Re: Huge performance issue with cciss driver on HP DL385 servers(2.6.13 -> 2.6.17)

From: Xavier Roche
Date: Tue Jul 11 2006 - 10:36:43 EST


Erik Mouw a écrit :
> AFAIK buffers for direct IO need to be *page* aligned. Use something
> like:
> I guess you got away with it cause your 512 byte alignment happened to
> align on a page, but you shouldn't count on that. However...

This example was actually just a minimal example - the running test is
correctly aligned.

> You should check the return value of pwrite().

Also checked in the "real" stress program.

Note that this problem is hard to reproduce, and might be related to
very strange I/O (and/or DMA access) operation ordering. The fact that
the program must be run just after a reboot (or the problem is not
easily reproductible) is something really fishy.
-
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/