Subject: Re: ksh lossage on sparc64
To: matthew green <>
From: Martin Husemann <>
List: tech-toolchain
Date: 11/20/2000 08:05:34
> if i compile ksh/jobs.c with -O0, ksh works.  yay compiler bugs!?!?

That's to be expected. The (volatile int) cast in the source is wrong
("implementation defined"), avoiding optimization makes all register
reload problems go away.

The question is why the (IMHO) right fix I tried (without that sprintf
block) does not work - thats a compiler bug.