Linux/AM5728: McSPI interface fails

mike A

Part Number: AM5728
Other Parts Discussed in Thread: AM3352

Tool/software: Linux

Hello,

We are trying to interface a MCP2515 to MCSPI4, where D0 is output and D1 is input. I can see that there is CLK, CS, MOSI and MISO, but the mcpsi controller driver always reads value 0x0 from MISO.

Just to add I have configured the pinmux correctly.

Thanks,

over 8 years ago

0 mike A over 8 years ago

Genius 3490 points

Hello,

Following is my DT:

SPI Pinmux:

                    pinmux_spi4 {
                       pinctrl-single,pins = <
                                   0x60 0x40008 /* spi4_sclk (gpmc_a8) */
                                   0x64 0x40008 /* spi4_d1 (gpmc_a9) MISO */
                                   0x68 0x10008 /* spi3_d0 (gpmc_a10) MOSI */
                                   0x6c 0x20008 /* spi4_cs0 (gpmc_a11) */
                                      >;
                       linux,phandle = <0x181>;
                       phandle = <0x181>;
                   };

SPI Node:

spi@480ba000 {
          compatible = "ti,omap4-mcspi";
          reg = <0x480ba000 0x200>;
          interrupts = <0x0 0x2b 0x4>;
          #address-cells = <0x1>;
          #size-cells = <0x0>;
          ti,hwmods = "mcspi4";
          ti,spi-num-cs = <0x1>;
          dmas = <0xc9 0x46 0xc9 0x47>;
          ti,pindir-d0-out-d1-in;
          dma-names = "tx0", "rx0";
          status = "okay";
          pinctrl-names = "default";
          pinctrl-0 = <0x181>;

          can3@0 {
               compatible = "microchip,mcp2515";
               spi-max-frequency = <10000000>;
               spi-cs-high;
               reg = <0>;
               interrupt-parent = <0xc5>;
               interrupts = <18 0x2>;
              clocks = <0x400>;
        };
};

Thanks,
-Mike

0 Biser Gatchev-XID over 8 years ago in reply to mike A

TI__Guru**** 393215 points

Do you have input enabled on the SCK pinmux? This is requires for proper functioning.

0 mike A over 8 years ago in reply to Biser Gatchev-XID

Genius 3490 points

Hi,

Yes I have enabled input to SCK,

Following is my pinmux to spi4 pins:

pinmux_spi4 {
pinctrl-single,pins = <
0x60 0x40008 /* spi4_sclk (gpmc_a8) */
0x64 0x40008 /* spi4_d1 (gpmc_a9) MISO */
0x68 0x10008 /* spi3_d0 (gpmc_a10) MOSI */
0x6c 0x20008 /* spi4_cs0 (gpmc_a11) */
>;
linux,phandle = <0x181>;
phandle = <0x181>;
};

As per above " 0x60 0x40008 /* spi4_sclk (gpmc_a8) */" BIT(18) is set to 1 which is input.

Thanks,
Mike

0 Biser Gatchev-XID over 8 years ago in reply to mike A

TI__Guru**** 393215 points

I have notified the software team to check this. They will respond here.

0 Yordan Kovachev over 8 years ago in reply to mike A

TI__Guru**** 161600 points

Which SDK is this?

Best Regards,
Yordan

0 mike A over 8 years ago in reply to Yordan Kovachev

Genius 3490 points

Hi,

I am using the latest sdk release 03.03.00.

Thanks,
Mike

0 mike A over 8 years ago in reply to Yordan Kovachev

Genius 3490 points

Hi,

I am using the yocto ti-layer which has 4.4 release. I also had a look at the 4.9 branch on git and I see no major difference in the omap2-spi driver and mc2515 driver.

I am trying to get it working in mode 0 , I also tried the mode 1 but with no luck. I even followed [1] seems like the controller driver works on am3352 but not on Am572x.

[1] e2e.ti.com/.../2113031

Thanks,
-Mike

0 Yordan Kovachev over 8 years ago in reply to mike A

TI__Guru**** 161600 points

Hi Mike,

Check the values of the CTRL_CORE_PAD_GPMC_A9, CTRL_CORE_PAD_GPMC_A10 & CTRL_CORE_PAD_GPMC_A11 registers with devmem2.

If you're using the latest SDK you should configure the SPI4 pinmux settings in u-boot/board/ti/am57xx/mux_data.h.

Latest ti linux releases changed & now the pinmux settings are done in u-boot, not in dts.

If you still wish to use dts for pinmux settings, then try the following settings:
spi4_pins: spi4_pins {
pinctrl-single,pins = <
DRA7XX_CORE_IOPAD(0x3460, (PIN_INPUT | MUX_MODE8))
DRA7XX_CORE_IOPAD(0x3464, (PIN_INPUT | MUX_MODE8))
DRA7XX_CORE_IOPAD(0x3468, (PIN_INPUT | MUX_MODE8))
DRA7XX_CORE_IOPAD(0x346C, (PIN_INPUT | MUX_MODE8))
>;
};

Best Regards,
Yordan

0 mike A over 8 years ago in reply to Yordan Kovachev

Genius 3490 points

Hi,

Thanks for the reply.

Yes I can confirm that dev2mem is reporting proper pinmux values.

Is the motrolla mode 0 0, different TI mode 0 0 as I came across this thread [1] . I thought it should just work as it just works straight of
referring to thread [2]

[1] e2e.ti.com/.../207305

[2] e2e.ti.com/.../2113031

Thanks,
Mike

0 Yordan Kovachev over 8 years ago in reply to mike A

TI__Guru**** 161600 points

Hi Mike,

TI Mode 0 0, means POL = 0 (MCSPI_CHxCONF[1] POL = 0x0) & PHA = 0 (MCSPI_CHxCONF[0] PHA = 0x0) or the SPI works with spi_sclk is active high and sampling occurs on the rising edge.

You can see all SPI modes (depending on spi_clk polarity & phase settings) in Table 24-291. Phase and Polarity Combinations. Try a different mode to see if it will impact the behavior of your system.

Best Regards,
Yordan

0 mike A over 8 years ago in reply to Yordan Kovachev

Genius 3490 points

Hi Yordan,

I have scoped the pin on the processor and I can see that there is valid data coming out (I used a logic analyzer), but I still see the mcspi driver reads 0x0. Is it I need to re-read the register again for Rx as its just one byte ?

could you please help me here.

Thanks,
Mike

0 mike A over 8 years ago in reply to Yordan Kovachev

Genius 3490 points

But I can see that mcspi driver is still reading 0x0.

Thanks,

Mike

0 Yordan Kovachev over 8 years ago in reply to mike A

TI__Guru**** 161600 points

Hi Mike,

Unfortunately I have no way of recreating your issue.

Is it I need to re-read the register again for Rx as its just one byte ?

Yes try this & see if the behaviour would change.

Can you share the SPI register values? Maybe there is something wrong with the configuration of the interface .

Best Regards,
Yordan

Processors

Processors forum

Linux/AM5728: McSPI interface fails