Hello,
syzbot found the following issue on:
HEAD commit: 4b0986a3613c Linux 5.18
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=125ba355f00000
kernel config: https://syzkaller.appspot.com/x/.config?x=1350d397b63b3036
dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1732a04df00000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=15189639f00000
The issue was bisected to:
commit 781d2a9a2fc7d0be53a072794dc03ef6de770f3d
Author: Jan Kara <[email protected]>
Date: Mon May 3 09:39:03 2021 +0000
udf: Check LVID earlier
bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=14deecd3f00000
final oops: https://syzkaller.appspot.com/x/report.txt?x=16deecd3f00000
console output: https://syzkaller.appspot.com/x/log.txt?x=12deecd3f00000
IMPORTANT: if you fix the issue, please add the following tag to the commit:
Reported-by: [email protected]
Fixes: 781d2a9a2fc7 ("udf: Check LVID earlier")
UDF-fs: error (device loop0): udf_fill_super: Error in udf_iget, block=96, partition=0
==================================================================
BUG: KASAN: use-after-free in udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
Write of size 1 at addr ffff8880839e0190 by task syz-executor234/3615
CPU: 1 PID: 3615 Comm: syz-executor234 Not tainted 5.18.0-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
<TASK>
__dump_stack lib/dump_stack.c:88 [inline]
dump_stack_lvl+0x1e3/0x2cb lib/dump_stack.c:106
print_address_description+0x65/0x4b0 mm/kasan/report.c:313
print_report+0xf4/0x210 mm/kasan/report.c:429
kasan_report+0xfb/0x130 mm/kasan/report.c:491
udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
udf_fill_super+0xde8/0x1b20 fs/udf/super.c:2309
mount_bdev+0x26c/0x3a0 fs/super.c:1367
legacy_get_tree+0xea/0x180 fs/fs_context.c:610
vfs_get_tree+0x88/0x270 fs/super.c:1497
do_new_mount+0x289/0xad0 fs/namespace.c:3040
do_mount fs/namespace.c:3383 [inline]
__do_sys_mount fs/namespace.c:3591 [inline]
__se_sys_mount+0x2e3/0x3d0 fs/namespace.c:3568
do_syscall_x64 arch/x86/entry/common.c:50 [inline]
do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80
entry_SYSCALL_64_after_hwframe+0x44/0xae
RIP: 0033:0x7fd64e59b08a
Code: 48 c7 c2 b8 ff ff ff f7 d8 64 89 02 b8 ff ff ff ff eb d2 e8 a8 00 00 00 0f 1f 84 00 00 00 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
RSP: 002b:00007fd64e546168 EFLAGS: 00000286 ORIG_RAX: 00000000000000a5
RAX: ffffffffffffffda RBX: 00007fd64e5461c0 RCX: 00007fd64e59b08a
RDX: 0000000020000000 RSI: 0000000020000700 RDI: 00007fd64e546180
RBP: 000000000000000e R08: 00007fd64e5461c0 R09: 00007fd64e5466b8
R10: 0000000000000810 R11: 0000000000000286 R12: 00007fd64e546180
R13: 0000000020000350 R14: 0000000000000003 R15: 0000000000000004
</TASK>
The buggy address belongs to the physical page:
page:ffffea00020e7800 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x839e0
flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff)
raw: 00fff00000000000 ffffea00020e7808 ffffea00020e7808 0000000000000000
raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
page dumped because: kasan: bad access detected
page_owner info is not present (never set?)
Memory state around the buggy address:
ffff8880839e0080: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
ffff8880839e0100: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
>ffff8880839e0180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
^
ffff8880839e0200: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
ffff8880839e0280: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
==================================================================
---
This report is generated by a bot. It may contain errors.
See https://goo.gl/tpsmEJ for more information about syzbot.
syzbot engineers can be reached at [email protected].
syzbot will keep track of this issue. See:
https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
For information about bisection process see: https://goo.gl/tpsmEJ#bisection
syzbot can test patches for this issue, for details see:
https://goo.gl/tpsmEJ#testing-patches
Hello,
syzbot tried to test the proposed patch but the build/boot failed:
fs/udf/super.c:113:7: error: no member named 'impUse' in 'logicalVolIntegrityDesc'
fs/udf/super.c:121:57: error: no member named 'impUse' in 'struct logicalVolIntegrityDesc'
Tested on:
commit: 143a6252 Merge tag 'arm64-upstream' of git://git.kerne..
