netdev - Re: [PATCH 3/5] genetlink: statically initialize families

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Tue, 25 Oct 2016 13:38:28 +0200
From:   Johannes Berg <johannes@...solutions.net>
To:     David Laight <David.Laight@...LAB.COM>,
        "netdev@...r.kernel.org" <netdev@...r.kernel.org>
Subject: Re: [PATCH 3/5] genetlink: statically initialize families

On Tue, 2016-10-25 at 11:25 +0000, David Laight wrote:

> > Actually, with the new system where it's not const, I could even
> > split this up and submit per subsystem, i.e. the fourth patch
> > doesn't depend on it. I thought it would, since I wanted to make it
> > const, but since I failed it doesn't actually have that dependency.
> 
> Actually, why aren't the structures 'const' ?
> 
> You could use a #define to set the .ops and .n_ops fields.
> (and maybe .module = THIS_MODULE as well).

This stuff isn't the problem - after this patch these are of course
statically initialized and const.

The problem is that the struct members family->id, family->mcgrp_offset 
and family->attrbuf, are only determined at genl_register_family().

I considered simply moving them into a new struct, that contains just
those along with a pointer to the family, but then I have essentially
two choices:

1) look up the registration struct by the family every time I need the
   family ID, which is all the time; that would be rather inefficient

2) change *all* genetlink code to not pass the family but rather pass a
   pointer returned by genl_register_family(); that's a massive change

So on the whole, I decided that __ro_after_init was entirely reasonable
and then even this patch isn't really necessary, but since I had it
anyway it still seemed to make sense, even if I had to add all those
forward declarations.

johannes