In message <[email protected]>,Alberto Ma ncini writes: >My concerns are: > > 1. as in previous code any preexistent keyring is replaced, > is this acceptable ?
not really, but i didnt see a good way to handle things at the time. keyring still isnt very mature and there isnt any standard userspace utility that runs at login to create an empty session keyring in the first place. ideally i guess you would create a new keyring and merge the existing keys on the right (excluding the pag key). > 2. is frustrating not to have exported necessary functions from > the kerel, a previous thread in this list concluded, as I > undestand, that there is no option, still true ? still true. it is unlikely that any GPL export would have the gpl tag removed. there was talking of creating a generic pag style interface and submitting it to the kernel. however, there wouldnt be any consumers in the kernel so it would likely be rejected. > 3. the name of the installed keyring is ``_afs.`` > nobody seems to care about the name as i know, > but the kernel code use different names for different > keryring types ... opinions ? prefixing with _ might have special meaning. i would check the code. > 4. what about other architectures ? it is called linux keyring support. if anyone else had something similar we could write a middleware i suppose. >To test the function i'm still using a syscall but i'd like to try a >new pioctl; any advice ? Really the code is not so easy so any help >would be good (e.g. where to start !) just look at setpag. its an ioctl. _______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
