Re: [PATCH 2/2] drivers/base/node.c: skip non-present sections in register_mem_sect_under_node

From: Jan Stancek
Date: Wed Sep 09 2015 - 07:11:39 EST


Greg,

any thoughts about the patch?

Regards,
Jan

----- Original Message -----
> From: "Dave Young" <dyoung@xxxxxxxxxx>
> To: "Jan Stancek" <jstancek@xxxxxxxxxx>
> Cc: gregkh@xxxxxxxxxxxxxxxxxxx, linux-kernel@xxxxxxxxxxxxxxx, linux-mm@xxxxxxxxx
> Sent: Tuesday, 1 September, 2015 9:15:53 AM
> Subject: Re: [PATCH 2/2] drivers/base/node.c: skip non-present sections in register_mem_sect_under_node
>
> On 08/27/15 at 04:43pm, Jan Stancek wrote:
> > Skip non-present sections in mem_blk to avoid crashing during boot
> > at register_mem_sect_under_node()->get_nid_for_pfn():
> >
> > Unable to handle kernel paging request for data at address
> > 0xf000000000080020
> > Faulting instruction address: 0xc00000000866b480
> > Oops: Kernel access of bad area, sig: 11 [#1]
> > SMP NR_CPUS=2048 NUMA pSeries
> > Modules linked in:
> > CPU: 14 PID: 1 Comm: swapper/14 Not tainted 4.2.0-rc8+ #6
> > task: c00000001e480000 ti: c00000001e500000 task.ti: c00000001e500000
> > NIP: c00000000866b480 LR: c00000000851aecc CTR: 0000000000000400
> > ...
> > NIP [c00000000866b480] get_nid_for_pfn+0x10/0x30
> > LR [c00000000851aecc] register_mem_sect_under_node+0x9c/0x190
> > Call Trace:
> > [c00000001e503b10] [c0000000084f89a4] put_device+0x24/0x40 (unreliable)
> > [c00000001e503b60] [c00000000851b3d4] register_one_node+0x2b4/0x390
> > [c00000001e503bc0] [c000000008ae7a50] topology_init+0x4c/0x1e8
> > [c00000001e503c30] [c00000000800b3bc] do_one_initcall+0x10c/0x260
> > [c00000001e503d00] [c000000008ae41b4] kernel_init_freeable+0x27c/0x364
> > [c00000001e503dc0] [c00000000800bc14] kernel_init+0x24/0x130
> > [c00000001e503e30] [c000000008009530] ret_from_kernel_thread+0x5c/0xac
> > Instruction dump:
> > 4e800020 60000000 60000000 60420000 3b80ffed 4bffffc8 00000000 00000000
> > 3920ffff 78633664 792900c4 7d434a14 <e94a0020> 2faa0000 41de0010 7c63482a
> > ---[ end trace e9ab4a173e0cee14 ]---
> >
> > This has been observed during kdump kernel boot on ppc64le KVM guest
> > (page size: 65536, sections_per_block: 16, PAGES_PER_SECTION: 256)
> > where kdump adds "rtas" to list of usable regions:
> > # hexdump -C /sys/firmware/devicetree/base/rtas/linux,rtas-base
> > 00000000 2f ff 00 00 |/...|
> >
> > [ 0.000000] Early memory node ranges
> > [ 0.000000] node 0: [mem 0x0000000000000000-0x000000001fffffff]
> > [ 0.000000] node 0: [mem 0x000000002fff0000-0x000000002fffffff]
> >
> > Crash happens when register_mem_sect_under_node goes over mem_blk that
> > spans sections 32-47, 32-46 are not present, 47 is present:
> > 32 * 256 * 65536 == 0x20000000
> > 47 * 256 * 65536 == 0x2f000000
> > It tries to access page for first pfn of this mem_blk (8192 == 32 * 256)
> > and crashes.
> >
> > Signed-off-by: Jan Stancek <jstancek@xxxxxxxxxx>
> > Cc: Greg Kroah-Hartman <gregkh@xxxxxxxxxxxxxxxxxxx>
> > ---
> > drivers/base/node.c | 3 +++
> > 1 file changed, 3 insertions(+)
> >
> > diff --git a/drivers/base/node.c b/drivers/base/node.c
> > index 4c7423a4b5f4..e638cfde7486 100644
> > --- a/drivers/base/node.c
> > +++ b/drivers/base/node.c
> > @@ -390,6 +390,9 @@ int register_mem_sect_under_node(struct memory_block
> > *mem_blk, int nid)
> > sect_no <= mem_blk->end_section_nr;
> > sect_no++) {
> >
> > + if (!present_section_nr(sect_no))
> > + continue;
> > +
> > sect_start_pfn = section_nr_to_pfn(sect_no);
> > sect_end_pfn = sect_start_pfn + PAGES_PER_SECTION - 1;
> >
> > --
> > 1.8.3.1
> >
> > --
> > To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> > the body of a message to majordomo@xxxxxxxxxxxxxxx
> > More majordomo info at http://vger.kernel.org/majordomo-info.html
> > Please read the FAQ at http://www.tux.org/lkml/
>
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/