On Tue, Jan 3, 2012 at 17:21, Greg KH [off-list ref] wrote:

On Wed, Dec 28, 2011 at 05:45:18PM +1100, Stephen Rothwell wrote:

quoted

Because of the powerpc problems above, I have used the driver-core tree
from next-20111222 for today.

Sorry about all of the problems, we tried to fix everything we could,
but your merges and cross-builds found stuff we missed :(

Kay, care to send me patches to fix this, and all of the other
linux-next-reported problems to me so we can get this resolved this
week?

I rather don't want to add error checking to stuff that doesn't do it
today. The sysdev stuff never had that forced checks, but the normal
device stuff has.

I think the force return value check is really a pretty misguided idea
in general, and it's up to the caller to do these checks and handle
rollbacks, not the driver core, I think.

Can't we just remove that forced check?

Kay