[Bug 215347] New: btintel: AX200 (8087:0029): hci crashed and can't recover after repeated rfkill on/off

https://bugzilla.kernel.org/show_bug.cgi?id=215347

            Bug ID: 215347
           Summary: btintel: AX200 (8087:0029): hci crashed and can't
                    recover after repeated rfkill on/off
           Product: Drivers
           Version: 2.5
    Kernel Version: 5.16-rc5
          Hardware: Intel
                OS: Linux
              Tree: Mainline
            Status: NEW
          Severity: blocking
          Priority: P1
         Component: Bluetooth
          Assignee: linux-bluetooth@vger.kernel.org
          Reporter: hui.wang@canonical.com
        Regression: No

Created attachment 300051
  --> https://bugzilla.kernel.org/attachment.cgi?id=300051&action=edit
testing script

We have a couple of Lenovo machines which have Bluetooth hci (8087:0029), If we
run a script like below:
for a in {1..100}
do
    echo $a
    rfkill block $HCI0_ID
    sleep 1
    rfkill unblock $HCI0_ID
    sleep 1
done

The kernel will print "hci0: Failed to read MSFT supported features (-110)"
after several round of rfkill block/unblock, and the bluetooth driver can't
work anymore from that moment. Sometimes 'rmmod btusb;modprobe btusb" could
make the bluetooth driver work again, sometimes need to reboot.

We already updated the kernel and linux-firmware to the latest, but still could
reproduce above issue.

And we run the same test on other machines with different Bluetooth hci like
8087:0026, they all work well with the testing script.

The 8087:0029 hci loads the intel/ibt-20-1-3.sfi, we suspect this is an issue
on the bluetooth firmware ibt-20-1-3.sfi, since 8087:0026 shares the same
kernel driver with 0029, but only 0029 has this issue.


And the testing script is attached.

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are the assignee for the bug.