Re: [PATCH] mm/page_alloc: fix boot hang in memmap_init_zone

[Jia He]

On 3/15/2018 11:39 PM, Daniel Vacek Wrote:
> On Thu, Mar 15, 2018 at 3:08 PM, Jia He <hejianet@xxxxxxxxx> wrote:
> > Hi Daniel
> >
> >
> >
> > On 3/14/2018 6:42 AM, Daniel Vacek Wrote:
> > > On some architectures (reported on arm64) commit 864b75f9d6b01
> > > ("mm/page_alloc: fix memmap_init_zone pageblock alignment")
> > > causes a boot hang. This patch fixes the hang making sure the alignment
> > > never steps back.
> > >
> > > Link:
> > > http://lkml.kernel.org/r/0485727b2e82da7efbce5f6ba42524b429d0391a.1520011945.git.neelx@xxxxxxxxxx
> > > Fixes: 864b75f9d6b01 ("mm/page_alloc: fix memmap_init_zone pageblock
> > > alignment")
> > > Signed-off-by: Daniel Vacek <neelx@xxxxxxxxxx>
> > > Tested-by: Sudeep Holla <sudeep.holla@xxxxxxx>
> > > Tested-by: Naresh Kamboju <naresh.kamboju@xxxxxxxxxx>
> > > Cc: Andrew Morton <akpm@xxxxxxxxxxxxxxxxxxxx>
> > > Cc: Mel Gorman <mgorman@xxxxxxxxxxxxxxxxxxx>
> > > Cc: Michal Hocko <mhocko@xxxxxxxx>
> > > Cc: Paul Burton <paul.burton@xxxxxxxxxx>
> > > Cc: Pavel Tatashin <pasha.tatashin@xxxxxxxxxx>
> > > Cc: Vlastimil Babka <vbabka@xxxxxxx>
> > > Cc: <stable@xxxxxxxxxxxxxxx>
> > > ---
> > >    mm/page_alloc.c | 7 ++++++-
> > >    1 file changed, 6 insertions(+), 1 deletion(-)
> > >
> > > diff --git a/mm/page_alloc.c b/mm/page_alloc.c
> > > index 3d974cb2a1a1..e033a6895c6f 100644
> > > --- a/mm/page_alloc.c
> > > +++ b/mm/page_alloc.c
> > > @@ -5364,9 +5364,14 @@ void __meminit memmap_init_zone(unsigned long size,
> > > int nid, unsigned long zone,
> > >                           * is not. move_freepages_block() can shift ahead
> > > of
> > >                           * the valid region but still depends on correct
> > > page
> > >                           * metadata.
> > > +                        * Also make sure we never step back.
> > >                           */
> > > -                       pfn = (memblock_next_valid_pfn(pfn, end_pfn) &
> > > +                       unsigned long next_pfn;
> > > +
> > > +                       next_pfn = (memblock_next_valid_pfn(pfn, end_pfn)
> > > &
> > >                                          ~(pageblock_nr_pages-1)) - 1;
> > > +                       if (next_pfn > pfn)
> > > +                               pfn = next_pfn;
> > It didn't resolve the booting hang issue in my arm64 server.
> > what if memblock_next_valid_pfn(pfn, end_pfn) is 32 and pageblock_nr_pages
> > is 8196?
> > Thus, next_pfn will be (unsigned long)-1 and be larger than pfn.
> > So still there is an infinite loop here.
> Hi Jia,
>
> Yeah, looks like another uncovered case. Noone reported this so far.
> Anyways upstream reverted all this for now and we're discussing the
> right approach here.
>
> In any case thanks for this report. Can you share something like below
> from your machine?
sure.
[ÂÂÂ 0.000000] NUMA: Faking a node at [mem 
0x0000000000000000-0x00000017ffffffff]
[ÂÂÂ 0.000000] NUMA: NODE_DATA [mem 0x17ffffcb80-0x17ffffffff]
[ÂÂÂ 0.000000] Zone ranges:
[ÂÂÂ 0.000000]ÂÂ DMA32ÂÂÂ [mem 0x0000000000200000-0x00000000ffffffff]
[ÂÂÂ 0.000000]ÂÂ NormalÂÂ [mem 0x0000000100000000-0x00000017ffffffff]
[ÂÂÂ 0.000000] Movable zone start for each node
[ÂÂÂ 0.000000] Early memory node ranges
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000000200000-0x000000000021ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000000820000-0x000000000307ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000003080000-0x000000000308ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000003090000-0x00000000031fffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000003200000-0x00000000033fffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000003410000-0x000000000563ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000005640000-0x000000000567ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000005680000-0x00000000056dffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x00000000056e0000-0x00000000086fffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008700000-0x000000000871ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008720000-0x000000000894ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008950000-0x0000000008baffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008bb0000-0x0000000008bcffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008bd0000-0x0000000008c4ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008c50000-0x0000000008e2ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008e30000-0x0000000008e4ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008e50000-0x0000000008fcffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000008fd0000-0x000000000910ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000009110000-0x00000000092effff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x00000000092f0000-0x000000000930ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000009310000-0x000000000963ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000009640000-0x000000000e61ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000000e620000-0x000000000e64ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000000e650000-0x000000000fffffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x0000000010800000-0x0000000017feffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000001c000000-0x000000001c00ffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000001c010000-0x000000001c7fffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000001c810000-0x000000007efbffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000007efc0000-0x000000007efdffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000007efe0000-0x000000007efeffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000007eff0000-0x000000007effffff]
[ÂÂÂ 0.000000]ÂÂ nodeÂÂ 0: [mem 0x000000007f000000-0x00000017ffffffff]
[ÂÂÂ 0.000000] Initmem setup node 0 [mem 
0x0000000000200000-0x00000017ffffffff]

--
Cheers,
Jia