Hi everybody,
I updated openSuse Tumbleweed 20230806 with a bluetooth keyboard
attached to the PC. After the update to 20230808 and rebooting bluetooth
was disabled and it was impossible to enable it again. Before the update
bluetooth was working fine.
The attached bluetooth keyboard caused the bluetooth module being added
to initrd. However, the corresponding firmware for MT7922 is not added.
dmesg shows:
[ 4.368031] bluetooth hci0: Direct firmware load for
mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin failed with error -2
In fact the file is present on the system
ls -l /usr/lib/firmware/mediatek/BT_RAM_CODE*
-rw-r--r-- 1 root root 512104 3. Aug 17:36
/usr/lib/firmware/mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin.xz
-rw-r--r-- 1 root root 343052 3. Aug 17:36
/usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
but missing in initrd
lsinitrd | grep BT_RAM_CODE
-rw-r--r-- 1 root root 343052 Aug 3 17:36
usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
Manually adding the file to initrd makes bluetooth work again.
I reported the bug here
https://bugzilla.suse.com/show_bug.cgi?id=1214133
and was requested to report here.
If I can be of any help or you need further information please let me know.
Thanks and kind regards
Jürgen
[Cc: +Qu, +Chris]
Dear Jürgen,
Am 10.08.23 um 22:24 schrieb Jürgen Hofmann:
> I updated openSuse Tumbleweed 20230806 with a bluetooth keyboard
> attached to the PC. After the update to 20230808 and rebooting bluetooth
> was disabled and it was impossible to enable it again. Before the update
> bluetooth was working fine.
Sorry, I do not know what software versions changed updating to openSUSE
Tumbleweed. Could you please mention that for the Linux kernel and BlueZ
and the initrd generator?
> The attached bluetooth keyboard caused the bluetooth module being added
> to initrd. However, the corresponding firmware for MT7922 is not added.
>
> dmesg shows:
>
> [ 4.368031] bluetooth hci0: Direct firmware load for mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin failed with error -2
>
> In fact the file is present on the system
>
> ls -l /usr/lib/firmware/mediatek/BT_RAM_CODE*
> -rw-r--r-- 1 root root 512104 3. Aug 17:36 /usr/lib/firmware/mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin.xz
> -rw-r--r-- 1 root root 343052 3. Aug 17:36 /usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
>
> but missing in initrd
>
> lsinitrd | grep BT_RAM_CODE
> -rw-r--r-- 1 root root 343052 Aug 3 17:36 usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
>
> Manually adding the file to initrd makes bluetooth work again.
>
> I reported the bug here
> https://bugzilla.suse.com/show_bug.cgi?id=1214133
> and was requested to report here.
>
> If I can be of any help or you need further information please let me
> know.
From Qu’s answer in the thread *[PATCH v3 1/2] Bluetooth: btusb: Add
new VID/PID 0489/e102 for MT7922* [1] it sounds to me, support for chip
was only added recently, and is going to be in Linux v6.6.
Kind regards,
Paul
[1]:
https://lore.kernel.org/linux-bluetooth/[email protected]/
On Sat, 12 Aug 2023 11:02:04 +0200,
Paul Menzel wrote:
>
> [Cc: +Qu, +Chris]
>
> Dear Jürgen,
>
>
> Am 10.08.23 um 22:24 schrieb Jürgen Hofmann:
>
> > I updated openSuse Tumbleweed 20230806 with a bluetooth keyboard
> > attached to the PC. After the update to 20230808 and rebooting
> > bluetooth was disabled and it was impossible to enable it
> > again. Before the update bluetooth was working fine.
>
> Sorry, I do not know what software versions changed updating to
> openSUSE Tumbleweed. Could you please mention that for the Linux
> kernel and BlueZ and the initrd generator?
>
> > The attached bluetooth keyboard caused the bluetooth module being
> > added to initrd. However, the corresponding firmware for MT7922 is
> > not added.
> >
> > dmesg shows:
> >
> > [ 4.368031] bluetooth hci0: Direct firmware load for mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin failed with error -2
> >
> > In fact the file is present on the system
> >
> > ls -l /usr/lib/firmware/mediatek/BT_RAM_CODE*
> > -rw-r--r-- 1 root root 512104 3. Aug 17:36 /usr/lib/firmware/mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin.xz
> > -rw-r--r-- 1 root root 343052 3. Aug 17:36 /usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
> >
> > but missing in initrd
> >
> > lsinitrd | grep BT_RAM_CODE
> > -rw-r--r-- 1 root root 343052 Aug 3 17:36 usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
> >
> > Manually adding the file to initrd makes bluetooth work again.
