headers
Jason Dravet | 2 Jan 2006 16:39
Picon
Favicon

[RFC]: add sysfs support to parport_pc, v3

Here is a new patch to parport_pc that adds sysfs and thus udev support to 
parport_pc.  I fixed my earilier problem of the kernel oops (I forgot the 
class_destory) and I can insmod and rmmod this module all day long with no 
side effects.  I do have one question why do both lp and parport nodes have 
to be created?

What do you think of this patch?  What would be the next step to get this 
into the kernel?

Thanks,
Jason Dravet

signed-off-by: Jason Dravet <dravet <at> hotmail.com>

--- /usr/src/linux-2.6.14/drivers/parport/parport_pc.c.orig	2005-12-30 
13:52:48.000000000 -0600
+++ /usr/src/linux-2.6.14/drivers/parport/parport_pc.c	2006-01-01 
11:29:05.000000000 -0600
 <at>  <at>  -14,6 +14,7  <at>  <at> 
  * More PCI support now conditional on CONFIG_PCI, 03/2001, Paul G.
  * Various hacks, Fred Barnes, 04/2001
  * Updated probing logic - Adam Belay <ambx1 <at> neo.rr.com>
+ * Added sysfs and udev - Jason Dravet <dravet <at> hotmail.com>
  */

/* This driver should work with any hardware that is broadly compatible
 <at>  <at>  -55,6 +56,7  <at>  <at> 
#include <linux/pci.h>
#include <linux/pnp.h>
#include <linux/sysctl.h>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Marko Kohtala | 2 Jan 2006 20:50
Picon

Re: [Linux-parport] [RFC]: add sysfs support to parport_pc, v3

2006/1/2, Jason Dravet <dravet <at> hotmail.com>:
> Here is a new patch to parport_pc that adds sysfs and thus udev support to
> parport_pc.  I fixed my earilier problem of the kernel oops (I forgot the
> class_destory) and I can insmod and rmmod this module all day long with no
> side effects.  I do have one question why do both lp and parport nodes have
> to be created?
>
> What do you think of this patch?  What would be the next step to get this
> into the kernel?

If your problem is that you do not get /dev/lp0 created automatically,
I think a better way to get that solved is to find out a way that the
module lp is loaded automatically when a printer is found on the
parport.

Maybe you should look into drivers/parport/probe.c and have it do
request_module("lp") when it finds a printer attached to the parport.

Some paride (or SCSI) devices could also be probed for using a
pre-IEEE1284 daisy chain command that gives a 16 bit device ID, and
proper drivers loaded with some module aliases including the device
ID. If something unidentifiable is found on the parport then maybe the
ppdev could be loaded so that userspace drivers will have an interface
to work with.

I fear that some people would find this too smart to be useful. I'd
expect them to request at least a module option to parport to disable
it.

Most people are happy with just "lp" in /etc/modules.
(Continue reading)

Permalink | Reply | 0 comments |
headers
Greg KH | 4 Jan 2006 02:08
Gravatar

Re: [RFC]: add sysfs support to parport_pc, v3

On Mon, Jan 02, 2006 at 09:39:50AM -0600, Jason Dravet wrote:
> Here is a new patch to parport_pc that adds sysfs and thus udev support to 
> parport_pc.  I fixed my earilier problem of the kernel oops (I forgot the 
> class_destory) and I can insmod and rmmod this module all day long with no 
> side effects.  I do have one question why do both lp and parport nodes have 
> to be created?
> 
> What do you think of this patch?  What would be the next step to get this 
> into the kernel?
> 
> Thanks,
> Jason Dravet
> 
> signed-off-by: Jason Dravet <dravet <at> hotmail.com>

"Signed-off-by:"

> 
> --- /usr/src/linux-2.6.14/drivers/parport/parport_pc.c.orig	2005-12-30 
> 13:52:48.000000000 -0600
> +++ /usr/src/linux-2.6.14/drivers/parport/parport_pc.c	2006-01-01 
> 11:29:05.000000000 -0600

Line wrapped so it can't be applied :(

>  <at>  <at>  -14,6 +14,7  <at>  <at> 
>  * More PCI support now conditional on CONFIG_PCI, 03/2001, Paul G.
>  * Various hacks, Fred Barnes, 04/2001
>  * Updated probing logic - Adam Belay <ambx1 <at> neo.rr.com>
> + * Added sysfs and udev - Jason Dravet <dravet <at> hotmail.com>
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jason Dravet | 4 Jan 2006 04:24
Picon
Favicon

Re: [RFC]: add sysfs support to parport_pc, v3

>From: Greg KH <greg <at> kroah.com>
> > +++ /usr/src/linux-2.6.14/drivers/parport/parport_pc.c	2006-01-01
> > 11:29:05.000000000 -0600
>
>Line wrapped so it can't be applied :(
>
> > + * Added sysfs and udev - Jason Dravet <dravet <at> hotmail.com>
> >  */
>
>Doesn't belong here, this goes in the change log.
My mistake.

