Re: [PATCH] page cache readahead implemented?

[Hugh Dickins]

On Sun, 22 Jun 2003, Andrew Morton wrote:
> Hugh Dickins <[e-mail]> wrote:
> > 
> > No prize for guessing it was tmpfs I found the problem with.
> 
> Yeah, the usual blot on the kernelscape.

Now, now, tmpfs is there to keep you honest!

> > Am I reading alloc_inode correctly, that the default it gives is
> > an empty_aops with NULL readpage, but a backing_dev_info with non-0
> > ra_pages?  How does your do_mmap_pgoff fare on a PROT_EXEC mapping
> > of one of those mmaping device drivers?
> 
> Probably explodes?  I'm not particularly serious about that change - it
> can be done in userspace.

You mean, by MADV_WILLNEED...?

> Is the do_mmap_pgoff() hack the only offender?  If so it would be better to
> localise the test in there too.

No, it's not quite the only (see original comments): I've not tried it,
but madvise_willneed (looks like it) suffers from the same assumption.

Hugh