- From: Frank Mori Hess <fmhess_at_users.sourceforge.net>
- Date: Thu, 1 May 2003 13:45:19 -0500
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
On Wednesday 30 April 2003 05:24 pm, Herman Bruyninckx wrote:
> I've been working some more days on extending the Comedi
> documentation. Basically I added a section on how to write a new
> device driver. Please, give feedback on the draft that you can find
> here: <http://people.mech.kuleuven.ac.be/~bruyninc/comedi/html>.
> (I notice that our server is currently not responding...)
>
> Herman
I have some feedback, listed in no particular order:
First off, good work! Now for the criticisms:
I don't think it needs a revision history section. We already have the cvs
logs, etc. A simple version number would be sufficient (set equal to
comedilib's version).
Where it says: 'Comedi has once and for all solved lots of boring but
indispensable infrastructural things, such as: PCI bus configuration...' it
should not include PCI bus configuration. The individual drivers need to
call the various kernel pci related functions directly, comedi doesn't help
much here in general (although the mite driver does some stuff that's only
for boards with mite chips).
This statement: 'When writing a device driver, the programmer should make use
of the API of the kernel-space kcomedilib' is misleading. The drivers
(except for a couple special exceptions) don't make use of kcomedilib
functions. kcomedilib is for people writing RT kernel modules. Rather,
comedidev.h is the place where driver-specific stuff is declared.
The description of the linux/comedi.h include file is missing.
Where it says 'The mydriver_attach() function must... announce that the
hardware driver has begun initialization by a printk("comedi%d: driver:
",minor);' should be removed. If we want to enforce a particular message to
be displayed on attach, we might as well just do it in the core.
where it says 'Drivers should report errors and warnings via a printk line
that starts with comedi%d: driver_name: where %d is the minor number of the
device.' You should just say to use comedi_error(), which is what is
actually generating that format of message.
- --
Frank
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.6 (GNU/Linux)
Comment: For info see http://www.gnupg.org
iD8DBQE+sWs/5vihyNWuA4URAl1fAJ47FArQNo98juzQQ49a26n43PI3tgCgjCQE
U1itLzJdxehZOUWJBZF/BM0=
=jm/Z
-----END PGP SIGNATURE-----
Received on 2003-05-01Z17:45:19