Re: [PATCH] sched: fix exec_start/task_hot on migrated tasks

[Peter Zijlstra <peterz@infradead.org>]

[-- Attachment #1: Type: text/plain, Size: 1320 bytes --]

On Fri, May 16, 2014 at 07:27:32PM +0530, Preeti Murthy wrote:
> > 0 isn't strictly the right thing to do here, since the clock can wrap,
> > but being wrong every ~585 years isn't too big an issue for this.
> 
> I don't understand this. Will setting it to 0 not indicate beginning
> of ticking? 

In modular spaces there is no beginnings nor endings.

> So when you find out for how long the task has run, the
> difference would be larger than what would have been had you
> let exec_start be at its previous value of the old cpu's clock_task right?

Nope, see the modular space thing, once every ~585 years we'll wrap the
clock and 0 is within range of recently ran.

> Will not setting exec_start to the clock_task of the destination rq
> during migration be better? This would be the closest we could
> come to estimating the amount of time the task has run on this new
> cpu while deciding task_hot or not no?

Setting it to the exact clock_task of the destination rq would make it
hot on that rq, even though it hasn't yet ran there, so you'd have to do
something like: rq_clock_task(dst_rq) - sysctl_sched_migration_cost.

But seeing as how that is far more work, and all this is heuristics
anyhow and an extra fail term of 1/585 years is far below the current
fail rate, all is well.

[-- Attachment #2: Type: application/pgp-signature, Size: 836 bytes --]