The purpose of this e-mail is to see if someone knows about known
problems that may be the cause of my problems so that I don't have to
waste reams of paper trying to track down the problem.
Not that long ago I managed to pick up a used (~100,000 pages printed)
HP LaserJet 4si. For the price and usage it looked like a good deal.
Indeed the engine itself has worked flawlessly, but I seem to have some
corruption problems somewhere before things get to the engine.
Here is an overview of my current configuration:
1. Linux box
a) 2.4.1 kernel optimized for Pentium Pro/Cellery/Pentium II
b) Athlon CPU w/ L2 ECC enabled
c) PC133 ECC RAM w/ ECC enabled
d) ASUS K7V w/ 1.007 BIOS (reflashed by myself)
e) Running SAMBA and CUPS
2. Windows box
a) Win ME installed
b) Windows 98 print driver for the 4si (For some reason I couldn't
get Win ME to install its own driver so I pulled out my 98 disk.)
c) Talks to printer over the network through my Linux box.
3. HP LaserJet 4si
a) 6 MB of RAM
b) Connected to Linux box with a relatively new IEEE 1284 compliant
parallel cable.
When I first noticed a problem I was running the 2.4.1 kernel with
Athlon optimizations and I was printing from my Windows box. When I
printed the operators manual for the printer from Windows, I found some
garbled characters on three different pages out of the 123 page manual.
I rebooted my Linux box into the default 2.2.17 kernel and I pulled the
token ring adapter out of the printer and tried to print again. The
first time around Windows glitched up and reported that it was out of
printer spooler memory. I checked both the Linux box and my Windows box
and they both seemed to have copious amounts of spool space. When I
tried once more the whole manual came out with no errors.
On further search away from the printer I noticed that a number of weird
things where going on with my Linux box at the time. I then read
somewhere that there where still some problems with gcc when it came to
optimizing for the Athlon. When I recompiled with different
optimizations, things started running much smoother. I am getting a
little spooked once again because on a recent print out from Netscape
under Linux I noticed some garbled characters on one page out of a multi
page print out. Also on another recent printout through Netscape under
Linux of a text file on the net one page suddenly cut out 3/4 of the way
down and continued on the next page. When I tried print again the exact
same thing happened again. My simple reasoning figures that that 6 MB
is enough to hold an entire page, plus a text file could be processed by
the printer at or greater than engine speed even through a parallel
port. I thought maybe I should turn on page protection under Linux, but
I seen an option for such a thing in the CUPS manual. (I also don't see
an option to reverse page order which I like to use under windows to do
manual duplexing. To think about it I just printed out the CUPS
software users manual under windows using this method without any
errors.)
