On Mon, Jun 04, 2018 at 01:59:09PM -0600, Keith Busch wrote: > > Per section 5.2 we need to issue the corresponding log page to clear an > > AEN, so for a namespace data changed AEN we need to read the changed > > namespace list log. And once we read that log anyway we might as well > > use it to optimize the rescan. > > > > Signed-off-by: Christoph Hellwig <h...@lst.de> > > I'm a little concerned about this. Userspace might be reading the same > log page. Since the contents of the page may change each time its read, > it's possible the driver will see only a subset of changed namespaces > at the point it gets to read the log page, missing a chance to > revalidate others.
I agree with the concern, but I don't think it really is uniqueue here. We allow userspace to send all kind sof harmful commands, even queue for queue creation and deletion. But let's assume we don't want to use the list due to this concern: we'd still have to read the log page, as per the NVMe spec the only think clearing a pending AEN is reading the associated log page. We'd then need to still do the full scan using identify. Is this what we want? If you think this is important for reliability we could just ignore the log page.
