The linux-next series "mm/memory_hotplug: Shrink zones before removing
memory" [1] seems make a crash easier to reproduce while reading
/proc/pagetypeinfo after offlining a memory section. Fix it by using
pfn_to_online_page() in the PFN walker.

[1] https://lore.kernel.org/linux-mm/20191006085646.5768-1-da...@redhat.com/

page:ffffea0021200000 is uninitialized and poisoned
raw: ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff
raw: ffffffffffffffff ffffffffffffffff ffffffffffffffff ffffffffffffffff
page dumped because: VM_BUG_ON_PAGE(PagePoisoned(p))
There is not page extension available.
------------[ cut here ]------------
kernel BUG at include/linux/mm.h:1107!
RIP: 0010:pagetypeinfo_showmixedcount_print+0x4fb/0x680
Call Trace:
 walk_zones_in_node+0x3a/0xc0
 pagetypeinfo_show+0x260/0x2c0
 seq_read+0x27e/0x710
 proc_reg_read+0x12e/0x190
 __vfs_read+0x50/0xa0
 vfs_read+0xcb/0x1e0
 ksys_read+0xc6/0x160
 __x64_sys_read+0x43/0x50
 do_syscall_64+0xcc/0xaec
 entry_SYSCALL_64_after_hwframe+0x49/0xbe

Signed-off-by: Qian Cai <c...@lca.pw>
---
 mm/page_owner.c | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/mm/page_owner.c b/mm/page_owner.c
index dee931184788..03a6b19b3cdd 100644
--- a/mm/page_owner.c
+++ b/mm/page_owner.c
@@ -296,11 +296,10 @@ void pagetypeinfo_showmixedcount_print(struct seq_file *m,
                pageblock_mt = get_pageblock_migratetype(page);
 
                for (; pfn < block_end_pfn; pfn++) {
-                       if (!pfn_valid_within(pfn))
+                       page = pfn_to_online_page(pfn);
+                       if (!page)
                                continue;
 
-                       page = pfn_to_page(pfn);
-
                        if (page_zone(page) != zone)
                                continue;
 
-- 
1.8.3.1

Reply via email to