Re: atomic_ops(3) and sync

[David Holland <dholland-mips%netbsd.org@localhost>]

On Wed, Jun 16, 2010 at 10:54:14AM -0400, Paul Koning wrote:
 > > > the assembler inserted a nop in the delay slot; I guess membar_ops.S
 > > > needs ".set noreorder" at the top.
 > 
 > In general, all assembly code should use .set noreorder.

That's generally my opinion too.

 > The fact the assembler does this stuff at all, never mind the fact
 > that it does it by default, is a horrible design error on the part
 > of MIPS Co.

So was the delay slot itself...

-- 
David A. Holland
dholland%netbsd.org@localhost