mbox series

[0/5] qcom: wcnss: Allow overriding firmware form DT

Message ID 20210312003318.3273536-1-bjorn.andersson@linaro.org
Headers show
Series qcom: wcnss: Allow overriding firmware form DT | expand

Message

Bjorn Andersson March 12, 2021, 12:33 a.m. UTC 
The wireless subsystem found in Qualcomm MSM8974 and MSM8916 among others needs
platform-, and perhaps even board-, specific firmware. Add support for
providing this in devicetree.

Bjorn Andersson (5):
  dt-bindings: soc: qcom: wcnss: Add firmware-name property
  wcn36xx: Allow firmware name to be overridden by DT
  soc: qcom: wcnss_ctrl: Introduce local variable "dev"
  soc: qcom: wcnss_ctrl: Allow reading firmware-name from DT
  arm64: dts: qcom: msm8916: Enable modem and WiFi

 .../devicetree/bindings/soc/qcom/qcom,wcnss.txt   |  7 +++++++
 arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi         | 12 ++++++++++++
 arch/arm64/boot/dts/qcom/msm8916.dtsi             |  2 +-
 drivers/net/wireless/ath/wcn36xx/main.c           |  7 +++++++
 drivers/net/wireless/ath/wcn36xx/smd.c            |  4 ++--
 drivers/net/wireless/ath/wcn36xx/wcn36xx.h        |  1 +
 drivers/soc/qcom/wcnss_ctrl.c                     | 15 ++++++++++-----
 7 files changed, 40 insertions(+), 8 deletions(-)

Comments

Bryan O'Donoghue March 15, 2021, 11:57 a.m. UTC | #1 
On 12/03/2021 00:33, Bjorn Andersson wrote:
> firmware-name = "qcom/msm8916/WCNSS_qcom_wlan_nv.bin";

On the Linaro debian build I have to do this

+       firmware-name = "wlan/prima/WCNSS_qcom_wlan_nv.bin";

---
bod
Bryan O'Donoghue March 15, 2021, 11:57 a.m. UTC | #2 
On 12/03/2021 00:33, Bjorn Andersson wrote:
> The WLAN NV firmware blob differs between platforms, and possibly
> devices, so add support in the wcnss_ctrl driver for reading the path of
> this file from DT in order to allow these files to live in a generic
> file system (or linux-firmware).
> 
> The new property is optional and the code falls back to the old filename
> if the property isn't specified.
> 
> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
> ---
>   drivers/soc/qcom/wcnss_ctrl.c | 10 +++++++---
>   1 file changed, 7 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/soc/qcom/wcnss_ctrl.c b/drivers/soc/qcom/wcnss_ctrl.c
> index 358526b9de06..2a06d631e415 100644
> --- a/drivers/soc/qcom/wcnss_ctrl.c
> +++ b/drivers/soc/qcom/wcnss_ctrl.c
> @@ -200,6 +200,7 @@ static int wcnss_download_nv(struct wcnss_ctrl *wcnss, bool *expect_cbc)
>   	struct wcnss_download_nv_req *req;
>   	const struct firmware *fw;
>   	struct device *dev = wcnss->dev;
> +	const char *nvbin = NVBIN_FILE;
>   	const void *data;
>   	ssize_t left;
>   	int ret;
> @@ -208,10 +209,13 @@ static int wcnss_download_nv(struct wcnss_ctrl *wcnss, bool *expect_cbc)
>   	if (!req)
>   		return -ENOMEM;
>   
> -	ret = request_firmware(&fw, NVBIN_FILE, dev);
> +	ret = of_property_read_string(dev->of_node, "firmware-name", &nvbin);
> +	if (ret < 0 && ret != -EINVAL)
> +		goto free_req;
> +
> +	ret = request_firmware(&fw, nvbin, dev);
>   	if (ret < 0) {
> -		dev_err(dev, "Failed to load nv file %s: %d\n",
> -			NVBIN_FILE, ret);
> +		dev_err(dev, "Failed to load nv file %s: %d\n", nvbin, ret);
>   		goto free_req;
>   	}
>   
> 

Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Bryan O'Donoghue March 15, 2021, 11:58 a.m. UTC | #3 
On 12/03/2021 00:33, Bjorn Andersson wrote:
> The WLAN NV firmware blob differs between platforms, and possibly
> devices, so add support in the wcn36xx driver for reading the path of
> this file from DT in order to allow these files to live in a generic
> file system (or linux-firmware).
> 
> For some reason the parent (wcnss_ctrl) also needs to upload this blob,
> so rather than specifying the same information in both nodes wcn36xx
> reads the string from the parent's of_node.
> 
> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
> ---
> 
> This patch can be applied independently of the others, but relates to the
> acceptance of the addition to the DT binding (in patch 1/5). So my suggestion
> is that this one goes through the ath tree and the others through the Qualcomm
> SoC tree.
> 
>   drivers/net/wireless/ath/wcn36xx/main.c    | 7 +++++++
>   drivers/net/wireless/ath/wcn36xx/smd.c     | 4 ++--
>   drivers/net/wireless/ath/wcn36xx/wcn36xx.h | 1 +
>   3 files changed, 10 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/net/wireless/ath/wcn36xx/main.c b/drivers/net/wireless/ath/wcn36xx/main.c
> index afb4877eaad8..87b5c0ff16c0 100644
> --- a/drivers/net/wireless/ath/wcn36xx/main.c
> +++ b/drivers/net/wireless/ath/wcn36xx/main.c
> @@ -1407,6 +1407,13 @@ static int wcn36xx_probe(struct platform_device *pdev)
>   		goto out_wq;
>   	}
>   
> +	wcn->nv_file = WLAN_NV_FILE;
> +	ret = of_property_read_string(wcn->dev->parent->of_node, "firmware-name", &wcn->nv_file);
> +	if (ret < 0 && ret != -EINVAL) {
> +		wcn36xx_err("failed to read \"firmware-name\" property\n");
> +		goto out_wq;
> +	}
> +
>   	wcn->smd_channel = qcom_wcnss_open_channel(wcnss, "WLAN_CTRL", wcn36xx_smd_rsp_process, hw);
>   	if (IS_ERR(wcn->smd_channel)) {
>   		wcn36xx_err("failed to open WLAN_CTRL channel\n");
> diff --git a/drivers/net/wireless/ath/wcn36xx/smd.c b/drivers/net/wireless/ath/wcn36xx/smd.c
> index d0c3a1557e8d..7b928f988068 100644
> --- a/drivers/net/wireless/ath/wcn36xx/smd.c
> +++ b/drivers/net/wireless/ath/wcn36xx/smd.c
> @@ -514,10 +514,10 @@ int wcn36xx_smd_load_nv(struct wcn36xx *wcn)
>   	u16 fm_offset = 0;
>   
>   	if (!wcn->nv) {
> -		ret = request_firmware(&wcn->nv, WLAN_NV_FILE, wcn->dev);
> +		ret = request_firmware(&wcn->nv, wcn->nv_file, wcn->dev);
>   		if (ret) {
>   			wcn36xx_err("Failed to load nv file %s: %d\n",
> -				      WLAN_NV_FILE, ret);
> +				    wcn->nv_file, ret);
>   			goto out;
>   		}
>   	}
> diff --git a/drivers/net/wireless/ath/wcn36xx/wcn36xx.h b/drivers/net/wireless/ath/wcn36xx/wcn36xx.h
> index 71fa9992b118..5977af2116e3 100644
> --- a/drivers/net/wireless/ath/wcn36xx/wcn36xx.h
> +++ b/drivers/net/wireless/ath/wcn36xx/wcn36xx.h
> @@ -185,6 +185,7 @@ struct wcn36xx {
>   	struct device		*dev;
>   	struct list_head	vif_list;
>   
> +	const char		*nv_file;
>   	const struct firmware	*nv;
>   
>   	u8			fw_revision;
> 

Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Bjorn Andersson March 18, 2021, 2:50 p.m. UTC | #4 
On Mon 15 Mar 07:01 CDT 2021, Bryan O'Donoghue wrote:

> On 12/03/2021 00:33, Bjorn Andersson wrote:
> > Enable the modem and WiFi subsystems and specify msm8916 specific
> > firmware path for these and the WCNSS control service.
> > 
> > Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
> > ---
> >   arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi | 12 ++++++++++++
> >   arch/arm64/boot/dts/qcom/msm8916.dtsi     |  2 +-
> >   2 files changed, 13 insertions(+), 1 deletion(-)
> > 
> > diff --git a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> > index 6aef0c2e4f0a..448e3561ef63 100644
> > --- a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> > +++ b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> > @@ -305,6 +305,12 @@ &mdss {
> >   	status = "okay";
> >   };
> > +&mpss {
> > +	status = "okay";
> > +
> > +	firmware-name = "qcom/msm8916/mba.mbn", "qcom/msm8916/modem.mbn";
> > +};
> > +
> >   &pm8916_resin {
> >   	status = "okay";
> >   	linux,code = <KEY_VOLUMEDOWN>;
> > @@ -312,6 +318,8 @@ &pm8916_resin {
> >   &pronto {
> >   	status = "okay";
> > +
> > +	firmware-name = "qcom/msm8916/wcnss.mbn";
> >   };
> 
> On Debian I have to do this
> 
> 
> index 2a6a23cb14ca..597cdc8f51cc 100644
> --- a/drivers/remoteproc/qcom_wcnss.c
> +++ b/drivers/remoteproc/qcom_wcnss.c
> @@ -33,7 +33,7 @@
>  #include "qcom_wcnss.h"
> 
>  #define WCNSS_CRASH_REASON_SMEM                422
> -#define WCNSS_FIRMWARE_NAME            "wcnss.mdt"
> +#define WCNSS_FIRMWARE_NAME            "qcom/msm8916/wcnss.mdt"
> 
> so I guess wcnss_probe() -> rproc_alloc() wants this fix too.
> 

Can you confirm that you're saying that you want below patch, which I
just merged?

https://lore.kernel.org/linux-remoteproc/20210312002441.3273183-1-bjorn.andersson@linaro.org/

(Which makes it possible to specify firmware name per platform/board)

Regards,
Bjorn
Vinod Koul March 18, 2021, 4:48 p.m. UTC | #5 
On 11-03-21, 16:33, Bjorn Andersson wrote:
> The wireless subsystem found in Qualcomm MSM8974 and MSM8916 among others needs
> platform-, and perhaps even board-, specific firmware. Add support for
> providing this in devicetree.

Reviewed-by: Vinod Koul <vkoul@kernel.org>
Jeffrey Hugo March 18, 2021, 4:56 p.m. UTC | #6 
form -> from in the subject?

On Thu, Mar 11, 2021 at 5:34 PM Bjorn Andersson
<bjorn.andersson@linaro.org> wrote:
>
> The wireless subsystem found in Qualcomm MSM8974 and MSM8916 among others needs
> platform-, and perhaps even board-, specific firmware. Add support for
> providing this in devicetree.
Bjorn Andersson March 18, 2021, 5:06 p.m. UTC | #7 
On Thu 18 Mar 11:56 CDT 2021, Jeffrey Hugo wrote:

> form -> from in the subject?
> 

Seems like I only failed in the cover letter, right?

Regards,
Bjorn

> On Thu, Mar 11, 2021 at 5:34 PM Bjorn Andersson
> <bjorn.andersson@linaro.org> wrote:
> >
> > The wireless subsystem found in Qualcomm MSM8974 and MSM8916 among others needs
> > platform-, and perhaps even board-, specific firmware. Add support for
> > providing this in devicetree.
Jeffrey Hugo March 18, 2021, 6:12 p.m. UTC | #8 
On Thu, Mar 18, 2021 at 11:06 AM Bjorn Andersson
<bjorn.andersson@linaro.org> wrote:
>
> On Thu 18 Mar 11:56 CDT 2021, Jeffrey Hugo wrote:
>
> > form -> from in the subject?
> >
>
> Seems like I only failed in the cover letter, right?

