Re: [PATCH v2] Bluetooth: Shutdown controller after workqueues are flushed or cancelled

From: Kai-Heng Feng
Date: Tue Aug 03 2021 - 02:45:30 EST


On Mon, Aug 2, 2021 at 11:05 AM Hillf Danton <hdanton@xxxxxxxx> wrote:
>
> On Fri, 14 May 2021 15:14:52 +0800 Kai-Heng Feng wrote:
> > Rfkill block and unblock Intel USB Bluetooth [8087:0026] may make it
> > stops working:
> > [ 509.691509] Bluetooth: hci0: HCI reset during shutdown failed
> > [ 514.897584] Bluetooth: hci0: MSFT filter_enable is already on
> > [ 530.044751] usb 3-10: reset full-speed USB device number 5 using xhci_hcd
> > [ 545.660350] usb 3-10: device descriptor read/64, error -110
> > [ 561.283530] usb 3-10: device descriptor read/64, error -110
> > [ 561.519682] usb 3-10: reset full-speed USB device number 5 using xhci_hcd
> > [ 566.686650] Bluetooth: hci0: unexpected event for opcode 0x0500
> > [ 568.752452] Bluetooth: hci0: urb 0000000096cd309b failed to resubmit (113)
> > [ 578.797955] Bluetooth: hci0: Failed to read MSFT supported features (-110)
> > [ 586.286565] Bluetooth: hci0: urb 00000000c522f633 failed to resubmit (113)
> > [ 596.215302] Bluetooth: hci0: Failed to read MSFT supported features (-110)
> >
> > Or kernel panics because other workqueues already freed skb:
> > [ 2048.663763] BUG: kernel NULL pointer dereference, address: 0000000000000000
> > [ 2048.663775] #PF: supervisor read access in kernel mode
> > [ 2048.663779] #PF: error_code(0x0000) - not-present page
> > [ 2048.663782] PGD 0 P4D 0
> > [ 2048.663787] Oops: 0000 [#1] SMP NOPTI
> > [ 2048.663793] CPU: 3 PID: 4491 Comm: rfkill Tainted: G W 5.13.0-rc1-next-20210510+ #20
> > [ 2048.663799] Hardware name: HP HP EliteBook 850 G8 Notebook PC/8846, BIOS T76 Ver. 01.01.04 12/02/2020
> > [ 2048.663801] RIP: 0010:__skb_ext_put+0x6/0x50
> > [ 2048.663814] Code: 8b 1b 48 85 db 75 db 5b 41 5c 5d c3 be 01 00 00 00 e8 de 13 c0 ff eb e7 be 02 00 00 00 e8 d2 13 c0 ff eb db 0f 1f 44 00 00 55 <8b> 07 48 89 e5 83 f8 01 74 14 b8 ff ff ff ff f0 0f c1
> > 07 83 f8 01
> > [ 2048.663819] RSP: 0018:ffffc1d105b6fd80 EFLAGS: 00010286
> > [ 2048.663824] RAX: 0000000000000000 RBX: ffff9d9ac5649000 RCX: 0000000000000000
> > [ 2048.663827] RDX: ffffffffc0d1daf6 RSI: 0000000000000206 RDI: 0000000000000000
> > [ 2048.663830] RBP: ffffc1d105b6fd98 R08: 0000000000000001 R09: ffff9d9ace8ceac0
> > [ 2048.663834] R10: ffff9d9ace8ceac0 R11: 0000000000000001 R12: ffff9d9ac5649000
> > [ 2048.663838] R13: 0000000000000000 R14: 00007ffe0354d650 R15: 0000000000000000
> > [ 2048.663843] FS: 00007fe02ab19740(0000) GS:ffff9d9e5f8c0000(0000) knlGS:0000000000000000
> > [ 2048.663849] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> > [ 2048.663853] CR2: 0000000000000000 CR3: 0000000111a52004 CR4: 0000000000770ee0
> > [ 2048.663856] PKRU: 55555554
> > [ 2048.663859] Call Trace:
> > [ 2048.663865] ? skb_release_head_state+0x5e/0x80
> > [ 2048.663873] kfree_skb+0x2f/0xb0
> > [ 2048.663881] btusb_shutdown_intel_new+0x36/0x60 [btusb]
> > [ 2048.663905] hci_dev_do_close+0x48c/0x5e0 [bluetooth]
> > [ 2048.663954] ? __cond_resched+0x1a/0x50
> > [ 2048.663962] hci_rfkill_set_block+0x56/0xa0 [bluetooth]
