On Wed, Sep 16, 2015 at 01:08:48PM -0400, Kyle Upton wrote:
> Applied a patch which changed value of OFPACT_SET_QUEUE to be greater
> than 32. Tracked down a subsequent autotest failure to invocation of
> rightmost_1bit_idx() with 64-bit argument 'opfacts_bitmap'.
> rightmost_1bit_idx() only works with 32-bit integers.
>
> Changed this and other occurrences where rightmost_1bit_idx() is
> invoked with a 64-bit argument to use ctz64() instead.
>
> Tested by running 'make check'.
>
> Signed-off-by: Kyle Upton <kup...@baymicrosystems.com>
After looking through the OVS code, I don't think that any of the users
of rightmost_1bit_idx() or leftmost_1bit_idx() rely on the case where
all of the bits are zero. That means that we can just change them to be
alternate names for ctz64() or log_2_floor(). Arguably we could get rid
of them entirely but personally I think the names are useful to make it
clear how the user is thinking of the function.
So, anyway, I sent an alternative patch:
http://openvswitch.org/pipermail/dev/2015-September/060257.html
What do you think?
Thanks,
Ben.
_______________________________________________
dev mailing list
dev@openvswitch.org
http://openvswitch.org/mailman/listinfo/dev
