Re: zero-filed page on VOP_PUTPAGES

[Matthias Scheler <tron%zhadum.org.uk@localhost>]

On Sun, Aug 07, 2011 at 12:29:48PM +0200, Emmanuel Dreyfus wrote:
> Matthias Scheler <tron%zhadum.org.uk@localhost> wrote:
> 
> > I remember an issue like that with "rump_smbfs" a while ago. The problem
> > was that a VOP_SETATTR() on PUFFS sets the file-size. And in this case
> > the backend in SMBFS used 32-bit file offset and therefore truncated
> > the file.
> 
> That is not something like this: it occurs at offsets as low as 16384,

That still can happen if the size calculation is wrong in another as
not using 64-bit math.

> and there is always valid data after the zeroed chunk.

But is the valid data at the correct offset?

        Kind regards

-- 
Matthias Scheler                                  http://zhadum.org.uk/