Re: [RFC 5/5] Bluetooth: hci_sync: Make use of hci_cmd_sync_queue set 3

[Luiz Augusto von Dentz <luiz.dentz@gmail.com>]

Hi Tedd,

On Tue, Jun 1, 2021 at 1:24 PM Tedd Ho-Jeong An <tedd.an@linux.intel.com> wrote:
>
> Hi Luiz,
>
> On Thu, 2021-05-27 at 17:01 -0700, Luiz Augusto von Dentz wrote:
> > From: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
> >
> > This make use of hci_cmd_sync_queue for the following MGMT commands:
> >
> >     Add Device
> >     Remove Device
> >
> > Tested with:
> >
> > mgmt-tester -s "Add Device"
> >
> > Test Summary
> > ------------
> > Add Device - Invalid Params 1                        Passed       0.017 seconds
> > Add Device - Invalid Params 2                        Passed       0.013 seconds
> > Add Device - Invalid Params 3                        Passed       0.013 seconds
> > Add Device - Invalid Params 4                        Passed       0.013 seconds
> > Add Device - Success 1                               Passed       0.014 seconds
> > Add Device - Success 2                               Passed       0.014 seconds
> > Add Device - Success 3                               Passed       0.014 seconds
> > Add Device - Success 4                               Passed       0.017 seconds
> > Add Device - Success 5                               Passed       0.017 seconds
> > Total: 9, Passed: 9 (100.0%), Failed: 0, Not Run: 0
> > Overall execution time: 0.14 seconds
> >
> > mgmt-tester -s "Remove Device"
> >
> > Test Summary
> > ------------
> > Remove Device - Invalid Params 1                     Passed       0.153 seconds
> > Remove Device - Invalid Params 2                     Passed       0.014 seconds
> > Remove Device - Invalid Params 3                     Passed       0.013 seconds
> > Remove Device - Success 1                            Passed       0.016 seconds
> > Remove Device - Success 2                            Passed       0.017 seconds
> > Remove Device - Success 3                            Passed       1.022 seconds
> > Remove Device - Success 4                            Passed       1.021 seconds
> > Remove Device - Success 5                            Passed       1.022 seconds
> > Total: 8, Passed: 8 (100.0%), Failed: 0, Not Run: 0
> > Overall execution time: 3.29 seconds
> >
> > Signed-off-by: Luiz Augusto von Dentz <luiz.von.dentz@intel.com>
> > ---
> >  net/bluetooth/hci_sync.c | 606 ++++++++++++++++++++++++++++++++++++++-
> >  net/bluetooth/hci_sync.h |   2 +
> >  net/bluetooth/mgmt.c     |  19 +-
> >  3 files changed, 622 insertions(+), 5 deletions(-)
> >
>
> While running new test cases for checking LL Privacy (submitted the series to mailing list),
> some test cases caused the kernel oops:
>
> general protection fault, probably for non-canonical address 0xdead000000000116: 0000 [#1] PTI
> CPU: 0 PID: 113 Comm: kworker/u3:2 Not tainted 5.12.0-g01861ba6bbe9-dirty #11
> Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.13.0-2.fc32 04/01/2014
> Workqueue: hci0 hci_cmd_sync_work
> RIP: 0010:hci_passive_scan_sync.part.0+0xed/0x820
> Code: 7c 24 13 00 75 12 48 8b 85 00 10 00 00 48 0f ba e0 29 0f 83 97 02 00 00 80 44 24 1e 01 4d 8b 3f 4c 39 3c 24 0f 84 25 01 00 00 <41> 0f b6 57 16 4d 8d 67 10 4c 89 ef 4c 89 e6 e8 2f 95 fb ff 41 0f
> RSP: 0018:ffffad9400187dc8 EFLAGS: 00010202
> RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
> RDX: ffff8d0a01850ca8 RSI: ffff8d0a0186a210 RDI: ffff8d0a01850000
> RBP: ffff8d0a01850000 R08: ffff8d0a01803ae6 R09: 0000000000004ffb
> R10: 0000000078563412 R11: 3fffffffffffffff R12: ffff8d0a0186a210
> R13: ffff8d0a01850cf8 R14: ffff8d0a01850d08 R15: dead000000000100
> FS:  0000000000000000(0000) GS:ffffffff87846000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000558641956130 CR3: 00000000018a2000 CR4: 00000000000006f0
> Call Trace:
>  ? unblock_device+0xe0/0xe0
>  hci_update_background_scan_sync+0x268/0x310
>  hci_cmd_sync_work+0x91/0xe0
>  process_one_work+0x19d/0x2f0
>  worker_thread+0x5a/0x3b0
>  ? rescuer_thread+0x330/0x330
>  kthread+0x108/0x120
>  ? __kthread_create_worker+0xf0/0xf0
>  ret_from_fork+0x22/0x30
> ---[ end trace efd7eab9e13c521e ]---
> RIP: 0010:hci_passive_scan_sync.part.0+0xed/0x820
> Code: 7c 24 13 00 75 12 48 8b 85 00 10 00 00 48 0f ba e0 29 0f 83 97 02 00 00 80 44 24 1e 01 4d 8b 3f 4c 39 3c 24 0f 84 25 01 00 00 <41> 0f b6 57 16 4d 8d 67 10 4c 89 ef 4c 89 e6 e8 2f 95 fb ff 41 0f
> RSP: 0018:ffffad9400187dc8 EFLAGS: 00010202
> RAX: 0000000000000000 RBX: 0000000000000000 RCX: 0000000000000000
> RDX: ffff8d0a01850ca8 RSI: ffff8d0a0186a210 RDI: ffff8d0a01850000
> RBP: ffff8d0a01850000 R08: ffff8d0a01803ae6 R09: 0000000000004ffb
> R10: 0000000078563412 R11: 3fffffffffffffff R12: ffff8d0a0186a210
> R13: ffff8d0a01850cf8 R14: ffff8d0a01850d08 R15: dead000000000100
> FS:  0000000000000000(0000) GS:ffffffff87846000(0000) knlGS:0000000000000000
> CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> CR2: 0000558641956130 CR3: 00000000018a2000 CR4: 00000000000006f0

