On Thu, 2019-03-21 at 23:57 +0500, Mikhail Gavrilov wrote:
> On Thu, 21 Mar 2019 at 20:48, Qian Cai <cai@xxxxxx> wrote:
> > OK, those pages look similar enough. If you add this to __init_single_page()
> > in
> > mm/page_alloc.c :
> >
> > if (page == (void *)0xffffdcd2607ce000 || page == (void *)0xffffe4b7607ce000
> > ||
> > page == (void *)0xffffd27aa07ce000 || page == (void *)0xffffcf49607ce000) {
> > printk("KK page = %px\n", page);
> > dump_stack();
> > }
> >
> > to see where those pages have been initialized in the first place.
>
> In the new kernel panics "page" also does not repeated.
>
> $ journalctl | grep "page:"
> Mar 21 20:46:56 localhost.localdomain kernel: page:fffffbbbe07ce000 is
> uninitialized and poisoned
> Mar 21 21:28:03 localhost.localdomain kernel: page:ffffdecc207ce000 is
> uninitialized and poisoned
> Mar 21 23:43:24 localhost.localdomain kernel: page:fffff91ce07ce000 is
> uninitialized and poisoned
That is OK. The above debug patch may still be useful to figure out where those
pages come from (or you could add those 3 pages address to the patch as well).
They may be initialized in a similar fashion or uninitialized to begin with.
