Re: Use of yield() in the kernel

[Ingo Molnar]

On Tue, 15 Oct 2002, Duncan Sands wrote:

> Maybe it is worth auditing the kernel source files using yield()?

most definitely.

> Here is the list of files using yield(), excluding non-i386 arch
> specific files:
> 
[...]
> mm/oom_kill.c

this one i think is OK.

> kernel/sched.c (in migration_call)

this is okay as well.

> kernel/softirq.c

these are okay too - both are nonperformance bits.

> arch/i386/mm/fault.c

okay as well, it's a last-ditch effort to not kill init, so yielding is
the right thing to do here.

the others i think should be fixed. (but there might be exceptions.)

	Ingo