Re: [PATCH] spinlock metering (2.3.42)

[Alexander Viro <viro@xxxxxxxxxxxx>]

On Wed, 9 Feb 2000, John Hawkes wrote:

> The kernel_flag is still the biggest culprit in 2.3.28 and it accounts
> for
> almost all of the witnessed wait-time cycles (157msec of the 160msec
> total).  A call in ext2_get_block() accounts for 45msec of this 157msec.
> The largest hold-time occurs in do_close() -- that 15msec mentioned

Almost definitely dput() work.

> earlier -- with a mean hold-time of 59usec (microsec).  Other long holds
> are done by sys_unlink() (10msec max), sys_open() (5.5msec), and
        Ditto.                                  Interesting...
> sys_execvs() (2msec).

Most likely we are dealing with dcache stuff. Sorry, it will have to wait.
Sigh...