git tree: upstream
dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
patch: https://syzkaller.appspot.com/x/patch.diff?x=15ff0355f00000
Hello,
syzbot has tested the proposed patch but the reproducer is still triggering an issue:
INFO: rcu detected stall in corrupted
rcu: INFO: rcu_preempt detected expedited stalls on CPUs/tasks: { P4150 } 4 jiffies s: 2565 root: 0x0/T
rcu: blocking rcu_node structures (internal RCU debug):
IPv6: ADDRCONF(NETDEV_CHANGE): team0: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): team_slave_1: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): hsr_slave_0: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): hsr_slave_1: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): hsr0: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): veth0_virt_wifi: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): veth0_vlan: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): vlan0: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): vlan1: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): batadv_slave_1: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): veth1_to_batadv: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): wlan0: link becomes ready
IPv6: ADDRCONF(NETDEV_CHANGE): wlan1: link becomes ready
Bluetooth: hci1: command 0x0409 tx timeout
Tested on:
commit: 143a6252 Merge tag 'arm64-upstream' of git://git.kerne..
git tree: upstream
console output: https://syzkaller.appspot.com/x/log.txt?x=147795e5f00000
kernel config: https://syzkaller.appspot.com/x/.config?x=dc73592fcf812ebf
dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
patch: https://syzkaller.appspot.com/x/patch.diff?x=147d0ad6f00000
Hello!
I had a look into this bug and I actually think this reproducer program does
something that is always going to be problematic. The reproducer has:
#{"threaded":true,"repeat":true,"procs":6,"slowdown":1,"sandbox":"","close_fds":false}
syz_mount_image$udf(...)
r0 = syz_open_dev$loop(&(0x7f0000000080), 0x0, 0x109002)
mmap(addr, 0x600000, PROT_WRITE, MAP_SHARED | MAP_FIXED, r0, 0x0)
pipe(addr)
So the reproducer effectively corrupts random loop devices with output from
pipe(2) syscall while there are filesystems mounted on them by other
threads. This is a guaranteed way to shoot yourself in the foot and crash
the kernel. You can say the kernel should not allow writing to mounted devices
and I'd generally agree but there are some cornercases (e.g. bootloaders or
lowlevel filesystem tools) which happen to do this so we cannot forbid that due
to compatibility reasons. So syzbot probably needs to implement some
internal logic not to futz with loop devices that are currently mounted.
Honza
On Mon 23-05-22 17:17:21, syzbot wrote:
> Hello,
>
> syzbot found the following issue on:
>
> HEAD commit: 4b0986a3613c Linux 5.18
> git tree: upstream
> console output: https://syzkaller.appspot.com/x/log.txt?x=125ba355f00000
> kernel config: https://syzkaller.appspot.com/x/.config?x=1350d397b63b3036
> dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
> compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1732a04df00000
> C reproducer: https://syzkaller.appspot.com/x/repro.c?x=15189639f00000
>
> The issue was bisected to:
>
> commit 781d2a9a2fc7d0be53a072794dc03ef6de770f3d
> Author: Jan Kara <[email protected]>
> Date: Mon May 3 09:39:03 2021 +0000
>
> udf: Check LVID earlier
>
> bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=14deecd3f00000
> final oops: https://syzkaller.appspot.com/x/report.txt?x=16deecd3f00000
> console output: https://syzkaller.appspot.com/x/log.txt?x=12deecd3f00000
>
> IMPORTANT: if you fix the issue, please add the following tag to the commit:
> Reported-by: [email protected]
> Fixes: 781d2a9a2fc7 ("udf: Check LVID earlier")
>
> UDF-fs: error (device loop0): udf_fill_super: Error in udf_iget, block=96, partition=0
> ==================================================================
> BUG: KASAN: use-after-free in udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> Write of size 1 at addr ffff8880839e0190 by task syz-executor234/3615
>
> CPU: 1 PID: 3615 Comm: syz-executor234 Not tainted 5.18.0-syzkaller #0
> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> Call Trace:
> <TASK>
> __dump_stack lib/dump_stack.c:88 [inline]
> dump_stack_lvl+0x1e3/0x2cb lib/dump_stack.c:106
> print_address_description+0x65/0x4b0 mm/kasan/report.c:313
> print_report+0xf4/0x210 mm/kasan/report.c:429
> kasan_report+0xfb/0x130 mm/kasan/report.c:491
> udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> udf_fill_super+0xde8/0x1b20 fs/udf/super.c:2309
> mount_bdev+0x26c/0x3a0 fs/super.c:1367
> legacy_get_tree+0xea/0x180 fs/fs_context.c:610
> vfs_get_tree+0x88/0x270 fs/super.c:1497
> do_new_mount+0x289/0xad0 fs/namespace.c:3040
> do_mount fs/namespace.c:3383 [inline]
> __do_sys_mount fs/namespace.c:3591 [inline]
> __se_sys_mount+0x2e3/0x3d0 fs/namespace.c:3568
> do_syscall_x64 arch/x86/entry/common.c:50 [inline]
> do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80
> entry_SYSCALL_64_after_hwframe+0x44/0xae
> RIP: 0033:0x7fd64e59b08a
> Code: 48 c7 c2 b8 ff ff ff f7 d8 64 89 02 b8 ff ff ff ff eb d2 e8 a8 00 00 00 0f 1f 84 00 00 00 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
> RSP: 002b:00007fd64e546168 EFLAGS: 00000286 ORIG_RAX: 00000000000000a5
> RAX: ffffffffffffffda RBX: 00007fd64e5461c0 RCX: 00007fd64e59b08a
> RDX: 0000000020000000 RSI: 0000000020000700 RDI: 00007fd64e546180
> RBP: 000000000000000e R08: 00007fd64e5461c0 R09: 00007fd64e5466b8
> R10: 0000000000000810 R11: 0000000000000286 R12: 00007fd64e546180
> R13: 0000000020000350 R14: 0000000000000003 R15: 0000000000000004
> </TASK>
>
> The buggy address belongs to the physical page:
> page:ffffea00020e7800 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x839e0
> flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff)
> raw: 00fff00000000000 ffffea00020e7808 ffffea00020e7808 0000000000000000
> raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
> page dumped because: kasan: bad access detected
> page_owner info is not present (never set?)
