Declare IPA firmware with MODULE_FIRMWARE(), so that initramfs tools can
build the firmware into initramfs image or warn on missing of the
firmware.
W: Possible missing firmware /lib/firmware/ipa_fws.mdt for module ipa
Signed-off-by: Shawn Guo <[email protected]>
---
drivers/net/ipa/ipa_main.c | 2 ++
1 file changed, 2 insertions(+)
diff --git a/drivers/net/ipa/ipa_main.c b/drivers/net/ipa/ipa_main.c
index cdfa98a76e1f..264bebc78d1e 100644
--- a/drivers/net/ipa/ipa_main.c
+++ b/drivers/net/ipa/ipa_main.c
@@ -545,6 +545,8 @@ static int ipa_firmware_load(struct device *dev)
return ret;
}
+MODULE_FIRMWARE(IPA_FW_PATH_DEFAULT);
+
static const struct of_device_id ipa_match[] = {
{
.compatible = "qcom,msm8998-ipa",
--
2.17.1
On Sun, Sep 26, 2021 at 02:55:29PM +0800, Shawn Guo wrote:
> Declare IPA firmware with MODULE_FIRMWARE(), so that initramfs tools can
> build the firmware into initramfs image or warn on missing of the
> firmware.
>
> W: Possible missing firmware /lib/firmware/ipa_fws.mdt for module ipa
>
> Signed-off-by: Shawn Guo <[email protected]>
> ---
> drivers/net/ipa/ipa_main.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/net/ipa/ipa_main.c b/drivers/net/ipa/ipa_main.c
> index cdfa98a76e1f..264bebc78d1e 100644
> --- a/drivers/net/ipa/ipa_main.c
> +++ b/drivers/net/ipa/ipa_main.c
> @@ -545,6 +545,8 @@ static int ipa_firmware_load(struct device *dev)
> return ret;
> }
>
> +MODULE_FIRMWARE(IPA_FW_PATH_DEFAULT);
> +
Just reminded by Steev, .mdt file conventionally means we are using
split firmware. Building only .mdt file into initramfs is not
sufficient. So please disregard the patch.
Shawn
> static const struct of_device_id ipa_match[] = {
> {
> .compatible = "qcom,msm8998-ipa",
> --
> 2.17.1
>
On 9/26/21 8:46 AM, Shawn Guo wrote:
> Just reminded by Steev, .mdt file conventionally means we are using
> split firmware. Building only .mdt file into initramfs is not
> sufficient. So please disregard the patch.
>
> Shawn
OK, will disregard. May I assume you are going to implement
a patch that does the right thing?
I had a note to myself to investigate using MODULE_FIRMWARE()
but hadn't gotten to it yet; I'm glad to have you do that
instead...
Thank you.
-Alex
On Sat 25 Sep 23:55 PDT 2021, Shawn Guo wrote:
> Declare IPA firmware with MODULE_FIRMWARE(), so that initramfs tools can
> build the firmware into initramfs image or warn on missing of the
> firmware.
>
> W: Possible missing firmware /lib/firmware/ipa_fws.mdt for module ipa
>
> Signed-off-by: Shawn Guo <[email protected]>
> ---
> drivers/net/ipa/ipa_main.c | 2 ++
> 1 file changed, 2 insertions(+)
>
> diff --git a/drivers/net/ipa/ipa_main.c b/drivers/net/ipa/ipa_main.c
> index cdfa98a76e1f..264bebc78d1e 100644
> --- a/drivers/net/ipa/ipa_main.c
> +++ b/drivers/net/ipa/ipa_main.c
> @@ -545,6 +545,8 @@ static int ipa_firmware_load(struct device *dev)
> return ret;
> }
>
> +MODULE_FIRMWARE(IPA_FW_PATH_DEFAULT);
I think it's fair to say that no device that is capable of running
mkinitcpio should actually use IPA_FW_PATH_DEFAULT, but rather some
device-specific firmware file.
Regards,
Bjorn
> +
> static const struct of_device_id ipa_match[] = {
> {
> .compatible = "qcom,msm8998-ipa",
> --
> 2.17.1
>
On 9/27/21 2:35 PM, Bjorn Andersson wrote:
> On Sat 25 Sep 23:55 PDT 2021, Shawn Guo wrote:
>
>> Declare IPA firmware with MODULE_FIRMWARE(), so that initramfs tools can
>> build the firmware into initramfs image or warn on missing of the
>> firmware.
>>
>> W: Possible missing firmware /lib/firmware/ipa_fws.mdt for module ipa
>>
>> Signed-off-by: Shawn Guo <[email protected]>
>> ---
>> drivers/net/ipa/ipa_main.c | 2 ++
>> 1 file changed, 2 insertions(+)
>>
>> diff --git a/drivers/net/ipa/ipa_main.c b/drivers/net/ipa/ipa_main.c
>> index cdfa98a76e1f..264bebc78d1e 100644
>> --- a/drivers/net/ipa/ipa_main.c
>> +++ b/drivers/net/ipa/ipa_main.c
>> @@ -545,6 +545,8 @@ static int ipa_firmware_load(struct device *dev)
>> return ret;
>> }
>>
>> +MODULE_FIRMWARE(IPA_FW_PATH_DEFAULT);
>
> I think it's fair to say that no device that is capable of running
> mkinitcpio should actually use IPA_FW_PATH_DEFAULT, but rather some
> device-specific firmware file.
Actually, yes, I should have thought of that too.
The default used here just specifies the path name
used *if the firmware file name is not specified in
the DTB*.
So I don't think this is correct regardless of whether
the MDT file is split or not. I think that's what
Bjorn is saying here.
-Alex
>
> Regards,
> Bjorn
>
>> +
>> static const struct of_device_id ipa_match[] = {
>> {
>> .compatible = "qcom,msm8998-ipa",
>> --
>> 2.17.1
>>