[PATCH v2] Revert "Bluetooth: btusb: mediatek: reset the controller before downloading the fw"

[PATCH v2] Revert "Bluetooth: btusb: mediatek: reset the controller before downloading the fw"

[Geoffrey D. Bennett]

This reverts commit ccfc8948d7e4d93cab341a99774b24586717d89a.

The MediaTek Bluetooth controller reset that was added increases the
Bluetooth device setup time from ~200ms to ~20s and interferes with
other devices on the bus.

Three users (with Focusrite Scarlett 2nd Gen 6i6 and 3rd Gen Solo and
4i4 audio interfaces) reported that since 6.11 (which added this
commit) their audio interface fails to initialise if connected during
boot. Two of the users confirmed they have an MT7922.

Errors like this are observed in dmesg for the audio interface:

  usb 3-4: parse_audio_format_rates_v2v3(): unable to find clock source (clock -110)
  usb 3-4: uac_clock_source_is_valid(): cannot get clock validity for id 41
  usb 3-4: clock source 41 is not valid, cannot use

The problem only occurs when both devices and kernel modules are
present and loaded during system boot, so it can be worked around by
connecting the audio interface after booting.

Fixes: ccfc8948d7e4 ("Bluetooth: btusb: mediatek: reset the controller before downloading the fw")
Closes: https://github.com/geoffreybennett/linux-fcp/issues/24
Bisected-by: Benedikt Ziemons <ben@rs485.network>
Tested-by: Benedikt Ziemons <ben@rs485.network>
Cc: stable@vger.kernel.org
Signed-off-by: Geoffrey D. Bennett <g@b4.vu>
---
Changelog:

v1 -> v2:

- Updated commit message with additional information.
- No change to this patch's diff.
- Dropped alternate patch that only reverted for 0x7922.
- Chris, Sean, Hao agreed to reverting the change:
  https://lore.kernel.org/linux-bluetooth/2025031352-octopus-quadrant-f7ca@gregkh/T/#m0b31a9a8e87b9499e1ec3370c08f03e43bfb54bf

 drivers/bluetooth/btmtk.c | 10 ----------
 1 file changed, 10 deletions(-)

diff --git a/drivers/bluetooth/btmtk.c b/drivers/bluetooth/btmtk.c
index 68846c5bd4f7..4390fd571dbd 100644
--- a/drivers/bluetooth/btmtk.c
+++ b/drivers/bluetooth/btmtk.c
@@ -1330,13 +1330,6 @@ int btmtk_usb_setup(struct hci_dev *hdev)
 		break;
 	case 0x7922:
 	case 0x7925:
-		/* Reset the device to ensure it's in the initial state before
-		 * downloading the firmware to ensure.
-		 */
-
-		if (!test_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags))
-			btmtk_usb_subsys_reset(hdev, dev_id);
-		fallthrough;
 	case 0x7961:
 		btmtk_fw_get_filename(fw_bin_name, sizeof(fw_bin_name), dev_id,
 				      fw_version, fw_flavor);
@@ -1345,12 +1338,9 @@ int btmtk_usb_setup(struct hci_dev *hdev)
 						btmtk_usb_hci_wmt_sync);
 		if (err < 0) {
 			bt_dev_err(hdev, "Failed to set up firmware (%d)", err);
-			clear_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags);
 			return err;
 		}

-		set_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags);
-
 		/* It's Device EndPoint Reset Option Register */
 		err = btmtk_usb_uhw_reg_write(hdev, MTK_EP_RST_OPT,
 					      MTK_EP_RST_IN_OUT_OPT);
--
2.45.0

RE: [v2] Revert "Bluetooth: btusb: mediatek: reset the controller before downloading the fw"

[bluez.test.bot]

[-- Attachment #1: Type: text/plain, Size: 553 bytes --]

This is an automated email and please do not reply to this email.

Dear Submitter,

Thank you for submitting the patches to the linux bluetooth mailing list.
While preparing the CI tests, the patches you submitted couldn't be applied to the current HEAD of the repository.

