Re: [Qemu-devel] [4367] Align file accesses with cache=off (Kevin Wolf, Laurent Vivier)

[Kevin Wolf <kwolf@suse.de>]

Jamie Lokier schrieb:
> Kevin Wolf wrote:
>> No, nobody mentioned the recursion problem.
> 
> Hmm.  I concede you're right in the sense that it was mentioned, but
> on a different thread about QEMU AIO recently :-)

Hm, okay. Will read that thread if I can find it. But I hope the 
recursion thing is fixed now anyway.

> Following the paragraph about two file descriptors, there was:
> 
>>> I'm not sure if that works, though.  On some OSes, if a file has any
>>> non-O_DIRECT open descriptor, all I/O is buffered ignoring the
>>> O_DIRECT flag.  If both are allowed simultaneously, I'm not sure what
>>> happens with cache-coherency between direct I/Os and buffered I/Os.
> 
> Not sure if that is quite the same thing :-)

Am I completely mistaken or is this still about two (or more) different 
file descriptors where one of them is non-O_DIRECT?

> I did miss that switching O_DIRECT on/off while AIOs are in flight on
> that descriptor might be dodgy (implementation dependent), and that it
> might not do the right things w.r.t. cohrency.

Me too. I just was too confident that it actually works when Laurent 
says it was better. You know, for every problem there is an answer that 
is clear, simple and wrong. ;-)

>> But even if so, this is more of a general feeling about how patches are 
>> handled and not only related to this patch.
> 
> I agree and have a similar feeling, though it's not a bad thing
> provided the issues are actually noticed, which they do seem to be.  

Actually, I do think it's a bad thing. Obviously, issues are noticed 
when the patch goes in and can be fixed then. Right. But what about the 
other 90% of the patches which don't get no attention at all? Nobody 
comments on them, they aren't committed, and after all they are wasted 
efforts.

It's a bad thing because it slows down qemu development by (passively) 
rejecting patches which are fine or could be fixed easily.

Kevin