[comp.sys.sgi] Flushing the icache

moore%cdr.utah.edu@cs.utah.edu (Tim Moore) (03/16/90)

Thanx to all those who responded to my previous MIPS question (about
lw). It turns out that the bug I'm battling is caused by an
icache/dcache consistency problem. So now I ask, is there any way to
flush the icache on sgi 4d machines? If the answer is no, is there a
way to find out the size of the icache at runtime?

Thanks much,
Tim Moore                     moore@cs.utah.edu {bellcore,hplabs}!utah-cs!moore
"Ah, youth. Ah, statute of limitations."
		-John Waters

jeffd@norge.sgi.com (Jeff Doughty) (03/16/90)

> Thanx to all those who responded to my previous MIPS question (about
> lw). It turns out that the bug I'm battling is caused by an
> icache/dcache consistency problem. So now I ask, is there any way to
> flush the icache on sgi 4d machines? If the answer is no, is there a
> way to find out the size of the icache at runtime?

The following code will do the job:

#include <sys/sysmips.h>

...

	sysmips(FLUSH_CACHE);
...

				Jeff Doughty
				IRIX group