For quite a while now I've been having some annoying
problems with my old parallel port Zip 100 under 2.2.x
(x>12) kernels. I'm using the ppa driver as a module, the BIOS
is in EPP mode and the driver is using EPP 32-bit mode (same thing
happens when in ECP mode and PS/2 mode). The machine is a pentium II
running Debian 2.1/2.2.
When writing a large amount of data to the disk (ext2)
various applications including netscape, window managers, du -ks
commands,..... will lockup up for extended periods of
time. In some cases they do not wake up until the
zip drive has finished. At the same time the data rate to
the zip disk seems to drop to virtually nothing, maybe taking
10 to 30 mins to write 80 MB for example. This makes the machine
very difficult to use while writing to the drive.
Reading from the drive is generally OK. Writing a few Mb to
maybe 10 - 20 MB is OK. It only happens when writing
30-90 Mb in one go. I suspect it has something to do with
the buffering. The machine has 512 MB for RAM and if writing
say a 80 Mb directory to the zip it is not uncommon for most
or all of the data to be written into the buffer very
quickly (a few seconds or so). If this happens the problem
is usually worse and may lock up applications for over 30 mins.
until the buffers are written to the disk. It also seems to be
worse when writing a few large files rather than a lot of small
ones. Once the writes are complete everything frees up.
The data is always written reliably.
I used to use the same zip drive on a 100 MHz 486 with 32 Mb
of RAM and a late 2.0.x kernel and it used to work perfectly
- usually write a full disk in about 5-8 min without bringing
the machine to its knees.
Any ideas would be gratefully accepted.
Dr Ian David Flintoft Email: firstname.lastname@example.org Applied Electromagnetics Group Tel: +44 (0)1904 432391 Department of Electronics Fax: +44 (0)1904 433224 University of York Heslington YORK, UK YO10 5DD Web: http://www-users.york.ac.uk/~idf1
-- To unsubscribe, send mail to: email@example.com -- -- with the single word "unsubscribe" in the body of the message. --
This archive was generated by hypermail 2b29 : Thu Oct 26 2000 - 12:00:47 EDT