--- Comment #27 from bonzini at gnu dot org 2009-07-01 16:52 ---
Created an attachment (id=18112)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18112&action=view)
... watch out, this is on top of mainline, not on top of pr40957-3.patch
(It would apply and just n
--- Comment #7 from bonzini at gnu dot org 2009-07-01 18:08 ---
*** Bug 40610 has been marked as a duplicate of this bug. ***
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #1 from bonzini at gnu dot org 2009-07-01 18:08 ---
*** This bug has been marked as a duplicate of 40607 ***
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #1 from bonzini at gnu dot org 2009-07-02 05:22 ---
can you bisect it?
--
bonzini at gnu dot org changed:
What|Removed |Added
CC
--- Comment #3 from bonzini at gnu dot org 2009-07-14 10:08 ---
One side problem here is that
mov r0, r2
mov r3, #1
eor r0, r0, r3
could become
mov r0, #1
eor r0, r0, r2
--
bonzini at gnu dot org changed
--- Comment #4 from bonzini at gnu dot org 2009-07-14 10:34 ---
... and besides that, we do not take rtx_costs into account.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #2 from bonzini at gnu dot org 2009-07-14 10:51 ---
Here combine simplifies the zero extension from
(lshiftrt:SI (ashift:SI (subreg:SI (reg:QI 138) 0)
(const_int 24 [0x18]))
(const_int 24 [0x18]))
to (ne:SI (subreg:SI (reg:QI 138)). From there it cannot derive
--- Comment #3 from bonzini at gnu dot org 2009-07-14 12:56 ---
Richard, is your testcase also a regression? In that case the culprit is
mostly
#if 0
/* Disabled to avoid exponential mutual recursion between nonzero_bits
and num_sign_bit_copies. */
if
--- Comment #1 from bonzini at gnu dot org 2009-07-14 20:35 ---
This fails for me with r149508 with a reload failure.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39716
--- Comment #1 from bonzini at gnu dot org 2009-07-14 20:36 ---
I meant that on most targets this testcase was improved by cond-optab, but not
on m68hc11.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39719
--- Comment #2 from bonzini at gnu dot org 2009-07-14 21:17 ---
(This is gcc.c-torture/compile/20071128-1.c).
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39722
--- Comment #1 from bonzini at gnu dot org 2009-07-14 21:25 ---
(This is peak-gcc-src/gcc/testsuite/gcc.c-torture/compile/pr38564.c).
I cannot reproduce this anymore.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #1 from bonzini at gnu dot org 2009-07-14 21:32 ---
Cannot reproduce this anymore.
--
bonzini at gnu dot org changed:
What|Removed |Added
Status
--- Comment #1 from bonzini at gnu dot org 2009-07-14 21:34 ---
I cannot reproduce this anymore.
--
bonzini at gnu dot org changed:
What|Removed |Added
--
bonzini at gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |bonzini at gnu dot org
|dot org
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|UNCONFIRMED |NEW
Ever Confirmed|0 |1
Last reconfirmed
--- Comment #3 from bonzini at gnu dot org 2009-07-14 22:38 ---
Fixed by the new SRA thanks to its usage of VIEW_CONVERT_EXPR at the tree
level.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #4 from bonzini at gnu dot org 2009-07-14 22:41 ---
I have a patch for both (two patches actually).
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39715
--- Comment #1 from bonzini at gnu dot org 2009-07-14 22:42 ---
I have a patch.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39726
--- Comment #5 from bonzini at gnu dot org 2009-07-14 22:42 ---
I have a patch.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40741
ned at gcc dot gnu dot org
ReportedBy: bonzini at gnu dot org
BugsThisDependsOn: 2161
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40760
--- Comment #32 from bonzini at gnu dot org 2009-07-15 06:05 ---
Yes, but I don't think it's infinite recursion. There are 11,000 else ifs in
the testcase.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40597
--- Comment #1 from bonzini at gnu dot org 2009-07-15 06:06 ---
A regression from when, well, there was no gimplifier.
