEFERRED need fixing, but I
Regtested successfully on x86_64-unknown-linux-gnu (except for failure
on entry_4 and select_type_12, which are known middle-end
regressions). Ok for trunk and 4.6?
OK. Thanks for the patch
Thanks for the review. Committed to trunk as r179660.
- and for telling me that select_type_12 is a
On Thu, Oct 6, 2011 at 4:25 PM, Kai Tietz kti...@redhat.com wrote:
Hi,
I modified the patch so, that it always just converts two leafs of a
TRUTH(AND|OR)IF chain into a TRUTH_(AND|OR) expression, if branch costs are
high and leafs are simple without side-effects.
Additionally I added some
On Fri, 7 Oct 2011, Tristan Gingold wrote:
Here is my patch with the option renamed.
Ok for trunk ?
OK with a spelling fix:
+@item -fallow-parameterless-variadic-functions
+Accept variadic functions without named parameters.
+
+Although it is possible to define such a function, this is
On 07/10/11 13:37, Paul Brook wrote:
Done, and attached.
Ok.
Two changes to the testcase that I'll pre-approve:
- Add a comment along the lines of
/* ARM has shift-and-alu insns. Depending on the ALU op GCC represents some
of these as a left shift, others as a multiply. Check that we
* config.gcc (x86_64-*): Add core-avx-i, core-avx2 for with_cpu.
diff --git a/gcc/config.gcc b/gcc/config.gcc
index 8b380ac..8099ed7 100644
--- a/gcc/config.gcc
+++ b/gcc/config.gcc
@@ -3302,7 +3302,7 @@ case ${target} in
| k8 | k8-sse3 | athlon64 | athlon64-sse3 |
-BEGIN PGP SIGNED MESSAGE-
Hash: SHA1
On 10/07/11 07:42, Tom Tromey wrote:
Jeff == Jeff Law l...@redhat.com writes:
Jeff Presumably it hasn't been included because not all gdb's
understand Jeff those bits and we typically don't build with -g3.
GCC is pretty much the perfect
Hi,
tree-emutls.c missed to called add_referenced_var for a variable that is
referenced. That always was a bug, but meanwhile is fatal (causing a
segfault). This fixes the problem. Okay for trunk if regstrapping
succeeds?
The related bugreport PR50640 (fortran segfaults for similar
Hello,
This patch adds the repropagation of expanded bitwise-not expressions.
ChangeLog
2011-10-07 Kai Tietz kti...@redhat.com
* tree-ssa-reassoc.c (walk_bitwise_stmt_elems): Helper
to collect different kinds of operands of a bitwise-binary
expression chain.
Hello,
This patch adds to the break-up pass the facility to sink bitwise-not operations
into bitwise-binary expressions. Additionally it handles special
cases for ~(~X),
and ~(X cmp Y).
ChangeLog
2011-10-07 Kai Tietz kti...@redhat.com
* tree-ssa-reassoc.c (remove_stmt_chain): Helper
Hello,
This patch adds to the break-up pass the facility to expand (X | Y) ==/!= 0
expression. This enables in later reassociation pass better results.
ChangeLog
2011-10-07 Kai Tietz kti...@redhat.com
* tree-ssa-reassoc.c (expand_cmp_ior): Helper for expanding
(X | Y) ==/!=
Hello,
This patch adds to the break-up code the conversion for X ==/!= ~0 to
~X ==/!= 0.
ChangeLog
2011-10-07 Kai Tietz kti...@redhat.com
* tree-ssa-reassoc.c (break_up_bitwise_combined_stmt): Add
handling for X !=/== 0 transformation to ~X !=/== 0.
2011-10-07 Kai Tietz
Hello,
This patch adds to the repropagation code the conversion for ~X ==/!=
CST to X ==/!= CST' (with CST' = ~ CST).
We need to do this back-conversion after initial bitwise-binary
repropagation loop, as otherwise
it would interfer.
ChangeLog
2011-10-07 Kai Tietz kti...@redhat.com
*
Hello,
This patch adds to the repropagation pass for bitwise-expression the
conversion of (X != 0) | (Y != 0) - (X | Y) != 0, and of (X == 0) (Y == 0)
- (X | Y) == 0.
