CVE-2024-56693
Severity CVSS v4.0:
Pending analysis
Type:
Unavailable / Other
Publication date:
28/12/2024
Last modified:
28/12/2024
Description
In the Linux kernel, the following vulnerability has been resolved:<br />
<br />
brd: defer automatic disk creation until module initialization succeeds<br />
<br />
My colleague Wupeng found the following problems during fault injection:<br />
<br />
BUG: unable to handle page fault for address: fffffbfff809d073<br />
PGD 6e648067 P4D 123ec8067 PUD 123ec4067 PMD 100e38067 PTE 0<br />
Oops: Oops: 0000 [#1] PREEMPT SMP KASAN NOPTI<br />
CPU: 5 UID: 0 PID: 755 Comm: modprobe Not tainted 6.12.0-rc3+ #17<br />
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS<br />
1.16.1-2.fc37 04/01/2014<br />
RIP: 0010:__asan_load8+0x4c/0xa0<br />
...<br />
Call Trace:<br />
<br />
blkdev_put_whole+0x41/0x70<br />
bdev_release+0x1a3/0x250<br />
blkdev_release+0x11/0x20<br />
__fput+0x1d7/0x4a0<br />
task_work_run+0xfc/0x180<br />
syscall_exit_to_user_mode+0x1de/0x1f0<br />
do_syscall_64+0x6b/0x170<br />
entry_SYSCALL_64_after_hwframe+0x76/0x7e<br />
<br />
loop_init() is calling loop_add() after __register_blkdev() succeeds and<br />
is ignoring disk_add() failure from loop_add(), for loop_add() failure<br />
is not fatal and successfully created disks are already visible to<br />
bdev_open().<br />
<br />
brd_init() is currently calling brd_alloc() before __register_blkdev()<br />
succeeds and is releasing successfully created disks when brd_init()<br />
returns an error. This can cause UAF for the latter two case:<br />
<br />
case 1:<br />
T1:<br />
modprobe brd<br />
brd_init<br />
brd_alloc(0) // success<br />
add_disk<br />
disk_scan_partitions<br />
bdev_file_open_by_dev // alloc file<br />
fput // won&#39;t free until back to userspace<br />
brd_alloc(1) // failed since mem alloc error inject<br />
// error path for modprobe will release code segment<br />
// back to userspace<br />
__fput<br />
blkdev_release<br />
bdev_release<br />
blkdev_put_whole<br />
bdev->bd_disk->fops->release // fops is freed now, UAF!<br />
<br />
case 2:<br />
T1: T2:<br />
modprobe brd<br />
brd_init<br />
brd_alloc(0) // success<br />
open(/dev/ram0)<br />
brd_alloc(1) // fail<br />
// error path for modprobe<br />
<br />
close(/dev/ram0)<br />
...<br />
/* UAF! */<br />
bdev->bd_disk->fops->release<br />
<br />
Fix this problem by following what loop_init() does. Besides,<br />
reintroduce brd_devices_mutex to help serialize modifications to<br />
brd_list.
Impact
References to Advisories, Solutions, and Tools
- https://git.kernel.org/stable/c/259bf925583ec9e3781df778cadf00594095090d
- https://git.kernel.org/stable/c/410896624db639500f24f46478b4bfa05c76bf56
- https://git.kernel.org/stable/c/41219c147df8bbd6591f59af5d695fb6c9a1cbff
- https://git.kernel.org/stable/c/63dfd728b30f79495dacc886127695a379805152
- https://git.kernel.org/stable/c/826cc42adf44930a633d11a5993676d85ddb0842
- https://git.kernel.org/stable/c/c0c2744cd2939ec5999c51dbaf2af16886548b7b