Other Parts Discussed in Thread: SK-AM62
Hello,
It looks like out of the box on SDK 9 the Wilink module is not enabled on the SK-AM62:
Compared to the previous SDK:
Do we have a guide on how to turn the wifi back on?
Munan
This thread has been locked.
If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.
Hi Munan,
Thanks for the question.
With SDK 9.0, are you trying to use yocto image or debian image?
Please have a look at the image below.
Thanks,
Vaibhav
Hi Munan,
This is a known issue and has been resolved in the device tree. It should be included into the next AM62 SDK, but in the meantime you can re-build the k3-am625-sk.dtb from ti-linux-kernel git repo branch ti-linux-6.1.y: https://git.ti.com/cgit/ti-linux-kernel/ti-linux-kernel/log/?h=ti-linux-6.1.y
bringing in this file did not help. I still cannot enable wifi. with this error.:
12-12 11:01:55.741 501 611 D WifiNl80211Manager: Setting up interface for client mode: wlan0
12-12 11:01:55.742 430 430 E wificond: Can't get wlan0 index: No such device
12-12 11:01:55.742 430 430 E wificond: Failed to get wiphy index
12-12 11:01:55.743 501 611 E WifiNl80211Manager: Could not get IClientInterface instance from wificond
12-12 11:01:55.743 501 611 E WifiNative: Failed to setup iface in wificond=w
Hi Noe,
It looks like the module "wlcore.ko" is not loading. Have you made any changes that would explain disabling the wlcore module?
Hi Noe, Sabeeh,
I did flash SDK 9.1 on the AM62-SK (GP EVM) and although the wifi did not come up by default I was able to get the interface to come up with ifconfig and the wilink driver initialized the interface. So it does appear that the change in the DTS got rolled into the 9.1 SDK.
I'm working on getting a kernel built on SDK 9 with the new DTS to see if that works.
Munan
Hey Sabeeh,
I asked the processors team to go ahead and migrate the dts back to the 9.0 sdk, but I'm seeing essentially the same issues as Noe.
I did also query the device tree in the /proc/device-tree sysfs debug directory and I do see that the wlan_lten does not show the regulator-always-on property so the device tree change was effective.
I do not see any of the wlcore.ko modules loaded and if I try to load them myself, I eventually end up with trying to load rfkill.ko which will not load. Not sure if that means anything. I also checked all the kernel configuration files for the expected wifi kernel modules and it looks like this is all included correctly and I see the required kernel modules in the filesystem.
here's my startup log
am62x-sk-sdk9.0_android.txtam62x-sk-sdk9.0_android_dmesg.txt
For what it's worth I do see some of the wilink modules try to load at startup but they were not present by the time the console came up.
Munan
Hi Munan,
Again, I'm not very familiar with Android build system (I'm slowly learning). But I see many "Unknown symbol" messages in your android_dmesg.txt. It seems like there are missing pieces to the kernel on your build. Can you check if you have CONFIG_MAC80211 installed on your kernel?
Is your customer seeing the same error messages?
Hey Sabeeh,
OK just to bring this thread up to date with the email conversation, there is an additional set of patches needed to list the rfkill kernel module. The reason why there are a bunch of unknown symbol messages in dmesg is because I was manually trying to load the various wlcore kernel modules etc. starting with wlcore, I was able to dig down to see which other dependencies it had (which are all the undefined symbol messages) and manually tried to load those until I got to rfkill.ko which would not load. I did manually check the kernel configuration and the CONFIG_MAC80211 is enabled.
This patch is supposed to allow the loading of the rfkill kernel module which I'll try today to see if it fixes things.
Munan