Re: [patch v3] splice: fix race with page invalidation

[Jamie Lokier <jamie@shareable.org>]

> > And by papering it over, it then just makes people less likely to bother 
> > with the real issue.
> 
> I think you are talking about a totally separate issue: that NFSD's
> use of splice can result in strange things if the file is truncated
> while being read.  But this is an NFSD issue and I don't see that it
> has _anything_ to do with the above bug in splice.  I think you are
> just confusing the two things.

I'm more concerned by sendfile() users like Apache, Samba, FTPd.  In
an earlier thread on this topic, I asked if the splice bug can also
result in sendfile() sending blocks of zeros, when a file is truncated
after it has been sent, and the answer was yes probably.

Not that I checked or anything.  But if it affects sendfile() it's a
bigger deal - that has many users.

Assuming it does affect sendfile(), it's exasperated by not being able
to tell when a sendfile() has finished with the pages its sending.
E.g. you can't lock the file or otherwise synchronise with another
program which wants to modify the file.

-- Jamie