http://www.riscos.info/bugzilla3/show_bug.cgi?id=243
--- Comment #3 from Lee <leenoar@sky.com> 2012-09-02 06:51:22 PDT ---
(In reply to comment #2)
> (In reply to comment #1)
> > This is caused by tail call optimisations clashing with the stack extension
> > routines in UnixLib. It should be fixed with r5990 and has already been fixed
> > at trunk.
>
> Is there somewhere to get a binary with r5990 in from?
> Alternatively do any of the earlier releases generate ARMv7 compatible
> loads/stores but simultaneously not have this bug?
I've placed a copy of the cc1 binary for you to try here:
<http://leenoar.yolasite.com/resources/gcc/cc1.zip>
This seems to be the main culprit and IIRC, the only binary in GCCSDK that I
have seen fail due to this (obviously cc1plus is a possibility too). It's
possible that TCOs in libunixlib may trigger the problem, but I think it's
unlikely.
It would be useful to know that this does in fact fix your problem.
Its location is !GCC/libexec/gcc/arm-unknown-riscos/4.1.2/cc1,e1f
As for previous versions, well I was hoping that this problem didn't go back
further that 4.6, but unfortunately, it looks like it may have been present
from the beginning. Perhaps the increasing complexity of GCC has made the rare
runtime conditions under which the problem occurs more likely.
--
Configure bugmail: http://www.riscos.info/bugzilla3/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are watching all bug changes.
_______________________________________________
GCCSDK mailing list gcc@gccsdk.riscos.info
Bugzilla: http://www.riscos.info/bugzilla/index.cgi
List Info: http://www.riscos.info/mailman/listinfo/gcc
Main Page: http://www.riscos.info/index.php/GCCSDK
No comments:
Post a Comment