[Solved] Target mvebu RTC can be wrong

For Developers
1

Re: issue opened in FS#831, the RTC on mvebu targets can be borked, causing some grief.

Information from a rango which currently has a borked RTC:

root@bsaedgy:/etc# uname -a
Linux bsaedgy 4.9.61 #0 SMP Fri Nov 10 13:53:04 2017 armv7l GNU/Linux
root@bsaedgy:/etc# cat openwrt_version
r5297-bddffc5
root@bsaedgy:/etc# adjtimex
    mode:         0
-o  offset:       8839 us
-f  freq.adjust:  267864 (65536 = 1ppm)
    maxerror:     16000000
    esterror:     16000000
    status:       65 (PLL | UNSYNC)
-p  timeconstant: 8
    precision:    1 us
    tolerance:    32768000
-t  tick:         10000 us
    time.tv_sec:  1510770059
    time.tv_usec: 702501
    return value: 5 (clock not synchronized)
root@bsaedgy:/etc# hwclock -r
Wed Dec 31 16:59:59 1969  0.000000 seconds
root@bsaedgy:/etc# date
Wed Nov 15 11:21:48 MST 2017
root@bsaedgy:/etc# hwclock -w
root@bsaedgy:/etc# hwclock -r
Wed Dec 31 16:59:59 1969  0.000000 seconds

Time is being set via NTP (busybox), but with a warning on reboot regarding the time discrepancy, and sysfixtime is unable to set the RTC.

Based on OOTB kernel configs, it appears that every 11 minutes the kernel should be taking the current time from userspace(NTP), and updating RTC. But the RTC cannot be set via hwclock. Based on information in FS#831, it appears that a reset of the date facility is required by way of uboot command. I currently do not have a serial connection to this device, so I am unable to confirm.

The solution presented in FS#831, removal of RTC, seems less than optimal. So, the question is, where does the issue lie? Is this truly a uboot issue or...

2

Can you try this patch:
https://drive.google.com/open?id=1bmvGk4UEYo6JK9WWO1zQBR5wmV_0Ol0i

This is based on the patches from russel kings clearfog repo.
With this i'm able to set the time and date correctly with hwlock now.
But the clock goes slow :disappointed:

//edit
root@LEDE:~# hwclock -u
Thu Mar 8 21:05:57 2018 0.000000 seconds
root@LEDE:~# hwclock -u -w
root@LEDE:~# hwclock -u
Thu Mar 8 21:07:03 2018 0.000000 seconds
root@LEDE:~# date
Thu Mar 8 21:07:12 CET 2018

So in ~10min it goes off by like 3-4min.

But close enough :wink:

3

There are a couple patchwork patches to test.

4

I have moved this fix into our 2018.01 u-boot build. It also includes a host of additional patches and functionality. This may be an easier solution than flashing a new image.

https://github.com/SolidRun/u-boot/commits/v2018.01-solidrun-a38x

Our documentation team is trying to catch up with our wiki but the notable user features for this release is. Kconfig menu option for choosing different images (SDHC, MMC, SPI). EMMC now can be flashed and run entirely from the BOOT hardware partitions. integrating the bubt command for simple u-boot updates to SPI or eMMC from either tftp or usb.