Re: linux kernel module programming guide

[John Levon <levon@movementarian.org>]

On Tue, Apr 01, 2003 at 11:48:03AM -0800, Peter Jay Salzman wrote:

> a new version of the linux kernel module programming guide (lkmpg) was
> submitted to tldp and is now available.

quick browsing ...

> enable CONFIG_MODVERSION in

it's CONFIG_MODVERSIONS

> One exception is chapter (fixme), 

...

> [2]	This file used to be called conf.modules before linux 2.0, but this name
> is now deprecated.

remove this note ?

> int init_module(void)
> void cleanup_module(void)

I think you should use module_init() etc. right up front. If you insist,
you could add a chapter on 2.2, but I do not see *any* point in
supporting 2.2 in this document.

>    printk("<1>Hello world 1.\n");

Why do you tell the user what the KERN_* macros do right up front ?
Instead, just use KERN_ALERT or whatever everywhere, and add a note on
what it actually does near the end, if you like.

>  __SMP__: This must be defined if the kernel was compiled to support symmetrical multiprocessing

Isn't used any more. Remove. This whole section is wrong, see
kernelnewbies.org/faq/#compmod

Section 3.1.1 explains something already explained in chapter 2

> You can even write modules to replace the kernel's system calls, which
> we'll do shortly.

This is not a sensible thing to cover in this document IMO.

> to be experts in memory managements

typo

> However, there's also a C99 way of assigning to elements of a
> structure. 

You should clarify that this is preferred

> 4.1.4. Unregistering A Device

Needs a discussion on .owner instead of the MOD_INC gunk


	#if defined(CONFIG_MODVERSIONS) && ! defined(MODVERSIONS)
	   #include <linux/modversions.h>
	   #define MODVERSIONS
	#endif

looks wrong

	static int Major;  

Kernel code doesn't look like this ... "major" instead. Pointing  to the
coding style docs would be good. Goes for all the code there.

I suspect the section on procfs should be deleted or drastically
shortened : pointing to erikm's guide would be sufficient, along with a
couple of do's and dont's on when to use procfs vs. sysctl in 2.4 (never ;)

> NULL, /* mknod */

Use C99 initialisers as already mentioned

You should strongly discourage the use of ioctl

> module_interruptible_sleep_on[1].

The sleep_on family should be used. instead use the wait_event family

regards
john
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive:       http://mail.nl.linux.org/kernelnewbies/
FAQ:           http://kernelnewbies.org/faq/