> >
> > I reported the bug here
> > https://bugzilla.suse.com/show_bug.cgi?id=1214133
> > and was requested to report here.
> >
> > If I can be of any help or you need further information please let me
> > know.
> From Qu’s answer in the thread *[PATCH v3 1/2] Bluetooth: btusb: Add
> new VID/PID 0489/e102 for MT7922* [1] it sounds to me, support for
> chip was only added recently, and is going to be in Linux v6.6.
The problem is rather the lack of MODULE_FIRMWARE() declarations for
the corresponding device. The device itself already works fine as is
with the current upstream kernel as long as the firmware file is
present on the system. But as dracut puts only the firmware file
listed in MODULE_FIRMWARE() into initrd, it starts failing once when
you build an initrd containing the BT.
thanks,
Takashi
On 2023/8/12 17:02, Paul Menzel wrote:
> [Cc: +Qu, +Chris]
>
> Dear Jürgen,
>
>
> Am 10.08.23 um 22:24 schrieb Jürgen Hofmann:
>
>> I updated openSuse Tumbleweed 20230806 with a bluetooth keyboard
>> attached to the PC. After the update to 20230808 and rebooting
>> bluetooth was disabled and it was impossible to enable it again.
>> Before the update bluetooth was working fine.
>
> Sorry, I do not know what software versions changed updating to openSUSE
> Tumbleweed. Could you please mention that for the Linux kernel and BlueZ
> and the initrd generator?
>
>> The attached bluetooth keyboard caused the bluetooth module being
>> added to initrd. However, the corresponding firmware for MT7922 is not
>> added.
>>
>> dmesg shows:
>>
>> [ 4.368031] bluetooth hci0: Direct firmware load for
>> mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin failed with error -2
>>
>> In fact the file is present on the system
>>
>> ls -l /usr/lib/firmware/mediatek/BT_RAM_CODE*
>> -rw-r--r-- 1 root root 512104 3. Aug 17:36
>> /usr/lib/firmware/mediatek/BT_RAM_CODE_MT7922_1_1_hdr.bin.xz
>> -rw-r--r-- 1 root root 343052 3. Aug 17:36
>> /usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
>>
>> but missing in initrd
>>
>> lsinitrd | grep BT_RAM_CODE
>> -rw-r--r-- 1 root root 343052 Aug 3 17:36
>> usr/lib/firmware/mediatek/BT_RAM_CODE_MT7961_1_2_hdr.bin.xz
>>
>> Manually adding the file to initrd makes bluetooth work again.
>>
>> I reported the bug here
>> https://bugzilla.suse.com/show_bug.cgi?id=1214133
>> and was requested to report here.
>>
>> If I can be of any help or you need further information please let me
>> know.
> From Qu’s answer in the thread *[PATCH v3 1/2] Bluetooth: btusb: Add
> new VID/PID 0489/e102 for MT7922* [1] it sounds to me, support for chip
> was only added recently, and is going to be in Linux v6.6.
I think this is a different one.
The new pid (0xe102) is only introduced in some recent laptops (with
ryzen 7040 series CPUs I guess?).
Unless the user is compiling a custom kernel with the extra vid/pid
added, it should not work from the very beginning.
Thus I believe it's something different.
Thanks,
Qu
>
>
> Kind regards,
>
> Paul
>
>
> [1]:
> https://lore.kernel.org/linux-bluetooth/[email protected]/
Dear Paul, dear Qu,
>> From Qu’s answer in the thread *[PATCH v3 1/2] Bluetooth: btusb: Add
>> new VID/PID 0489/e102 for MT7922* [1] it sounds to me, support for
>> chip was only added recently, and is going to be in Linux v6.6.
>
> The problem is rather the lack of MODULE_FIRMWARE() declarations for
> the corresponding device. The device itself already works fine as is
> with the current upstream kernel as long as the firmware file is
> present on the system. But as dracut puts only the firmware file
> listed in MODULE_FIRMWARE() into initrd, it starts failing once when
> you build an initrd containing the BT.
from all I can tell (and understand), the analysis from Takashi about
the missing MODULE_FIRMWARE() declarations is correct.
Thanks,
Jürgen