Messages in this thread | | | Date | Tue, 2 Dec 2008 14:12:05 +0000 (GMT) | From | Hugh Dickins <> | Subject | Re: [PATCH 1/8] badpage: simplify page_alloc flag check+clear |
| |
On Tue, 2 Dec 2008, Christoph Lameter wrote: > On Tue, 2 Dec 2008, Hugh Dickins wrote: > > > > But they are always clear on free. The checking is irrelevant. > > > > How about CHECK_PAGE_FLAGS_CLEAR_AT_FREE? > > Strange name.
Looks like I'm not going to be able to satisfy you then. I didn't introduce the names in the patch, so let's leave them as is for now, and everybody can muse on what they should get called in the end.
> > > If (page->flags & (all the flags including dirty and SwapBacked)) > > > zap-em. > > > > That's exactly what I did, isn't it? > > Yes but you added another instance of this.
Did I? Whereabouts? I wonder if you're thinking of the + page->flags &= ~PAGE_FLAGS_CHECK_AT_PREP; in prep_new_page(), which replaces the clearing of another collection of flags which somehow didn't get named before.
That clearing is a temporary measure, to keep the handling of PageReserved unchanged in that patch; then it vanishes in the next patch, where we treat all bad_page candidates the same way.
> Can you consolidate all the check and clears into one?
You mean one test_and_clear_bits() that somehow covers the different cases of what we expect at free time and what we need at alloc time? I don't think so.
Hugh
| |