El 24/05/15 a les 16:19, Antti Kantee ha escrit:
On 24/05/15 14:15, Antti Kantee wrote:
However, no races will
occur as a result, i.e. kpreempt_disable() being a nop is not a bug.
Hmm, there might be some interaction issues with external timing-sensitive
parties such as hardware. But I'd rather see the specific code in the specific
device driver that depends on that before thinking about it more.
Well, so far I haven't found any evidence that my other fixes for
hardware-related timing problems
require this. I guess if problems arise we can revisit the matter :)
--
Robert Millan
