On Tue, 3 Nov 2015, Boaz Harrosh wrote: > On 11/03/2015 04:49 AM, Hugh Dickins wrote: > > On Mon, 2 Nov 2015, Boaz Harrosh wrote: > >> > >> Do you think the code is actually wrong as is? > > > > Not that I know of: just a little too complicated and confusing. > > > > But becomes slightly wrong if my simplification to page migration > > goes through, since that introduces an instant when PageDirty is set > > before the new page contains the correct data and is marked Uptodate. > > Hence my patch. > > > >> > >> BTW: Very similar code is in fs/nfs/objlayout/objio_osd.c::__r4w_get_page > > > > Indeed, the patch makes the same adjustment to that code too. > > > > OK thanks. Let me setup and test your patch. On top of 4.3 is good? > I'll send you a tested-by once I'm done. Great, thanks Boaz, that will help a lot. On top of 4.3 very good. (Of course, that will not test the rare page migration race I shall introduce later; but it will test that you're doing the right thing with this change at your end, when you will be safe against the race.) Hugh -- 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>