On Wed, Mar 7, 2018 at 8:23 AM, Leon Romanovsky <l...@kernel.org> wrote:
> On Tue, Mar 06, 2018 at 10:59:02PM -0800, syzbot wrote:
>> Hello,
>> syzbot hit the following crash on upstream commit
>> ce380619fab99036f5e745c7a865b21c59f005f6 (Tue Mar 6 04:31:14 2018 +0000)
>> Merge tag 'please-pull-ia64_misc' of
>> git://git.kernel.org/pub/scm/linux/kernel/git/aegl/linux
>> So far this crash happened 52 times on upstream.
>> C reproducer is attached.
>> syzkaller reproducer is attached.
>> Raw console output is attached.
>> compiler: gcc (GCC) 7.1.1 20170620
>> .config is attached.
>> IMPORTANT: if you fix the bug, please add the following tag to the commit:
>> Reported-by: syzbot+a38b0e9f694c379ca...@syzkaller.appspotmail.com
>> It will help syzbot understand when the bug is fixed. See footer for
>> details.
>> If you forward the report, please keep this part and the footer.
>> audit: type=1400 audit(1520367364.281:6): avc:  denied  { map } for
>> pid=4138 comm="bash" path="/bin/bash" dev="sda1" ino=1457
>> scontext=unconfined_u:system_r:insmod_t:s0-s0:c0.c1023
>> tcontext=system_u:object_r:file_t:s0 tclass=file permissive=1
>> audit: type=1400 audit(1520367370.605:7): avc:  denied  { map } for
>> pid=4152 comm="syzkaller100190" path="/root/syzkaller100190328" dev="sda1"
>> ino=16481 scontext=unconfined_u:system_r:insmod_t:s0-s0:c0.c1023
>> tcontext=unconfined_u:object_r:user_home_t:s0 tclass=file permissive=1
>> WARNING: CPU: 0 PID: 4152 at mm/slab_common.c:1012 kmalloc_slab+0x5d/0x70
>> mm/slab_common.c:1012
>> Kernel panic - not syncing: panic_on_warn set ...
>> CPU: 0 PID: 4152 Comm: syzkaller100190 Not tainted 4.16.0-rc4+ #343
>> Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS
>> Google 01/01/2011
>> Call Trace:
>>  __dump_stack lib/dump_stack.c:17 [inline]
>>  dump_stack+0x194/0x24d lib/dump_stack.c:53
>>  panic+0x1e4/0x41c kernel/panic.c:183
>>  __warn+0x1dc/0x200 kernel/panic.c:547
>>  report_bug+0x211/0x2d0 lib/bug.c:184
>>  fixup_bug.part.11+0x37/0x80 arch/x86/kernel/traps.c:178
>>  fixup_bug arch/x86/kernel/traps.c:247 [inline]
>>  do_error_trap+0x2d7/0x3e0 arch/x86/kernel/traps.c:296
>>  do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:315
>>  invalid_op+0x1b/0x40 arch/x86/entry/entry_64.S:986
>> RIP: 0010:kmalloc_slab+0x5d/0x70 mm/slab_common.c:1012
>> RSP: 0018:ffff8801bf76f970 EFLAGS: 00010246
>> RAX: 0000000000000000 RBX: fffffffffffffff4 RCX: ffffffff819733cb
>> RDX: ffffffff8423372f RSI: 0000000000000000 RDI: 000000003efef4b4
>> RBP: ffff8801bf76f970 R08: 0000000000000000 R09: 0000000000000000
>> R10: ffffffff88613380 R11: 0000000000000000 R12: 000000003efef4b4
>> R13: 0000000020000080 R14: 00000000014200c0 R15: ffff8801bf76fa68
>>  __do_kmalloc mm/slab.c:3700 [inline]
>>  __kmalloc_track_caller+0x21/0x760 mm/slab.c:3720
>>  memdup_user+0x2c/0x90 mm/util.c:160
>>  ucma_set_option+0x11f/0x4d0 drivers/infiniband/core/ucma.c:1297
>>  ucma_write+0x2d6/0x3d0 drivers/infiniband/core/ucma.c:1627
>>  __vfs_write+0xef/0x970 fs/read_write.c:480
>>  vfs_write+0x189/0x510 fs/read_write.c:544
>>  SYSC_write fs/read_write.c:589 [inline]
>>  SyS_write+0xef/0x220 fs/read_write.c:581
>>  do_syscall_64+0x281/0x940 arch/x86/entry/common.c:287
>>  entry_SYSCALL_64_after_hwframe+0x42/0xb7
>> RIP: 0033:0x43fe69
>> RSP: 002b:00007ffe099a6388 EFLAGS: 00000217 ORIG_RAX: 0000000000000001
>> RAX: ffffffffffffffda RBX: 00000000004002c8 RCX: 000000000043fe69
>> RDX: 000000000000006b RSI: 00000000200000c0 RDI: 0000000000000003
>> RBP: 00000000006ca018 R08: 00000000004002c8 R09: 00000000004002c8
>> R10: 00000000004002c8 R11: 0000000000000217 R12: 0000000000401790
>> R13: 0000000000401820 R14: 0000000000000000 R15: 0000000000000000
>> Dumping ftrace buffer:
>>    (ftrace buffer empty)
>> Kernel Offset: disabled
>> Rebooting in 86400 seconds..
> I'm surprised that it surfed only now.
> It is clear bug, user's input wasn't checked.

This is very simple. syzkaller did not test rdma_cm before. Just
yesterday I added descriptions for /dev/infiniband/rdma_cm API:
This gave me ~10 different crashes immediately, but syzkaller wasn't
able to progress too far because for now all VMs crash on these
previous bugs within seconds.

I am pretty sure syzkaller still does not test lots of other
rdma-related things, but there is no reason to believe that they
contain fewer bugs (like these easily exploitable bugs on
world-writable device).
In order to teach syzkaller to test other rdma stuff one needs to add
descriptions similar to the one above.

> But it is not clear to me why optval wasn't declared as u64.

After deciphering the API (headers and sources really) I came to
conclusion that this is a pointer declared as u64 so that compat
interface is not different from normal one.

This API deciphering is hard for somebody who has absolutely no idea
what's rdma whatsoever. So syzkaller descriptions not written by you
(rdma developers) tend to be low quality, e.g. one needs to figure out
that rdma_ucm_destroy_id.id needs to be an id of rdma_cm context of
mcast group id depending on command that that's previously returned in
this and that fields of responses to this and that commands. And if
this is messed up, syzkaller won't be able to chain meaningful syscall

Reply via email to