--- Comment #19 from rguenth at gcc dot gnu dot org 2005-11-17 11:36
---
Fixed.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
Status|ASSIGNE
--- Comment #18 from rguenth at gcc dot gnu dot org 2005-11-17 11:35
---
Subject: Bug 24851
Author: rguenth
Date: Thu Nov 17 11:35:00 2005
New Revision: 107117
URL: http://gcc.gnu.org/viewcvs?root=gcc&view=rev&rev=107117
Log:
2005-11-16 Richard Guenther <[EMAIL PROTECTED]>
--- Comment #17 from paolo dot bonzini at lu dot unisi dot ch 2005-11-16
09:41 ---
Subject: Re: [4.1 Regression] f2c miscompilation
rguenth at gcc dot gnu dot org wrote:
>--- Comment #16 from rguenth at gcc dot gnu dot org 2005-11-16 09:39
>---
>Is the second reduced testca
--- Comment #16 from rguenth at gcc dot gnu dot org 2005-11-16 09:39
---
Is the second reduced testcase not fine from a standards POV? I.e.
void abort(void);
int main()
{
int a[10], *p, *q;
q = &a[1];
p = &q[-1];
if (p >= &a[9])
abort ();
return 0;
}
or does "array" in
--- Comment #15 from ian at airs dot com 2005-11-16 01:09 ---
Does it bother us that legal ISO C programs are not permitted to say &a[-1]?
You are permitted to take the address of the element immediately after the
array, but you aren't permitted to take the address of the element before
--- Comment #14 from pinskia at gcc dot gnu dot org 2005-11-14 19:09
---
(In reply to comment #13)
> I have a fix for the only valid testcase (comment #7) here.
s/valid/defined/
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24851
--- Comment #13 from pinskia at gcc dot gnu dot org 2005-11-14 19:08
---
I have a fix for the only valid testcase (comment #7) here.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
---
--- Comment #12 from pinskia at gcc dot gnu dot org 2005-11-14 18:47
---
Note we haveyyp = &yys[-1]; (where yys is an array) in
the orginal testcase so that is undefined.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24851
--- Comment #11 from pinskia at gcc dot gnu dot org 2005-11-14 17:31
---
div_if_zero_remainder assumes that the type of the agruments comming in are the
same.
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24851
--- Comment #10 from pinskia at gcc dot gnu dot org 2005-11-14 17:29
---
Actually I was wrong on which patch caused/exposed this,
This was caused by:
2005-05-14 Richard Guenther <[EMAIL PROTECTED]>
* fold-const.c (div_if_zero_remainder): New function.
(try_move_mult_t
--- Comment #9 from pinskia at gcc dot gnu dot org 2005-11-14 17:04 ---
(In reply to comment #8)
> Patch to avoid the situation posted.
So this patch makes the real bug latent.
--
pinskia at gcc dot gnu dot org changed:
What|Removed |Added
---
--- Comment #8 from rguenth at gcc dot gnu dot org 2005-11-14 17:03 ---
Patch to avoid the situation posted.
--
rguenth at gcc dot gnu dot org changed:
What|Removed |Added
--- Comment #7 from rguenth at gcc dot gnu dot org 2005-11-14 16:27 ---
Valid testcase (?):
void abort(void);
int main()
{
int a[10], *p, *q;
q = &a[1];
p = &q[-1];
if (p >= &a[9])
abort ();
return 0;
}
fold_stmt is doing the &a[0] + -4B to &a[1073741823] transformation.
--- Comment #6 from pinskia at gcc dot gnu dot org 2005-11-14 15:56 ---
The problem here is that we are comparing in unsigned when we shoud be
comparing in a signed type.
This was introduced by:
2005-01-29 Richard Guenther <[EMAIL PROTECTED]>
PR tree-optimization/15791
--- Comment #5 from pinskia at gcc dot gnu dot org 2005-11-14 15:42 ---
Confirmed, short testcase:
int f(void)
{
int i = -1;
int a[10];
int *b = &a[2];
return &a[i]>=b;
}
int main(void)
{
if (f())
abort();
}
--
pinskia at gcc dot gnu dot org changed:
What
--- Comment #4 from rguenth at gcc dot gnu dot org 2005-11-14 15:42 ---
C++ frontend is fine. With C out of
(.03.gimple)
p = &a + -4B;
D.1282 = &a + 36B;
we (fold?) get
(.10.cleanup_cfg)
p = &a[1073741823];
D.1282 = &a[9];
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=2
--- Comment #3 from rguenth at gcc dot gnu dot org 2005-11-14 15:40 ---
Reduced testcase:
void abort(void);
int main()
{
int a[10], *p;
p = &a[-1];
if (p >= &a[9])
abort ();
return 0;
}
--
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=24851
17 matches
Mail list logo
