Re: [rfc patch 3/4] splice: remove confirm frompipe_buf_operations

[Miklos Szeredi]

> > 
> > Let's start with page_cache_pipe_buf_confirm().  How should we deal
> > with finding an invalidated page (!PageUptodate(page) &&
> > !page->mapping)?
> 
> I suspect we just have to use it. After all, it was valid when the read 
> was done. The fact that it got invalidated later is kind of immaterial.

Right.  But what if it's invalidated *before* becoming uptodate (if
you'd read my mail further, I discussed this).

Why does invalidate_complete_page2() do ClearPageUptodate()?  Dunno,
maybe it shoulnd't.  But that would need a rather thorough audit of
all code checking PageUptodate()...

> splice() is an optimized read+write. The read reads it into a temporary 
> buffer. The fact that it's a zero-copy buffer and basically just re-uses 
> the source doesn't really change that.

Agreed.

Miklos
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/