--- "Raghu R. Arur" <rra2002@aria.ncl.cs.columbia.edu> wrote: > > Not exactly. It does set the pte in some cases: > > > > > page = virt_to_page(__va(phys_addr)); > > > if ((!VALID_PAGE(page)) || PageReserved(page)) > > > set_pte(pte, mk_pte_phys(phys_addr, prot)); > > > ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ > > > > Yes i see that it is setting the pte. But it is setting > it only if the page is reserved or if the page is not > valid. It is setting the pte entry which it had cleared > it before. Are you saying the old and new entries are same? They are not. > So it is clearing the pte of those pages which > are not reserved and which are valid. No. PTEs for all addresses in the range are cleared. remap_page_range() is mostly called from mmap routines of drivers. The virtual to physical mapping may or may not be known at the time of mmap. If the mapping is known, the physical pages will be reserved and the PTE can be set. If the mapping is not known, then the PTE should be set to zero (or some other invalid value) so that accessing the address will cause a page fault. Instead, if the page is valid but not reserved, the PTE should not be set. There is no guarantee that the page will not be reused. Ideally, drivers calling remap_page_range should not pass such a page as parameter. Hope this helps, Ravi. __________________________________ Do you Yahoo!? Yahoo! SiteBuilder - Free, easy-to-use web site design software http://sitebuilder.yahoo.com -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
