[openib-general] Re: mthca_arbel_post_srq_recv/mthca_tavor_post_srq_recv

[Roland Dreier]

    Michael> Hi, Roland!  The code in
    Michael> mthca_arbel_post_srq_recv/mthca_tavor_post_srq_recv looks
    Michael> very strange: there seems to be unreacheable code,
    Michael> spinlocks dont seem to be dropped on error, etc.

    Michael> Further, it seems that the functions return the number of
    Michael> posted descriptors on error. This differs from post_recv
    Michael> which alwasy returns an error code on error. Is that
    Michael> intentional?  Am I missing something? Could you comment
    Michael> on this design please?

No, you're right, it's absolutely broken.  The code is the result of a
screwed up merge from an earlier version.  I fixed it up now.

 - R.