[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus@mc.pp.se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus@mc.pp.se  2003-06-01 17:57 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing


"dhazeghi@yahoo.com" <gcc-bugzilla@gcc.gnu.org> writes:

> Hello,
> 
> I can confirm that on gcc 3.3 branch and mainline (20030527), the same faulty code is still 
> produced. Which previous snapshot do you find this to be a regression against?

To the best of my knowledge, the 20030113 snapshot did not have this
problem.  (Sorry that I can't just test to make sure, but I don't have
the binaries anymore.)


  // Marcus






------- You are receiving this mail because: -------
You are on the CC list for the bug, or are watching someone who is.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[pinskia@physics.uc.edu]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


pinskia@physics.uc.edu changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |3.3.1

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[giovannibajo at libero dot it]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


giovannibajo at libero dot it changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |critical
           Priority|P3                          |P1

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[mmitchel at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.3.1                       |3.3.2


------- Additional Comments From mmitchel at gcc dot gnu dot org  2003-07-11 22:55 -------
Postponed until GCC 3.3.2: SH is not a primary platform.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |ebotcazou at gcc dot gnu dot
                   |                            |org
  GCC build triplet|sparc-sun-solaris2.8        |
   GCC host triplet|sparc-sun-solaris2.8        |


------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-07 14:28 -------
Confirmed on i586-redhat-linux-gnu as of gcc version 3.3.2 20030904
(prerelease). This is tricky reload problem, which is related to the range of
immediate constants accepted by the 'add' instructions (so the bug doesn't show
up with buf[126] and below).

I cleared the 'host' field since the bug is not SPARC-specific.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|ebotcazou at gcc dot gnu dot|
                   |org                         |
         AssignedTo|unassigned at gcc dot gnu   |ebotcazou at gcc dot gnu dot
                   |dot org                     |org
             Status|NEW                         |ASSIGNED


------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-12 06:56 -------
Trying to fix the reload pass.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


dank at kegel dot com changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dank at kegel dot com


------- Additional Comments From dank at kegel dot com  2003-09-12 07:42 -------
Thanks!  BTW are you going to submit a testcase for the testsuite,
too (or is there already one)?

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-12 08:18 -------
It would be nice to have a testcase (a runnable one of course). Could you write
it and post it here (I don't have access to SH hardware)? I'll commit it along
with my patch.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus at mc dot pp dot se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus at mc dot pp dot se  2003-09-12 19:06 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing


Here is the reproduction recipe reworked as a runnable testsuite test.
If the program exits with code 0, the compiler is ok.  If it exits
with code 1 or segfaults, the compiler is not ok.

---8<--- test.c ---8<---
#include <stdio.h>
#include <string.h>

char res[128];

char *use(char *z)
{
  strncat(res, z, 10);
  return "X";
}

void func(char *a, char *b)
{
  char buf[128];
  unsigned char i; 
  char *item[] = {
    "one",
    "two",
  };
  strcpy(buf, "buf");
  for(i=0; i<2; i++) {
    char *x;
    use(item[i]);
    x = use(buf);
    use(a);
    use(b);
    use(x);
  }
}

int main()
{
  strcpy(res, "<");
  func("A", "B");
  strcat(res, ">");
  /* printf("res=\"%s\"\n", res); */
  return strcmp(res, "<onebufABXtwobufABX>")? 1 : 0;
}
---8<---

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus at mc dot pp dot se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus at mc dot pp dot se  2003-09-12 19:12 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

The flags -O2 -m4 will be needed for the testcase as well, of course.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-12 21:02 -------
Thanks, but testcases should be self-contained, that is they shouldn't include
header files. Testing the first character sent to use() and aborting if it is
unexpected should be good enough.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dkegel at ixiacom dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dkegel at ixiacom dot com  2003-09-12 23:08 -------
OK, how 'bout this?  This reproduces the crash nicely,
and should be ready to check in as a dg testcase.

/* PR optimization/10392
 * [3.3/3.4 regression] [SH] optimizer generates faulty array indexing
 * The address calculation of an index operation on an array on the stack 
 * can _under some conditions_ get messed up completely
*/
/* { dg-do run } */

const char *foo;

const char *use(const char *str)
{
	foo = str;
	return str[0] ? "apple" : "bannana";
}

const char *func(char *a, char *b)
{
	char buf[128];
	unsigned char i;
	const char *result;

	char *item[] = {
		"one",
		"two",
	};

	buf[0] = 'b';
	buf[1] = 0;

	for (i = 0; i < 2; i++) {
		/* bug is: following line crashes with SIGSEGV on sh4 -O2 */
		result = use(item[i]);

		use(buf);
		use(a);
		use(b);
		result = use(result);
	}
	return result;
}

int main()
{
	/* func never returns zero, so this will always exit with status 0,
	 * but the optimizer doesn't know that 
	 */
	return func("a", "b") != 0;
}

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dkegel at ixiacom dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dkegel at ixiacom dot com  2003-09-12 23:10 -------
Er, except maybe the dg-do header needs to be changed a bit, I dunno.  
I haven't tested it with dg, just standalone.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-13 03:41 -------
Looks good. Thanks!

