This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: RFC: default insn length should be 1 but that breaks m68hc11and regresses mn10300.
- From: Stephane Carrez <stcarrez at nerim dot fr>
- To: Hans-Peter Nilsson <hans-peter dot nilsson at axis dot com>
- Cc: gcc-patches at gcc dot gnu dot org
- Date: Sun, 03 Apr 2005 21:30:02 +0200
- Subject: Re: RFC: default insn length should be 1 but that breaks m68hc11and regresses mn10300.
- References: <200504020908.j3298iwH019737@ignucius.se.axis.com>
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Hi!
Hans-Peter Nilsson wrote:
> [...]
>
> For m68hc11, it's a SEGV building newlib (the -m68hc12 -mshort
> multilib building newlib/libc/stdio/getdelim.c), caused by
> dereferencing NULL. It happens in m68hc11_reorg calling
> compute_bb_for_insn, where it seems BB_END of a block points to
> a deleted insn which is not in the chain linked from BB_HEAD
> (causing the loop to run into the NULL at the end of insns):
>
> [...]
The m68hc11 port does not use the length attribute at all.
I suspect the problem is present even without your patch.
>
> It seems m68hc11-elf is in a shoddy shape at the moment, given
> these baseline numbers (and you need to edit the baseboard file
> which references the nonexistent sim-valid.x script, replacing
> with I guess sim-valid-m68hc11.ld):
the script is part of newlib.
>
> === gcc Summary ===
>
> # of expected passes 22265
> # of unexpected failures 5618
> # of unexpected successes 1
> # of expected failures 74
> # of unresolved testcases 4754
> # of untested testcases 28
> # of unsupported tests 600
> /home/hp/cvs_areas/combined/m68hc11-regobj/gcc/xgcc version 4.1.0 20050329 (experimental)
>
> I'm assuming the port is to blame, but I won't insist on
Concerning these bad validation results:
- There are several fixes in the reload pass which are necessary.
- Several tests must be fixed as they are not valid on a 16-bit target
(http://m68hc11.serveftp.org/m68hc11_src.php).
On gcc 3.3.5 and with the fixes, I obtain these results,
with a site.exp that stresses more gcc options than the
default (on Sat Feb 12 22:47:45 2005).
=== gcc Summary ===
# of expected passes 200003
# of unexpected failures 112
# of unexpected successes 1
# of expected failures 363
# of unresolved testcases 226
# of unsupported tests 395
Stephane
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFCUEQ6yiaNLe8yxdARAudEAJ9VbLiZXNwJ1eGOPDIS0IewiuiBjgCcC789
73SIdTZKKnTpZ8bSUf5r8xY=
=BwsV
-----END PGP SIGNATURE-----