hpoj-devel

[hpoj-devel] SMP lockup

[Dr. J. W. G. <dr...@dr...>]

I have seen old reports of SMP issues but nothing recently, so for the 
record for a PSC 750 running WhiteBox Linux (kernel 
2.4.21-15.0.4.ELsmp) on a SMP machine, after a tarball install of  
hpoj-0.91 I find that the computer locks up (i.e. frozen screen, no 
response to keyboard/mouse, requires hitting "reset" button to re-boot) 
when I try to use xsane.  If I boot in non-SMP mode then xsane works 
normally.

I was hoping that I could just re-boot into non-SMP modes when I  need 
to scan, but find that with pt-al active in SMP my machine locks up in 
the middle of the night!  Apparently this is related to the following 
messages I found in /var/log/messages (but don't understand):
   04:02:03 localhost kernel: printer.c: usblp0: nonzero read/write bulk 
status received: -2
   04:02:03 localhost kernel: printer.c: usblp0: error -2 reading from 
printer

So I've had to remove start-up of pt-al at boot time (i.e. from 
/etc/init.d) and the middle-of-the-night lock-ups then cease.  I assume 
that if I need to use the scanner I can reboot into non_SMP mode and 
then start pt-al manually but have not tried this (after wasting 7 
hours of my life trying to get things working, I reached my frustration 
limit.)

Printing works find throughout all this in SMP and non-SMP, which I 
believe occurs through CUPS, which  is started prior to pt-al in the 
boot sequence.

-- 
Dr. John W. (Jack) Glendening   Meteorologist

Re: [hpoj-devel] SMP lockup

[James B. H. <jh...@vi...>]

Hi!

> I have seen old reports of SMP issues but nothing recently, so for the 
> record for a PSC 750 running WhiteBox Linux (kernel 
> 2.4.21-15.0.4.ELsmp) on a SMP machine, after a tarball install of  
> hpoj-0.91 I find that the computer locks up (i.e. frozen screen, no 
> response to keyboard/mouse, requires hitting "reset" button to re-boot) 
> when I try to use xsane.  If I boot in non-SMP mode then xsane works 
> normally.

The SMP-with-USB bug was an issue until the last few 2.4.X kernels
(believe the latest is 2.4.28).  I don't know exactly which 2.4.2X version
corrected it, but I do have further notes that is was corrected in
(of course) the later 2.5.X development kernels, and thus is gone by
2.6.0.  One thing that was not done was extensive testing among these
components (that is, 2.6.X kernels, hpoj-0.90/0.91, and the libusb
releases of that time and moving forward).  However, the bug should
be gone.

If your intent was to document the specific offending items, then I
apologize for the noise, as I'm answering a question not asked.  If
you're interested in a solution approach, recommend an upgrade until
at least one of the last couple 2.4.X kernels to give it a shot, and
if that doesn't get it, then into 2.6.0.

Another approach, if that's not a good trade, would be to use a parallel
port adapter for the printer, assuming it will support one.  I've done
that with my OJ 7130, and all is fine.  (The bug never existed with
hpoj using parallel; just USB).

Hope this is of some use!

jbh