>
> Memory state around the buggy address:
> ffff8880839e0080: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> ffff8880839e0100: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> >ffff8880839e0180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> ^
> ffff8880839e0200: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> ffff8880839e0280: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> ==================================================================
>
>
> ---
> This report is generated by a bot. It may contain errors.
> See https://goo.gl/tpsmEJ for more information about syzbot.
> syzbot engineers can be reached at [email protected].
>
> syzbot will keep track of this issue. See:
> https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
> For information about bisection process see: https://goo.gl/tpsmEJ#bisection
> syzbot can test patches for this issue, for details see:
> https://goo.gl/tpsmEJ#testing-patches
--
Jan Kara <[email protected]>
SUSE Labs, CR
On Tuesday 24 May 2022 12:32:46 Jan Kara wrote:
>
> Hello!
>
> I had a look into this bug and I actually think this reproducer program does
> something that is always going to be problematic. The reproducer has:
>
> #{"threaded":true,"repeat":true,"procs":6,"slowdown":1,"sandbox":"","close_fds":false}
>
> syz_mount_image$udf(...)
> r0 = syz_open_dev$loop(&(0x7f0000000080), 0x0, 0x109002)
> mmap(addr, 0x600000, PROT_WRITE, MAP_SHARED | MAP_FIXED, r0, 0x0)
> pipe(addr)
>
> So the reproducer effectively corrupts random loop devices with output from
> pipe(2) syscall while there are filesystems mounted on them by other
> threads. This is a guaranteed way to shoot yourself in the foot and crash
> the kernel. You can say the kernel should not allow writing to mounted devices
> and I'd generally agree but there are some cornercases (e.g. bootloaders or
> lowlevel filesystem tools) which happen to do this so we cannot forbid that due
> to compatibility reasons. So syzbot probably needs to implement some
> internal logic not to futz with loop devices that are currently mounted.
>
> Honza
Hello! Bootloaders and other similar software in most cases needs to
update first sector or sectors with bootloader data or main superblock
of filesystem (where are metadata like label or UUIDs)... In most cases
these "write" operations do not touch filesystem data.
So I'm thinking if it would not make sense to add some kernel config
option to disallow write operations on mounted block devices, with some
filesystem hook/callback which can allow writing to specific block.
E.g. UDF filesystem does not use first 32kB of disk and userspace
software can overwrite it as it wants even when fs is mounted, without
crashing kernel. So that udf hook/callback would allow write access to
this area.
Userspace applications always invent "smart" things and I think it is a
good idea to protect kernel if it is possible.
I understand that there is need to overwrite mounted block device.
Updating bootloader stored at the beginning of the rootfs disk is
important operation. Also changing filesystem label at runtime / mounted
fs is something which users want and it is legitimate requirement. For
UDF this change is not easy operation and userspace software (e.g.
udflabel) needs to update lot of blocks on device, which can really
break mounted udf fs.
