Accessing D_CAN on Beaglebone

Hi Heinz,

After searching a while I found that it could be that not all clocks are enabled. Can someone please point me how I can enable the CAN module clocks. What are the clocks I have to enable and how?

You can find all the module clock definitions & declarations at    arch/arm/mach-omap2/clock3xxx_data.c

Please refer the following TI wiki links for enabling D_CAN driver in linux kernel with debugging.

http://e2e.ti.com/cfs-file.ashx/__key/telligent-evolution-components-attachments/00-791-01-00-00-83-15-24/d_5F00_can_5F00_on_5F00_Beaglebone.pptx

http://processors.wiki.ti.com/index.php/AM335X_DCAN_Driver_Guide

Titusrathinaraj Stalin said:

You can find all the module clock definitions & declarations at    arch/arm/mach-omap2/clock3xxx_data.c

Please refer the following TI wiki links for enabling D_CAN driver in linux kernel with debugging.

http://e2e.ti.com/cfs-file.ashx/__key/telligent-evolution-components-attachments/00-791-01-00-00-83-15-24/d_5F00_can_5F00_on_5F00_Beaglebone.pptx

http://processors.wiki.ti.com/index.php/AM335X_DCAN_Driver_Guide

Thanks,

 arch/arm/mach-omap2/clock3xxx_data.c seems to be
arch/arm/mach-omap2/ccock3xxx_data.c in may case, the Beaglebone Black and kernel 3.8.

The number of clocks defined in this file confuses me. I can not see which one is related to the DCAN module.
There is only:

   CLK(NULL,    "hecc_ck",    &hecc_ck,    CK_AM35XX),
But the 335 has DCAN and not HECC.

The other documents describe the SocketCAN driver and it's initialization. Thats a level above what I like to do, getting CAN working with the can4linux driver.

It would help getting some lines of code showing what a kernel module has to doe to get access to the CAN registers. May be thats only three to 4 calls. But I'm not able to get it correct and hope someone else can help. Hopefilly with some lines of code.

Thanks

 Heinz

Hi Heinz,

Heinz said:

 arch/arm/mach-omap2/clock3xxx_data.c seems to be
arch/arm/mach-omap2/ccock3xxx_data.c in may case, the Beaglebone Black and kernel 3.8.

Sorry for the mis-interpretation,

1) Please refer the below location for your clock declarations, 

arch/arm/mach-omap2/clock33xx_data.c

    CLK("d_can.0",    "fck",            &dcan0_fck,    CK_AM33XX),
    CLK("d_can.1",    "fck",            &dcan1_fck,    CK_AM33XX),
    CLK("d_can.0",    "ick",            &dcan0_ick,    CK_AM33XX),
    CLK("d_can.1",    "ick",            &dcan1_ick,    CK_AM33XX),

2) D_CAN pinmuxing done at board file at "arch/arm/mach-omap2/board-am335xevm.c"

static struct pinmux_config d_can_gp_pin_mux[] = {
    {"uart0_ctsn.d_can1_tx", OMAP_MUX_MODE2 | AM33XX_PULL_ENBL},
    {"uart0_rtsn.d_can1_rx", OMAP_MUX_MODE2 | AM33XX_PIN_INPUT_PULLUP},
    {NULL, 0},
};

static struct pinmux_config d_can_ia_pin_mux[] = {
    {"uart0_rxd.d_can0_tx", OMAP_MUX_MODE2 | AM33XX_PULL_ENBL},
    {"uart0_txd.d_can0_rx", OMAP_MUX_MODE2 | AM33XX_PIN_INPUT_PULLUP},
    {NULL, 0},
};

3) You can see the pin mux entry for AM33xx (for 8 types of modes) at at "arch/arm/mach-omap2/mux33xx.c" ,

    _AM33XX_MUXENTRY(UART0_CTSN, 0,
        "uart0_ctsn", NULL, "d_can1_tx", "i2c1_sda",
        "spi1_d0", NULL, NULL, "gpio1_8"),
    _AM33XX_MUXENTRY(UART0_RTSN, 0,
        "uart0_rtsn", NULL, "d_can1_rx", "i2c1_scl",
        "spi1_d1", "spi1_cs0", NULL, "gpio1_9"),
    _AM33XX_MUXENTRY(UART0_RXD, 0,
        "uart0_rxd", "spi1_cs0", "d_can0_tx", "i2c2_sda",
        NULL, NULL, NULL, "gpio1_10"),
    _AM33XX_MUXENTRY(UART0_TXD, 0,
        "uart0_txd", "spi1_cs1", "d_can0_rx", "i2c2_scl",
        NULL, NULL, NULL, "gpio1_11"),

Q1) Which SDK version are you using ?

TI provided ?

As per the TI release, kernel version should be linux_v3.2.

Download AM335x SDK from following link for beagle bone black(AM335x) .

http://software-dl.ti.com/sitara_linux/esd/AM335xSDK/latest/index_FDS.html

http://www.ti.com/tool/linuxezsdk-sitara

http://www.elinux.org/Beagleboard:BeagleBoneBlack#TI_EZSDK

http://www.elinux.org/Beagleboard:3.2_SDK_On_BeagleBone_Black

Titusrathinaraj

I'm working with the 3.8 kernel with a patch for The BBB applied.
I don't know where to search but can not find the clock definitions you have mentioned

File arch/arm/mach-omap2/clock33xx_data.c is not in place. instead I have
kernel/arch/arm/mach-omap2/cclock33xx_data.c

 * AM33XX Clock data
 *
 * Copyright (C) 2012 Texas Instruments Incorporated - http://www.ti.com/
 * Vaibhav Hiremath <hvaibhav@ti.com>
 ....
Interestingly SocketCAN compiles and works.

 Heinz

Hello,

I have a similar problem  with non-linefetch unhandled faults:

Unhandled fault: external abort on non-linefetch (0x1028) at 0xfa1d0008

But in my case the problem is quite undetermined, we have a custom hardware based on the BBB and on one prototypes we face this problems with a certain probability. We use the CAN as socket CAN. and what I am seeing is the following:

1) The kernel is booting, during driver probing the CAN driver is accessing a CAN register without problems

2) Before the CAN driver is configured via canconfig "something" (I guess a socket daemon) is addressing the CAN hardware to read a CAN hardware register.

3) This operation could lead in seldom, undetermined cases to the above mentioned kernel crash.

4) When I suppress the first access to the CAN register the problem is gone. 

So my best guess is also the clock configuration of the CAN hardware module. But I am very confused why this problem is only occurring during bootup and not in normal CAN operation mode.

best

Matthias