> I don't see how this could influence the standard I2C communication at
> all. If change in em_i2c_unreg_slave() is excluded, all that was changed
> is moving irq number from local variable to the em_i2c_device struct
> which is also not used outside of the em_i2c_unreg_slave() appart from
> logging :)

I agree. Still, I do have brown-paper-bag experiences caused by wrong
assumptions like "this cannot fail". And we are changing the way
interrupts are acquired. So, if it is not too hard, I'd prefer to have
patches tested, too. I'd still apply the patch if it turns out to be too
complicated to test (given the reviews raise the trust). Yet, also on
the pro-side, it doesn't hurt to test a newer kernel on a packed-away
system once in a while.

Thanks for the reviews!