[openib-general] (was: slab error while removing ib_mad) testing IB of a kernel before its release

[Or Gerlitz]

Roland Dreier wrote:
>     Or> I think you were on vacation when i posted this, there were
>     Or> two responses saying they were not able to reproduce it, but
>     Or> no one was trying 2.6.17-X
> 
> Not sure why you expect me to solve this -- other than the fact that I
> am a great debugger ;)

Let me clarify a little:

The test case for itself (probing out a module loaded by the pci hotplug 
subsystem) is kind of rare and its not the issue (I am doing it when 
replacing the ib stack with newer code)

When i posted the original report, i got responses from two people both 
saying they have tried it with this or that flavor of the current stable 
kernel (2.6.16) and that the problem does not reproduce (sure...).

The impression i was getting from those responses and the luck of others 
is that (say) almost no one of the openib maintainers test infiniband 
with the "next" kernel which is not released yet.

And if we don't test it, sure we can't expect it to work, no less.

That's the point i wanted to make later in that thread. I opted for 
deferring this discussion for a time you are around, so this is why i 
write it only now.

Or.