[parisc-linux] Re: RFC: mmap patch

[David S. Miller]

   From: Randolph Chung <tausq@debian.org>
   Date: Sat, 8 Mar 2003 15:31:41 -0800

   > You need merely 9MB of address space (2 * 4MB) if you implement
   > my {copy,clear}_user_page() dynamic mapping hack, that will be
   > tons more cheaper than any kmap based scheme and also be nicer
   > on the TLB as there will be zero TLB changes occurring around
   > the copy/clear.
   
   i hope i'm not missing something obvious, but from what i understand
   from this discussion, {copy,user}_user_page helps with the initial
   syncing up of new mappings that are created for a process... but how
   does this deal with syncing up changes afterwards? for that we still
   need the flush_dcache_page to walk the user mappings right?
   
Yes, the {copy,clear}_user_page() thing is an optimization for
handling of cache alias issues wrt. anonymous pages.

The PA-RISC mmap/write bug is something entirely different,
flush_dcache_page() just isn't doing what it is defined to
do :-)