--
bonzini at gnu dot org changed:
What|Removed |Added
ssigned at gcc dot gnu dot org
ReportedBy: bonzini at gnu dot org
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40761
--- Comment #3 from bonzini at gnu dot org 2009-07-15 11:11 ---
do while does not have the same behavior, so the loop "shape" is important.
the following is as bad and does not have function calls.
#define ONE while (x-- > y)
#define TEN ONE ONE ONE ONE ONE ONE ON
--- Comment #13 from bonzini at gnu dot org 2009-07-15 16:20 ---
For the record, it's actually somewhat related to PR39726 (a m68k
pessimization), not PR39715. However, because of the way combine canonicalizes
the resulting expression, the patch for that bug does not fix the tes
--- Comment #36 from bonzini at gnu dot org 2009-07-23 23:01 ---
No, all patches were committed.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40597
--- Comment #2 from bonzini at gnu dot org 2009-08-01 21:25 ---
Hmm, the patch changes a canonicalization in a way that may break some patterns
on some targets. :-(
Not sure how to proceed.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=39726
--- Comment #7 from bonzini at gnu dot org 2009-08-01 22:57 ---
Hmm, Pentium is not a cmove target and it doesn't like sahf, so
ix86_fp_comparison_operator matches all comparison operators. I would have
thought then that fp_jcc_3_387 would match:
(define_insn "*fp_jcc_3_38
--- Comment #9 from bonzini at gnu dot org 2009-08-01 23:41 ---
I confirm that the SAHF instruction sequence is generated because of
optimize_insn_for_size_p () == 1.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40934
--- Comment #10 from bonzini at gnu dot org 2009-08-01 23:59 ---
Minimized testcase:
extern double host_frametime;
extern float pitchvel;
V_DriftPitch (float delta, float move)
{
if (!delta)
move = host_frametime;
if (delta > 0)
;
else if (delta < 0 &&
--- Comment #11 from bonzini at gnu dot org 2009-08-02 00:15 ---
This is the simplest patch that can possibly work:
Index: ../../gcc/config/i386/i386.c
===
--- ../../gcc/config/i386/i386.c(revision 150334
--- Comment #1 from bonzini at gnu dot org 2009-08-03 20:54 ---
Well, stage2 is better than stage3 :-)
Can you post a preprocessed testcase?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40952
--- Comment #2 from bonzini at gnu dot org 2009-08-04 06:37 ---
I found this pasto, can you check?
Index: gcc/calls.c
===
--- gcc/calls.c (revision 150359)
+++ gcc/calls.c (working copy)
@@ -2353,9 +2353,9 @@ expand_call
--- Comment #8 from bonzini at gnu dot org 2009-08-04 22:30 ---
Created an attachment (id=18304)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18304&action=view)
plausible patch
The patch adds back the mode of operation corresponding to
promote_nominal_mode's ass
--- Comment #10 from bonzini at gnu dot org 2009-08-05 06:09 ---
preprocessed testcase?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40952
--- Comment #12 from bonzini at gnu dot org 2009-08-05 23:09 ---
Created an attachment (id=18309)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18309&action=view)
another patch
This failure is related to NRV, which is why there is no C failure.
If I take care of it the p
--- Comment #15 from bonzini at gnu dot org 2009-08-07 07:24 ---
Subject: Re: version 150336 broke bootstrap on ia64-hp-hpux11.23
> After applying the most recent patch the ia64 bootstrap started working but
> the
> pa bootstrap (hppa2.0w-hp-hpux11.11) started failing. T
--- Comment #5 from bonzini at gnu dot org 2009-08-07 09:46 ---
No, the build system should support everything; not just host==target != build
which is not so uncommon -- for example this is how cygwin worked before it
could host GCC -- but even the admittedly crazy target==build
--- Comment #7 from bonzini at gnu dot org 2009-08-07 10:45 ---
What's the content of gcc/fenv.h in the build directory? If I have to guess,
I'd suppose it's generated by fixincludes.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40974
--- Comment #16 from bonzini at gnu dot org 2009-08-08 07:57 ---
Can you try running the testsuite on a non-bootstrapped hppa-hpux compiler?
