Re: [PATCH 2.5.66-mm2] Fix page_convert_anon locking issues

[Hugh Dickins]

On Thu, 3 Apr 2003, Dave McCracken wrote:
> 
> No, try_to_unmap will claim success when in fact there are still mappings.
> It'd be all right if it failed, but there's no way to tell it to fail.  The
> page will be freed by kswapd based on try_to_unmap's claim of success.

No: see the various checks on page_count(page) in vmscan.c:
though page_convert_anon temporarily leaves a page with neither
mapcount nor the right number of pte pointers, page_count is unaffected.

Hugh