Re: lp.c in 2.0.x

Wolfgang Walter (wolfgang.walter@stusta.mhn.de)
Sat, 25 Apr 1998 16:38:45 +0200


On Sat, Apr 25, 1998 at 03:21:57PM +0200, Andrea Arcangeli wrote:
> On Sat, 25 Apr 1998, Wolfgang Walter wrote:
>
> >Hi !
> >
> >I have a problem with lp in 2.0.x when printing with plp spooler system on
> >faster machines to a FS-600 using IRQ's (normal and epp/ecp modes). I don't
> >have this problems when I use a pentium 120 or when I use polling mode. The
> >problem is that only parts get printed. The following patch fixes the problem,
> >so I'm not shure why it exactly works. My real intention for this patch was to
> >observe the interrupts more easily. Can it be that on faster machines plp
> >can close the line bevor all is printed?
> >
> >If anyone has similar problems, does this help, too?
>
> Yes, some people reported similar problems with the 2.1.98 drivers and I
> fixed the irq handling now (yesterday!!). Expect the fix in the official
> kernel soon (I trust Linus will agree with the patch of course). Now the
> patch is under testing (I just have some winner reports but I' d like to
> wait a bit more).
>
> >Here's the patch:
>
> It force lp to don' t care about the interrupt so it' s the same of
> working in polling, take a look at /proc/interrupt and you will see.

Not really, as I use tunelp to switch interrupt on. And /proc/interrupts
show that the interrupt is installed and works.

>
> If you think that having lp interrupt driven working in all kind of
> hardware is relevant for 2.0.x I could backport the patch for 2.0.34preX.
> My patch also improve performance a lot, lp load my P266 CPU of 0.8%
> during the printing! I' d like to be sure my interrupt fix will be applyed
> to the official kernel before starting to backporting a patch with 2.2
> soon.

Yes, I would really appreciate a backport. I think printing is one of the
basic features, and I like the idea that printing does not consume a lot
of cpu time.

>
> Andrea[s] Arcangeli
>
>

Wolfgang Walter

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu