Re: drivers/tee double-free of page when optee_shm_register fails

[Patrik Lantz <Patrik.Lantz@axis.com>]

[-- Attachment #1: Type: text/plain, Size: 2283 bytes --]

Hi,


I'm attaching a reproducer (only CA, there is no TA used in this case ) and crash log for reference.


Regards,

Patrik

________________________________
From: Sumit Garg <sumit.garg@linaro.org>
Sent: Tuesday, December 14, 2021 8:41:20 AM
To: Lars Persson
Cc: op-tee(a)lists.trustedfirmware.org; Patrik Lantz
Subject: Re: drivers/tee double-free of page when optee_shm_register fails

Hi Lars,

On Fri, 10 Dec 2021 at 20:23, Lars Persson <larper@axis.com> wrote:
>
> Hi
>
> Me and Patrik have been tracing a kernel memory corruption bug that is
> triggered when op-tee runs out of resources and returns an error from
> the OPTEE_MSG_CMD_REGISTER_SHM call. This is yet another fall-out from
> Patrik's fuzzing of the TEE subsystem.
>
> The symptoms would look like this when page debugging is enabled:
> BUG: Bad page state in process optee_example_h  pfn:46bb0
> page:(ptrval) refcount:-1 mapcount:0 mapping:00000000 index:0x0 pfn:0x46bb0
> flags: 0x0(zone=0)
>
> Our reproducer runs a loop with the TEE_IOC_SHM_ALLOC until memory runs
> out at the optee-os end (dynamic SHM enabled). The error is 100%
> reproducible with such a loop.

Can you share a simple reproducer test application (CA and TA)?

>
> We have traced this down to what seems to be a miss in the memory
> ownership contract during the call to OPTEE_MSG_CMD_REGISTER_SHM.
>
> When pool_op_alloc() detects that optee_shm_register() has failed, it
> will free the allocated page at the very end of the function.
> Unfortunately that page has already been freed because OP-TEE has sent a
> OPTEE_RPC_CMD_SHM_FREE for this shm object before returning from
> OPTEE_MSG_CMD_REGISTER_SHM. This is my conclusion based on prints added
> to the code.

I can't see any RPC free command from OP-TEE in case the
OPTEE_MSG_CMD_REGISTER_SHM fails. Have you compared kernel addresses
for SHM during OPTEE_MSG_CMD_REGISTER_SHM and OPTEE_RPC_CMD_SHM_FREE?

>
> I cannot write a patch for this because I am at a loss of who actually
> is supposed to trigger the free of the pages in this situation. Is there
> an API spec that makes this clear ?

It should only be the pool_op_alloc() API in case
OPTEE_MSG_CMD_REGISTER_SHM fails.

-Sumit

>
> BR,
>   Lars



[-- Attachment #2: crashlog.txt --]
[-- Type: text/plain, Size: 2347 bytes --]

==================================================================
BUG: KASAN: double-free or invalid-free in tee_shm_alloc (/mnt/build/optee-qemu/linux/drivers/tee/tee_shm.c:210) 

CPU: 0 PID: 310 Comm: optee_example_h Not tainted 5.14.0 #10
Hardware name: Generic DT based system
(unwind_backtrace) from show_stack (/mnt/build/optee-qemu/linux/arch/arm/kernel/traps.c:254) 
(show_stack) from dump_stack_lvl (/mnt/build/optee-qemu/linux/lib/dump_stack.c:106 (discriminator 1)) 
(dump_stack_lvl) from print_address_description.constprop.0 (/mnt/build/optee-qemu/linux/mm/kasan/report.c:234) 
(print_address_description.constprop.0) from kasan_report_invalid_free (/mnt/build/optee-qemu/linux/mm/kasan/report.c:359) 
(kasan_report_invalid_free) from kfree (/mnt/build/optee-qemu/linux/./include/linux/vmstat.h:523 /mnt/build/optee-qemu/linux/mm/slub.c:3248 /mnt/build/optee-qemu/linux/mm/slub.c:4264) 
(kfree) from tee_shm_alloc (/mnt/build/optee-qemu/linux/drivers/tee/tee_shm.c:210) 
(tee_shm_alloc) from tee_ioctl (/mnt/build/optee-qemu/linux/drivers/tee/tee_core.c:296 /mnt/build/optee-qemu/linux/drivers/tee/tee_core.c:862) 
(tee_ioctl) from sys_ioctl (/mnt/build/optee-qemu/linux/fs/ioctl.c:52 /mnt/build/optee-qemu/linux/fs/ioctl.c:1029 /mnt/build/optee-qemu/linux/fs/ioctl.c:1067 /mnt/build/optee-qemu/linux/fs/ioctl.c:1055) 
(sys_ioctl) from ret_fast_syscall (/mnt/build/optee-qemu/linux/arch/arm/kernel/entry-common.S:51) 
Exception stack(0x86293fa8 to 0x86293ff0)
3fa0:                   00000016 000001a0 00000003 c010a401 20000080 00000001
3fc0: 00000016 000001a0 0049151d 00000036 47ba12d0 47ba0e10 47ba12d0 6ecaec88
3fe0: 47ba0cd0 47ba0cc0 00490e69 66c74d72

The buggy address belongs to the page:
page:(ptrval) refcount:0 mapcount:0 mapping:(ptrval) index:0x0 pfn:0x4686b
aops:0x40 ino:0
flags: 0x0(zone=0)
raw: 00000000 c0888d84 b80654bc 82401200 00000000 80200020 ffffffff 00000000
page dumped because: kasan: bad access detected

Memory state around the buggy address:
8686af80: 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
8686b000: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
>8686b080: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
^
8686b100: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
8686b180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
==================================================================