On Wed, Sep 26, 2012 at 10:47:54AM +0800, Fengguang Wu wrote: > On Tue, Sep 25, 2012 at 10:06:08PM -0400, Naoya Horiguchi wrote: > > On Tue, Sep 25, 2012 at 05:20:48PM -0700, David Rientjes wrote: > > > On Tue, 25 Sep 2012, Naoya Horiguchi wrote: > > > > > > > KPF_THP can be set on non-huge compound pages like slab pages, because > > > > PageTransCompound only sees PG_head and PG_tail. Obviously this is a bug > > > > and breaks user space applications which look for thp via /proc/kpageflags. > > > > Currently thp is constructed only on anonymous pages, so this patch makes > > > > KPF_THP be set when both of PageAnon and PageTransCompound are true. > > > > > > > > Changelog in v2: > > > > - add a comment in code > > > > > > > > Signed-off-by: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx> > > > > > > Wouldn't PageTransCompound(page) && !PageHuge(page) && !PageSlab(page) be > > > better for a future extension of thp support? > > > > Yes, this saves us an additional change when thp starts handling pagecaches. > > Andrew, can you replace the previous version in -mm tree with new one below? > > > > Thanks, > > Naoya > > --- > > From: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx> > > Date: Tue, 25 Sep 2012 21:30:25 -0400 > > Subject: [PATCH v3] kpageflags: fix wrong KPF_THP on slab pages > > > > KPF_THP can be set on non-huge compound pages like slab pages, because > > PageTransCompound only sees PG_head and PG_tail. Obviously this is a bug > > s/sees/checks/ > > > and breaks user space applications which look for thp via /proc/kpageflags. > > This patch rules out setting KPF_THP wrongly by additional PageSlab check. > > > > Changelog in v3: > > - check PageSlab instead of PageAnon > > - fix patch subject > > > > Changelog in v2: > > - add a comment in code > > > > Signed-off-by: Naoya Horiguchi <n-horiguchi@xxxxxxxxxxxxx> > > --- > > fs/proc/page.c | 7 ++++++- > > 1 file changed, 6 insertions(+), 1 deletion(-) > > > > diff --git a/fs/proc/page.c b/fs/proc/page.c > > index 7fcd0d6..e36d1f3 100644 > > --- a/fs/proc/page.c > > +++ b/fs/proc/page.c > > @@ -115,7 +115,12 @@ u64 stable_page_flags(struct page *page) > > u |= 1 << KPF_COMPOUND_TAIL; > > if (PageHuge(page)) > > u |= 1 << KPF_HUGE; > > - else if (PageTransCompound(page)) > > + /* > > + * PageTransCompound can be true for slab pages because it just sees > > s/sees/checks/ > > > + * PG_head/PG_head, so we need to check PageSlab to make sure the given > > PG_head/PG_head should be PG_head/PG_tail. Ah, sorry for my carelessness. > > + * page is a thp, not a non-huge compound page. > > + */ > > + else if (PageTransCompound(page) && !PageSlab(page)) > > u |= 1 << KPF_THP; > > Good catch! > > Will this report THP for the various drivers that do __GFP_COMP > page allocations? I'm afraid it will. I think of checking PageLRU as an alternative, but it needs compound_head() to report tail pages correctly. In this context, pages are not pinned or locked, so it's unsafe to use compound_head() because it can return a dangling pointer. Maybe it's a thp's/hugetlbfs's (not kpageflags specific) problem, so going forward with compound_head() expecting that it will be fixed in the future work can be an option. Thanks, Naoya -- 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>