----- Output -----

error: patch failed: drivers/bluetooth/btmtk.c:1330
error: drivers/bluetooth/btmtk.c: patch does not apply
hint: Use 'git am --show-current-patch' to see the failed patch

Please resolve the issue and submit the patches again.


---
Regards,
Linux Bluetooth

Re: [PATCH v2] Revert "Bluetooth: btusb: mediatek: reset the controller before downloading the fw"

[Geoffrey D. Bennett]

Hi all,

Sorry, I see that an identical patch has already been applied to
bluetooth-next
https://lore.kernel.org/linux-bluetooth/20250315022730.11071-1-hao.qin@mediatek.com/

While I'm glad the issue is being addressed, my original patch
https://lore.kernel.org/linux-bluetooth/Z8ybV04CVUfVAykH@m.b4.vu/
contained useful context and tags that didn't make it into the final
commit.

For getting this fix into current kernel releases 6.12/6.13/6.14, I
think the patch needs the "Cc: stable@vger.kernel.org" tag that was in
my original submission but missing from Hao's. Since this is causing
significant issues for users on kernels 6.11+ (audio interfaces
failing to work), it's important this gets backported.

Hao, is this something you can do? I think the instructions at
https://www.kernel.org/doc/html/v6.14/process/stable-kernel-rules.html#option-3
need to be followed, but I've not done this before.

Thanks,
Geoffrey.

On Fri, Mar 28, 2025 at 07:22:06AM +1030, Geoffrey D. Bennett wrote:
> This reverts commit ccfc8948d7e4d93cab341a99774b24586717d89a.
> 
> The MediaTek Bluetooth controller reset that was added increases the
> Bluetooth device setup time from ~200ms to ~20s and interferes with
> other devices on the bus.
> 
> Three users (with Focusrite Scarlett 2nd Gen 6i6 and 3rd Gen Solo and
> 4i4 audio interfaces) reported that since 6.11 (which added this
> commit) their audio interface fails to initialise if connected during
> boot. Two of the users confirmed they have an MT7922.
> 
> Errors like this are observed in dmesg for the audio interface:
> 
>   usb 3-4: parse_audio_format_rates_v2v3(): unable to find clock source (clock -110)
>   usb 3-4: uac_clock_source_is_valid(): cannot get clock validity for id 41
>   usb 3-4: clock source 41 is not valid, cannot use
> 
> The problem only occurs when both devices and kernel modules are
> present and loaded during system boot, so it can be worked around by
> connecting the audio interface after booting.
> 
> Fixes: ccfc8948d7e4 ("Bluetooth: btusb: mediatek: reset the controller before downloading the fw")
> Closes: https://github.com/geoffreybennett/linux-fcp/issues/24
> Bisected-by: Benedikt Ziemons <ben@rs485.network>
> Tested-by: Benedikt Ziemons <ben@rs485.network>
> Cc: stable@vger.kernel.org
> Signed-off-by: Geoffrey D. Bennett <g@b4.vu>
> ---
> Changelog:
> 
> v1 -> v2:
> 
> - Updated commit message with additional information.
> - No change to this patch's diff.
> - Dropped alternate patch that only reverted for 0x7922.
> - Chris, Sean, Hao agreed to reverting the change:
>   https://lore.kernel.org/linux-bluetooth/2025031352-octopus-quadrant-f7ca@gregkh/T/#m0b31a9a8e87b9499e1ec3370c08f03e43bfb54bf
> 
>  drivers/bluetooth/btmtk.c | 10 ----------
>  1 file changed, 10 deletions(-)
> 
> diff --git a/drivers/bluetooth/btmtk.c b/drivers/bluetooth/btmtk.c
> index 68846c5bd4f7..4390fd571dbd 100644
> --- a/drivers/bluetooth/btmtk.c
> +++ b/drivers/bluetooth/btmtk.c
> @@ -1330,13 +1330,6 @@ int btmtk_usb_setup(struct hci_dev *hdev)
>  		break;
>  	case 0x7922:
>  	case 0x7925:
> -		/* Reset the device to ensure it's in the initial state before
> -		 * downloading the firmware to ensure.
> -		 */
> -
> -		if (!test_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags))
> -			btmtk_usb_subsys_reset(hdev, dev_id);
> -		fallthrough;
>  	case 0x7961:
>  		btmtk_fw_get_filename(fw_bin_name, sizeof(fw_bin_name), dev_id,
>  				      fw_version, fw_flavor);
> @@ -1345,12 +1338,9 @@ int btmtk_usb_setup(struct hci_dev *hdev)
>  						btmtk_usb_hci_wmt_sync);
>  		if (err < 0) {
>  			bt_dev_err(hdev, "Failed to set up firmware (%d)", err);
> -			clear_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags);
>  			return err;
>  		}
> 
> -		set_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags);
> -
>  		/* It's Device EndPoint Reset Option Register */
>  		err = btmtk_usb_uhw_reg_write(hdev, MTK_EP_RST_OPT,
>  					      MTK_EP_RST_IN_OUT_OPT);
> --
> 2.45.0
> 
> 
> 

