On 18-Jul-14, at 5:55 PM, Carlos O'Donell wrote:
This discussion has got me thinking that the char and short
implementations
may be broken (i.e,
the CAS operation may clobber data adjacent to location being
operated on).
I'm thinking we need
separate LWS calls for them. Thoughts?
Yes, you need masking or short load variants. How did you test the
char and short implementations? ;-)
There is no testing beyond the general testing of atomics in GCC.
There is masking
in the GCC code but I think this is racy. Another thread might modify
the masked data
during the CAS operation and this might be lost. I've never seen it
in the real world though.
The implementation was originally copied from arm.
Dave
--
John David Anglin dave.anglin@xxxxxxxx
--
To unsubscribe from this list: send the line "unsubscribe linux-parisc" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