>
> On Mon 23-05-22 17:17:21, syzbot wrote:
> > Hello,
> >
> > syzbot found the following issue on:
> >
> > HEAD commit: 4b0986a3613c Linux 5.18
> > git tree: upstream
> > console output: https://syzkaller.appspot.com/x/log.txt?x=125ba355f00000
> > kernel config: https://syzkaller.appspot.com/x/.config?x=1350d397b63b3036
> > dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
> > compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
> > syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1732a04df00000
> > C reproducer: https://syzkaller.appspot.com/x/repro.c?x=15189639f00000
> >
> > The issue was bisected to:
> >
> > commit 781d2a9a2fc7d0be53a072794dc03ef6de770f3d
> > Author: Jan Kara <[email protected]>
> > Date: Mon May 3 09:39:03 2021 +0000
> >
> > udf: Check LVID earlier
> >
> > bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=14deecd3f00000
> > final oops: https://syzkaller.appspot.com/x/report.txt?x=16deecd3f00000
> > console output: https://syzkaller.appspot.com/x/log.txt?x=12deecd3f00000
> >
> > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > Reported-by: [email protected]
> > Fixes: 781d2a9a2fc7 ("udf: Check LVID earlier")
> >
> > UDF-fs: error (device loop0): udf_fill_super: Error in udf_iget, block=96, partition=0
> > ==================================================================
> > BUG: KASAN: use-after-free in udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> > Write of size 1 at addr ffff8880839e0190 by task syz-executor234/3615
> >
> > CPU: 1 PID: 3615 Comm: syz-executor234 Not tainted 5.18.0-syzkaller #0
> > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> > Call Trace:
> > <TASK>
> > __dump_stack lib/dump_stack.c:88 [inline]
> > dump_stack_lvl+0x1e3/0x2cb lib/dump_stack.c:106
> > print_address_description+0x65/0x4b0 mm/kasan/report.c:313
> > print_report+0xf4/0x210 mm/kasan/report.c:429
> > kasan_report+0xfb/0x130 mm/kasan/report.c:491
> > udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> > udf_fill_super+0xde8/0x1b20 fs/udf/super.c:2309
> > mount_bdev+0x26c/0x3a0 fs/super.c:1367
> > legacy_get_tree+0xea/0x180 fs/fs_context.c:610
> > vfs_get_tree+0x88/0x270 fs/super.c:1497
> > do_new_mount+0x289/0xad0 fs/namespace.c:3040
> > do_mount fs/namespace.c:3383 [inline]
> > __do_sys_mount fs/namespace.c:3591 [inline]
> > __se_sys_mount+0x2e3/0x3d0 fs/namespace.c:3568
> > do_syscall_x64 arch/x86/entry/common.c:50 [inline]
> > do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80
> > entry_SYSCALL_64_after_hwframe+0x44/0xae
> > RIP: 0033:0x7fd64e59b08a
> > Code: 48 c7 c2 b8 ff ff ff f7 d8 64 89 02 b8 ff ff ff ff eb d2 e8 a8 00 00 00 0f 1f 84 00 00 00 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
> > RSP: 002b:00007fd64e546168 EFLAGS: 00000286 ORIG_RAX: 00000000000000a5
> > RAX: ffffffffffffffda RBX: 00007fd64e5461c0 RCX: 00007fd64e59b08a
> > RDX: 0000000020000000 RSI: 0000000020000700 RDI: 00007fd64e546180
> > RBP: 000000000000000e R08: 00007fd64e5461c0 R09: 00007fd64e5466b8
> > R10: 0000000000000810 R11: 0000000000000286 R12: 00007fd64e546180
> > R13: 0000000020000350 R14: 0000000000000003 R15: 0000000000000004
> > </TASK>
> >
> > The buggy address belongs to the physical page:
> > page:ffffea00020e7800 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x839e0
> > flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff)
> > raw: 00fff00000000000 ffffea00020e7808 ffffea00020e7808 0000000000000000
> > raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
> > page dumped because: kasan: bad access detected
> > page_owner info is not present (never set?)
> >
> > Memory state around the buggy address:
> > ffff8880839e0080: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > ffff8880839e0100: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > >ffff8880839e0180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > ^
> > ffff8880839e0200: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > ffff8880839e0280: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > ==================================================================
> >
> >
> > ---
> > This report is generated by a bot. It may contain errors.
> > See https://goo.gl/tpsmEJ for more information about syzbot.
> > syzbot engineers can be reached at [email protected].
> >
> > syzbot will keep track of this issue. See:
> > https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
> > For information about bisection process see: https://goo.gl/tpsmEJ#bisection
> > syzbot can test patches for this issue, for details see:
> > https://goo.gl/tpsmEJ#testing-patches
> --
> Jan Kara <[email protected]>
> SUSE Labs, CR
On Tue 24-05-22 12:46:58, Pali Roh?r wrote:
> On Tuesday 24 May 2022 12:32:46 Jan Kara wrote:
> >
> > Hello!
> >
> > I had a look into this bug and I actually think this reproducer program does
> > something that is always going to be problematic. The reproducer has:
> >
> > #{"threaded":true,"repeat":true,"procs":6,"slowdown":1,"sandbox":"","close_fds":false}
> >
> > syz_mount_image$udf(...)
> > r0 = syz_open_dev$loop(&(0x7f0000000080), 0x0, 0x109002)
> > mmap(addr, 0x600000, PROT_WRITE, MAP_SHARED | MAP_FIXED, r0, 0x0)
> > pipe(addr)
> >
> > So the reproducer effectively corrupts random loop devices with output from
> > pipe(2) syscall while there are filesystems mounted on them by other
> > threads. This is a guaranteed way to shoot yourself in the foot and crash
> > the kernel. You can say the kernel should not allow writing to mounted devices
> > and I'd generally agree but there are some cornercases (e.g. bootloaders or
> > lowlevel filesystem tools) which happen to do this so we cannot forbid that due
> > to compatibility reasons. So syzbot probably needs to implement some
> > internal logic not to futz with loop devices that are currently mounted.