However, I have an idea and will provide a patch soon.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=40952
--- Comment #17 from bonzini at gnu dot org 2009-08-08 08:18 ---
Try squashing this in:
Index: pa/pa.c
===
--- pa/pa.c (revision 150359)
+++ pa/pa.c (working copy)
@@ -9199,7 +9199,7 @@ pa_promote_function_mode
--- Comment #20 from bonzini at gnu dot org 2009-08-10 20:09 ---
Created an attachment (id=18337)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=18337&action=view)
patch fixing hpux differences
This patch squashes in the following too. It is a stupid used-uninitialized
b
--- Comment #24 from bonzini at gnu dot org 2009-08-12 16:29 ---
patch committed as r150700
--
bonzini at gnu dot org changed:
What|Removed |Added
Status
--- Comment #5 from bonzini at gnu dot org 2005-11-25 19:12 ---
Another failure from glibc's tests is more severe: __DBL_MAX__ * __DBL_MAX__ -
__DBL_MAX * __DBL_MAX__ is turned into an INF rather than a NAN (as would
happen on most targets) or zero (as would happen on x87 becau
--- Comment #6 from bonzini at gnu dot org 2005-11-25 19:13 ---
CCing Eric since he recently had problems with ia64 fma
(http://gcc.gnu.org/ml/gcc/2005-10/msg01036.html).
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #2 from bonzini at gnu dot org 2005-11-28 10:53 ---
Created an attachment (id=10352)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10352&action=view)
patch to fix the bug
Kaz, can you please test this patch?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25115
--- Comment #5 from bonzini at gnu dot org 2005-11-29 17:55 ---
Frederic, "-Wstrict-aliasing" is only working for C up until 4.1 (included).
Your code, rewritten to C, would issue a warning with 4.1 and earlier compilers
as well. Also note that -Wstrict-aliasing does not p
--- Comment #2 from bonzini at gnu dot org 2005-11-30 08:15 ---
I have an obvious patch, which is to invert the CONSTANT_CLASS_P and
TREE_CONSTANT_OVERFLOW tests in line 3330.
Will commit after bootstrapping/testing succeeds.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25169
--- Comment #8 from bonzini at gnu dot org 2005-11-30 16:11 ---
Reduced testcase:
typedef unsigned long long u64;
extern u64 f (u64 x);
int g (unsigned x, u64 *z)
{
u64 w = *z;
u64 h = f (w) << 32;
u64 l = f (w);
u64 g = h | l;
unsigned p = g;
if (p == x)
--- Comment #9 from bonzini at gnu dot org 2005-11-30 17:36 ---
Created an attachment (id=10369)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10369&action=view)
proposed patch
This patch enables the peephole2 only if operands[1] and operands[2] are
constant.
An alte
--
bonzini at gnu dot org changed:
What|Removed |Added
AssignedTo|paolo dot bonzini at lu dot |bonzini at gnu dot org
|unisi dot ch
--- Comment #1 from bonzini at gnu dot org 2005-12-05 09:09 ---
Created an attachment (id=10406)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10406&action=view)
patch sketch
A very experimental (not even built, let alone bootstrapped/regtested) patch
--
http://gcc.
--- Comment #11 from bonzini at gnu dot org 2005-12-05 17:21 ---
Patch applied to mainline. I will ask for 4.1 approval in a few days, most
likely at the beginning of next week.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #2 from bonzini at gnu dot org 2005-12-06 12:28 ---
Created an attachment (id=10417)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10417&action=view)
proposed patch
I bootstrapped this on i686-pc-linux-gnu, all languages. Eric, can you test it
on a non-C99 p
--- Comment #7 from bonzini at gnu dot org 2005-12-15 14:43 ---
Created an attachment (id=10496)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10496&action=view)
one more update
This one includes the change to the usage of __extension__ that Andrew pointed
out.
