Hi,
On 2015-12-03 16:10:51 -0600, Kevin Grittner wrote:
> Is the c.h change above on anything resembling the right track for
> a patch for this? If not, what would such a patch look like?
I think a better path would be to add fallback support for 64bit atomics
- like we already have for 32bit. T
On Fri, Dec 4, 2015 at 6:35 AM, Greg Stark wrote:
> On Thu, Dec 3, 2015 at 10:10 PM, Kevin Grittner wrote:
>> Is the c.h change above on anything resembling the right track for
>> a patch for this? If not, what would such a patch look like?
>
> It would be nicer if we could come up with an inter
On Thu, Dec 3, 2015 at 10:10 PM, Kevin Grittner wrote:
> Is the c.h change above on anything resembling the right track for
> a patch for this? If not, what would such a patch look like?
It would be nicer if we could come up with an interface that didn't
require #ifdefs everywhere it's used.
So
The "snapshot too old" patch has an XXX comment about being able to
drop a spinlock usage from a frequently called "get" method if the
64-bit read could be trusted to be atomic. There is no need for a
barrier in this case, because a stale value just means we won't be
quite as aggressive about prun
