ICE in extract_insn with gcc-20040620

ICE in extract_insn with gcc-20040620

[Tim Janik]

[-- Attachment #1: Type: TEXT/PLAIN, Size: 1242 bytes --]

hi Matthias and gcc team,

i'm mailing you because i ran into an ICE with the debian unstable
package of gcc-3.5, specfically gcc-snapshot_20040620-1_i386.deb:

 gcc-snapshot (20040620-1) unstable; urgency=low

   * CVS 20040620, taken from HEAD (gcc-3.5).

  -- Matthias Klose <doko@debian.org>  Sun, 20 Jun 2004 00:39:00 -0700

i'm not sure whether that package is a stock compiler from
CVS or if additional patches are applied and if the ice can simply
be reproduced with a recent CVS version, so the report goes to
matthias and the gcc-bugs team.

i could reproduce the ice under two distinct x86 machines with
linux-2.4.20 and linux-2.6.6, libc-2.3.2 on both systems.

aparently, -ftracer triggers an ice in extract_insn with the
attached preprocessed source, the exact output is:

$ /usr/lib/gcc-snapshot/bin/gcc -ftracer -c ice1.c
sfiglue.c: In function `sfi_glue_gc_run':
sfiglue.c:880: error: unrecognizable insn:
(jump_insn 92 91 93 4 (return) -1 (nil)
    (nil))
sfiglue.c:880: internal compiler error: in extract_insn, at recog.c:2068
Please submit a full bug report,
with preprocessed source if appropriate.
See <URL:http://gcc.gnu.org/bugs.html> for instructions.


let me know if you need any further information.

---
ciaoTJ


[-- Attachment #2: ice1.c.bz2 --]
[-- Type: APPLICATION/octet-stream, Size: 33877 bytes --]

Re: ICE in extract_insn with gcc-20040620

[Andrew Pinski]

> aparently, -ftracer triggers an ice in extract_insn with the
> attached preprocessed source, the exact output is:
> 
> $ /usr/lib/gcc-snapshot/bin/gcc -ftracer -c ice1.c
> sfiglue.c: In function `sfi_glue_gc_run':
> sfiglue.c:880: error: unrecognizable insn:
> (jump_insn 92 91 93 4 (return) -1 (nil)
>     (nil))
> sfiglue.c:880: internal compiler error: in extract_insn, at recog.c:2068
> Please submit a full bug report,
> with preprocessed source if appropriate.
> See <URL:http://gcc.gnu.org/bugs.html> for instructions.
> 
> 
> let me know if you need any further information.

No but this should already be fixed in the latest GCC from the CVS, there
was a bug like this for many differnet problems.

Thanks,
Andrew Pinski

Re: ICE in extract_insn with gcc-20040620

[Falk Hueffner]

Tim Janik <timj@gtk.org> writes:

> i'm mailing you because i ran into an ICE with the debian unstable
> package of gcc-3.5, specfically gcc-snapshot_20040620-1_i386.deb:
>
> i'm not sure whether that package is a stock compiler from
> CVS or if additional patches are applied and if the ice can simply
> be reproduced with a recent CVS version, so the report goes to
> matthias and the gcc-bugs team.

You should just file a bug with the Debian bug tracking system then.
I can, however, reproduce this with an unpatched 3.5 20040627, so it's
not Debian-specific, and you might as well report it directly at
http://gcc.gnu.org/bugs.html.

-- 
	Falk

Re: ICE in extract_insn with gcc-20040620

[Matthias Klose]

Tim Janik writes:
> 
>  gcc-snapshot (20040620-1) unstable; urgency=low
> 
>    * CVS 20040620, taken from HEAD (gcc-3.5).
> 
>   -- Matthias Klose <doko@debian.org>  Sun, 20 Jun 2004 00:39:00 -0700
> 
> i'm not sure whether that package is a stock compiler from
> CVS or if additional patches are applied and if the ice can simply
> be reproduced with a recent CVS version, so the report goes to
> matthias and the gcc-bugs team.

just as a note, the gcc-snapshot's are unpatched upstream versions
(well, on hppa-linux java is enabled).

	Matthias