Looks like.  I didn't even parse that this was the cover letter.

Nothing to see here.
Bryan O'Donoghue March 19, 2021, 2:41 p.m. UTC | #9 
On 18/03/2021 14:50, Bjorn Andersson wrote:
> On Mon 15 Mar 07:01 CDT 2021, Bryan O'Donoghue wrote:
> 
>> On 12/03/2021 00:33, Bjorn Andersson wrote:
>>> Enable the modem and WiFi subsystems and specify msm8916 specific
>>> firmware path for these and the WCNSS control service.
>>>
>>> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
>>> ---
>>>    arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi | 12 ++++++++++++
>>>    arch/arm64/boot/dts/qcom/msm8916.dtsi     |  2 +-
>>>    2 files changed, 13 insertions(+), 1 deletion(-)
>>>
>>> diff --git a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
>>> index 6aef0c2e4f0a..448e3561ef63 100644
>>> --- a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
>>> +++ b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
>>> @@ -305,6 +305,12 @@ &mdss {
>>>    	status = "okay";
>>>    };
>>> +&mpss {
>>> +	status = "okay";
>>> +
>>> +	firmware-name = "qcom/msm8916/mba.mbn", "qcom/msm8916/modem.mbn";
>>> +};
>>> +
>>>    &pm8916_resin {
>>>    	status = "okay";
>>>    	linux,code = <KEY_VOLUMEDOWN>;
>>> @@ -312,6 +318,8 @@ &pm8916_resin {
>>>    &pronto {
>>>    	status = "okay";
>>> +
>>> +	firmware-name = "qcom/msm8916/wcnss.mbn";
>>>    };
>>
>> On Debian I have to do this
>>
>>
>> index 2a6a23cb14ca..597cdc8f51cc 100644
>> --- a/drivers/remoteproc/qcom_wcnss.c
>> +++ b/drivers/remoteproc/qcom_wcnss.c
>> @@ -33,7 +33,7 @@
>>   #include "qcom_wcnss.h"
>>
>>   #define WCNSS_CRASH_REASON_SMEM                422
>> -#define WCNSS_FIRMWARE_NAME            "wcnss.mdt"
>> +#define WCNSS_FIRMWARE_NAME            "qcom/msm8916/wcnss.mdt"
>>
>> so I guess wcnss_probe() -> rproc_alloc() wants this fix too.
>>
> 
> Can you confirm that you're saying that you want below patch, which I
> just merged?
> 
> https://lore.kernel.org/linux-remoteproc/20210312002441.3273183-1-bjorn.andersson@linaro.org/
> 
> (Which makes it possible to specify firmware name per platform/board)
> 
> Regards,
> Bjorn
> 

yep

Tested-by: Bryan O'Donoghue <bryan.odonoghue@linaro.org>
Stephan Gerhold March 19, 2021, 3:19 p.m. UTC | #10 
Hi Bjorn,

On Thu, Mar 11, 2021 at 04:33:18PM -0800, Bjorn Andersson wrote:
> Enable the modem and WiFi subsystems and specify msm8916 specific
> firmware path for these and the WCNSS control service.
> 
> Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>

The changes itself look good to me, but the commit message is really
misleading. It does not mention anywhere that the change actually just
enables "modem" on apq8016-sbc instead of "msm8916". :)

Also, WCNSS was actually enabled before already (with the default
firmware path). In my opinion, it would be clearer to change the
firmware-name for it in an extra patch.

