lkml.org 
[lkml]   [2019]   [Oct]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Date
SubjectUAF read in print_binder_transaction_log_entry() on ANDROID_BINDERFS kernels
Hi!

There is a use-after-free read in print_binder_transaction_log_entry()
on ANDROID_BINDERFS kernels because
print_binder_transaction_log_entry() prints the char* e->context_name
as string, and if the transaction occurred on a binder device from
binderfs, e->context_name belongs to the binder device and is freed
when the inode disappears.

Luckily this shouldn't have security implications, since:

a) reading the binder transaction log is already a pretty privileged operation
b) I can't find any actual users of ANDROID_BINDERFS

I guess there are three ways to fix it:
1) Create a new shared global spinlock for binderfs_evict_inode() and
binder_transaction_log_show(), and let binderfs_evict_inode() scan the
transaction log for pointers to its name and replace them with
pointers to a statically-allocated string "{DELETED}" or something
like that.
2) Let the transaction log contain non-reusable device identifiers
instead of name pointers, and let print_binder_transaction_log_entry()
look them up in something like a hashtable.
3) Just copy the name into the transaction log every time.

I'm not sure which one is better, or whether there's a nicer fourth
option, so I'm leaving writing a patch for this to y'all.


Trigger instructions (requires you to have some helpers that can
register a context manager and send some transaction to it):
==============
root@test:/home/user# mkdir /tmp/binder
root@test:/home/user# mount -t binder -o stats=global /dev/null /tmp/binder
root@test:/home/user# ls -l /tmp/binder
total 0
crw------- 1 root root 248, 1 Oct 7 20:34 binder
crw------- 1 root root 248, 0 Oct 7 20:34 binder-control
drwxr-xr-x 3 root root 0 Oct 7 20:34 binder_logs
crw------- 1 root root 248, 2 Oct 7 20:34 hwbinder
crw------- 1 root root 248, 3 Oct 7 20:34 vndbinder
root@test:/home/user# ln -s /tmp/binder/binder /dev/binder
[run some simple binder demo code to temporarily register a context
manager and send a binder transaction]
root@test:/home/user# rm /tmp/binder/binder
root@test:/home/user# cat /tmp/binder/binder_logs/transaction_log
2: call from 2277:2277 to 2273:0 context @������� node 1 handle 0
size 24:8 ret 0/0 l=0
5: call from 2273:2273 to 2277:2277 context @������� node 3 handle 1
size 0:0 ret 0/0 l=0
6: reply from 2277:2277 to 2273:2273 context @������� node 0 handle 0
size 4:0 ret 0/0 l=0
7: reply from 2273:2273 to 2277:2277 context @������� node 0 handle 0
size 4:0 ret 0/0 l=0
root@test:/home/user#
==============

ASAN splat:
[ 333.300753] ==================================================================
[ 333.303197] BUG: KASAN: use-after-free in string_nocheck+0x9d/0x160
[ 333.305081] Read of size 1 at addr ffff8880b0981258 by task cat/2279

[ 333.307415] CPU: 1 PID: 2279 Comm: cat Not tainted 5.4.0-rc1+ #513
[ 333.309304] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996),
BIOS 1.12.0-1 04/01/2014
[ 333.310987] Call Trace:
[ 333.312032] dump_stack+0x7c/0xc0
[ 333.312581] ? string_nocheck+0x9d/0x160
[ 333.313157] print_address_description.constprop.7+0x36/0x50
[ 333.314030] ? string_nocheck+0x9d/0x160
[ 333.314603] ? string_nocheck+0x9d/0x160
[ 333.315236] __kasan_report.cold.10+0x1a/0x35
[ 333.315972] ? string_nocheck+0x9d/0x160
[ 333.316545] kasan_report+0xe/0x20
[ 333.317104] string_nocheck+0x9d/0x160
[ 333.317652] ? widen_string+0x160/0x160
[ 333.318270] ? string_nocheck+0x160/0x160
[ 333.318857] ? unwind_get_return_address+0x2a/0x40
[ 333.319636] ? profile_setup.cold.9+0x96/0x96
[ 333.320359] string+0xb6/0xc0
[ 333.320800] ? hex_string+0x280/0x280
[ 333.321398] vsnprintf+0x20c/0x780
[ 333.321898] ? num_to_str+0x180/0x180
[ 333.322503] ? __kasan_kmalloc.constprop.6+0xc1/0xd0
[ 333.323235] ? vfs_read+0xbc/0x1e0
[ 333.323814] ? ksys_read+0xb5/0x150
[ 333.324323] ? do_syscall_64+0xb9/0x3b0
[ 333.324948] ? entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 333.325756] seq_vprintf+0x78/0xb0
[ 333.326253] seq_printf+0x96/0xc0
[ 333.327132] ? seq_vprintf+0xb0/0xb0
[ 333.327678] ? match_held_lock+0x2e/0x240
[ 333.328450] binder_transaction_log_show+0x237/0x2d0
[ 333.329163] seq_read+0x266/0x690
[ 333.329705] vfs_read+0xbc/0x1e0
[ 333.330178] ksys_read+0xb5/0x150
[ 333.330724] ? kernel_write+0xb0/0xb0
[ 333.331257] ? trace_hardirqs_off_caller+0x57/0x130
[ 333.332045] ? mark_held_locks+0x29/0xa0
[ 333.332678] ? do_syscall_64+0x6b/0x3b0
[ 333.333235] do_syscall_64+0xb9/0x3b0
[ 333.333856] entry_SYSCALL_64_after_hwframe+0x49/0xbe
[ 333.334635] RIP: 0033:0x7fbbb95d4461
[ 333.335153] Code: fe ff ff 50 48 8d 3d fe d0 09 00 e8 e9 03 02 00
66 0f 1f 84 00 00 00 00 00 48 8d 05 99 62 0d 00 8b 00 85 c0 75 13 31
c0 0f 05 <48> 3d 00 f0 ff ff 77 57 c3 66 0f 1f 44 00 00 41 54 49 89 d4
55 48
[ 333.337950] RSP: 002b:00007ffcbe6438e8 EFLAGS: 00000246 ORIG_RAX:
0000000000000000
[ 333.339072] RAX: ffffffffffffffda RBX: 0000000000020000 RCX: 00007fbbb95d4461
[ 333.340157] RDX: 0000000000020000 RSI: 00007fbbb9324000 RDI: 0000000000000003
[ 333.341320] RBP: 00007fbbb9324000 R08: 00000000ffffffff R09: 0000000000000000
[ 333.342454] R10: fffffffffffffb9c R11: 0000000000000246 R12: 00007fbbb9324000
[ 333.343550] R13: 0000000000000003 R14: 0000000000000fff R15: 0000000000020000

