>
>
> The FS block size is not a driver issue. The command you use (cat
> /dev/zero > some_file) is an extremally high stress for the caching policy
> in some Linux kernel versions. Based on the fiew information your report,
> I donnot see any obvious problems related to the SCSI sub-system.
I dunno if it's really SCSI related (that was the reason of my resend to
the linux-kernel list). But I can't believe that the system becomes so
extremly busy that no other programs could be started during creating
cd-images or similar things. On my former system with 1k blocks this was
no problem at all (it was a bit slow but usable). Even during copy of
large files from my (in comparisation to hard disks) slow MO drive with 2k
blocks to a hard disk with 1k blocks the system locks sometimes for one or
more seconds. For example it is wise to not use the MO drive during CD
burning at the moment. For quite some time I mailed the author of the
ext2-2k-sector patch for the 2.0.x kernels about this problem. He guessed
a bug in the kernels buffer cache that occure with blocks >1k that may be
fixed in 2.[12].x but it's still there.
> Did you look into the system log to see if some not usual messages
> were written during that IO load that seems to you not normal?
No messages. The load is not really the problem (since this is only a
number ;-) - it's only very high in comparisation to the same operations
on 1k block filesystems. But if I can't do other disk i/o (i.e. starting
programs, use programs that do some disk i/o) then i call this NOT normal.
For real multitasking this problem should be fixed.
ron
-- /\/\ Ronald Wahl -- rwahl@gmx.net /\/\ C S N /\/\ \/\/ ------------------------------- \/\/ ---------------------- \/\/ /\/\ http://www.tu-chemnitz.de/~row/ /\/\ network administration /\/\ \/\/ PGP key available \/\/ system management \/\/
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/