> ---
>  arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi | 12 ++++++++++++
>  arch/arm64/boot/dts/qcom/msm8916.dtsi     |  2 +-
>  2 files changed, 13 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> index 6aef0c2e4f0a..448e3561ef63 100644
> --- a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> +++ b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> @@ -305,6 +305,12 @@ &mdss {
>  	status = "okay";
>  };
>  
> +&mpss {
> +	status = "okay";
> +
> +	firmware-name = "qcom/msm8916/mba.mbn", "qcom/msm8916/modem.mbn";
> +};
> +
>  &pm8916_resin {
>  	status = "okay";
>  	linux,code = <KEY_VOLUMEDOWN>;
> @@ -312,6 +318,8 @@ &pm8916_resin {
>  
>  &pronto {
>  	status = "okay";
> +
> +	firmware-name = "qcom/msm8916/wcnss.mbn";
>  };
>  

How do I get a .mbn from the wcnss.{mdt,.b??} files provided in the
DB410c firmware package? I guess I should just run them through
https://github.com/andersson/pil-squasher?

Also, is the single file format (mbn) preferred now? Not sure if there
is any significant difference except having less files laying around.

Thanks,
Stephan
Bjorn Andersson May 31, 2021, 4:33 p.m. UTC | #11 
On Fri 19 Mar 10:19 CDT 2021, Stephan Gerhold wrote:

> Hi Bjorn,
> 

Hi Stephan,

I thought I wrote you a reply, but I don't see it on lore or in my
mailbox...

> On Thu, Mar 11, 2021 at 04:33:18PM -0800, Bjorn Andersson wrote:
> > Enable the modem and WiFi subsystems and specify msm8916 specific
> > firmware path for these and the WCNSS control service.
> > 
> > Signed-off-by: Bjorn Andersson <bjorn.andersson@linaro.org>
> 
> The changes itself look good to me, but the commit message is really
> misleading. It does not mention anywhere that the change actually just
> enables "modem" on apq8016-sbc instead of "msm8916". :)
> 
> Also, WCNSS was actually enabled before already (with the default
> firmware path). In my opinion, it would be clearer to change the
> firmware-name for it in an extra patch.
> 

You're right, I will rewrite the commit message.

> > ---
> >  arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi | 12 ++++++++++++
> >  arch/arm64/boot/dts/qcom/msm8916.dtsi     |  2 +-
> >  2 files changed, 13 insertions(+), 1 deletion(-)
> > 
> > diff --git a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> > index 6aef0c2e4f0a..448e3561ef63 100644
> > --- a/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> > +++ b/arch/arm64/boot/dts/qcom/apq8016-sbc.dtsi
> > @@ -305,6 +305,12 @@ &mdss {
> >  	status = "okay";
> >  };
> >  
> > +&mpss {
> > +	status = "okay";
> > +
> > +	firmware-name = "qcom/msm8916/mba.mbn", "qcom/msm8916/modem.mbn";
> > +};
> > +
> >  &pm8916_resin {
> >  	status = "okay";
> >  	linux,code = <KEY_VOLUMEDOWN>;
> > @@ -312,6 +318,8 @@ &pm8916_resin {
> >  
> >  &pronto {
> >  	status = "okay";
> > +
> > +	firmware-name = "qcom/msm8916/wcnss.mbn";
> >  };
> >  
> 
> How do I get a .mbn from the wcnss.{mdt,.b??} files provided in the
> DB410c firmware package? I guess I should just run them through
> https://github.com/andersson/pil-squasher?
> 

Yes, .mdt + .bNN files can be put back together as a .mbn by running
them through the pil-squasher.

It's worth mentioning that the suffix doesn't matter for the loader, so
symlinking the .mdt to .mbn (in either way) will keep the kernel
happy - if that simplifies transitions in either way.

> Also, is the single file format (mbn) preferred now? Not sure if there
> is any significant difference except having less files laying around.
> 

We've had numerous cases where people has managed to mix the parts from
different versions of the firmware, sometimes with the nice result of
the loader complaining that some particular .bNN file is missing
other times just with the signature check (mysteriously) failing.

Lumping the parts into a single .mbn resolves all these issues and
hence I prefer this. It is however not a requirement.

Regards,
Bjorn