> > +#include <linux/sysfs.h>
>
>Your email client also ate the leading spaces :(
My email client is hotmail.  I have come to realize how much hotmail sucks 
for email.  It eats the leading spaces, converts tabs to spaces, and line 
wraps.  I will work on getting a new mail account.

> > +	class_device_create(parallel_class, NULL, MKDEV(6, countports),
> > NULL, "lp%i", countports);
> > +	class_device_create(parallel_class, NULL, MKDEV(99, countports),
> > NULL, "parport%i", countports);
> > +	countports++;
>
>What does the 6 and 99 come from?  Aren't these #defined in a header file 
>somewhere?
Good question.  The answer is I have no idea.  I booted Fedora Core and I 
did a ls -l /dev and wrote down the specs for lp0 and parport0.  lp0 looked 
like crw-rw----  root   lp   6,  0  date  lp0 so that is what I used.  After
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jan Engelhardt | 4 Jan 2006 10:14
Picon

Re: [RFC]: add sysfs support to parport_pc, v3

>> somewhere?
> Good question.  The answer is I have no idea.  I booted Fedora Core and I did a
> ls -l /dev and wrote down the specs for lp0 and parport0.  lp0 looked like
> crw-rw----  root   lp   6,  0  date  lp0 so that is what I used.  After reading
> Linux Device Drivers I found out that 6 is the major number and 0 is the minor
> number.  The goal was to make sure that the nodes generated by my patch were
> the same as the nodes generated before my patch so I used those numbers.
> Should I have used different major numbers?

You should have at least used the constants from <linux/major.h>

Jan Engelhardt
--

-- 
| Alphagate Systems, http://alphagate.hopto.org/
| jengelh's site, http://jengelh.hopto.org/
Permalink | Reply | 0 comments |
headers
Andrew Morton | 4 Jan 2006 23:31

Re: [RFC]: add sysfs support to parport_pc, v3

"Jason Dravet" <dravet <at> hotmail.com> wrote:
>
> >From: Greg KH <greg <at> kroah.com>
> > > +++ /usr/src/linux-2.6.14/drivers/parport/parport_pc.c	2006-01-01
> > > 11:29:05.000000000 -0600
> >
> >Line wrapped so it can't be applied :(
> >
> > > + * Added sysfs and udev - Jason Dravet <dravet <at> hotmail.com>
> > >  */
> >
> >Doesn't belong here, this goes in the change log.
> My mistake.
> 
> > > +#include <linux/sysfs.h>
> >
> >Your email client also ate the leading spaces :(
> My email client is hotmail.  I have come to realize how much hotmail sucks 
> for email.  It eats the leading spaces, converts tabs to spaces, and line 
> wraps.  I will work on getting a new mail account.
> 
> > > +	class_device_create(parallel_class, NULL, MKDEV(6, countports),
> > > NULL, "lp%i", countports);
> > > +	class_device_create(parallel_class, NULL, MKDEV(99, countports),
> > > NULL, "parport%i", countports);
> > > +	countports++;
> >
> >What does the 6 and 99 come from?  Aren't these #defined in a header file 
> >somewhere?
> Good question.  The answer is I have no idea.  I booted Fedora Core and I
(Continue reading)

Permalink | Reply | 0 comments |
headers
Marko Kohtala | 5 Jan 2006 17:17
Picon

Re: [RFC]: add sysfs support to parport_pc, v3

2006/1/5, Andrew Morton <akpm <at> osdl.org>:
> "Jason Dravet" <dravet <at> hotmail.com> wrote:
> > >From: Greg KH <greg <at> kroah.com>
> > > > + * Added sysfs and udev - Jason Dravet <dravet <at> hotmail.com>
> > > >  */
>
> 6 is OK - it's LP_MAJOR.
>
> However 99 is JSFD_MAJOR, used by drivers/sbus/char/jsflash.c.  And yet my
> /dev/parport0 is also 99:0 (RH 7.3 and RH FC1).  I've no idea how that came
> about??
>
> bix:/home/akpm> grep parport /etc/makedev.d/*
> /etc/makedev.d/generic:a generic parport
> /etc/makedev.d/linux-2.4.x:c $PRINTER              99   0  1   8 parport%d

JSFD is a block device so tha majors are ok. I'm not just sure if the
PP_MAJOR from linux/ppdev.h should be moved to major.h.

