Thank you for all the replies I received. However, since all of them were
incorrect, I feel it is my duty to clarify the situation and answer my own
question as I have figured it out at home where I examined the source more
closely.
The trick here is done via module_init()/module_exit() macros which for a
module (if MODULE defined) evaluates to the
init_module()/cleanup_module(), i.e. init_elf_binfmt()/exit_elf_binfmt()
become init_module()/cleanup_module().
If the module (i.e. compilation unit) is compiled into the kernel
statically, then module_init() generates some static variable which is
initialized with the return value of the function. And module_exit(), of
course, generates /* Nothing */ :)
Very impressive but not immediately obvious.
Regards,
------
Tigran A. Aivazian | http://www.sco.com
Escalations Research Group | tel: +44-(0)1923-813796
Santa Cruz Operation Ltd | http://www.ocston.org/~tigran
On Mon, 13 Sep 1999, Tigran Aivazian wrote:
> Hi,
>
> If binfmt_elf is a module then register_binfmt() called from
> init_elf_binfmt does the job. But if it is not a module how is elf_format
> linked to formats?
>
> I grepped for it and couldn't find it. The formats is initialized to NULL
> so it can't be done the way it is done for default_exec_domain (i.e.
> initializaing a variable instead of calling register_exec_domain()). And
> elf_formats is static to binfmt_elf.c so I am a bit lost...
>
> Regards,
> Tigran.
>
>
>
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/