lkml.org 
[lkml]   [2002]   [Aug]   [4]   [last100]   RSS Feed
Views: [more markup]  [less markup]  [headers]  [forward] 
 
Messages in this thread
/
DateSun, 4 Aug 2002 08:14:55 +0200 (CEST)
FromRoland Kuhn <>
SubjectRe: large file IO starving ls -l
On Sat, 3 Aug 2002, Linus Torvalds wrote:

> In article <Pine.LNX.4.44.0208032253260.23040-100000@pc40.e18.physik.tu-muenchen.de>,
> Roland Kuhn  <rkuhn@e18.physik.tu-muenchen.de> wrote:
> >
> >Now the question is: who keeps ls from returning? The command never hits 
> >the disk (reads in above histogram do not increase), but stays for many 
> >seconds (up to one minute) in state D.
> 
> ext2 used to have similar issues with the superblock lock - where things
> like block allocation (very much in the write path) would grab the
> superblock lock, and completely destroy interactive feel even for
> processes that didn't need to do IO, because the superblock lock was
> often grabbed even if the data was actually cached (sb locking needed
> just to _look_up_ the physical block so that you could look up the
> cached data in the buffer cache). 
> 
> Al Viro largely fixed in for ext2, which now uses lock_super() a lot
> less. But a lot of filesystems are based on the old ext2 locking, and
> may have inherited some of the worst parts..
> 
Thanks for the hint, I will try to have a look at the reiserfs code. Could 
you give me a hint where this lock usually is taken?

Ciao,
					Roland

+---------------------------+-------------------------+
|    TU Muenchen            |                         |
|    Physik-Department E18  |  Raum    3558           |
|    James-Franck-Str.      |  Telefon 089/289-12592  |
|    85747 Garching         |                         |
+---------------------------+-------------------------+
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/

\
 
 \ /
  Last update: 2005-03-22 12:27    [from the cache]
©2003-2008