rpi3 clock drift

[James Shuriff]

I dumped the FDT from the system I built from source and I cannot find any reference to 0x337F980 (54 MHz). I looked at the FreeBSD source file generic_timer.c and it looks like it's grabbing the frequency from FDT but I cannot figure out where it's getting the 54 MHz from. clk_osc in the FDT is still showing 0x124f800 (19.2 MHz) on the system with the bad clock.

- James Shuriff

-----Original Message-----
From: James Shuriff
Sent: Friday, November 29, 2019 1:36 PM
To: 'Ian Lepore' <ian at freebsd.org>
Cc: Ross Alexander <rwa at athabascau.ca>; freebsd-arm at freebsd.org
Subject: RE: rpi3 clock drift

Thanks! I pulled the FDT from the Raspberry Pi running FreeBSD's RPI3 image (with the clock functioning). I think I found it on clock at 3:

clocks {

compatible = "simple-bus";
#address-cells = <0x1>;
#size-cells = <0x0>;
clock at 3 {

compatible = "fixed-clock";
reg = <0x3>;
#clock-cells = <0x0>;
clock-output-names = "osc";
clock-frequency = <0x124f800>;
phandle = <0x4>;
};
clock at 4 {

compatible = "fixed-clock";
reg = <0x4>;
#clock-cells = <0x0>;
clock-output-names = "otg";
clock-frequency = <0x1c9c3800>;
phandle = <0x18>;
};
};

clock at 3 is clk_osc and clock at 4 is clk_usb. I believe this is coming from bcm2710-rpi-3-b.dtb on the FAT partition, as I decompiled it and found the same clock defs. However the clock defs are the same in the file that's on my busted system.

- James Shuriff

-----Original Message-----
From: Ian Lepore <ian at freebsd.org>
Sent: Friday, November 29, 2019 1:16 PM
To: James Shuriff <james at opentech.cc>; Ross Alexander <rwa at athabascau.ca>; freebsd-arm at freebsd.org
Subject: Re: rpi3 clock drift

On Fri, 2019-11-29 at 18:06 +0000, James Shuriff wrote:
> Can you help me determine where the timecounter frequency is being
> pulled from? I believe that 54 MHz value is being pulled from the FDT
> I just don't know where specifically. My board is supposed to be 19.2
> MHz. I booted up FreeBSD's own RPI3 image and I got 19.2 MHz and
> didn't have any drift issues. Only when I use the environment I built
> from source do I get 54 MHz, as if an RPI 4 DTB somehow made it into
> the build.
>

You can get a definitive look at the fdt data in use by decompiling it on the running system, like this:

  sysctl -b hw.fdt.dtb | dtc -I dtb -O dts >/tmp/active.dts

That would help you confirm your theory by finding which node has the 54mhz value in it.  What to do beyond that, I'm not sure.  RPi fdt data always confuses me.  I think these days it's loaded from the msdos partition by uboot, but I'm not sure of that.

> I'm also interested in getting an RTC hat. I know how to build support
> for it in the bootloader and OS but how do you tell FreeBSD to start
> using it for system timing once it's connected?
>
> - James Shuriff
>

The only thing an RTC does is initially set the system time once at boot.  (On systems that support suspend/resume, it also resets the kernel clock on resume from sleep/low-power modes.)  There is no way to use an RTC as the system timecounter.

To use an RTC to set the time intially at boot, you only need to load its driver via the loader.conf setting and enable the i2c device with an fdt overlay.  The device driver registers itself as an RTC, and the kernel automatically sets the time from the registered RTC that has the best resolution.

-- Ian

________________________________
 DISCLAIMER: This message and any attachments are intended solely for the use of the recipient and may contain confidential information. If you have received this message in error please delete it and promptly notify the sender, James Shuriff (james at opentech.cc<mailto:james at opentech.cc>).