The patch by Jason however is not ok. He had another problem and this
is not the fix. What he tries to do is already in lp and ppdev, where
I think they belong.

There is also something weird: why does RedHat create these nodes in
/dev when udev already does that.
Permalink | Reply | 0 comments |
headers
Kay Sievers | 5 Jan 2006 22:25

Re: [RFC]: add sysfs support to parport_pc, v3

On Thu, Jan 05, 2006 at 06:17:51PM +0200, Marko Kohtala wrote:
> 2006/1/5, Andrew Morton <akpm <at> osdl.org>:
> > "Jason Dravet" <dravet <at> hotmail.com> wrote:
> > > >From: Greg KH <greg <at> kroah.com>
> > > > > + * Added sysfs and udev - Jason Dravet <dravet <at> hotmail.com>
> > > > >  */
> >
> > 6 is OK - it's LP_MAJOR.
> >
> > However 99 is JSFD_MAJOR, used by drivers/sbus/char/jsflash.c.  And yet my
> > /dev/parport0 is also 99:0 (RH 7.3 and RH FC1).  I've no idea how that came
> > about??
> >
> > bix:/home/akpm> grep parport /etc/makedev.d/*
> > /etc/makedev.d/generic:a generic parport
> > /etc/makedev.d/linux-2.4.x:c $PRINTER              99   0  1   8 parport%d
> 
> JSFD is a block device so tha majors are ok. I'm not just sure if the
> PP_MAJOR from linux/ppdev.h should be moved to major.h.
> 
> The patch by Jason however is not ok. He had another problem and this
> is not the fix. What he tries to do is already in lp and ppdev, where
> I think they belong.

Yeah, I was wondering about it too.

> There is also something weird: why does RedHat create these nodes in
> /dev when udev already does that.

Probably cause they want to be safe if nothing from their sysconfig loads
(Continue reading)

Permalink | Reply | 0 comments |
headers
marco | 6 Jan 2006 15:49
Picon

Hp 5100c not recognized

Hello,

(first, sorry for my bad english)

I tried in this week to get my parallel port scanner (HP 5100c) working on my system.
I am working on a Pentium III PC with ubuntu Breezy 5.10 and a 2.6.12 kernel.
I've followed this post: 

http://lists.infradead.org/pipermail/linux-parport/2005-October/000355.html

and everything works properly:
i sucessfully installed the modules with modprobe (with insmod i've had many problems):

modprobe scsi_mod
modprobe sg
modprobe parport
modprobe parport_pc
modprobe ppscsi

with this result (see with dmesg):

[4294970.382000] SCSI subsystem initialized
[4294970.434000] ppSCSI 0.92 (0.92) installed

but when i try to loading the last modules the shell response is:

modprobe onscsi
FATAL: Error inserting onscsi (/lib/modules/2.6.12-10-686/kernel/drivers/parport/onscsi.ko): No
such device
(Continue reading)

Permalink | Reply | 6 comments |
headers
J Fairbairn | 6 Jan 2006 17:09
X-Face
Favicon

Re: Hp 5100c not recognized

On 2006-01-06 at 15:49+0100 marco wrote:
> Hello,
> 
> (first, sorry for my bad english)
> 
> I tried in this week to get my parallel port scanner (HP
> 5100c) working on my system.  I am working on a Pentium
> III PC with ubuntu Breezy 5.10 and a 2.6.12 kernel.  I've
> followed this post:
> 
> http://lists.infradead.org/pipermail/linux-parport/2005-October/000355.html
> [...]
> but when i try to loading the last modules the shell response is:
> 
> modprobe onscsi

modprobe epst 

is likely to work better.

Cheers,

  Jón

--

-- 
Jón Fairbairn                              Jon.Fairbairn at cl.cam.ac.uk
Permalink | Reply | 5 comments |

Gmane