[ 333.344845] Allocated by task 2259:
[ 333.345416] save_stack+0x19/0x80
[ 333.345899] __kasan_kmalloc.constprop.6+0xc1/0xd0
[ 333.346636] __kmalloc_track_caller+0xf4/0x2e0
[ 333.347271] kmemdup+0x17/0x40
[ 333.347796] binderfs_binder_device_create.isra.6+0x217/0x530
[ 333.348674] binderfs_fill_super+0x486/0x81e
[ 333.349309] mount_nodev+0x41/0xb0
[ 333.349860] legacy_get_tree+0x7b/0xc0
[ 333.350398] vfs_get_tree+0x40/0x130
[ 333.350970] do_mount+0xacb/0xea0
[ 333.351449] ksys_mount+0xb1/0xd0
[ 333.352007] __x64_sys_mount+0x5d/0x70
[ 333.352545] do_syscall_64+0xb9/0x3b0
[ 333.353144] entry_SYSCALL_64_after_hwframe+0x49/0xbe

[ 333.354144] Freed by task 2278:
[ 333.354598] save_stack+0x19/0x80
[ 333.355135] __kasan_slab_free+0x12e/0x180
[ 333.355734] kfree+0xe6/0x310
[ 333.356234] binderfs_evict_inode+0xb8/0xd0
[ 333.356831] evict+0x16f/0x290
[ 333.358081] do_unlinkat+0x2f6/0x420
[ 333.358593] do_syscall_64+0xb9/0x3b0
[ 333.359176] entry_SYSCALL_64_after_hwframe+0x49/0xbe

[ 333.360196] The buggy address belongs to the object at ffff8880b0981258
which belongs to the cache kmalloc-8 of size 8
[ 333.361991] The buggy address is located 0 bytes inside of
8-byte region [ffff8880b0981258, ffff8880b0981260)
[ 333.363796] The buggy address belongs to the page:
[ 333.364538] page:ffffea0002c26040 refcount:1 mapcount:0
mapping:ffff8880b6c03c80 index:0x0
[ 333.365765] flags: 0x1fffc0000000200(slab)
[ 333.366402] raw: 01fffc0000000200 ffffea0002cb1d80 0000001400000014
ffff8880b6c03c80
[ 333.367546] raw: 0000000000000000 0000000000aa00aa 00000001ffffffff
0000000000000000
[ 333.369030] page dumped because: kasan: bad access detected

[ 333.370095] Memory state around the buggy address:
[ 333.370824] ffff8880b0981100: fc fb fc fc fb fc fc fb fc fc fb fc
fc fb fc fc
[ 333.371907] ffff8880b0981180: fb fc fc fb fc fc fb fc fc fb fc fc
fb fc fc fb
[ 333.372969] >ffff8880b0981200: fc fc fb fc fc fb fc fc fb fc fc fb
fc fc fb fc
[ 333.374033] ^
[ 333.374884] ffff8880b0981280: fc fb fc fc fb fc fc fb fc fc fb fc
fc fb fc fc
[ 333.375957] ffff8880b0981300: fb fc fc fb fc fc fb fc fc fb fc fc
fb fc fc fb
[ 333.377013] ==================================================================

\
 
 \ /
  Last update: 2019-10-07 22:50    [W:0.116 / U:2.480 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site