|
Message
From: Heiko Panther <heiko.panther@w...>
Date: Wed, 12 Nov 2003 10:52:00 +0100
Subject: [openrisc] Re: More information on optimisation problem
I came up with a self-contained demo. You'll probably have to change my
platform specific print command to compile it. I also included asm dumps
of the problematic routine.
There is one "tight" for-loop in _cvt, when it is removed, the bug
appears. When it is there, no problems even with optimization on.
BTW, the cvt code is from the ecos diagnostic print routines.
Heiko
o2bug.tgz
|
 |
Browse
