Re: [PATCH V4 libibverbs 2/7] Add member functions to poll an extended CQ

On Mon, May 30, 2016 at 06:01:24PM +0300, Matan Barak (External) wrote:

So, you suggest that "simple" attributes will take into account byte_offset,
bit_offset, bits_size (or mask) and endianity. While the right endianity
conversion function will be chosen by inspecting the bits_size?
Of course, this is in addition to the attribute's nature (simple/possible
simple/complex), which we need to condition on that too.
This doesn't sound simple for me and I'm afraid it will affect performance
more than just a real simple function call.

Right, that is exactly the point - if the inlined version requires a
branch to choose betwen simple/complex then it will be about the same
cost to just branch to the correct code in the driver.

Jason
--
To unsubscribe from this list: send the line "unsubscribe linux-rdma" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html