Re: udevd is killing file write performance.

[Benjamin LaHaise]

On Wed, Feb 22, 2006 at 03:12:23PM -0800, Andrew Morton wrote:
> It's not a terribly bad hack - it's just poor-man's hashing, and it's
> reasonably well-suited to the sorts of machines and workloads which we
> expect will hit this problem.

The dnotify/inotify wakeups are a problem, namely because the implementation 
is braindead: it makes the wrong part of the interface fast (setting up 
notify entries) at the expense of making the rest of the kernel slow (adding 
locks to read()/write()).  read() and write() are incredibly hot paths in 
the kernel and should be optimized at the expense of dnotify and inotify, 
which are uncommon operations.

		-ben
-- 
"Ladies and gentlemen, I'm sorry to interrupt, but the police are here 
and they've asked us to stop the party."  Don't Email: <dont@xxxxxxxxx>.
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/