Re: [PATCH v2 1/2] mm: uninitialized struct page poisoning sanity checking

[Sasha Levin]

On Tue, Mar 13, 2018 at 08:38:57PM -0400, Pavel Tatashin wrote:
>Hi Sasha,
>
>It seems the patch is doing the right thing, and it catches bugs. Here
>we access uninitialized struct page. The question is why this happens?

Not completely; note that we die on an invalid reference rather than
assertion failure.

>register_mem_sect_under_node(struct memory_block *mem_blk, int nid)
>   page_nid = get_nid_for_pfn(pfn);
>
>node id is stored in page flags, and since struct page is poisoned,
>and the pattern is recognized, the panic is triggered.
>
>Do you have config file? Also, instructions how to reproduce it?

Attached the config. It just happens on boot.

>Thank you,
>Pasha
>
>
>On Tue, Mar 13, 2018 at 7:43 PM, Sasha Levin
><Alexander.Levin@xxxxxxxxxxxxx> wrote:
>> On Wed, Jan 31, 2018 at 04:02:59PM -0500, Pavel Tatashin wrote:
>>>During boot we poison struct page memory in order to ensure that no one is
>>>accessing this memory until the struct pages are initialized in
>>>__init_single_page().
>>>
>>>This patch adds more scrutiny to this checking, by making sure that flags
>>>do not equal to poison pattern when the are accessed. The pattern is all
>>>ones.
>>>
>>>Since, node id is also stored in struct page, and may be accessed quiet
>>>early we add the enforcement into page_to_nid() function as well.
>>>
>>>Signed-off-by: Pavel Tatashin <pasha.tatashin@xxxxxxxxxx>
>>>---
>>
>> Hey Pasha,
>>
>> This patch is causing the following on boot:
>>
>> [    1.253732] BUG: unable to handle kernel paging request at fffffffffffffffe
>> [    1.254000] PGD 2284e19067 P4D 2284e19067 PUD 2284e1b067 PMD 0
>> [    1.254000] Oops: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI
>> [    1.254000] Modules linked in:
>> [    1.254000] CPU: 1 PID: 1 Comm: swapper/0 Not tainted 4.16.0-rc5-next-20180313 #10
>> [    1.254000] Hardware name: Microsoft Corporation Virtual Machine/Virtual Machine, BIOS 090007  06/02/2017
>> [    1.254000] RIP: 0010:__dump_page (??:?)
>> [    1.254000] RSP: 0000:ffff881c63c17810 EFLAGS: 00010246
>> [    1.254000] RAX: dffffc0000000000 RBX: ffffea0084000000 RCX: 1ffff1038c782f2b
>> [    1.254000] RDX: 1fffffffffffffff RSI: ffffffff9e160640 RDI: ffffea0084000000
>> [    1.254000] RBP: ffff881c63c17c00 R08: ffff8840107e8880 R09: ffffed0802167a4d
>> [    1.254000] R10: 0000000000000001 R11: ffffed0802167a4c R12: 1ffff1038c782f07
>> [    1.254000] R13: ffffea0084000020 R14: fffffffffffffffe R15: ffff881c63c17bd8
>> [    1.254000] FS:  0000000000000000(0000) GS:ffff881c6ac00000(0000) knlGS:0000000000000000
>> [    1.254000] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>> [    1.254000] CR2: fffffffffffffffe CR3: 0000002284e16000 CR4: 00000000003406e0
>> [    1.254000] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>> [    1.254000] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>> [    1.254000] Call Trace:
>> [    1.254000] dump_page (/mm/debug.c:80)
>> [    1.254000] get_nid_for_pfn (/./include/linux/mm.h:900 /drivers/base/node.c:396)
>> [    1.254000] register_mem_sect_under_node (/drivers/base/node.c:438)
>> [    1.254000] link_mem_sections (/drivers/base/node.c:517)
>> [    1.254000] topology_init (/./include/linux/nodemask.h:271 /arch/x86/kernel/topology.c:164)
>> [    1.254000] do_one_initcall (/init/main.c:835)
>> [    1.254000] kernel_init_freeable (/init/main.c:901 /init/main.c:909 /init/main.c:927 /init/main.c:1076)
>> [    1.254000] kernel_init (/init/main.c:1004)
>> [    1.254000] ret_from_fork (/arch/x86/entry/entry_64.S:417)
>> [ 1.254000] Code: ff a8 01 4c 0f 44 f3 4d 85 f6 0f 84 31 0e 00 00 4c 89 f2 48 b8 00 00 00 00 00 fc ff df 48 c1 ea 03 80 3c 02 00 0f 85 2d 11 00 00 <49> 83 3e ff 0f 84 a9 06 00 00 4d 8d b7 c0 fd ff ff 48 b8 00 00
>> All code
>> ========
>>    0:   ff a8 01 4c 0f 44       ljmp   *0x440f4c01(%rax)
>>    6:   f3 4d 85 f6             repz test %r14,%r14
>>    a:   0f 84 31 0e 00 00       je     0xe41
>>   10:   4c 89 f2                mov    %r14,%rdx
>>   13:   48 b8 00 00 00 00 00    movabs $0xdffffc0000000000,%rax
>>   1a:   fc ff df
>>   1d:   48 c1 ea 03             shr    $0x3,%rdx
>>   21:   80 3c 02 00             cmpb   $0x0,(%rdx,%rax,1)
>>   25:   0f 85 2d 11 00 00       jne    0x1158
>>   2b:*  49 83 3e ff             cmpq   $0xffffffffffffffff,(%r14)               <-- trapping instruction
>>   2f:   0f 84 a9 06 00 00       je     0x6de
>>   35:   4d 8d b7 c0 fd ff ff    lea    -0x240(%r15),%r14
>>   3c:   48                      rex.W
>>   3d:   b8                      .byte 0xb8
>>         ...
>>
>> Code starting with the faulting instruction
>> ===========================================
>>    0:   49 83 3e ff             cmpq   $0xffffffffffffffff,(%r14)
>>    4:   0f 84 a9 06 00 00       je     0x6b3
>>    a:   4d 8d b7 c0 fd ff ff    lea    -0x240(%r15),%r14
>>   11:   48                      rex.W
>>   12:   b8                      .byte 0xb8
>>         ...
>> [    1.254000] RIP: __dump_page+0x1c8/0x13c0 RSP: ffff881c63c17810 (/./include/asm-generic/sections.h:42)
>> [    1.254000] CR2: fffffffffffffffe
>> [    1.254000] ---[ end trace e643dfbc44b562ca ]---
>>
>> --
>>
>> Thanks,
>> Sasha

-- 

Thanks,
Sasha

Attachment: config-sasha.gz
Description: config-sasha.gz