On Tuesday 10 January 2012 21:12:00 Lucas De Marchi wrote: > On Tue, Jan 10, 2012 at 11:40 PM, Dave Reisner <d@xxxxxxxxxxxxxx> wrote: > > kmod_module_insert_module will never return -EEXIST given the check for > > initstate prior to the call. Check for EPERM instead and throw an > > appropriate error when an unprivileged user attempts to load a module. > > As we talked on IRC: > > There is a race condition that we want to avoid. This is why we check > before and after we try to remove the module, since it could be > removed between the check and the syscall. be good to add a comment to the source about this -mike
Attachment:
signature.asc
Description: This is a digitally signed message part.