PROCESSOR-SDK-AM335X: Regarding journal log messages

TI doesn't support kernel v5.10 for AM335x yet, we won't be able to comment on these log messages. But I can tell the message "usb: IRQ vbus not found" can be ignored, it only means your platform doesn't have external VBUS interrupt. The USB driver should be improved to not print this message.

Where do you get the v5.10 kernel? Have you tried the one on git.ti.com to see if it makes any difference?

https://git.ti.com/cgit/ti-linux-kernel/ti-linux-kernel/commit/?h=ti-linux-5.10.y

Hi,

I just built the TI v5.10 kernel and ran it on AM335x GP EVM, I can see the same log messages except "usb: IRQ vbus not found" and "timedout waiting for gfx:0".

I also checked the kernel v5.4 in Processor SDK release v7.3.0.5, I can still see these messages except those two as in v5.10.

So I would think these log messages are not critical.

BTY, I also checked the kernel v4.19 which is in Processor SDK v6.3.0.106, which log only has "omap_voltage_late_init: Voltage driver support not added" though.

Hi Bin Liu,

Thanks for the reply. I saw "omap_reset_deassert: timedout waiting for gfx:0" message is coming from omap_prm.c driver. Can I know is this driver required for system and Who is enabling it? and for what omap_prm driver is used?

Thanks & Regards,

Durga Prasad

Hi,

What is the AM335x device part number you use? Does it have the GPU module?

Hi,

We are using sitara AM3352BZCZA100.

Thanks & Reagrds

Durga Prasad

Hi Bin Liu,

Can I make below messages from error to warning severity. So that it will not log into journal.

 l3-aon-clkctrl:0000:0: failed to disable

debugfs: Directory '49000000.dma' with parent 'dmaengine' already present!

musb-dsps 47401400.usb: IRQ vbus not found

omap_voltage_late_init: Voltage driver support not added

Thanks & Regards,

Durga Prasad.

Hi Durga,

What is the problem with logging these messages? that is the purpose of logging, right? Why to hide them?

Anyway, if you really don't want to see them by default, you can modify the kernel source code, to replace function name pr_err() with pr_debug() in the places where these messages are generated.

Hi Bin Liu,

If these message are getting logged every time our journal will be filled with these messages only and get overflowed and how ever we are saving these in nand. Unnecessary nand R/W cycles will happen.

So to avoid those just we need confirmation like if these are critical we have to leave as it is and fix it , if not change it to warning. So that later for debugging it will be used but for final build we will disable, If it is not that critical.

Can we upstream the changes by changing it to pr_debug or pr_warn as well or Your team will do from your end?

Thanks & Regards,

Durga Prasad

Hi Durga,

Understood the reason.

Our dev team is till working on migrating kernel v5.10 to AM335x, so these messages might go away eventually in the next Processor SDK Linux release for AM335x. However some of them might still exist if it is not critical, as the last message already exists in Processor SDK v6.3. Whatever code changes made by our dev team will be upstreamed to the mainline.

Hi Bin Liu,

Thanks for reply. How about this "omap_reset_deassert: timedout waiting for gfx:0". How to disable gfx from kernel. Since GFX is not present in our part number.

I tried by disabling CONFIG_ARCH_OMAP2PLUS_TYPICAL in kernel config to disable omap_prm driver. But still I use to see above message. Is my understanding correct. What is the correct way to disable gfx from kernel?

Thanks & Regards

Durga Prasad.

Hello,

Please confirm if you have disabled the GPU in the DTS? The GPU node is defined in the following file: https://git.ti.com/cgit/ti-linux-kernel/ti-linux-kernel/tree/arch/arm/boot/dts/am33xx.dtsi?h=ti-linux-5.4.y#n476. In your DTS file, I would recommend adding something like the following: &gpu { status = "disabled";};

Regards,
Krunal

Hello Bhargav,

I am using 5.10 kernel, where in gpu label is not present in am33xx.dtsi.

https://git.kernel.org/pub/scm/linux/kernel/git/stable/linux.git/tree/arch/arm/boot/dts/am33xx.dtsi?h=v5.10.83

If I try to disable in dts my build will fail with error: Label or path gpu not found

target-module@56000000 {
                        compatible = "ti,sysc-omap4", "ti,sysc";
                        reg = <0x5600fe00 0x4>,
                              <0x5600fe10 0x4>;
                        reg-names = "rev", "sysc";
                        ti,sysc-midle = <SYSC_IDLE_FORCE>,
                                        <SYSC_IDLE_NO>,
                                        <SYSC_IDLE_SMART>;
                        ti,sysc-sidle = <SYSC_IDLE_FORCE>,
                                        <SYSC_IDLE_NO>,
                                        <SYSC_IDLE_SMART>;
                        clocks = <&gfx_l3_clkctrl AM3_GFX_L3_GFX_CLKCTRL 0>;
                        clock-names = "fck";
                        power-domains = <&prm_gfx>;
                        resets = <&prm_gfx 0>;
                        reset-names = "rstctrl";
                        #address-cells = <1>;
                        #size-cells = <1>;
                        ranges = <0 0x56000000 0x1000000>;

                        /*
                         * Closed source PowerVR driver, no child device
                         * binding or driver in mainline
                         */
                };

Thanks & Regards

Durga Prasad.

Hello,

That's strange because on TI git repo, the SGX node is defined even for the 5.10 branch. As an experiement, could you remove the target-module@56000000 node. The address 56000000 is for the GPU and if there is no GPU, I do not think the clock configuration is required and I would just remove that entry. I will also check with our developers and get back to you. 

Regards,
Krunal