ChangeLog
2011-10-07 Kai Tietz kti...@redhat.com
* tree-ssa-reassoc.c (walk_bitwise_stmt_elems): Add new argument
On Oct 7, 2011, at 4:53 AM, Rainer Orth wrote:
gcc.c-torture/execute/vector-compare-2.c was FAILing on Solaris 8 and
9/x86 with ABI warnings
Closer investigation revealed a mess:
So, the compile testsuite does:
gcc-dg-runtest [lsort [glob -nocomplain $srcdir/$subdir/*.\[cS\]]] -w
in
When Jakub added the 'i' alternatives for avx2, these broke.
r~
* gcc.target/i386/avx256-unaligned-load-2.c: Tweek vinsert pattern
match for avx2.
* gcc.target/i386/avx256-unaligned-store-2.c: Similarly.
diff --git
Committed to trunk.
gcc/
PR 50655
* configure.ac: Add .register directives to VIS3 test.
* configure: Regenerate.
gcc/testsuite/
PR 50655
* gcc.target/sparc/sparc.exp: Add vis3 target test.
* gcc.target/sparc/cmask.c: Use it.
*
2011/10/7 Georg-Johann Lay a...@gjlay.de:
Fix of wrong data_section_start for ATmega164A.
Ok?
Johann
PR target/50652
* config/avr/avr-mcus.def (AVR_MCU): Set .data_section_start of
atmega164a to 0x100.
Approved.
Denis.
On 10/05/2011 12:07 PM, Uros Bizjak wrote:
We already have V2TImode, but hidden in VIMAX_AVX2 mode iterator.
Based on that, I would suggest that we model correct insn
functionality and try to avoid unspec. On the related note, there is
no move insn for V2TImode, so V2TI should be added to V16
Move several static inline functions from pph-streamer.h to
pph-streamer-{out,in}.c. This is part 1 of several patches,
and minimizes diffs.
Index: gcc/cp/ChangeLog.pph
2011-10-07 Lawrence Crowl cr...@google.com
* pph-streamer.h (pph_out_tree_array): Remove unused.
Test gcc.dg/graphite/id-pr46845.c turns off warnings for x86 targets.
powerpc targets warn about ABI issues with vector arguments for this
test, so this patch includes powerpc*-*-* in the list of targets to
turn off warnings. OK for trunk?
2011-10-07 Janis Johnson jani...@codesourcery.com
On Oct 7, 2011, at 12:50 PM, Janis Johnson wrote:
Test gcc.dg/graphite/id-pr46845.c turns off warnings for x86 targets.
powerpc targets warn about ABI issues with vector arguments for this
test, so this patch includes powerpc*-*-* in the list of targets to
turn off warnings. OK for trunk?
On Friday 30 September 2011 18:51:21 Steve Kargl wrote:
Mikael,
I've finally made it through the set of patches,
and did not find anything that raised a red flag.
I'll note that I did not study the issue/question
you raised with patch 6. Tobias is probably the
best person to offer an
2011/10/7 Janus Weil ja...@gcc.gnu.org:
Regtested successfully on x86_64-unknown-linux-gnu (except for failure
on entry_4 and select_type_12, which are known middle-end
regressions). Ok for trunk and 4.6?
OK. Thanks for the patch
Thanks for the review. Committed to trunk as r179660.
...
A representation with vec_merge and a 0..255 constant is incorrect.
Both 128-bit lanes are merged with the same pattern, thus the rtl-level
vec_merge operand should replicate the imm8 operand into two bytes.
I created an expander with the old name and old interface so that we
can continue using
I have just committed as obvious a one-line patch to fix a regression
which is triggered by -fwhole-file:
http://gcc.gnu.org/viewcvs?root=gccview=revrev=179413
Should I backport to 4.6? And also to 4.5?
Yes, please backport to 4.6. If you want, you can also backport to 4.5.
However, I do
Move several static inline functions from pph-streamer.h to
pph-streamer-{out,in}.c. This is part 2 of several patches,
and minimizes diffs.
Index: gcc/cp/ChangeLog.pph
2011-10-07 Lawrence Crowl cr...@google.com
* pph-streamer.h (pph_in_chain): Moved to pph-streamer-in.c as extern.
