Linux/AM4378: Kernel Hangs During Boot

Tanner,

I have also checked rest_init() function on the AM437x EVM. I made the below update of the default kernel source code:

linux-kernel/init/main.c

asmlinkage __visible void __init start_kernel(void)

{

........

/* Do the rest non-__init'ed, we're now alive */

       + printk("before rest_init()\n");

rest_init();

       + printk("after rest_init()\n");

}

I can see that "before rest_init()" is printed at very later stage of the kernel boot flow, I have so many console messages before that message. I can not see "after rest_init()" message, which means that it is normal that the flow never goes out of rest_init(). I am using PSDK v5.01, but should be similar for v4 of the PSDK.

Please find the console log I have attached as txt file:

https://e2e.ti.com/cfs-file/__key/communityserver-discussions-components-files/791/am437x_5F00_boot_5F00_log

So your kernel is booting, but you do not see console messages. Please check and verify your UART console.

Regards,
Pavel

Pavel,

I have verified both on the oscilloscope and in the CTRL_STS register that the value of SYS_BOOT[15:14] is set to 0x0.

It seems like from your second post that even during a normal boot flow does not get past the function rest_init(), so it seems like that might be a dead end...

I have attached the stripped down dts file that I am currently using. Our custom board also uses UART0 for the console and MMC0 for the SD card. U-Boot is printing everything out, so the console has to be set up correctly, right? Or is there a different spot in the kernel where you also have to set up the UART console?

Another thing we have done is use all of our software configuration stuff (U-Boot configuration, dts file, kernel config file), load it onto the SD card, and plug the SD card into the EVM and boot. It fails terribly obviously, but the kernel still prints out all of its messages. Our board I'm sure is failing somewhere during the kernel boot process, but I just can't understand why it isn't printing out anything at all.

/*
 * Copyright (C) 2013 Texas Instruments Incorporated - http://www.ti.com/
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License version 2 as
 * published by the Free Software Foundation.
 */

/* AM437x GP EVM */

/dts-v1/;

#include "am4372.dtsi"
#include <dt-bindings/pinctrl/am43xx.h>
#include <dt-bindings/pwm/pwm.h>
#include <dt-bindings/gpio/gpio.h>

/ {
	model = "TI AM437x UTI BOARD";
	compatible = "ti,am437x-UTI", "ti,am4372","ti,am43";

	chosen {
		stdout-path = &uart0;
	};
};

&am43xx_pinmux {
	uart_0_pins_default: uart_0_pins_default {
		pinctrl-single,pins = <
			AM4372_IOPAD(0x970, PIN_INPUT | MUX_MODE0)            // (K25) uart0_rxd.uart0_rxd
			AM4372_IOPAD(0x974, PIN_OUTPUT_PULLDOWN | MUX_MODE0)  // (J24) uart0_txd.uart0_txd
			/*AM4372_IOPAD(0x968, PIN_INPUT_PULLUP | MUX_MODE0)     // (L25) uart0_ctsn.uart0_ctsn
			AM4372_IOPAD(0x96c, PIN_OUTPUT_PULLDOWN | MUX_MODE0)  // (J25) uart0_rtsn.uart0_rtsn*/
		>;
	};

	// Optional sleep pin settings. Must manually enter values in the below skeleton.
	uart_0_pins_sleep: uart_0_pins_sleep {
		pinctrl-single,pins = <
			AM4372_IOPAD(0x970, PIN_INPUT | MUX_MODE0)            // (K25) uart0_rxd.uart0_rxd
			AM4372_IOPAD(0x974, PIN_OUTPUT_PULLDOWN | MUX_MODE0)  // (J24) uart0_txd.uart0_txd
			/*AM4372_IOPAD(0x968, PIN_INPUT_PULLUP | MUX_MODE0)     // (L25) uart0_ctsn.uart0_ctsn
			AM4372_IOPAD(0x96c, PIN_OUTPUT_PULLDOWN | MUX_MODE0)  // (J25) uart0_rtsn.uart0_rtsn*/
		>;
	};

	mmc_0_pins_default: mmc_0_pins_default {
		pinctrl-single,pins = <
			AM4372_IOPAD(0x960, PIN_INPUT | MUX_MODE7) // (R25) spi0_cs1.mmc0_sdcd
		>;
	};
};

&mmc1 {
	status = "okay";
	bus-width = <4>;
	pinctrl-names = "default";
	pinctrl-0 = <&mmc_0_pins_default>;
	broken-cd;
};

&uart0 {
	status = "okay";
	pinctrl-names = "default", "sleep";
	pinctrl-0 = <&uart_0_pins_default>;
	pinctrl-1 = <&uart_0_pins_sleep>;
};

Tanner,

Tanner Hunsucker73 said:

I tried making the pin settings exactly as you have listed above, but still nothing. When I run the md command in U-Boot, I got the following:

=> md 0x44e10970
44e10970: 000e0000 000b0000 08060007 08060007

I think those settings look right? Based on what the register values are set to in mux.c, although mux.c doesn't set the ctsn or rtsn registers, so I'm not sure what those values should be.

Please align your kernel DTS UART0 pinmux settings, with thse from u-boot, as these are known working settings.

Tanner Hunsucker73 said:

When you say "Make sure you have setup 'console=ttyO0,115200n8' in your bootargs for the kernel command line", does that mean just having console=ttyO0,115200n8 in the U-Boot environment? Because I do have that in there.

Yes, u-boot environment.

Tanner Hunsucker73 said:

Also, I can see that execution gets into the smp_setup_processor_id() function, and to the printk line, but nothing prints. So it seems like what you said in your previous response, that the kernel is booting but not printing anything, is correct. It seems like printk functions are being called, but not printing anything.

Have you mage any significant changes in your defconfig file? Can you attach your.config file, may be you have some UART misconfiguration there.

What else I can suggest you is to:

- probe the UART0 signals with scope and compare the signals with u-boot stage

- dump UART0 module registers settings in u-boot and in kernel and compare for differences

Also make sure you have remove any reference to RTC, see below e2e thread for details:

Regards,
Pavel

Hi,

For the moment the code will need to be commented out in the kernel to allow you continued development. I will need to research this a little further with the driver development team to understand if there is a config option or if commenting out this code is the best solution.

The other question asked by a user on this thread about the configurability of this RTC-less use case is one that I will also research where the DTS board process happens in relation to the low-level init process called hwmod. 

Best Regards,

Schuyler