> >
> > Honza
>
> Hello! Bootloaders and other similar software in most cases needs to
> update first sector or sectors with bootloader data or main superblock
> of filesystem (where are metadata like label or UUIDs)... In most cases
> these "write" operations do not touch filesystem data.
Well, we generally try to move away from updating superblock directly by
userspace - e.g. these days we have ioctls to update label or uuid so that
coordination with the kernel is done properly. But there are still cases
where this happens.
> So I'm thinking if it would not make sense to add some kernel config
> option to disallow write operations on mounted block devices, with some
> filesystem hook/callback which can allow writing to specific block.
>
> E.g. UDF filesystem does not use first 32kB of disk and userspace
> software can overwrite it as it wants even when fs is mounted, without
> crashing kernel. So that udf hook/callback would allow write access to
> this area.
>
> Userspace applications always invent "smart" things and I think it is a
> good idea to protect kernel if it is possible.
>
> I understand that there is need to overwrite mounted block device.
> Updating bootloader stored at the beginning of the rootfs disk is
> important operation. Also changing filesystem label at runtime / mounted
> fs is something which users want and it is legitimate requirement. For
> UDF this change is not easy operation and userspace software (e.g.
> udflabel) needs to update lot of blocks on device, which can really
> break mounted udf fs.
Well, the difficulty is with identifying where writing is OK and where not.
It is not always is simple range at the beginning of the device where
writes can happen (although that probably covers majority of cases). But
thinking more about it the problem is not so much with applications
modifying disk contents (we don't trust disk contents much) but with
applications modifying buffer cache where we believe we have validated data
structures. So we could somehow disallow buffer cache modification under
mounted filesystem. Either the filesystem and app view of buffer cache
would be inconsistent or apps would be doing direct IO when the filesystem
is mounted. Either option has its consequences but maybe we could create
something working out of that.
Honza
> > On Mon 23-05-22 17:17:21, syzbot wrote:
> > > Hello,
> > >
> > > syzbot found the following issue on:
> > >
> > > HEAD commit: 4b0986a3613c Linux 5.18
> > > git tree: upstream
> > > console output: https://syzkaller.appspot.com/x/log.txt?x=125ba355f00000
> > > kernel config: https://syzkaller.appspot.com/x/.config?x=1350d397b63b3036
> > > dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
> > > compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
> > > syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1732a04df00000
> > > C reproducer: https://syzkaller.appspot.com/x/repro.c?x=15189639f00000
> > >
> > > The issue was bisected to:
> > >
> > > commit 781d2a9a2fc7d0be53a072794dc03ef6de770f3d
> > > Author: Jan Kara <[email protected]>
> > > Date: Mon May 3 09:39:03 2021 +0000
> > >
> > > udf: Check LVID earlier
> > >
> > > bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=14deecd3f00000
> > > final oops: https://syzkaller.appspot.com/x/report.txt?x=16deecd3f00000
> > > console output: https://syzkaller.appspot.com/x/log.txt?x=12deecd3f00000
> > >
> > > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > > Reported-by: [email protected]
> > > Fixes: 781d2a9a2fc7 ("udf: Check LVID earlier")
> > >
> > > UDF-fs: error (device loop0): udf_fill_super: Error in udf_iget, block=96, partition=0
> > > ==================================================================
> > > BUG: KASAN: use-after-free in udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> > > Write of size 1 at addr ffff8880839e0190 by task syz-executor234/3615
> > >
> > > CPU: 1 PID: 3615 Comm: syz-executor234 Not tainted 5.18.0-syzkaller #0
> > > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> > > Call Trace:
> > > <TASK>
> > > __dump_stack lib/dump_stack.c:88 [inline]
> > > dump_stack_lvl+0x1e3/0x2cb lib/dump_stack.c:106
> > > print_address_description+0x65/0x4b0 mm/kasan/report.c:313
> > > print_report+0xf4/0x210 mm/kasan/report.c:429
> > > kasan_report+0xfb/0x130 mm/kasan/report.c:491
> > > udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> > > udf_fill_super+0xde8/0x1b20 fs/udf/super.c:2309
> > > mount_bdev+0x26c/0x3a0 fs/super.c:1367
> > > legacy_get_tree+0xea/0x180 fs/fs_context.c:610
> > > vfs_get_tree+0x88/0x270 fs/super.c:1497
> > > do_new_mount+0x289/0xad0 fs/namespace.c:3040
> > > do_mount fs/namespace.c:3383 [inline]
> > > __do_sys_mount fs/namespace.c:3591 [inline]
> > > __se_sys_mount+0x2e3/0x3d0 fs/namespace.c:3568
> > > do_syscall_x64 arch/x86/entry/common.c:50 [inline]
> > > do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80
> > > entry_SYSCALL_64_after_hwframe+0x44/0xae
> > > RIP: 0033:0x7fd64e59b08a
> > > Code: 48 c7 c2 b8 ff ff ff f7 d8 64 89 02 b8 ff ff ff ff eb d2 e8 a8 00 00 00 0f 1f 84 00 00 00 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
> > > RSP: 002b:00007fd64e546168 EFLAGS: 00000286 ORIG_RAX: 00000000000000a5
> > > RAX: ffffffffffffffda RBX: 00007fd64e5461c0 RCX: 00007fd64e59b08a
> > > RDX: 0000000020000000 RSI: 0000000020000700 RDI: 00007fd64e546180
> > > RBP: 000000000000000e R08: 00007fd64e5461c0 R09: 00007fd64e5466b8
> > > R10: 0000000000000810 R11: 0000000000000286 R12: 00007fd64e546180
> > > R13: 0000000020000350 R14: 0000000000000003 R15: 0000000000000004
> > > </TASK>
> > >
> > > The buggy address belongs to the physical page:
> > > page:ffffea00020e7800 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x839e0
> > > flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff)
> > > raw: 00fff00000000000 ffffea00020e7808 ffffea00020e7808 0000000000000000
> > > raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
> > > page dumped because: kasan: bad access detected
> > > page_owner info is not present (never set?)
