Drop mxm-mwifiex-setup recipe by quaresmajose · Pull Request #998 · foundriesio/meta-lmp

quaresmajose · 2023-01-04T18:37:39Z

We can use yocto variables for this propose:

MrCry0

.

angolini

Do you know if the KERNEL_MODULE_* variables also deal with firmware?

I think I still prefer to keep this as a MACHINE_FEATURE

angolini · 2023-01-04T20:13:12Z

+KERNEL_MODULE_AUTOLOAD:imx8mm-lpddr4-evk-ebbr += "moal"
+KERNEL_MODULE_PROBECONF:imx8mm-lpddr4-evk-ebbr += "moal"


Is there a reason to add this driver only for the ebbr variation of the board?

Right, it shouldn't be ebbr specific.

VARIABLE:board += ""

IIRC this will also potentially cause issues with the assignment (var ?=, var:board = ""), which is why we usually just append here instead.

Is there a reason to add this driver only for the ebbr variation of the board?

This replace the previous MACHINE_FEATURES:append:imx8mm-lpddr4-evk = " mxm-mwifiex-load" that uses the imx8mm-lpddr4-evk machine override, a few lines below there is another but now for imx8mp-lpddr4-evk.

IIRC this will also potentially cause issues with the assignment (var ?=, var:board = ""), which is why we usually just append here instead.

Changed to use the append

Is there a reason to add this driver only for the ebbr variation of the board?

This replace the previous MACHINE_FEATURES:append:imx8mm-lpddr4-evk = " mxm-mwifiex-load" that uses the imx8mm-lpddr4-evk machine override, a few lines below there is another but now for imx8mp-lpddr4-evk.

misunderstood this part and ebbr variation is my fault! now fixed

quaresmajose · 2023-01-05T11:18:12Z

Do you know if the KERNEL_MODULE_* variables also deal with firmware?

I think not, this variables is handled on kernel-module-split.bbclass that don't know nothing about firmware.

I think I still prefer to keep this as a MACHINE_FEATURE

Is this configuration to auto load the kernel module still required? Why the moal can't be probed by the kernel and loaded by udev? I don't have any board to test this.

angolini · 2023-01-05T12:20:51Z

As far as I understand, MACHINE_FEATURES describes the hardware. There are firmwares that are only installed when some tag is present in the MACHINE_FEATURES

As far as I understand, moal module is a generic wifi driver, and I would say it is a common dependency for other drivers.

So I understand we can configure the moal driver probe using the OE variable, but I would not remove the MACHINE_FEATURES.

Are you facing some log errors due to changes in some firmware due to the latest meta-freescale bump, @quaresmajose ?

quaresmajose · 2023-01-05T13:10:59Z

As far as I understand, MACHINE_FEATURES describes the hardware. There are firmwares that are only installed when some tag is present in the MACHINE_FEATURES

As far as I understand, moal module is a generic wifi driver, and I would say it is a common dependency for other drivers.

So I understand we can configure the moal driver probe using the OE variable, but I would not remove the MACHINE_FEATURES.

Are you facing some log errors due to changes in some firmware due to the latest meta-freescale bump, @quaresmajose ?

The MACHINE_FEATURES have now all the pieces on meta-freescale and we have our adaptation in #997 so LmP is aligned.

This PR is for auto load the kernel modules in this two machines, I can change it to be done with a MACHINE_FEATURES.
My question is why we need this auto load and modprobe config in LmP and meta-freescale machines works out of the box?

angolini · 2023-01-05T15:37:30Z

I'm not entirely aware of the context of this PR. I thought it was only an enhancement, not a bug fix. I think the suggestion was to have the OE variables ready to load the module instead of something else.

We can discuss this more offline, but I would need to understand what is happening.

quaresmajose · 2023-01-05T16:32:46Z

I'm not entirely aware of the context of this PR. I thought it was only an enhancement, not a bug fix. I think the suggestion was to have the OE variables ready to load the module instead of something else.

We can discuss this more offline, but I would need to understand what is happening.

and you are right, this is a different way of doing the same thing and is not a bug. we can talk offline

MrCry0 · 2023-01-05T16:53:48Z

I'm not entirely aware of the context of this PR. I thought it was only an enhancement, not a bug fix. I think the suggestion was to have the OE variables ready to load the module instead of something else.
We can discuss this more offline, but I would need to understand what is happening.

and you are right, this is a different way of doing the same thing and is not a bug. we can talk offline

@angolini has a reason. I agree it would be better to keep a machine feature and mxm-mwifiex-setup_0.1.bb, just replacing the whole recipe with OE variables. And we can drop our conf files and replace the whole recipe mxm-mwifiex-setup with:

KERNEL_MODULE_AUTOLOAD:append = " moal"
KERNEL_MODULE_PROBECONF:append = " moal"
module_conf_moal = "options moal mod_para=nxp/wifi_mod_para.conf drvdbg=0x6"

ricardosalveti · 2023-01-05T17:59:12Z

-MACHINE_FEATURES:append:imx8mm-lpddr4-evk = " mxm-mwifiex-load"
+KERNEL_MODULE_AUTOLOAD:append:imx8mm-lpddr4-evk = " moal"
+KERNEL_MODULE_PROBECONF:append:imx8mm-lpddr4-evk = " moal"
+module_conf_moal:imx8mm-lpddr4-evk = "options moal mod_para=nxp/wifi_mod_para.conf drvdbg=0x6"


Would be good to confirm indeed if this customization for mod_para is still needed as it seems it is not required anymore by looking at the BSP (maybe the default path it looks now is nxp/wifi_mod_para.conf?).

