linux-kernel - Re: [PATCH] To fix the below failure of handling page fault caused by the invalid input from user.

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <56982041-c7ac-fc68-f5df-12a11d351ed6@colorfullife.com>
Date:   Sat, 29 Jan 2022 16:59:47 +0100
From:   Manfred Spraul <manfred@...orfullife.com>
To:     Dongyang Wang <dongyang.wang@...driver.com>
Cc:     akpm@...ux-foundation.org, christian.brauner@...ntu.com,
        dave@...olabs.net, ebiederm@...ssion.com, legion@...nel.org,
        linux-kernel@...r.kernel.org, varad.gautam@...e.com
Subject: Re: [PATCH] To fix the below failure of handling page fault caused by
 the invalid input from user.

Hi Dongyang,

On 1/29/22 16:02, Dongyang Wang wrote:
> Hi Manfred and Andrew,
>
> Thanks for your feedback.
>
>> Hi Dongyang,
>>
>> On 1/26/22 03:42, Andrew Morton wrote:
>>> On Tue, 18 Jan 2022 17:19:52 +0800 Dongyang Wang <dongyang.wang@...driver.com> wrote:
>>>
>>>> [786058.308965] Unable to handle kernel paging request at virtual address 01000004
>>>> [786058.316286] pgd = 38a99693
>>>> [786058.319080] [01000004] *pgd=07800003, *pmd=00000000
>>>> [786058.324056] Internal error: Oops: 206 [#1] PREEMPT SMP ARM
>>>> [786058.324100] CPU:  PID: Comm:  Tainted: G         C
>>>> [786058.324102] Hardware name:
>>>> [786058.324114] PC is at __copy_to_user_std+0x4c/0x3c4
>>>> [786058.324120] LR is at store_msg+0xc0/0xe8
>>>> [786058.324124] pc : [<c0c0587c>]    lr : [<c0871d04>]    psr: 20010013
>>>> [786058.324126] sp : c3503ec4  ip : 00000000  fp : b4c9a660
>>>> [786058.324129] r10: c4228dc0  r9 : c3502000  r8 : 00000ffc
>>>> [786058.324132] r7 : 01000000  r6 : 546d3f8b  r5 : b4911690  r4 : 00000ffc
>>>> [786058.324134] r3 : 00000000  r2 : 00000f7c  r1 : 01000004  r0 : b4911690
>>>> [786058.324139] Flags: nzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
>>>> [786058.324142] Control: 30c5387d  Table: 0edc2040  DAC: 55555555
>>>> [786058.324145] Process  (pid: , stack limit = 0x25018bdf)
>> Why is process and pid: empty? Is this some kind of kernel process calling?
> The pid is 8369, it's a userspace app.
>
>>>> [786058.324148] Stack: (0xc3503ec4 to 0xc3504000)
>>>> [786058.324153] 3ec0:          b4911690 546d3f8b 01000000 00000ffc b4911690 00000ffc 00000000
>>>> [786058.324157] 3ee0: 00000ffc c0871d04 546d4f73 c3407801 c3503f28 c3407800 00000000 b49106a8
>>>> [786058.324161] 3f00: c4228dc0 c087abd4 00000002 b49106a8 617b9d03 00000000 00000000 c121d508
>>>> [786058.324165] 3f20: 00000000 bf06a1a8 d1b634cc 16b26e77 c5af5280 00000100 00000200 db806540
>>>> [786058.324170] 3f40: 00000001 c121d508 00000008 0000005c 00000000 00010008 b49106a8 c0601208
>>>> [786058.324173] 3f60: c3502000 00000040 b4c9a660 c087b474 c3503f78 c121d508 617b9d03 00000000
>>>> [786058.324177] 3f80: 2303d6cc 00000115 c0601208 c121d508 b4c9a660 b4c9a660 00000001 b49106a8
>>>> [786058.324181] 3fa0: 00000115 c06011dc b4c9a660 00000001 0000005c b49106a8 00010008 00000000
>>>> [786058.324185] 3fc0: b4c9a660 00000001 b49106a8 00000115 00000000 b4c9b400 00000000 b4c9a660
>>>> [786058.324189] 3fe0: 00000115 b4c9a650 b6b253bd b6b254b6 800d0030 0000005c 00000000 00000000
>>>> [786058.324201] [<c0c0587c>] (__copy_to_user_std) from [<c0871d04>] (store_msg+0xc0/0xe8)

I would search here: __copy_to_user_std should fail if the address is 
invalid.

For whatever reasons, it produces a page fault.

First: Is this reproducible? Does it fail immediately if you pass an 
invalid value to mq_timedreceive()?

https://elixir.bootlin.com/linux/v4.18.20/source/arch/arm/include/asm/uaccess.h#L464

It seems ARM has special optimizations (CONFIG_UACCESS_WITH_MEMCPY), and 
I cannot see if this is MMU or NO_MMU


> The kernel version is 4.18.

Ok.


--

     Manfred