Re: delay execution in the kernel level code.

[Andrew Philip Bell]

On Mon, 12 Oct 1998, Andrea Arcangeli wrote:
>On Mon, 12 Oct 1998, Andrew Philip Bell wrote:
>
>>rdtsc()), then several context switches could (and would) throw off this check.
>
>rdtsc is the asm instruction available on >=Pentium arch. Other arch has
>their own way to read the timestamp counter.
>
>The rdtsc instruction is not linux specific and has nothing to do with
>scheduling. 

My point was that this call must override the scheduler.

>Using it you' ll stall the machine but it' s the most precise
>real-time delay you can achieve with an x86 at least.
>

OK, then according to this, rdtsc() does what the fellow was originally looking
for (at the cost of pre-empting every other process).   

>Andrea Arcangeli