And also if autoloading is still required as well, @MrCry0 should be able to confirm.

quaresmajose · 2023-01-05T18:03:13Z

Convert this to WIP because the variable KERNEL_MODULE_AUTOLOAD and KERNEL_MODULE_PROBECONF in yocto needs be adapted because they use "/etc" hard-coded in kernel-module-split.bbclass and we need to have this in ${libdir}

Use yocto variables for this propose: - KERNEL_MODULE_AUTOLOAD https://docs.yoctoproject.org/ref-manual/variables.html?highlight=kernel_module_autoload#term-KERNEL_MODULE_AUTOLOAD - KERNEL_MODULE_PROBECONF https://docs.yoctoproject.org/ref-manual/variables.html?highlight=kernel_module_autoload#term-KERNEL_MODULE_PROBECONF Signed-off-by: Jose Quaresma <[email protected]>

Signed-off-by: Jose Quaresma <[email protected]>

Use yocto variables for this propose: - KERNEL_MODULE_AUTOLOAD https://docs.yoctoproject.org/ref-manual/variables.html?highlight=kernel_module_autoload#term-KERNEL_MODULE_AUTOLOAD - KERNEL_MODULE_PROBECONF https://docs.yoctoproject.org/ref-manual/variables.html?highlight=kernel_module_autoload#term-KERNEL_MODULE_PROBECONF Signed-off-by: Jose Quaresma <[email protected]>

quaresmajose · 2023-02-09T11:54:00Z

Convert this to WIP because the variable KERNEL_MODULE_AUTOLOAD and KERNEL_MODULE_PROBECONF in yocto needs be adapted because they use "/etc" hard-coded in kernel-module-split.bbclass and we need to have this in ${libdir}

I have a patch prepared for oe-core so i will merge this in main-next to test

quaresmajose requested review from MrCry0, angolini and ricardosalveti January 4, 2023 18:37

MrCry0 approved these changes Jan 4, 2023

View reviewed changes

angolini reviewed Jan 4, 2023

View reviewed changes

MrCry0 self-requested a review January 5, 2023 10:45

quaresmajose force-pushed the nxp89xx branch from 707a397 to 13f2b0a Compare January 5, 2023 11:08

quaresmajose force-pushed the nxp89xx branch from 13f2b0a to 4d9d8fe Compare January 5, 2023 11:30

ricardosalveti reviewed Jan 5, 2023

View reviewed changes

quaresmajose marked this pull request as draft January 5, 2023 18:03

quaresmajose added 4 commits February 9, 2023 11:51

bsp: lmp-machine-custom: drop mxm-mwifiex-load

4edc6db

Signed-off-by: Jose Quaresma <[email protected]>

base: lmp-feature-wifi: drop mxm-mwifiex-setup

81ff109

Signed-off-by: Jose Quaresma <[email protected]>

quaresmajose force-pushed the nxp89xx branch from 4d9d8fe to e444ce4 Compare February 9, 2023 11:52

quaresmajose changed the base branch from main to main-next February 9, 2023 11:52

quaresmajose marked this pull request as ready for review February 9, 2023 11:54

quaresmajose merged commit 3a3ce8c into foundriesio:main-next Feb 9, 2023

quaresmajose deleted the nxp89xx branch February 9, 2023 11:55

		KERNEL_MODULE_AUTOLOAD:imx8mm-lpddr4-evk-ebbr += "moal"
		KERNEL_MODULE_PROBECONF:imx8mm-lpddr4-evk-ebbr += "moal"

Conversation

quaresmajose commented Jan 4, 2023

Uh oh!

MrCry0 left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

angolini left a comment

Choose a reason for hiding this comment

Uh oh!

angolini Jan 4, 2023

Choose a reason for hiding this comment

Uh oh!

ricardosalveti Jan 4, 2023

Choose a reason for hiding this comment

Uh oh!

quaresmajose Jan 5, 2023

Choose a reason for hiding this comment

Uh oh!

quaresmajose Jan 5, 2023

Choose a reason for hiding this comment

Uh oh!

quaresmajose Jan 5, 2023

Choose a reason for hiding this comment

Uh oh!

quaresmajose commented Jan 5, 2023

Uh oh!

angolini commented Jan 5, 2023

Uh oh!

quaresmajose commented Jan 5, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

angolini commented Jan 5, 2023

Uh oh!

quaresmajose commented Jan 5, 2023

Uh oh!

MrCry0 commented Jan 5, 2023 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ricardosalveti Jan 5, 2023

Choose a reason for hiding this comment

Uh oh!

ricardosalveti Jan 5, 2023

Choose a reason for hiding this comment

Uh oh!

quaresmajose commented Jan 5, 2023

Uh oh!

quaresmajose commented Feb 9, 2023

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

MrCry0 left a comment •

edited

Loading

quaresmajose commented Jan 5, 2023 •

edited

Loading

MrCry0 commented Jan 5, 2023 •

edited

Loading