Remove Device - Success 7 - Remove from whitelist - setup complete
Remove Device - Success 7 - Remove from whitelist - run
  Registering Device Removed notification
  Test condition added, total 1
  Registering HCI command callback
  Test condition added, total 2
  Sending Remove Device (0x0034)
  Test condition added, total 3
  Remove Device (0x0034): Success (0x00)
  Test condition complete, 2 left
  New Device Removed event received
  Test condition complete, 1 left
  HCI Command 0x2042 length 6
[   58.818744] ==================================================================
[   58.819253] BUG: KASAN: use-after-free in
hci_passive_scan_sync.part.0+0x52e/0xaf0
[   58.819833] Read of size 6 at addr ffff8880019bca10 by task kworker/u3:0/49
[   58.820305]
[   58.820584]
[   58.820708] Allocated by task 92:
[   58.820979]
[   58.821092] Freed by task 93:
[   58.821345]
[   58.821489] The buggy address belongs to the object at ffff8880019bca00
[   58.821489]  which belongs to the cache kmalloc-32 of size 32
[   58.822488] The buggy address is located 16 bytes inside of
[   58.822488]  32-byte region [ffff8880019bca00, ffff8880019bca20)
[   58.823363] The buggy address belongs to the page:
[   58.823773]
[   58.823881] Memory state around the buggy address:
[   58.824215]  ffff8880019bc900: fb fb fb fb fc fc fc fc fa fb fb fb
fc fc fc fc
[   58.824771]  ffff8880019bc980: 00 00 00 00 fc fc fc fc fb fb fb fb
fc fc fc fc
[   58.825275] >ffff8880019bca00: fa fb fb fb fc fc fc fc fa fb fb fb
fc fc fc fc
[   58.825884]                          ^
[   58.826169]  ffff8880019bca80: fb fb fb fb fc fc fc fc fb fb fb fb
fc fc fc fc
[   58.826790]  ffff8880019bcb00: fb fb fb fb fc fc fc fc fb fb fb fb
fc fc fc fc
[   58.827330] ==================================================================
  HCI Command 0x2012 length 7
  Test condition complete, 0 left

Btw, it is a good idea to enable KSAN when testing, Im afraid this
might be related to hdev_lock.

>
> However, it is not seen on the current bluetooth-next tree.
>
> Regards,
>
> Tedd
>


-- 
Luiz Augusto von Dentz