> I've recently noticed that calling sys_delete_module(NULL) (in
> kernel/module.c), which is supposed to remove all automatically loaded
> modules that aren't used, doesn't always work as expected.
It always work as expected.
> It requires several calls to sys_delete_module to reliably remove all
> unused modules.
It was designed this way.
> For example, I got the 'lp' device auto-loaded (with parport_pc and
> parport). lsmod says:
> Module Size Used by
> parport_pc 5520 1 (autoclean)
> lp 4616 0 (autoclean)
> parport 6868 1 (autoclean) [parport_pc lp]
> After 'rmmod -a' once, lsmod still says:
> Module Size Used by
> parport_pc 5520 1 (autoclean)
> lp 4616 0 (autoclean)
> parport 6868 1 (autoclean) [parport_pc lp]
> After another 'rmmod -a', lsmod says:
> Module Size Used by
> parport_pc 5520 0 (autoclean)
> parport 6868 0 (autoclean) [parport_pc]
> And finally, after another try, lsmod happily announces that no modules are
> loaded.
> Is this a bug? Is this a feature? I don't know.
Feature. Back from 2.0 days...
> The reason this happens, however, is the MOD_VISITED flag. It prevents
> modules from immediately being released after performing a MOD_INC_USE_COUNT
> or MOD_DEC_USE_COUNT on them.
> I can't say that I understand why MOD_VISITED is required, and even if this
> thing is a feature, I'd love to know what is the reason for it.
Think more about lp example, auto-unloading and lpr'ing of postscript files on
non-postscript printer :-))
-
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/