> > >
> > > Memory state around the buggy address:
> > > ffff8880839e0080: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > ffff8880839e0100: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > >ffff8880839e0180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > ^
> > > ffff8880839e0200: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > ffff8880839e0280: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > ==================================================================
> > >
> > >
> > > ---
> > > This report is generated by a bot. It may contain errors.
> > > See https://goo.gl/tpsmEJ for more information about syzbot.
> > > syzbot engineers can be reached at [email protected].
> > >
> > > syzbot will keep track of this issue. See:
> > > https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
> > > For information about bisection process see: https://goo.gl/tpsmEJ#bisection
> > > syzbot can test patches for this issue, for details see:
> > > https://goo.gl/tpsmEJ#testing-patches
> > --
> > Jan Kara <[email protected]>
> > SUSE Labs, CR
--
Jan Kara <[email protected]>
SUSE Labs, CR
On Tue, 24 May 2022 at 13:37, Jan Kara <[email protected]> wrote:
>
> On Tue 24-05-22 12:46:58, Pali Rohár wrote:
> > On Tuesday 24 May 2022 12:32:46 Jan Kara wrote:
> > >
> > > Hello!
> > >
> > > I had a look into this bug and I actually think this reproducer program does
> > > something that is always going to be problematic. The reproducer has:
> > >
> > > #{"threaded":true,"repeat":true,"procs":6,"slowdown":1,"sandbox":"","close_fds":false}
> > >
> > > syz_mount_image$udf(...)
> > > r0 = syz_open_dev$loop(&(0x7f0000000080), 0x0, 0x109002)
> > > mmap(addr, 0x600000, PROT_WRITE, MAP_SHARED | MAP_FIXED, r0, 0x0)
> > > pipe(addr)
> > >
> > > So the reproducer effectively corrupts random loop devices with output from
> > > pipe(2) syscall while there are filesystems mounted on them by other
> > > threads. This is a guaranteed way to shoot yourself in the foot and crash
> > > the kernel. You can say the kernel should not allow writing to mounted devices
> > > and I'd generally agree but there are some cornercases (e.g. bootloaders or
> > > lowlevel filesystem tools) which happen to do this so we cannot forbid that due
> > > to compatibility reasons. So syzbot probably needs to implement some
> > > internal logic not to futz with loop devices that are currently mounted.
> > >
> > > Honza
> >
> > Hello! Bootloaders and other similar software in most cases needs to
> > update first sector or sectors with bootloader data or main superblock
> > of filesystem (where are metadata like label or UUIDs)... In most cases
> > these "write" operations do not touch filesystem data.
>
> Well, we generally try to move away from updating superblock directly by
> userspace - e.g. these days we have ioctls to update label or uuid so that
> coordination with the kernel is done properly. But there are still cases
> where this happens.
>
> > So I'm thinking if it would not make sense to add some kernel config
> > option to disallow write operations on mounted block devices, with some
> > filesystem hook/callback which can allow writing to specific block.
> >
> > E.g. UDF filesystem does not use first 32kB of disk and userspace
> > software can overwrite it as it wants even when fs is mounted, without
> > crashing kernel. So that udf hook/callback would allow write access to
> > this area.
> >
> > Userspace applications always invent "smart" things and I think it is a
> > good idea to protect kernel if it is possible.
> >
> > I understand that there is need to overwrite mounted block device.
> > Updating bootloader stored at the beginning of the rootfs disk is
> > important operation. Also changing filesystem label at runtime / mounted
> > fs is something which users want and it is legitimate requirement. For
> > UDF this change is not easy operation and userspace software (e.g.
> > udflabel) needs to update lot of blocks on device, which can really
> > break mounted udf fs.
>
> Well, the difficulty is with identifying where writing is OK and where not.