[STABLE 6.12/6.14] Bluetooth MediaTek controller reset fixes

[Geoffrey D. Bennett]

Hi stable@vger.kernel.org,

Could you please apply:

1. Commit a7208610761ae ("Bluetooth: btmtk: Remove resetting mt7921
before downloading the fw") to v6.12.x (it's already in
v6.14).

2. Commit 33634e2ab7c6 ("Bluetooth: btmtk: Remove the resetting step
before downloading the fw") to v6.12.x and v6.14.x.

These fixes address an issue with some audio interfaces failing to
initialise during boot on kernels 6.11+. As noted in my original
analysis below, the MediaTek Bluetooth controller reset increases the
device setup time from ~200ms to ~20s and can interfere with other USB
devices on the bus.

Thanks,
Geoffrey.

On Fri, Mar 28, 2025 at 08:44:49AM +1030, Geoffrey D. Bennett wrote:
> Hi all,
> 
> Sorry, I see that an identical patch has already been applied to
> bluetooth-next
> https://lore.kernel.org/linux-bluetooth/20250315022730.11071-1-hao.qin@mediatek.com/
> 
> While I'm glad the issue is being addressed, my original patch
> https://lore.kernel.org/linux-bluetooth/Z8ybV04CVUfVAykH@m.b4.vu/
> contained useful context and tags that didn't make it into the final
> commit.
> 
> For getting this fix into current kernel releases 6.12/6.13/6.14, I
> think the patch needs the "Cc: stable@vger.kernel.org" tag that was in
> my original submission but missing from Hao's. Since this is causing
> significant issues for users on kernels 6.11+ (audio interfaces
> failing to work), it's important this gets backported.
> 
> Hao, is this something you can do? I think the instructions at
> https://www.kernel.org/doc/html/v6.14/process/stable-kernel-rules.html#option-3
> need to be followed, but I've not done this before.
> 
> Thanks,
> Geoffrey.
> 
> On Fri, Mar 28, 2025 at 07:22:06AM +1030, Geoffrey D. Bennett wrote:
> > This reverts commit ccfc8948d7e4d93cab341a99774b24586717d89a.
> > 
> > The MediaTek Bluetooth controller reset that was added increases the
> > Bluetooth device setup time from ~200ms to ~20s and interferes with
> > other devices on the bus.
> > 
> > Three users (with Focusrite Scarlett 2nd Gen 6i6 and 3rd Gen Solo and
> > 4i4 audio interfaces) reported that since 6.11 (which added this
> > commit) their audio interface fails to initialise if connected during
> > boot. Two of the users confirmed they have an MT7922.
> > 
> > Errors like this are observed in dmesg for the audio interface:
> > 
> >   usb 3-4: parse_audio_format_rates_v2v3(): unable to find clock source (clock -110)
> >   usb 3-4: uac_clock_source_is_valid(): cannot get clock validity for id 41
> >   usb 3-4: clock source 41 is not valid, cannot use
> > 
> > The problem only occurs when both devices and kernel modules are
> > present and loaded during system boot, so it can be worked around by
> > connecting the audio interface after booting.
> > 
