[openib-general] [PATCH] mthca: memfree completion with error workaround

[Roland Dreier]

    Fabian> Why not just have multiple implemenations of the function,
    Fabian> and setup the function pointer in the verbs according to
    Fabian> what firmware and device is in use?  That way devices not
    Fabian> affected could continue to use the optimized version...

Yup, that's the obvious solution.  But I'm not sure it's worth that
much bloat in this case.