> It is not always is simple range at the beginning of the device where
> writes can happen (although that probably covers majority of cases). But
> thinking more about it the problem is not so much with applications
> modifying disk contents (we don't trust disk contents much) but with
> applications modifying buffer cache where we believe we have validated data
> structures. So we could somehow disallow buffer cache modification under
> mounted filesystem. Either the filesystem and app view of buffer cache
> would be inconsistent or apps would be doing direct IO when the filesystem
> is mounted. Either option has its consequences but maybe we could create
> something working out of that.
Did not know it's not OK to write loop-associated fd.
It's quite hard to prohibit the fuzzer from doing this. It's possible
to prohibit whole syscalls, or opening particular files, or executing
particular ioctl commands. But there are lots of interesting ways how
an fd can be shared/messed with.
We could prohibit opening /dev/loop*, or associating loop with fd, but
it will inhibit all useful loop testing. And will also prohibit
testing mounting of all filesystems.
It will be good if the kernel provides some config that prohibits such
unsafe operation. Or, of course, if it prohibits only "buffer cache
modification".
> > > On Mon 23-05-22 17:17:21, syzbot wrote:
> > > > Hello,
> > > >
> > > > syzbot found the following issue on:
> > > >
> > > > HEAD commit: 4b0986a3613c Linux 5.18
> > > > git tree: upstream
> > > > console output: https://syzkaller.appspot.com/x/log.txt?x=125ba355f00000
> > > > kernel config: https://syzkaller.appspot.com/x/.config?x=1350d397b63b3036
> > > > dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
> > > > compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
> > > > syz repro: https://syzkaller.appspot.com/x/repro.syz?x=1732a04df00000
> > > > C reproducer: https://syzkaller.appspot.com/x/repro.c?x=15189639f00000
> > > >
> > > > The issue was bisected to:
> > > >
> > > > commit 781d2a9a2fc7d0be53a072794dc03ef6de770f3d
> > > > Author: Jan Kara <[email protected]>
> > > > Date: Mon May 3 09:39:03 2021 +0000
> > > >
> > > > udf: Check LVID earlier
> > > >
> > > > bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=14deecd3f00000
> > > > final oops: https://syzkaller.appspot.com/x/report.txt?x=16deecd3f00000
> > > > console output: https://syzkaller.appspot.com/x/log.txt?x=12deecd3f00000
> > > >
> > > > IMPORTANT: if you fix the issue, please add the following tag to the commit:
> > > > Reported-by: [email protected]
> > > > Fixes: 781d2a9a2fc7 ("udf: Check LVID earlier")
> > > >
> > > > UDF-fs: error (device loop0): udf_fill_super: Error in udf_iget, block=96, partition=0
> > > > ==================================================================
> > > > BUG: KASAN: use-after-free in udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> > > > Write of size 1 at addr ffff8880839e0190 by task syz-executor234/3615
> > > >
> > > > CPU: 1 PID: 3615 Comm: syz-executor234 Not tainted 5.18.0-syzkaller #0
> > > > Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
> > > > Call Trace:
> > > > <TASK>
> > > > __dump_stack lib/dump_stack.c:88 [inline]
> > > > dump_stack_lvl+0x1e3/0x2cb lib/dump_stack.c:106
> > > > print_address_description+0x65/0x4b0 mm/kasan/report.c:313
> > > > print_report+0xf4/0x210 mm/kasan/report.c:429
> > > > kasan_report+0xfb/0x130 mm/kasan/report.c:491
> > > > udf_close_lvid+0x68a/0x980 fs/udf/super.c:2072
> > > > udf_fill_super+0xde8/0x1b20 fs/udf/super.c:2309
> > > > mount_bdev+0x26c/0x3a0 fs/super.c:1367
> > > > legacy_get_tree+0xea/0x180 fs/fs_context.c:610
> > > > vfs_get_tree+0x88/0x270 fs/super.c:1497
> > > > do_new_mount+0x289/0xad0 fs/namespace.c:3040
> > > > do_mount fs/namespace.c:3383 [inline]
> > > > __do_sys_mount fs/namespace.c:3591 [inline]
> > > > __se_sys_mount+0x2e3/0x3d0 fs/namespace.c:3568
> > > > do_syscall_x64 arch/x86/entry/common.c:50 [inline]
> > > > do_syscall_64+0x2b/0x70 arch/x86/entry/common.c:80
> > > > entry_SYSCALL_64_after_hwframe+0x44/0xae
> > > > RIP: 0033:0x7fd64e59b08a
> > > > Code: 48 c7 c2 b8 ff ff ff f7 d8 64 89 02 b8 ff ff ff ff eb d2 e8 a8 00 00 00 0f 1f 84 00 00 00 00 00 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 c7 c1 b8 ff ff ff f7 d8 64 89 01 48
> > > > RSP: 002b:00007fd64e546168 EFLAGS: 00000286 ORIG_RAX: 00000000000000a5
> > > > RAX: ffffffffffffffda RBX: 00007fd64e5461c0 RCX: 00007fd64e59b08a
> > > > RDX: 0000000020000000 RSI: 0000000020000700 RDI: 00007fd64e546180
> > > > RBP: 000000000000000e R08: 00007fd64e5461c0 R09: 00007fd64e5466b8
> > > > R10: 0000000000000810 R11: 0000000000000286 R12: 00007fd64e546180
> > > > R13: 0000000020000350 R14: 0000000000000003 R15: 0000000000000004
> > > > </TASK>
> > > >
> > > > The buggy address belongs to the physical page:
> > > > page:ffffea00020e7800 refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x839e0
> > > > flags: 0xfff00000000000(node=0|zone=1|lastcpupid=0x7ff)
> > > > raw: 00fff00000000000 ffffea00020e7808 ffffea00020e7808 0000000000000000
> > > > raw: 0000000000000000 0000000000000000 00000000ffffffff 0000000000000000
> > > > page dumped because: kasan: bad access detected
> > > > page_owner info is not present (never set?)