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus at mc dot pp dot se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus at mc dot pp dot se  2003-09-13 09:58 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

Two comments about this test case:

1) It does not SIGSEGV on NetBSD/dreamcast.  The argument to use()
   will point into the environment buffer, in my case it points to
   "VENDOR=unknown".  But as long as it actually segfaults on the
   machine you're using to run these tests, I suppose that's ok...

2) Because func() never returns zero, the program always exits with
   status _1_, not 0.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-13 10:33 -------
It doesn't segfault on my machine since I don't have SH hardware :-)

I'd then strongly suggest to test str[0] in use() and put an abort(). And the
main function should always return 0 on success.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-09-13 14:14 -------
Thanks.  I'll update my copy of the testcase with those corrections.
(I plan to submit a patch in a couple days with testcases for
this and a bunch other sh PRs.)

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus at mc dot pp dot se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus at mc dot pp dot se  2003-09-13 21:13 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

What I don't get is why the testcase shouldn't include stuff that's
part of the C standard.  Several other testcases already do.  In
gcc.dg alone I count to 61 of them.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-09-15 17:40 -------
I've prepared a patch to add a testcase for this PR to the testsuite; see
http://www.kegel.com/crosstool/current/patches/gcc-3.3.1/pr10392-1-test.patch
I'll submit it to gcc-patches in a couple days if nobody squawks.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus at mc dot pp dot se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus at mc dot pp dot se  2003-09-15 18:14 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

I have tested this testcase on NetBSD/dreamcast, and it seems ok.  One
thing though:  Shouldn't dg-options include "-m4"?  The bug will not
be trigged without this flag.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-09-16 01:23 -------
Perhaps.  But then you'd need to restrict the test to only run
on sh4 targets, and on those targets, -m4 is given anyway, isn't it?

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[marcus at mc dot pp dot se]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From marcus at mc dot pp dot se  2003-09-16 11:07 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing


On NetBSD/dreamcast -m4 is not default even though the CPU is a SH4,
since the binaries should be interoperable with other NetBSD/sh
ports.  The question is of course: Does the machine which will be
running this regression test to detect the regression default to -m4?

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-09-16 16:15 -------
Unfortunately, I can't think of a safe and effective way to specify -m4 right in
the testcase. 

I could restrict the testcase to sh*-*-* and avoid failing
non-SH processors, but what about SH processors that can't run code
compiled with -m4?  

I could restrict the testcase to sh4-*-*, and specify -m4; that would 
make the intent clear, but wouldn't have run the test when e.g. 
http://mc.pp.se/dc/files/gcc3.3toolchain_local.tar.bz2 was built,
as that used target shl-unknown-netbsdelf1.6T.

Any opinions?

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-17 08:19 -------
I think

/* { dg-options "-O2" } */
/* { dg-options "-O2 -m4" { target sh4-*-* } } */

is a good solution.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-09-17 17:09 -------
Roger, wilco!

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-09-18 04:57 -------
What are your fullnames? I want to put them in the ChangeLog entry for the testcase.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-09-18 05:09 -------
I'm dank@kegel.com (Dan Kegel)

Google says the submitter is  marcus@mc.pp.se (Marcus Comstedt)

Thanks for committing the testcase!

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[mmitchel at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


mmitchel at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|3.3.2                       |3.3.3


------- Additional Comments From mmitchel at gcc dot gnu dot org  2003-10-16 02:58 -------
It looks like we build a test-case here -- but never fixed the bug.

Postponed until GCC 3.3.3.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |mark at codesourcery dot com


------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-10-16 04:29 -------
Well, we fixed the bug:
patch: http://gcc.gnu.org/ml/gcc-patches/2003-09/msg01083.html
approval: http://gcc.gnu.org/ml/gcc-patches/2003-10/msg00433.html
but: http://gcc.gnu.org/ml/gcc-patches/2003-09/msg02076.html
and I still don't have any testcase...

