When running iotop on Lucid
Andy Whitcroft wrote:
> On Fri, Dec 04, 2009 at 01:10:42PM -0700, Tim Gardner wrote:
>> CONFIG_TASK_DELAY_ACCT not enabled in kernel, cannot determine SWAPIN
>> and IO %
>> Suppose we should enable it?
> Sounds fine in principle. I am a little concerned that it adds a
> spinlock on and update of a task specific structure for every IO by the
> looks of it.
> I'll enable this for the first kernel after alpha-1 and perhaps you can
> compare the performance of the two kernels.
Guess I should have looked at it closer. In retrospect, and upon closer
inspection, lets leave this feature disabled.
Tim Gardner email@example.com
kernel-team mailing list