Andres Freund <andres@2ndquadrant.com> writes:
> On 2013-11-28 10:31:21 -0500, Tom Lane wrote:
>> The only remaining risk is that, if pointer
>> fetch/store isn't atomic, we might fetch a half-updated pointer; which
>> will be non-null, but not something we can use to reach the list. Since
>> we do purport to support such architectures, we'd better apply the patch.
> We do support such architectures? Don't we already assume we can store
> xids atomically (c.f. GetOldestActiveTransactionId())? Do we support a
> 64bit arch, that has a atomic 4byte store, but not atomic 8byte stores?
Dunno whether there are any in practice, but it's not an assumption
we make anywhere.
regards, tom lane