On Thu, Jul 16, 2015 at 09:38:56PM +0200, Oleg Nesterov wrote: > On 07/16, Benjamin LaHaise wrote: > > > > On Thu, Jul 16, 2015 at 09:05:03PM +0200, Oleg Nesterov wrote: > > > Thanks! > > ... > > > but the problem looks clear: CONFIG_MMU is not set, so we need > > > a dummy filemap_page_mkwrite() along with generic_file_mmap() and > > > generic_file_readonly_mmap(). > > > > > > I'll send the fix, but... > > > > > > Benjamin, Jeff, shouldn't AIO depend on MMU? Or it can actually work even > > > if CONFIG_MMU=n? > > > > It should work when CONFIG_MMU=n, > > Really? I am just curious. I am not saying the code currently does, but that the use of the functionality is valid. The code clearly doesn't at the moment, and given that nobody has complained, it's seems unlikely that there are any active users. -ben > alloc_anon_inode() doesn't set S_IFREG, it seems that nommu.c:do_mmap_pgoff() > should just fail in validate_mmap_request() ? > > Even if not, it should fail because of MAP_SHARED && !NOMMU_MAP_DIRECT? > > I am just trying to understand, could you explain? > > Oleg. -- "Thought is the essence of where you are now." -- To unsubscribe, send a message with 'unsubscribe linux-mm' in the body to majordomo@xxxxxxxxx. For more info on Linux MM, see: http://www.linux-mm.org/ . Don't email: <a href=mailto:"dont@xxxxxxxxx";> email@xxxxxxxxx </a>