[Bug 218264] New: Potential kernel regression with bluetooth pairing on specific radios

[Bug 218264] New: Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

            Bug ID: 218264
           Summary: Potential kernel regression with bluetooth pairing on
                    specific radios
           Product: Drivers
           Version: 2.5
          Hardware: All
                OS: Linux
            Status: NEW
          Severity: normal
          Priority: P3
         Component: Bluetooth
          Assignee: linux-bluetooth@vger.kernel.org
          Reporter: marco.rodolfi@tuta.io
        Regression: No

I had a Logitech K480 paired with my Steam Deck (RTL8822CE) from a previous
successful pair on Linux 6.4.x/6.5.x, not sure on the specific version. Since
Fedora updated to 6.6.x and I accidentally lost the pairing key I can't in no
way be able to associate. It connects fine but just fails with
bluez.authenticationfailure. I've opened a bug report in the bluez issues page
https://github.com/bluez/bluez/issues/654 and after testing on another computer
with a different bluetooth radio (RTL8771) + kernel 6.6.5 and bluez 5.70 the
keyboard connected immediately and requiring me to insert the passcode
correctly. I'm writing this from this specific keyboard.

I'm still not sure if it's the kernel driver fault or bluez fault.

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

The Linux kernel's regression tracker (Thorsten Leemhuis) (regressions@leemhuis.info) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |regressions@leemhuis.info

--- Comment #1 from The Linux kernel's regression tracker (Thorsten Leemhuis) (regressions@leemhuis.info) ---
(In reply to Marco from comment #0)

> I'm still not sure if it's the kernel driver fault or bluez fault.

To check that I suggest you install a 6.5.y Fedora kernel again and check if it
works there; if it does boot a 6.6.y kernel again and recheck if things still
are broken there.

Old kernels can be found here:
https://koji.fedoraproject.org/koji/packageinfo?packageID=8 , like these ones
for f38 (https://koji.fedoraproject.org/koji/buildinfo?buildID=2322803 ) and
f39 (https://koji.fedoraproject.org/koji/buildinfo?buildID=2322801 )

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

--- Comment #2 from Marco (marco.rodolfi@tuta.io) ---
tested now with a live usb with 6.4.15 + bluez 5.69 and it works perfectly
fine, I'll try to find a live usb with 6.5 and test again. To me it seems more
and more a kernel regression now.

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

--- Comment #3 from Marco (marco.rodolfi@tuta.io) ---
6.5.6 + bluez 5.69 works 100% fine. I would reduce the commit space further
down, but since I'm on a Deck + on an immutable distro (Bazzite which is Fedora
Kinoite 39 based) bisecting is a pain.

I stole the pairing keys made now from the live USB and I'll place them into
the corresponding bluez folders, but to me and from what I've read around on
the net this seems a kernel regressione introduced in the 6.6 cycle.

If I could reproduce this on my workstation it would not be a problem to
bisect, sadly it works fine on there.

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

--- Comment #4 from Marco (marco.rodolfi@tuta.io) ---
Oddly even the copied keys make it fails on connect with
br-connection-create-socket.

I have no clue what is screwed up at this point, sadly even this workaround is
not doable.

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

--- Comment #5 from Marco (marco.rodolfi@tuta.io) ---
Today I managed to pair, finally, but I do not get any input from the keyboard.
in dmesg I do get this:

[ 6539.432356] Bluetooth: HIDP (Human Interface Emulation) ver 1.2
[ 6539.432370] Bluetooth: HIDP socket layer initialized
[ 6673.460129] hid-generic 0005:046D:B33D.0006: unknown main item tag 0x0
[ 6673.460148] hid-generic 0005:046D:B33D.0006: item fetching failed at offset
271/272
[ 6673.460171] hid-generic: probe of 0005:046D:B33D.0006 failed with error -22
[ 6706.511559] hid-generic 0005:046D:B33D.0007: unknown main item tag 0x0
[ 6706.511586] hid-generic 0005:046D:B33D.0007: item fetching failed at offset
271/272
[ 6706.511625] hid-generic: probe of 0005:046D:B33D.0007 failed with error -22
[ 6741.248981] hid-generic 0005:046D:B33D.0008: unknown main item tag 0x0
[ 6741.248997] hid-generic 0005:046D:B33D.0008: item fetching failed at offset
271/272
[ 6741.249017] hid-generic: probe of 0005:046D:B33D.0008 failed with error -22

Bluetooth regression AND hid regression or this can be caused only by bluetooth
driver shenanigans?

Marco

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

--- Comment #6 from The Linux kernel's regression tracker (Thorsten Leemhuis) (regressions@leemhuis.info) ---
(In reply to Marco from comment #5)

> Bluetooth regression AND hid regression or this can be caused only by
> bluetooth driver shenanigans?

Hard to tell. But I fear that no developer will look into this unless you
perform a bisection. But before you do so you likely want to test mainline and
a working kernel on the same installation (e.g. not with live linux or anything
where more than the kernel changed)

Side note, FWIW: I have no idea if the bluetooth devs are even listening here.
To udnerstand why, search for "bugzilla" in
https://docs.kernel.org/admin-guide/reporting-issues.html and
https://linux-regtracking.leemhuis.info/post/frequent-reasons-why-linux-kernel-bug-reports-are-ignored/

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

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

[Bug 218264] Potential kernel regression with bluetooth pairing on specific radios

[bugzilla-daemon]

https://bugzilla.kernel.org/show_bug.cgi?id=218264

Marco (marco.rodolfi@tuta.io) changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|---                         |INVALID

--- Comment #7 from Marco (marco.rodolfi@tuta.io) ---
Everything seem to be back working on the latest builds of Bluez, so this might
just be a Bluez and not a kernel regression after all.

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

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