Re: stat benchmark

Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]
From: Zach Brown
Date: Monday, April 28, 2008 - 10:46 am

> I guess it would be difficult to get close to the optimal disk schedule by

It wouldn't even get to the I/O scheduler.   The VFS stat path (see
real_lookup()) is synchronous and serialized.  Each stat will hold the
i_mutex of the parent directory while it waits for the file system's
lookup method to populate the inode from disk.

- z
--
Previous message: [thread] [date] [author]
Next message: [thread] [date] [author]

Messages in current thread:
stat benchmark, Soeren Sandmann, (Thu Apr 24, 1:59 pm)
Re: stat benchmark, Carl Henrik Lunde, (Thu Apr 24, 2:42 pm)
Re: stat benchmark, Jan Engelhardt, (Thu Apr 24, 2:44 pm)
Re: stat benchmark, Justin Banks, (Thu Apr 24, 7:27 pm)
Re: stat benchmark, Christoph Hellwig, (Fri Apr 25, 12:01 am)
Re: stat benchmark, Theodore Tso, (Fri Apr 25, 12:48 pm)
Re: stat benchmark, Carl Henrik Lunde, (Sun Apr 27, 3:40 pm)
Re: stat benchmark, Soeren Sandmann, (Sun Apr 27, 4:29 pm)
Re: stat benchmark, Carl Henrik Lunde, (Sun Apr 27, 5:13 pm)
Re: stat benchmark, Theodore Tso, (Sun Apr 27, 7:10 pm)
Re: stat benchmark, Ulrich Drepper, (Sun Apr 27, 9:43 pm)
Re: stat benchmark, Theodore Tso, (Mon Apr 28, 4:53 am)
Re: stat benchmark, Avi Kivity, (Mon Apr 28, 4:59 am)
Re: stat benchmark, Theodore Tso, (Mon Apr 28, 6:31 am)
Re: stat benchmark, J. Bruce Fields, (Mon Apr 28, 9:18 am)
Re: stat benchmark, Zach Brown, (Mon Apr 28, 10:46 am)
Re: stat benchmark, Alexander Larsson, (Mon Apr 28, 12:41 pm)