This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
Re: regscan x86 abort
- To: Richard Henderson <rth at cygnus dot com>
- Subject: Re: regscan x86 abort
- From: Jeffrey A Law <law at cygnus dot com>
- Date: Tue, 08 Sep 1998 00:50:43 -0600
- cc: egcs-patches at cygnus dot com
- Reply-To: law at cygnus dot com
In message <19980905225433.A25953@dot.cygnus.com>you write:
>
> --huq684BweRXVnRxX
> Content-Type: text/plain; charset=us-ascii
>
> The attached function causes a segv in reg_scan_mark_refs because
> we traverse a REG_WAS_0 note, and find a nil in the INSN stored
> there. Scanning the attached insn seems wrong.
>
>
> r~
>
> --huq684BweRXVnRxX
> Content-Type: text/plain; charset=us-ascii
> Content-Disposition: attachment; filename=z
>
> * regclass.c (reg_scan_mark_refs): Don't traverse attached insns.
I'm not sure this is right.
If you look down in reg_scan_mark_refs, it handles INSN_LISTs explicitly.
I would be worried that there are cases where we should be looking
at INSNs.
If no such cases exist, then we should also ignore JUMP_INSNs,
CALL_INSNs and resolve what to do with INSN_LISTs.
Did the nil occur in the insn pointed to by the REG_WAS_0 note, or
was there a nil in the actual note itself?
jeff