> > > >
> > > > Memory state around the buggy address:
> > > > ffff8880839e0080: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > > ffff8880839e0100: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > > >ffff8880839e0180: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > > ^
> > > > ffff8880839e0200: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > > ffff8880839e0280: ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff ff
> > > > ==================================================================
> > > >
> > > >
> > > > ---
> > > > This report is generated by a bot. It may contain errors.
> > > > See https://goo.gl/tpsmEJ for more information about syzbot.
> > > > syzbot engineers can be reached at [email protected].
> > > >
> > > > syzbot will keep track of this issue. See:
> > > > https://goo.gl/tpsmEJ#status for how to communicate with syzbot.
> > > > For information about bisection process see: https://goo.gl/tpsmEJ#bisection
> > > > syzbot can test patches for this issue, for details see:
> > > > https://goo.gl/tpsmEJ#testing-patches
> > > --
> > > Jan Kara <[email protected]>
> > > SUSE Labs, CR
> --
> Jan Kara <[email protected]>
> SUSE Labs, CR
>
> --
> You received this message because you are subscribed to the Google Groups "syzkaller-bugs" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
> To view this discussion on the web visit https://groups.google.com/d/msgid/syzkaller-bugs/20220524113704.gdwvao43b23hyf7z%40quack3.lan.
Hello,
syzbot has tested the proposed patch and the reproducer did not trigger any issue:
Reported-and-tested-by: [email protected]
Tested on:
commit: 97fa5887 USB: new quirk for Dell Gen 2 devices
git tree: https://git.kernel.org/pub/scm/linux/kernel/git/gregkh/usb.git
kernel config: https://syzkaller.appspot.com/x/.config?x=cacd11c8fed2980d
dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
compiler: Debian clang version 13.0.1-++20220126092033+75e33f71c2da-1~exp1~20220126212112.63, GNU ld (GNU Binutils for Debian) 2.35.2
patch: https://syzkaller.appspot.com/x/patch.diff?x=1373055bf00000
Note: testing is done by a robot and is best-effort only.
syzbot suspects this issue was fixed by commit:
commit 6f861765464f43a71462d52026fbddfc858239a5
Author: Jan Kara <[email protected]>
Date: Wed Nov 1 17:43:10 2023 +0000
fs: Block writes to mounted block devices
bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=12c941c2180000
start commit: f016f7547aee Merge tag 'gpio-fixes-for-v6.7-rc8' of git://..
git tree: upstream
kernel config: https://syzkaller.appspot.com/x/.config?x=f8e72bae38c079e4
dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
syz repro: https://syzkaller.appspot.com/x/repro.syz?x=13c2caf9e80000
C reproducer: https://syzkaller.appspot.com/x/repro.c?x=11f518f9e80000
If the result looks correct, please mark the issue as fixed by replying with:
#syz fix: fs: Block writes to mounted block devices
For information about bisection process see: https://goo.gl/tpsmEJ#bisection
On Sun 18-02-24 10:20:04, syzbot wrote:
> syzbot suspects this issue was fixed by commit:
>
> commit 6f861765464f43a71462d52026fbddfc858239a5
> Author: Jan Kara <[email protected]>
> Date: Wed Nov 1 17:43:10 2023 +0000
>
> fs: Block writes to mounted block devices
>
> bisection log: https://syzkaller.appspot.com/x/bisect.txt?x=12c941c2180000
> start commit: f016f7547aee Merge tag 'gpio-fixes-for-v6.7-rc8' of git://..
> git tree: upstream
> kernel config: https://syzkaller.appspot.com/x/.config?x=f8e72bae38c079e4
> dashboard link: https://syzkaller.appspot.com/bug?extid=60864ed35b1073540d57
> syz repro: https://syzkaller.appspot.com/x/repro.syz?x=13c2caf9e80000
> C reproducer: https://syzkaller.appspot.com/x/repro.c?x=11f518f9e80000
>
> If the result looks correct, please mark the issue as fixed by replying with:
Yes, this was expected :)
#syz fix: fs: Block writes to mounted block devices
Honza
--
Jan Kara <[email protected]>
SUSE Labs, CR