--
bon
--- Comment #1 from bonzini at gnu dot org 2005-12-16 09:00 ---
A "make" from a toplevel is equivalent to the old "make bubblestrap" or "make
-C bubblestrap". In practice "make" just does the right thing, compiling all
that is needed to not hav
--
bonzini at gnu dot org changed:
What|Removed |Added
AssignedTo|unassigned at gcc dot gnu |bonzini at gnu dot org
|dot org
--- Comment #9 from bonzini at gnu dot org 2005-12-17 10:19 ---
Should work on the mainline now.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #5 from bonzini at gnu dot org 2005-12-17 10:27 ---
Well, another work around should probably be (untested)
Confirmed, reduced testcase:
void f(void*);
void *g(void);
void _aie_malloc(unsigned int size)
{
void *aie_memory_heap_ptr;
if (__builtin_expect(size != 0
--- Comment #9 from bonzini at gnu dot org 2005-12-17 13:31 ---
Created an attachment (id=10518)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10518&action=view)
patch suggested by Roger
This patch uses the approach, suggested by Roger Sayle on gcc-patches, to Look
for loo
--- Comment #3 from bonzini at gnu dot org 2005-12-19 08:28 ---
Sorry to ask the question. Are you sure that this is a 4.2 regression and not
a 4.1 regression as well? The fixincludes module is not bootstrapped, so there
is very small (if any) chance that the toplevel changes in 4.2
--- Comment #2 from bonzini at gnu dot org 2005-12-19 08:40 ---
I think this was some kind of merging hiccup because I found this in the
ChangeLog:
2004-08-30 Paolo Bonzini <[EMAIL PROTECTED]>
* configure.ac: Do not run fixincludes after stage1 during
to
--- Comment #3 from bonzini at gnu dot org 2005-12-19 08:47 ---
Indeed, Geoff removed the snippet in revision 90117 probably because he did not
know what it was good for. I will reapply the patch.
--
bonzini at gnu dot org changed:
What|Removed
--- Comment #13 from bonzini at gnu dot org 2005-12-19 14:24 ---
David, ok to commit to 4.1 now?
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #15 from bonzini at gnu dot org 2005-12-19 17:46 ---
patch committed to 4.1 branch as well.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #9 from bonzini at gnu dot org 2005-12-20 14:15 ---
Created an attachment (id=10535)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10535&action=view)
fix the #ifndef -> use #ifdef instead
--
bonzini at gnu dot org changed:
What
--- Comment #11 from bonzini at gnu dot org 2005-12-20 17:22 ---
patch committed
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #11 from bonzini at gnu dot org 2005-12-21 08:25 ---
Created an attachment (id=10541)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10541&action=view)
fix pasto in int16_t definition
This might be the last one...
--
bonzini at gnu dot org changed:
--- Comment #4 from bonzini at gnu dot org 2005-12-28 15:46 ---
I will be able to fix this at the beginning of January. If anybody wants to
revert my combine.c patch before that, go ahead.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25578
--- Comment #5 from bonzini at gnu dot org 2005-12-29 07:53 ---
I had an obvious fix yesterday but I had no time to bootstrap/regtest it. So
if possible please don't revert...
The reduced C testcase is
int f(int x)
{
return x >> 31 ? -1 : 1;
}
The MPFR_SIGN macro (
--- Comment #14 from bonzini at gnu dot org 2006-01-02 18:55 ---
patch committed, please verify it works now
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #5 from bonzini at gnu dot org 2006-01-02 19:31 ---
Reconfirmed after the first load PRE patch went in, as it does not handle
globals.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #3 from bonzini at gnu dot org 2006-01-02 19:37 ---
Just for the record, I'm not sure about how much it is worth to have fold
perform all the various sorts of shift/and simplifications that combine can do.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=19986
--- Comment #8 from bonzini at gnu dot org 2006-01-03 08:04 ---
> Ehm, wouldn't "unsigned char *" alias everything? GCSE doesn't do load PRE
> for
> me on the original test case, either.
Yes, I was checking if load PRE removes the redundant load of t3
--- Comment #8 from bonzini at gnu dot org 2006-01-03 14:58 ---
patch committed
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #23 from bonzini at gnu dot org 2006-01-03 15:30 ---
> For PowerPC, it is effective to use the instruction if
> there are multiple divides, such as the three divisions mentioned above. The
> IBM XLC compiler propagates the reciprocal and numerator pair th
--- Comment #4 from bonzini at gnu dot org 2006-01-04 08:50 ---
H.J., can we close this?
