RE: [hpoj-devel] Sane owner?

[PASCHAL,DAVID (HP-Roseville,ex1) <dav...@hp...>]

Obi-Wan [mailto:bv...@in...] wrote:
> Tap tap tap... is this thing on?  I haven't received anything 
> since Tuesday.
Yes, "this thing" is "on", but I haven't been "on" the last several days.
:-) Sorry about that.

> So, David, are you the sane owner, too?  If so, I just 
> noticed a couple
> problems tonight.
> 
> First, with xsane, the size rulers along the top & left of the preview
> window don't seem to mesh.  When set to inches, the top seems OK, but
> the side is garbage.  The numbers are roughly 1.5 X too high.
I only "own" the changes I made to the SANE HP backend to support the SCL
OfficeJets.  A problem like this could either be an xsane bug, an HP backend
bug, or a scanner bug.  If you can figure out a way to reproduce it then
I'll look into it.  Since Oliver Rauch (the xsane developer) makes frequent
updates, you should make sure you're running the latest version, just in
case that has anything to do with it.  You could also try "xscanimage" and
see if it acts strangely too.

> I also had sane lock up on me while scanning.  I was getting a preview
> scan of the entire bed after putting in a new photo.  It had scanned
> almost the entire thing, when it just stopped.  XSane wouldn't refresh
> its windows.  There were no syslog messages & nothing to 
> stdout/stderr.
> After a minute or two, the scanner light returned to home, but still
> no refresh from xsane.  This happened just as cron launched a very
> large tar followed by gzip, but that may be coincidence.
If the lockup happened right when the cron job started, then I suspect it's
more than coincidence, because ptal-mlcd does its parallel-port signalling
in user mode and could theoretically be affected if it's starved for CPU
time.  If this happens again, it would be helpful if you could attach gdb to
the running(?) ptal-mlcd process and get a stack trace ("bt" I think) to see
what it's blocked on.  Note that when you attach gdb to a running process
like this, the process will be suspended until you exit gdb or issue the
"continue" (I think) command.

> When attempting to restart xsane, it brought up the "scanning for
> devices" window and hung.  I had to kill -9 xsane.  I stopped &
> restarted the ptal brothers.  I then noticed my PSC 500's LCD said
> "Scanner system failure.  Turn system off and on again," 
> which I did.  
> That all seemed to fix the problem.  Still no log entries other than
> the startup banners.  (Perhaps a log entry for every shutdown would
> also be in order.)
I've thought about that, but the problem is that there may be situations,
such as a "kill -9", where it's not possible to log a message.  ptal-mlcd
logs a message when it hits a "fatal error", so the only remaining
situations I can think of without a log message are where it segfaults or
you kill it.

> Upon restarting xsane, I see that the vertical ruler is now correct.
> Hmmm...
Perhaps the peripheral got confused and power cycling it got it back in a
good state.  Do you remember what you had been doing before this problem
happened?

David