Hello,
this is the updated version with the suggestion
2011/10/7 Richard Guenther richard.guent...@gmail.com:
On Thu, Oct 6, 2011 at 4:25 PM, Kai Tietz kti...@redhat.com wrote:
+ ((TREE_CODE_CLASS (TREE_CODE (arg1)) != tcc_comparison
+ TREE_CODE (arg1) != TRUTH_NOT_EXPR
+
On Fri, Oct 07, 2011 at 02:13:40PM -0700, Richard Henderson wrote:
+(define_insn *avx2_pblendw
+ [(set (match_operand:V16HI 0 register_operand =x)
+ (vec_merge:V16HI
+ (match_operand:V16HI 2 nonimmediate_operand xm)
+ (match_operand:V16HI 1 register_operand x)
+
On 10/07/2011 02:40 PM, Jakub Jelinek wrote:
On Fri, Oct 07, 2011 at 02:13:40PM -0700, Richard Henderson wrote:
+(define_insn *avx2_pblendw
+ [(set (match_operand:V16HI 0 register_operand =x)
+(vec_merge:V16HI
+ (match_operand:V16HI 2 nonimmediate_operand xm)
+
First, thanks so much for tackling this review. I don't think anyone's
overly enthusiastic about reviewing dejagnu crap^H^H^H^Hcode.
On 10/04/11 08:43, Aldy Hernandez wrote:
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg01272.html
So others have already mentioned the name memmodel as being
Remove #if 0 scaffolding. This is part 3 of several patches,
and minimizes diffs.
Index: gcc/cp/ChangeLog.pph
2011-10-07 Lawrence Crowl cr...@google.com
* pph-streamer.h: Remove #if 0 scaffolding.
Index: gcc/cp/pph-streamer.h
I would like to bring Simon's patch for the
--with-native-system-header-dir configure option from the
google/integration branch into gcc mainline. This patch permits
changing the system header directory from /usr/include to something
else. It's mainly useful in conjunction with sysroot, so that
On Fri, 7 Oct 2011, Ian Lance Taylor wrote:
2011-10-07 Simon Baldwin sim...@google.com
* configure.ac: Add --with-native-system-header-dir option.
* configure: Rebuild from configure.ac.
* Makefile.in: Support --with-native-system-header-dir.
* doc/install.texi:
Our existing vector permutation support uses the name vec_perm.
The new support added for __builtin_shuffle used a variety of
names. This brings all the internals into line with vec_perm.
I've added documentation for the (existing) variable permute
rtl named pattern (vec_perm). I've also added
On Oct 7, 2011, at 2:52 PM, Aldy Hernandez wrote:
First, thanks so much for tackling this review. I don't think anyone's
overly enthusiastic about reviewing dejagnu crap^H^H^H^Hcode.
I'll review it. :-) The last version looks fine to me, watch out for failures.
How do you like this
The -fsplit-stack runtime support code has a bug reported in PR 46093:
if the size of the stack parameters is exceptionally large, the
parameters can overrun the entire newly allocated stack segment. This
patch fixes the problem. Bootstrapped and tested on
x86_64-unknown-linux-gnu. Committed to
The file libgcc/generic-morestack-thread.c refers to EAGAIN, so for
cleanliness it should really #include errno.h. This does not
introduce any new build problems, as generic-morestack.c already
#includes errno.h anyhow. This patch adds the #include. Bootstrapped
and tested on
Hi,
this diagnostic PR is about duplicate inform messages for this kind of
testcase, where C has a cloned destructor:
class A {};
struct C : A
{
virtual ~C () = 0;
} c;
The fix seems easy: output a cloned destructor only once (+ output any
other member functions normally). Patch tested
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg01293.html
Yes, this is useful for static prediction tuning.
David
On Fri, Oct 7, 2011 at 6:40 PM, Dehao Chen de...@google.com wrote:
http://gcc.gnu.org/ml/gcc-patches/2011-09/msg01293.html
This patch makes pph.h the single interface header for PPH.
The pph-streamer.h header is now an internal header for use
within the pph*.c files. Some declarations have moved from
one header to the other to make that possible.
Index: gcc/cp/ChangeLog.pph
2011-10-07 Lawrence Crowl
86 matches
Mail list logo
