I'm new to the list, hope my question is not a FAQ.
The problem is with the nanosleep call, who is very very long. On
laita:~# time sleep 1
That's seem normal to me. But on the uml host:
douron:~# time sleep 1
it is very very long! Problem occurs whith all nanosleep system call, ie
vi call nanosleep to make the screen blink.
Despite of this problem, performance seems normal, but each time
nanosleep is called, there's a performance probleme.
I'm running kernel 2.6.17 patched with skas 3-1, on a debian sid.
Thanks for help
From: Jeff Dike <jdike@ad...> - 2006-11-01 00:55:06
On Tue, Oct 31, 2006 at 09:32:02PM +0100, Olivier Archer wrote:
> That's seem normal to me. But on the uml host:
> douron:~# time sleep 1
> real 1m22.724s
> user 0m0.000s
> sys 0m0.000s
There was a fix for this, I think after 2.6.17.