* Re: [syzbot] BUG: bad usercopy in io_openat2_prep
2023-02-11 16:36 ` [syzbot] BUG: bad usercopy in io_openat2_prep Kees Cook
@ 2023-02-12 15:47 ` Pavel Begunkov
2023-02-12 16:04 ` syzbot
2023-02-13 11:27 ` Dmitry Vyukov
1 sibling, 1 reply; 4+ messages in thread
From: Pavel Begunkov @ 2023-02-12 15:47 UTC (permalink / raw)
To: Kees Cook, syzbot, akpm, keescook, linux-hardening, linux-kernel,
linux-mm, syzkaller-bugs, io-uring
On 2/11/23 16:36, Kees Cook wrote:
> On February 11, 2023 8:08:52 AM PST, syzbot <[email protected]> wrote:
>> Hello,
>>
>> syzbot found the following issue on:
>>
>> HEAD commit: ca72d58361ee Merge branch 'for-next/core' into for-kernelci
>> git tree: git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci
>> console output: https://syzkaller.appspot.com/x/log.txt?x=14a882f3480000
>> kernel config: https://syzkaller.appspot.com/x/.config?x=f3e78232c1ed2b43
>> dashboard link: https://syzkaller.appspot.com/bug?extid=cdd9922704fc75e03ffc
>> compiler: Debian clang version 15.0.7, GNU ld (GNU Binutils for Debian) 2.35.2
>> userspace arch: arm64
>> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1203777b480000
>> C reproducer: https://syzkaller.appspot.com/x/repro.c?x=124c1ea3480000
I couldn't reproduce it, let's try latest io_uring first
#syz test: https://git.kernel.dk/linux.git for-6.3/io_uring
>> Downloadable assets:
>> disk image: https://storage.googleapis.com/syzbot-assets/e2c91688b4cd/disk-ca72d583.raw.xz
>> vmlinux: https://storage.googleapis.com/syzbot-assets/af105438bee6/vmlinux-ca72d583.xz
>> kernel image: https://storage.googleapis.com/syzbot-assets/4a28ec4f8f7e/Image-ca72d583.gz.xz
>>
>> IMPORTANT: if you fix the issue, please add the following tag to the commit:
>> Reported-by: [email protected]
>>
>> usercopy: Kernel memory overwrite attempt detected to SLUB object 'pid' (offset 24, size 24)!
>
> This looks like some serious memory corruption. The pid slab is 24 bytes in size, but struct io_open is larger... Possible UAF after the memory being reallocated to a new slab??
>
> -Kees
>
>> [...]
>> Call trace:
>> usercopy_abort+0x90/0x94
>> __check_heap_object+0xa8/0x100
>> __check_object_size+0x208/0x6b8
>> io_openat2_prep+0xcc/0x2b8
>> io_submit_sqes+0x338/0xbb8
>> __arm64_sys_io_uring_enter+0x168/0x1308
>> invoke_syscall+0x64/0x178
>> el0_svc_common+0xbc/0x180
>> do_el0_svc+0x48/0x110
>> el0_svc+0x58/0x14c
>> el0t_64_sync_handler+0x84/0xf0
>> el0t_64_sync+0x190/0x194
>
>
>
--
Pavel Begunkov
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [syzbot] BUG: bad usercopy in io_openat2_prep
2023-02-12 15:47 ` Pavel Begunkov
@ 2023-02-12 16:04 ` syzbot
0 siblings, 0 replies; 4+ messages in thread
From: syzbot @ 2023-02-12 16:04 UTC (permalink / raw)
To: akpm, asml.silence, io-uring, kees, keescook, linux-hardening,
linux-kernel, linux-mm, syzkaller-bugs
Hello,
syzbot has tested the proposed patch but the reproducer is still triggering an issue:
BUG: bad usercopy in io_openat2_prep
usercopy: Kernel memory overwrite attempt detected to SLUB object 'pid' (offset 24, size 24)!
------------[ cut here ]------------
kernel BUG at mm/usercopy.c:102!