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[dank at kegel dot com]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From dank at kegel dot com  2003-10-16 06:40 -------
Oh, sorry.  You mean a testcase derived from
ftp://shell14.ba.best.com/pub.t/tm2/stress-1.17.tar.gz
I would try, but I can't contact that host, and I can't find
another source.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392



------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-10-16 06:52 -------
Yes. I think we should be very cautious before letting the patch go in, because
reload is a very sensitive pass of the compiler.

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

PLEASE REPLY TO gcc-bugzilla@gcc.gnu.org ONLY, *NOT* gcc-bugs@gcc.gnu.org.

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392


ebotcazou at gcc dot gnu dot org changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |joern dot rennecke at superh
                   |                            |dot com
         AssignedTo|ebotcazou at gcc dot gnu dot|unassigned at gcc dot gnu
                   |org                         |dot org
             Status|ASSIGNED                    |NEW


------- Additional Comments From ebotcazou at gcc dot gnu dot org  2003-10-25 13:11 -------
I don't really know what to do with this PR. The current situation is:
- the patch was approved for 3.3 branch,
- the patch reportedly introduces a regression in a particular package on SH4,
- the bug doesn't show up on mainline (but is latent).

I give up, all the more that I don't have access to SH hardware. If a SH
maintainer wants to take care of it...

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[gdr at gcc dot gnu dot org]

------- Additional Comments From gdr at gcc dot gnu dot org  2003-12-21 18:45 -------
Joern --

This bug is present on both 3.3.x branch and mainline.
It has "Severity:" set to critical.  Do you think we should
keep it open for 3.3.3? (There is virtually no chance I fix 
this myself as I don't speak SH).


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[gdr at gcc dot gnu dot org]

------- Additional Comments From gdr at gcc dot gnu dot org  2004-01-14 08:08 -------
There does not seem to be anything one can do for this before 3.3.3.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|critical                    |minor
           Priority|P1                          |P3


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[joern dot rennecke at superh dot com]

------- Additional Comments From joern dot rennecke at superh dot com  2004-01-14 19:03 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

> ------- Additional Comments From gdr at gcc dot gnu dot org  2003-12-21 18:45 -------
> Joern --
> 
> This bug is present on both 3.3.x branch and mainline.
> It has "Severity:" set to critical.  Do you think we should
> keep it open for 3.3.3? (There is virtually no chance I fix 
> this myself as I don't speak SH).

AFAICS the problem is no longer present in mainline; I've attached the output
I get with yesterday's compiler.  When did you last reconfirm the bug?

The incorrect use of reload registers observed is consistent with
a problem in reg_overlap_mentioned_for_reload_p that I've fixed with
this patch:

2003-12-15  J"orn Rennecke <joern.rennecke@superh.com>

        * reload.c (reg_overlap_mentioned_for_reload_p):
        When looking at a PLUS in X, avoid spuriously returning nonzero
        when IN is a REG or another simple PLUS, or a MEM containing one.

http://gcc.gnu.org/ml/gcc-patches/2003-09/msg01299.html

	.file	"10392.c"
	.text
	.section	.rodata.str1.4,"aMS",@progbits,1
	.align 2
.LC0:
	.string	"one"
	.align 2
.LC1:
	.string	"two"
	.text
	.align 1
	.align 5
	.global	_func
	.type	_func, @function
_func:
	mov.l	r8,@-r15
	mov.l	r9,@-r15
	mov.l	r10,@-r15
	mov	#0,r10
	mov.l	r11,@-r15
	mov	r5,r11
	mov.l	r12,@-r15
	mov	r4,r12
	mov.l	r13,@-r15
	mov	#1,r13
	mov.l	r14,@-r15
	sts.l	pr,@-r15
	mov.w	.L13,r0
	add	#-68,r15
	mov.l	.L10,r1
	add	#-68,r15
	mov	r15,r14
	add	r14,r0
	mov.l	r1,@r0
	mov.l	.L11,r9
	mov.l	.L12,r1
	mov.l	r1,@(4,r0)
.L5:
	mov.w	.L13,r1
	mov	r10,r2
	shll2	r2
	mov	r14,r0
	add	r2,r1
	jsr	@r9
	mov.l	@(r0,r1),r4
	jsr	@r9
	mov	r14,r4
	mov	r12,r4
	jsr	@r9
	mov	r0,r8
	jsr	@r9
	mov	r11,r4
	jsr	@r9
	mov	r8,r4
	mov	r10,r1
	add	#1,r1
	extu.b	r1,r10
	cmp/hi	r13,r10
	bf	.L5
	add	#68,r14
	add	#68,r14
	mov	r14,r15
	lds.l	@r15+,pr
	mov.l	@r15+,r14
	mov.l	@r15+,r13
	mov.l	@r15+,r12
	mov.l	@r15+,r11
	mov.l	@r15+,r10
	mov.l	@r15+,r9
	rts	
	mov.l	@r15+,r8
	.align 1
.L13:
	.short	128
.L14:
	.align 2
.L10:
	.long	.LC0
.L11:
	.long	_use
.L12:
	.long	.LC1
	.size	_func, .-_func
	.ident	"GCC: (GNU) 3.4.0 20040113 (experimental)"