> > Fixes: ccfc8948d7e4 ("Bluetooth: btusb: mediatek: reset the controller before downloading the fw")
> > Closes: https://github.com/geoffreybennett/linux-fcp/issues/24
> > Bisected-by: Benedikt Ziemons <ben@rs485.network>
> > Tested-by: Benedikt Ziemons <ben@rs485.network>
> > Cc: stable@vger.kernel.org
> > Signed-off-by: Geoffrey D. Bennett <g@b4.vu>
> > ---
> > Changelog:
> > 
> > v1 -> v2:
> > 
> > - Updated commit message with additional information.
> > - No change to this patch's diff.
> > - Dropped alternate patch that only reverted for 0x7922.
> > - Chris, Sean, Hao agreed to reverting the change:
> >   https://lore.kernel.org/linux-bluetooth/2025031352-octopus-quadrant-f7ca@gregkh/T/#m0b31a9a8e87b9499e1ec3370c08f03e43bfb54bf
> > 
> >  drivers/bluetooth/btmtk.c | 10 ----------
> >  1 file changed, 10 deletions(-)
> > 
> > diff --git a/drivers/bluetooth/btmtk.c b/drivers/bluetooth/btmtk.c
> > index 68846c5bd4f7..4390fd571dbd 100644
> > --- a/drivers/bluetooth/btmtk.c
> > +++ b/drivers/bluetooth/btmtk.c
> > @@ -1330,13 +1330,6 @@ int btmtk_usb_setup(struct hci_dev *hdev)
> >  		break;
> >  	case 0x7922:
> >  	case 0x7925:
> > -		/* Reset the device to ensure it's in the initial state before
> > -		 * downloading the firmware to ensure.
> > -		 */
> > -
> > -		if (!test_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags))
> > -			btmtk_usb_subsys_reset(hdev, dev_id);
> > -		fallthrough;
> >  	case 0x7961:
> >  		btmtk_fw_get_filename(fw_bin_name, sizeof(fw_bin_name), dev_id,
> >  				      fw_version, fw_flavor);
> > @@ -1345,12 +1338,9 @@ int btmtk_usb_setup(struct hci_dev *hdev)
> >  						btmtk_usb_hci_wmt_sync);
> >  		if (err < 0) {
> >  			bt_dev_err(hdev, "Failed to set up firmware (%d)", err);
> > -			clear_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags);
> >  			return err;
> >  		}
> > 
> > -		set_bit(BTMTK_FIRMWARE_LOADED, &btmtk_data->flags);
> > -
> >  		/* It's Device EndPoint Reset Option Register */
> >  		err = btmtk_usb_uhw_reg_write(hdev, MTK_EP_RST_OPT,
> >  					      MTK_EP_RST_IN_OUT_OPT);
> > --
> > 2.45.0
> > 
> > 
> > 

Re: [STABLE 6.12/6.14] Bluetooth MediaTek controller reset fixes

[Greg KH]

On Thu, May 08, 2025 at 05:00:44PM +0930, Geoffrey D. Bennett wrote:
> Hi stable@vger.kernel.org,
> 
> Could you please apply:
> 
> 1. Commit a7208610761ae ("Bluetooth: btmtk: Remove resetting mt7921
> before downloading the fw") to v6.12.x (it's already in
> v6.14).
> 
> 2. Commit 33634e2ab7c6 ("Bluetooth: btmtk: Remove the resetting step
> before downloading the fw") to v6.12.x and v6.14.x.
> 
> These fixes address an issue with some audio interfaces failing to
> initialise during boot on kernels 6.11+. As noted in my original
> analysis below, the MediaTek Bluetooth controller reset increases the
> device setup time from ~200ms to ~20s and can interfere with other USB
> devices on the bus.

Now queued up, thanks.

greg k-h