Internal error: Oops - BUG: 00000000f2000800 [#1] PREEMPT SMP
Modules linked in:
CPU: 0 PID: 4995 Comm: syz-executor.0 Not tainted 6.2.0-rc6-syzkaller-00050-gfbe870a72fd1 #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/21/2023
pstate: 60400005 (nZCv daif +PAN -UAO -TCO -DIT -SSBS BTYPE=--)
pc : usercopy_abort+0x90/0x94
lr : usercopy_abort+0x90/0x94
sp : ffff800012dd3be0
x29: ffff800012dd3bf0 x28: 000000000000001c x27: ffff0000d0e13400
x26: 00000000200000c0 x25: ffff80000cf51000 x24: fffffc0000000000
x23: 05ffc00000000200 x22: fffffc0003108280 x21: ffff0000c420a118
x20: 0000000000000000 x19: 0000000000000018 x18: 0000000000000000
x17: 0000000000000000 x16: ffff0000d0e13df8 x15: ffff80000dbd1118
x14: ffff0000d0e13400 x13: 00000000ffffffff x12: ffff0000d0e13400
x11: ff808000081bd5b0 x10: 0000000000000000 x9 : 4c3aa38d2e853f00
x8 : 4c3aa38d2e853f00 x7 : ffff800008162dbc x6 : 0000000000000000
x5 : 0000000000000080 x4 : 0000000000000001 x3 : 0000000000000000
x2 : ffff0001fefbff08 x1 : 0000000100000000 x0 : 000000000000005d
Call trace:
usercopy_abort+0x90/0x94
__check_heap_object+0xa8/0x100
__check_object_size+0x208/0x6b8
io_openat2_prep+0xcc/0x2f0
io_submit_sqes+0x330/0xba8
__arm64_sys_io_uring_enter+0x168/0x9b0
invoke_syscall+0x64/0x178
el0_svc_common+0xbc/0x180
do_el0_svc+0x48/0x150
el0_svc+0x58/0x14c
el0t_64_sync_handler+0x84/0xf0
el0t_64_sync+0x190/0x194
Code: 911d2800 aa0903e1 f90003e8 94e6d3da (d4210000)
---[ end trace 0000000000000000 ]---
Tested on:
commit: fbe870a7 io_uring,audit: don't log IORING_OP_MADVISE
git tree: https://git.kernel.dk/linux.git for-6.3/io_uring
console output: https://syzkaller.appspot.com/x/log.txt?x=17241257480000
kernel config: https://syzkaller.appspot.com/x/.config?x=22fc000172595f28
dashboard link: https://syzkaller.appspot.com/bug?extid=cdd9922704fc75e03ffc
compiler: Debian clang version 15.0.7, GNU ld (GNU Binutils for Debian) 2.35.2
userspace arch: arm64
Note: no patches were applied.
^ permalink raw reply [flat|nested] 4+ messages in thread
* Re: [syzbot] BUG: bad usercopy in io_openat2_prep
2023-02-11 16:36 ` [syzbot] BUG: bad usercopy in io_openat2_prep Kees Cook
2023-02-12 15:47 ` Pavel Begunkov
@ 2023-02-13 11:27 ` Dmitry Vyukov
1 sibling, 0 replies; 4+ messages in thread
From: Dmitry Vyukov @ 2023-02-13 11:27 UTC (permalink / raw)
To: Kees Cook
Cc: syzbot, akpm, keescook, linux-hardening, linux-kernel, linux-mm,
syzkaller-bugs, io-uring, Aleksandr Nogikh
On Mon, 13 Feb 2023 at 12:05, Kees Cook <[email protected]> wrote:
>
> On February 11, 2023 8:08:52 AM PST, syzbot <[email protected]> wrote:
> >Hello,
> >
> >syzbot found the following issue on:
> >
> >HEAD commit: ca72d58361ee Merge branch 'for-next/core' into for-kernelci
> >git tree: git://git.kernel.org/pub/scm/linux/kernel/git/arm64/linux.git for-kernelci
> >console output: https://syzkaller.appspot.com/x/log.txt?x=14a882f3480000
> >kernel config: https://syzkaller.appspot.com/x/.config?x=f3e78232c1ed2b43
> >dashboard link: https://syzkaller.appspot.com/bug?extid=cdd9922704fc75e03ffc
> >compiler: Debian clang version 15.0.7, GNU ld (GNU Binutils for Debian) 2.35.2
> >userspace arch: arm64
> >syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1203777b480000
> >C reproducer: https://syzkaller.appspot.com/x/repro.c?x=124c1ea3480000
> >
> >Downloadable assets:
> >disk image: https://storage.googleapis.com/syzbot-assets/e2c91688b4cd/disk-ca72d583.raw.xz
> >vmlinux: https://storage.googleapis.com/syzbot-assets/af105438bee6/vmlinux-ca72d583.xz
> >kernel image: https://storage.googleapis.com/syzbot-assets/4a28ec4f8f7e/Image-ca72d583.gz.xz
> >
> >IMPORTANT: if you fix the issue, please add the following tag to the commit:
> >Reported-by: [email protected]
> >
> >usercopy: Kernel memory overwrite attempt detected to SLUB object 'pid' (offset 24, size 24)!
>
> This looks like some serious memory corruption. The pid slab is 24 bytes in size, but struct io_open is larger... Possible UAF after the memory being reallocated to a new slab??
We've just noticed that some of syzbot arm64 configs did not enable
KASAN, so it could produce false one-off reports caused by previous
silent memory corruptions.
So if you don't see anything obvious, don't spend too much time looking at it.
#syz invalid
> -Kees
>
> > [...]
> >Call trace:
> > usercopy_abort+0x90/0x94
> > __check_heap_object+0xa8/0x100
> > __check_object_size+0x208/0x6b8
> > io_openat2_prep+0xcc/0x2b8
> > io_submit_sqes+0x338/0xbb8
> > __arm64_sys_io_uring_enter+0x168/0x1308
> > invoke_syscall+0x64/0x178
> > el0_svc_common+0xbc/0x180
> > do_el0_svc+0x48/0x110
> > el0_svc+0x58/0x14c
> > el0t_64_sync_handler+0x84/0xf0
> > el0t_64_sync+0x190/0x194
>
>
>
> --
> Kees Cook
^ permalink raw reply [flat|nested] 4+ messages in thread