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25435
--- Comment #5 from bonzini at gnu dot org 2006-01-04 08:57 ---
Janis, can you confirm this is fixed on 4.0?
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #2 from bonzini at gnu dot org 2006-01-04 12:21 ---
Patch committed, also fixes race conditions in bootstrap.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #1 from bonzini at gnu dot org 2006-01-04 15:56 ---
Accepting this, but it is not a blocker as all-gcc is not even documented
anywhere.
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #5 from bonzini at gnu dot org 2006-01-04 16:38 ---
Patch posted. Andrew suggested another approach, but it is more complex and
I'm not going to work on it until more important issues are sorted out.
--
bonzini at gnu dot org changed:
What|Re
--- Comment #2 from bonzini at gnu dot org 2006-01-05 15:10 ---
If it's still there, assign it to me.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25677
--- Comment #4 from bonzini at gnu dot org 2006-01-05 16:59 ---
Combine does not try to combine across a call. But it does not invalidate
equivalences and, in this case, it tries to replace v (the result of a function
call) with the return value register... even though in the meanwhile
--- Comment #5 from bonzini at gnu dot org 2006-01-05 16:59 ---
Created an attachment (id=10585)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10585&action=view)
patch to try
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=25677
--- Comment #3 from bonzini at gnu dot org 2006-01-07 10:01 ---
*** This bug has been marked as a duplicate of 25670 ***
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #2 from bonzini at gnu dot org 2006-01-07 10:01 ---
*** Bug 25694 has been marked as a duplicate of this bug. ***
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #2 from bonzini at gnu dot org 2006-01-07 10:01 ---
Huh? We do in the toplevel make target "compare".
Paolo
--
bonzini at gnu dot org changed:
What|Removed
--- Comment #4 from bonzini at gnu dot org 2006-01-09 07:47 ---
Changing the summary then.
--
bonzini at gnu dot org changed:
What|Removed |Added
AssignedTo
--- Comment #5 from bonzini at gnu dot org 2006-01-09 13:14 ---
Of course all-gcc is still doing the same thing for a cross compiler. Anyway I
have a patch to restore the old functionality and, after that, I'll submit doc
fixes.
My comment was just meant to justify the loweri
--- Comment #17 from bonzini at gnu dot org 2006-01-11 14:30 ---
patch committed to 4.1 too
--
bonzini at gnu dot org changed:
What|Removed |Added
Status
--- Comment #28 from bonzini at gnu dot org 2006-01-11 14:30 ---
patch committed to both branches
--
bonzini at gnu dot org changed:
What|Removed |Added
--- Comment #24 from bonzini at gnu dot org 2006-01-11 14:31 ---
patch committed
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #2 from bonzini at gnu dot org 2006-01-12 18:58 ---
The compiler surely converts &data[i] to data+i or the other way round even at
no optimization level and so does with &*&*&*&*&*a that is equivalent to a.
You cannot expect the preprocessor to do
--- Comment #7 from bonzini at gnu dot org 2006-01-16 18:21 ---
Patch committed
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--- Comment #1 from bonzini at gnu dot org 2006-01-20 19:56 ---
Created an attachment (id=10692)
--> (http://gcc.gnu.org/bugzilla/attachment.cgi?id=10692&action=view)
prototype patch
This is a prototype patch to fix the bug using autoconf to compute the
necessary sizes/ali
--- Comment #15 from bonzini at gnu dot org 2006-01-20 19:58 ---
libgomp should use GCC_HEADER_STDINT too. See the patch for PR25884 which does
so.
--
bonzini at gnu dot org changed:
What|Removed |Added
--
bonzini at gnu dot org changed:
What|Removed |Added
AssignedTo|bonzini at gnu dot org |unassigned at gcc dot gnu
--- Comment #15 from bonzini at gnu dot org 2006-07-25 08:34 ---
patch committed
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|ASSIGNED
--
bonzini at gnu dot org changed:
What|Removed |Added
Status|NEW |ASSIGNED
Last reconfirmed|2006-03-30 20:09:49 |2006-07-25 14:53:34
201 - 300 of 1283 matches
Mail list logo
