Re: Regression: NULL pointer dereference after NFS_V4_2_READ_PLUS (commit 7fd461c47)

[Krzysztof Kozlowski <krzysztof.kozlowski@linaro.org>]

On 10/02/2023 21:55, Anna Schumaker wrote:
> Hi Krzysztof,
> 
> On Fri, Feb 10, 2023 at 3:53 AM Krzysztof Kozlowski
> <krzysztof.kozlowski@linaro.org> wrote:
>>
>> On 10/02/2023 09:41, Krzysztof Kozlowski wrote:
>>>
>>>
>>>>
>>>> I was also wondering if it would be possible to turn on KASAN in your
>>>> kernel, which should give us a little more info?
>>>
>>> I'll try with KASAN.
>>
>> Not much from the KASAN, except that kernel
>> continues to boot and runs:
>>
>> [   44.722846] vdd_vmem: disabling
>> [   44.793465] systemd[1]: Hostname set to <odroidhc1>.
>> [   45.357929] systemd[1]: bpf-lsm: BPF LSM hook not enabled in the kernel, BPF LSM not supported
>> [   45.980476] ==================================================================
>> [   45.986372] BUG: KASAN: null-ptr-deref in xdr_inline_decode+0x140/0x200
>> [   45.992929] Read of size 4092 at addr 00000004 by task kworker/u16:3/71
>> [   45.999513]
>> [   46.000940] CPU: 6 PID: 71 Comm: kworker/u16:3 Not tainted 6.2.0-rc7-00018-g0983f6bf2bfc #222
>> [   46.009504] Hardware name: Samsung Exynos (Flattened Device Tree)
>> [   46.015542] Workqueue: rpciod rpc_async_schedule
>> [   46.020123]  unwind_backtrace from show_stack+0x10/0x14
>> [   46.025323]  show_stack from dump_stack_lvl+0x58/0x70
>> [   46.030301]  dump_stack_lvl from kasan_report+0xa8/0xe0
>> [   46.035501]  kasan_report from kasan_check_range+0x94/0x1a0
>> [   46.041048]  kasan_check_range from memcpy+0x28/0x68
>> [   46.045985]  memcpy from xdr_inline_decode+0x140/0x200
>> [   46.051098]  xdr_inline_decode from nfs4_xdr_dec_read_plus+0x298/0x5b8
>> [   46.057602]  nfs4_xdr_dec_read_plus from call_decode+0x39c/0x530
>> [   46.063581]  call_decode from __rpc_execute+0x1f4/0xc5c
>> [   46.068776]  __rpc_execute from rpc_async_schedule+0x2c/0x4c
>> [   46.074411]  rpc_async_schedule from process_one_work+0x51c/0xc44
>> [   46.080478]  process_one_work from worker_thread+0x9c/0x7c0
>> [   46.086022]  worker_thread from kthread+0x16c/0x1b8
>> [   46.090872]  kthread from ret_from_fork+0x14/0x2c
>> [   46.095550] Exception stack(0xf0ba3fb0 to 0xf0ba3ff8)
>> [   46.100580] 3fa0:                                     00000000 00000000 00000000 00000000
>> [   46.108740] 3fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
>> [   46.116885] 3fe0: 00000000 00000000 00000000 00000000 00000013 00000000
>> [   46.123461] ==================================================================
>> [   46.130774] Disabling lock debugging due to kernel taint
>> [   50.848579] systemd[1]: Queued start job for default target Graphical Interface.
>> [   50.877068] systemd[1]: Created slice Slice /system/getty.
>>
>>
>> Decoded stacktrace is (this is master branch):
>>
>> [   46.020123] unwind_backtrace from show_stack (arch/arm/kernel/traps.c:258)
>> [   46.025323] show_stack from dump_stack_lvl (lib/dump_stack.c:107 (discriminator 4))
>> [   46.030301] dump_stack_lvl from kasan_report (mm/kasan/report.c:184 mm/kasan/report.c:519)
>> [   46.035501] kasan_report from kasan_check_range (mm/kasan/generic.c:173 mm/kasan/generic.c:189)
>> [   46.041048] kasan_check_range from memcpy (mm/kasan/shadow.c:65)
>> [   46.045985] memcpy from xdr_inline_decode (net/sunrpc/xdr.c:1419 net/sunrpc/xdr.c:1454)
> 
> Actually, this part is really useful. net/sunrpc/xdr.c:1419 points to
> the memcpy in xdr_copy_to_scratch(), which has me wondering if I'm
> incorrectly setting up the xdr scratch buffer that READ_PLUS uses for
> decoding. Can you try this patch and let me know if it helps?
> 
> From ac2d6c501dbcdb306480edaee625b5496f1fb4f5 Mon Sep 17 00:00:00 2001
> From: Anna Schumaker <Anna.Schumaker@Netapp.com>
> Date: Fri, 10 Feb 2023 15:50:22 -0500
> Subject: [PATCH] NFSv4.2: Rework scratch handling for READ_PLUS
> 

Patch is corrupted - maybe mail program reformatted it when sending:

Applying: NFSv4.2: Rework scratch handling for READ_PLUS
error: corrupt patch at line 12
Patch failed at 0001 NFSv4.2: Rework scratch handling for READ_PLUS


Best regards,
Krzysztof