AM3352: resume from standby problem with kernel 6.1.46

Hi Angel,

Can you review these two threads? There was a similar issue where the config file needs to be updated.

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1333959/beagl-bone-black-sdk-9-1-suspend-not-working

https://e2e.ti.com/support/processors-group/processors/f/processors-forum/1354190/re-beagl-bone-black-sdk-9-1-suspend-not-working

Best Regards,

Anshu

Hi Anshu,

I have double checked instructions from TI documentation and config file seems to be fine:

CONFIG_FW_LOADER_USER_HELPER=y
CONFIG_FW_LOADER_USER_HELPER_FALLBACK=y
CONFIG_MAILBOX=y
CONFIG_OMAP2PLUS_MBOX=y
CONFIG_WKUP_M3_RPROC=y
CONFIG_SOC_TI=y
CONFIG_WKUP_M3_IPC=y
CONFIG_TI_EMIF_SRAM=y
CONFIG_AMX3_PM=y
CONFIG_RTC_DRV_OMAP=y
CONFIG_SUSPEND=y
CONFIG_SUSPEND_FREEZER=y

also elf FW is loaded:

dmesg | grep pm
root@A137K3:~# dmesg | grep pm
[ 3.102358] remoteproc remoteproc0: Booting fw image am335x-pm-firmware.elf, size 230332

dmesg | grep CM3
[ 3.125411] wkup_m3_ipc 44e11324.wkup_m3_ipc: CM3 Firmware Version = 0x192

Last reply from Bin Liu in the second thread says that he found the same problem, is there any solution now? 

We also need to resume from gpio0_3. We defined this input in our .dts file and it is available in the FS:

root@A137K3:/home# cat /sys/devices/platform/gpio_buttons\@0/power/wakeup
enabled

But again it is not possible to resume. I think it could be the same problem.

Best regards and than you

Angel

Hi Angel,

Thanks for the update. Looks like there is a bug for UART wakeup source. I'll file this bug to the SW development team.

Best Regards,

Anshu

Hi Anshu,

Thank you for your answer.

I have also checked BBB to wake up with gpio0_26. It works fine with kernel version 4.19.94, but not with kernel 6.1.80.

It is a key feature for our product to wake from UART and from gpio0_x, so could you please also inform your SW development team about the gpio0_x wakeup bug?

It appears that SDK 09.01.00.001 is the latest official release from TI, and it contains kernel 6.1.46. Is it possible to use a more recent kernel version from TI that fixes these bugs?

Best regards and thank you

Angel

Hi Angel,

Thanks for letting me know. I'll update the bug report about the GPIO issue as well.

TI will only provide support for the kernel version used in the SDK. You're welcome to review kernel changes to see if it helps fix the bugs, but it would be outside our support scope.

Best Regards,

Anshu

Hi Anshu,

I understand that you only support official releases. I was thinking of the versions you release on the official TI github
git.ti.com/.../

I did a quick search here and found no fix for this bug up to version 6.1.91.
If you do make a fix, how can I know about that and where can I find the new version fixing the bug?

Best regards and thank you

Angel

Hi Angel,

I'm not aware of any way to track the bug. You can check on the issues tracker, but I'm not sure how often its updated: https://software-dl.ti.com/processor-sdk-linux/esd/AM335X/09_01_00_001/exports/docs/devices/AM335X/linux/Release_Specific_Release_Notes.html#issues-tracker

I've made a note to update this thread if a solution is found.

Thanks,

Anshu

Hi Anshu,

Is it possible to know when you are planning to find a solution for this suspend/resume bug in your roadmap?

It is an important feature and none of the 6.x.x kernel versions seem to work.

As we said before, for our product power management is a key feature.

Best regards and thank you

Angel

Hi Angel,

There's no update as of now. I'll message the software development team if they have a plan

Best Regards,

Anshu

Hello Angel 

Just wanted to update this thread, to let you know  that this is being actively debugged and is planned to be sorted for an upcoming SDK9.3 release in Dec.  We are hoping that we will find the issue and fix in the next 1-2 weeks and if there is any incremental update we can share before the SDK release, we will post an update here in another week or so. 

Hello,

I tested again with the new kernel version 6.1.112 and still cannot get it to work.
I built the kernel config with the command "make ARCH=arm CROSS_COMPILE=arm-none-linux-gnueabihf- multi_v7_defconfig ti_multi_v7_prune.config no_smp.config" and then added CONFIG_DEBUG_GPIO and CONFIG_GPIO_SYSFS.
The device tree file configuration is
&uart0 {
pinctrl-names = "default";
pinctrl-0 = <&uart0_pins>;
status = "okay";
wakeup source;
}; /* uart0 */

gpio_buttons: gpio_buttons@0 {
compatible = "gpio-keys";
#address-cells = <1>;
#size-cells = <0>;
autorepeat;
switch@1 {
label = "MYbutton
linux,code = <0x102>;
gpios = <&gpio0 3 GPIO_ACTIVE_LOW>;
gpio-key,wakeup;
};

Then I put the system into standby mode with the command "echo standby > /sys/power/state" and the system does not wake up. I check that uart0 is receiving data and gpio0_3 is toggling.
The same test is done with the older 4.19.94 kernel and it works fine.

Anyway, we have a few questions after checking the fix you gave us.
- The fix is only implemented in a serial driver file, so should this change also fix the gpio0 wakeup error?
- What if we are not using the 8250_omap.c serial driver? Up to know we have been using CONFIG_SERIAL_OMAP instead of CONFIG_SERIAL_8250_OMAP.

If you need something else from us please let us know

Best regards

Angel

Hi Angel,

Bin verified this driver patch, but he is currently out of office this week. Please expect a response next week when he returns.

Feel free to provide any updates in the mean time.

Best,

Anshu