Re: [PATCH 05/14] fs: don't allow kernel reads and writes without iter ops

[Eric Biggers]

On Fri, Oct 02, 2020 at 09:27:09AM -0700, Linus Torvalds wrote:
> On Thu, Oct 1, 2020 at 3:41 PM Al Viro <viro@xxxxxxxxxxxxxxxxxx> wrote:
> >
> > Better
> >         loff_t dummy = 0;
> > ...
> >                 wr = __kernel_write(file, data, bytes, &dummy);
> 
> No, just fix __kernel_write() to work correctly.
> 
> The fact is, NULL _is_ the right pointer for ppos these days.
> 
> That commit by Christoph is buggy: it replaces new_sync_write() with a
> buggy open-coded version.
> 
> Notice how new_sync_write does
> 
>         kiocb.ki_pos = (ppos ? *ppos : 0);
> ,,,
>         if (ret > 0 && ppos)
>                 *ppos = kiocb.ki_pos;
> 
> but the open-coded version doesn't.
> 
> So just fix that in linux-next. The *last* thing we want is to have
> different semantics for the "same" kernel functions.

It's a bit unintuitive that ppos=NULL means "use pos 0", not "use file->f_pos".

Anyway, it works.  The important thing is, this is still broken in linux-next...

- Eric