Strange clock behavior for am33xx

Diego Sueiro

Folks,

I'm having an strange behavior on the clock used for TIMER1.

When I boot the kernel on BBB I have:

OMAP clocksource: timer1 at 24000000 Hz

And when I boot the same kernel and a dtb with few differences (mmc interface) on a custom am33xx board (based on BBB and the clock circuit is the same), I have:

OMAP clocksource: timer1 at 26000000 Hz

This is causing the system clock to drift a lot.

How kernel chooses the clock frequency giving the following dtsi and it's parameters?

http://lxr.free-electrons.com/source/arch/arm/boot/dts/am33xx-clocks.dtsi?v=3.14

http://lxr.free-electrons.com/source/arch/arm/boot/dts/am33xx-clocks.dtsi?v=3.14#L368

http://lxr.free-electrons.com/source/arch/arm/boot/dts/am33xx-clocks.dtsi?v=3.14#L14

I'm using Yocto Project and meta-ti on daisy branch and kernel 3.14.

I'm a lit bit lost on how to debug this. Any tips should be helpful.

over 10 years ago

0 Biser Gatchev-XID over 10 years ago

TI__Guru**** 393215 points

Hi Diego,

Please check the pullup/pulldown resistor configuration on SYSBOOT[15:14]. For 24MHz it should be 01b, for 26MHz it's 11b. You may have something wrong there. This configuration must match your main oscillator frequency.

0 Diego Sueiro over 10 years ago in reply to Biser Gatchev-XID

Prodigy 180 points

Thanks for the tip.
I've checked the resistors and it's configured for 24MHz (SYS_BOOT[15:14]=01b).

But on my dts I have:
uart5_pins: pinmux_uart5_pins {
pinctrl-single,pins = <
0x0D8 (PIN_INPUT | MUX_MODE6) /* lcd_data14.uart5_ctsn */
0x0DC (PIN_OUTPUT_PULLDOWN | MUX_MODE6) /* lcd_data15.uart5_rtsn */
0x0C0 (PIN_OUTPUT_PULLDOWN | MUX_MODE4) /* lcd_data8.uart5_txd */
0x0C4 (PIN_INPUT_PULLUP | MUX_MODE4) /* lcd_data9.uart5_rxd */
>;
};

Maybe this is interfering on the clock frequency detection?
I'm going to disabe uart5 and see what happens.

0 Biser Gatchev-XID over 10 years ago in reply to Diego Sueiro

TI__Guru**** 393215 points

Software won't help here. You should rather check if some external device isn't pulling LCD_DATA15 high while the processor is coming out of reset.

0 Diego Sueiro over 10 years ago in reply to Biser Gatchev-XID

Prodigy 180 points

Yes, I have a device connected on these pins and it's changing it's status.

I have followed the instructions on the link below and it's working as expected:

http://comments.gmane.org/gmane.linux.ports.arm.omap/126334

Thanks for your help.

0 Biser Gatchev-XID over 10 years ago in reply to Diego Sueiro

TI__Guru**** 393215 points

Glad you solved it and thanks for sharing your results on the forum!

Processors

Processors forum

Strange clock behavior for am33xx