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

[joern dot rennecke at superh dot com]

------- Additional Comments From joern dot rennecke at superh dot com  2004-01-14 19:37 -------
Subject: Re:  [3.3/3.4 regression] [SH] optimizer generates faulty array indexing

P.S.: Closer inspection showed that the patch that fixed this
on mainline was actually this one:

2003-06-26  J"orn Rennecke <joern.rennecke@superh.com>

        * reload.c (can_reload_into): New function.
        (push_reload): Use it.

http://gcc.gnu.org/ml/gcc-patches/2003-06/msg02964.html

It applies cleanly to 3.3 with:
cvs -z 9 update -j1.214 -j1.215 reload.c
and fixes the testcase.


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2004-01-14 19:49 -------
Only a 3.3 regression.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
            Summary|[3.3/3.4 regression] [SH]   |[3.3 regression] [SH]
                   |optimizer generates faulty  |optimizer generates faulty
                   |array indexing              |array indexing


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-01-15 20:21 -------
Subject: Bug 10392

CVSROOT:	/cvs/gcc
Module name:	gcc
Branch: 	gcc-3_3-branch
Changes by:	amylaar@gcc.gnu.org	2004-01-15 20:21:52

Modified files:
	gcc            : ChangeLog reload.c 

Log message:
	PR optimization/10392
	* reload.c (can_reload_into): New function.
	(push_reload): Use it.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/ChangeLog.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.16114.2.882&r2=1.16114.2.883
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/reload.c.diff?cvsroot=gcc&only_with_tag=gcc-3_3-branch&r1=1.199.2.10&r2=1.199.2.11



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[cvs-commit at gcc dot gnu dot org]

------- Additional Comments From cvs-commit at gcc dot gnu dot org  2004-01-16 18:24 -------
Subject: Bug 10392

CVSROOT:	/cvs/gcc
Module name:	gcc
Changes by:	amylaar@gcc.gnu.org	2004-01-16 18:24:09

Modified files:
	gcc/testsuite  : ChangeLog 
Added files:
	gcc/testsuite/gcc.dg: pr10392-1.c 

Log message:
	PR 10392
	From Marcus Comstedt / Dan Kegel:
	* gcc.dg/pr10392-1.c: New test.

Patches:
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/ChangeLog.diff?cvsroot=gcc&r1=1.3385&r2=1.3386
http://gcc.gnu.org/cgi-bin/cvsweb.cgi/gcc/gcc/testsuite/gcc.dg/pr10392-1.c.diff?cvsroot=gcc&r1=NONE&r2=1.1



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[amylaar at gcc dot gnu dot org]

------- Additional Comments From amylaar at gcc dot gnu dot org  2004-01-16 18:28 -------
Patch from mainline merged into 3.3 branch, testcase checked in on mainline.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
         AssignedTo|unassigned at gcc dot gnu   |amylaar at gcc dot gnu dot
                   |dot org                     |org
             Status|REOPENED                    |ASSIGNED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-01-21 12:45 -------
There is a problem on the 3.3 branch:

../../gcc/reload.c: In function `can_reload_into':
../../gcc/reload.c:859: warning: traditional C rejects ISO C style function
definitions



-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |REOPENED
         Resolution|FIXED                       |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[gdr at integrable-solutions dot net]

------- Additional Comments From gdr at integrable-solutions dot net  2004-01-21 14:09 -------
Subject: Re:  [3.3 regression] [SH] optimizer generates faulty array indexing

"ebotcazou at gcc dot gnu dot org" <gcc-bugzilla@gcc.gnu.org> writes:

| There is a problem on the 3.3 branch:
| 
| ../../gcc/reload.c: In function `can_reload_into':
| ../../gcc/reload.c:859: warning: traditional C rejects ISO C style function
| definitions

Odd.  I bootstrapped the compiler at least about 20 times this night,
I did not notice this.  I'll fix it.

-- Gaby


-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392

[Bug optimization/10392] [3.3 regression] [SH] optimizer generates faulty array indexing

[ebotcazou at gcc dot gnu dot org]

------- Additional Comments From ebotcazou at gcc dot gnu dot org  2004-01-25 07:14 -------
The last problem has been fixed.


-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |FIXED


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=10392