Re: [RFC] Maintaining persistent kernel data across module

[Ph. Marek]

>Some modules claim to have a need to maintain persistent data in the
>kernel even when the module is unloaded and reloaded.  I can see four
>approaches to fixing this problem, there may be others.  Note: this
>only applies to kernel data, it says nothing about maintaining external
>data across reboots.
>
>Comments please.
>....
there is another option available: let the modules use files.
then the module compiled as eth0 could get one set of data (ip-adress, ...),
eth1 could load another file, but the random-"module" uses only one data file
(storing entropy).

I know that that doesn't sound pretty - BUT:
It could be combined with some of the other options (preferable 4).
eg: "eth0"-data requested, flag PERSISTENT set: file is created and mmap()ed,
pointer returned to module - looks like normal memory.
another module requests run-time data - is not stored on disk, but kmalloc() used.

that is a generic interface which allowes a common interface for both types of 
storage - and the path for the data could be given to the kernel like the path to 
insmod today.

how about this?