Re: Async direct IO write vs buffered read race

[Eric Sandeen <esandeen@redhat.com>]

On 6/23/17 2:59 AM, Jan Kara wrote:
> On Thu 22-06-17 12:55:50, Jeff Moyer wrote:


>> Christoph or Jan, any thoughts on this?
> 
> So our stance has been: Do not ever mix buffered and direct IO! Definitely
> not on the same file range, most definitely not at the same time.

FWIW, I'd always known that concurrent buffered & direct wasn't
particularly deterministic, i.e. the racing buffered read may get old or
new data at the time.

I was surprised to learn that the stale file data would linger
indefinitely in the page cache though.  Maybe I was just blissfully
unaware.  :)

-Eric

> The thing we do is a best effort thing that more or less guarantees that if
> you do say buffered IO and direct IO after that, it will work reasonably.
> However if direct and buffered IO can race, bad luck for your data. I don't
> think we want to sacrifice any performance of AIO DIO (and offloading of
> direct IO completion to a workqueue so that we can do invalidation costs
> noticeable mount of performance) for supporting such usecase.
> 
> 								Honza
>