> > [ 2048.664007] rfkill_set_block+0x98/0x170
> > [ 2048.664016] rfkill_fop_write+0x136/0x1e0
> > [ 2048.664022] vfs_write+0xc7/0x260
> > [ 2048.664030] ksys_write+0xb1/0xe0
> > [ 2048.664035] ? exit_to_user_mode_prepare+0x37/0x1c0
> > [ 2048.664042] __x64_sys_write+0x1a/0x20
> > [ 2048.664048] do_syscall_64+0x40/0xb0
> > [ 2048.664055] entry_SYSCALL_64_after_hwframe+0x44/0xae
> > [ 2048.664060] RIP: 0033:0x7fe02ac23c27
> > [ 2048.664066] Code: 0d 00 f7 d8 64 89 02 48 c7 c0 ff ff ff ff eb b7 0f 1f 00 f3 0f 1e fa 64 8b 04 25 18 00 00 00 85 c0 75 10 b8 01 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 51 c3 48 83 ec 28 48 89 54 24 18 48 89 74 24
> > [ 2048.664070] RSP: 002b:00007ffe0354d638 EFLAGS: 00000246 ORIG_RAX: 0000000000000001
> > [ 2048.664075] RAX: ffffffffffffffda RBX: 0000000000000001 RCX: 00007fe02ac23c27
> > [ 2048.664078] RDX: 0000000000000008 RSI: 00007ffe0354d650 RDI: 0000000000000003
> > [ 2048.664081] RBP: 0000000000000000 R08: 0000559b05998440 R09: 0000559b05998440
> > [ 2048.664084] R10: 0000000000000000 R11: 0000000000000246 R12: 0000000000000003
> > [ 2048.664086] R13: 0000000000000000 R14: ffffffff00000000 R15: 00000000ffffffff
> >
> > So move the shutdown callback to a place where workqueues are either
> > flushed or cancelled to resolve the issue.
> >
> > Signed-off-by: Kai-Heng Feng <kai.heng.feng@xxxxxxxxxxxxx>
> > ---
> > v2:
> > - Rebased on bluetooth-next.
> >
> > net/bluetooth/hci_core.c | 16 ++++++++--------
> > 1 file changed, 8 insertions(+), 8 deletions(-)
> >
> > diff --git a/net/bluetooth/hci_core.c b/net/bluetooth/hci_core.c
> > index 7baf93eda936..6eedf334f943 100644
> > --- a/net/bluetooth/hci_core.c
> > +++ b/net/bluetooth/hci_core.c
> > @@ -1716,14 +1716,6 @@ int hci_dev_do_close(struct hci_dev *hdev)
> >
> > BT_DBG("%s %p", hdev->name, hdev);
> >
> > - if (!hci_dev_test_flag(hdev, HCI_UNREGISTER) &&
> > - !hci_dev_test_flag(hdev, HCI_USER_CHANNEL) &&
> > - test_bit(HCI_UP, &hdev->flags)) {
> > - /* Execute vendor specific shutdown routine */
> > - if (hdev->shutdown)
> > - hdev->shutdown(hdev);
> > - }
> > -
> > cancel_delayed_work(&hdev->power_off);
> > cancel_delayed_work(&hdev->ncmd_timer);
> >
> > @@ -1801,6 +1793,14 @@ int hci_dev_do_close(struct hci_dev *hdev)
> > clear_bit(HCI_INIT, &hdev->flags);
> > }
> >
> > + if (!hci_dev_test_flag(hdev, HCI_UNREGISTER) &&
> > + !hci_dev_test_flag(hdev, HCI_USER_CHANNEL) &&
> > + test_bit(HCI_UP, &hdev->flags)) {
> > + /* Execute vendor specific shutdown routine */
> > + if (hdev->shutdown)
> > + hdev->shutdown(hdev);
> > + }
> > +
> > /* flush cmd work */
> > flush_work(&hdev->cmd_work);
> >
> > --
> > 2.30.2
>
> btusb_shutdown_intel_new
> __hci_cmd_sync(hdev, HCI_OP_RESET, 0, NULL, HCI_INIT_TIMEOUT);
> __hci_cmd_sync_ev(hdev, opcode, plen, param, 0, timeout);
> hci_req_run_skb(&req, hci_req_sync_complete);
>
> hci_req_sync_complete
> if (skb)
> hdev->req_skb = skb_get(skb);
>
> Given the skb_get in hci_req_sync_complete makes it safe to free skb on
> driver side, I doubt this patch is the correct fix as it is.

Some workqueues are still active.
The shutdown() should be called at least after hci_request_cancel_all().

Kai-Heng