Messages in this thread | | | Date | Wed, 3 Aug 2011 12:04:01 +0200 | Subject | Re: [PATCH] shm: fix a race between shm_exit() and shm_init() | From | Manuel Lauss <> |
| |
On Wed, Aug 3, 2011 at 10:19 AM, Linus Torvalds <torvalds@linux-foundation.org> wrote: > On Tue, Aug 2, 2011 at 10:05 PM, Marc Zyngier <marc.zyngier@arm.com> wrote: >> >> Reordering the initcalls seems the easiest solution, but it is still very >> fragile... > > So that's what I tried to do, by making it a "pure_initcall()". Even > that didn't seem to be enough according to Manuel. > > Can you try my patch (that makes just that ipc ns init be a > pure_initcall(), together with your hack on top of Andrew's? What is > it that happens so early that even pure_initcall() hasn't been done > yet?
I stuck a few printk's in the init path, the first "schedule()" in init/main.c::rest_init() starts it. I don't know enough kernel (yet) to trace it further, sorry.
Manuel
| |