Kernel patch request for AFS
Arjan van de Ven
arjan at fenrus.demon.nl
Mon Mar 13 21:10:25 UTC 2006
On Mon, 2006-03-13 at 15:50 -0500, Monty wrote:
> On Mon, Mar 13, 2006 at 09:27:53PM +0100, Arjan van de Ven wrote:
> > because it's going away! And very soon. It's already a NOP. And Rusty
> > sent a patch to kill it off for real; that's probably going to make
> > 2.6.17.
> ...eliminating a kernel interface, even a noop, in a subminor
> release!? Dear God... this sort of thing makes *me* want to move to
> [assuming I read that right, it's slightly ambiguous if 'kill it off'
> means the bug or the API]
it means the API.
The interface is on the "to be removed" list for a long time already;
it's not working anymore, a better one exists for 3 years now.
Note that this is an INTERNAL interface; those change all the time in
the kernel. Removing old should-be-unused ones is one of the ways needed
to keep the kernel clean and relatively small. You can argue long and
hard about each individual one and how much it costs; but trust me, it
adds up bigtime.
> Of course, they're running RHEL not Fedora, but what we're talking
> about is not specific to Fedora
actually it is since RH is not removing interfaces in the same RHEL
but that's expensive in general. (retrofitting stuff into old kernels is
a LOT of pain and really hard work: think about it: there is a lot of
danger in just moving code about, because of all the changes
internally). Thankfully Fedora chooses the "safe and easy" way of going
to newer upstream kernels all the time ;)
More information about the fedora-test-list