Hello Rick, On Sun, Sep 29, 2019 at 07:47:40PM -0700, Rick Thomas wrote: > > On Sep 29, 2019, at 1:05 PM, Uwe Kleine-König <uwe@kleine-koenig.org> wrote: > > > > The step from 4.19.67-2 to 4.19.67-2+deb10u1 is harmless for this > > problem. The issue that John is describing comes from a change by Roger > > Shimizu to reduce the kernel size[1]. I guess there is some difference > > in userspace that is relevant for the problem not occuring for Rick. > > https://bugs.debian.org/855203 seems relevant here. > > > >>> yet installed on this machine. I think I’ll hold off on installing > >>> that update for a while! > >>> > >>> And, FWIW: > >>> > >>> rbthomas@sheeva:~$ grep RTC_DRV_CMOS /boot/config-4.19.0-6-marvell > >>> CONFIG_RTC_DRV_CMOS=m > > > > Note that CONFIG_RTC_DRV_CMOS isn't relevant for the Sheevaplug. The > > relevant symbol is CONFIG_RTC_DRV_MV there. > > So here’s what I get for that: > rbthomas@sheeva:~$ grep RTC_DRV_MV /boot/config-4.19.0-6-marvell > CONFIG_RTC_DRV_MV=m > > > I wonder if it would be sensible to implement the logic for hctosys > > (the "during boot" part that is) in rtc_register instead. This way it > > would not trigger to early if the relevant driver was modular (as is the > > case here). > > Could you fill in the details a bit more here? I assume this is > something to do with the systemd way of initializing the system, which > I’m afraid I don’t understand as well as I probably should… please > forgive my ignorance! IIRC systemd-timesyncd fast forwards the time at boot time to be not earlier than before the last shutdown. Maybe this is not active for you? (There is /lib/systemd/system/systemd-timesyncd.service.d/disable-with-time-daemon.conf on my system that disables timesyncd when ntpd or similar software is installed.) > In particular, where/when is the “hwclock hctosys” being done during > boot currently? And where/when would it be done under your proposal? Currently it is tried to be done in a late initcall in the kernel. This is where hctosys: unable to open rtc device (rtc0) is emitted. This is however before modules are loaded, so the rtc-mv driver isn't available yet. The rtc-mv driver calls rtc_register_device() when it was loaded and the device is found. This would IMHO be a more suitable point in time to sync the clock. There might however be some side effects I'm missing as userspace is already running then. I added Alexandre Belloni (= Linux RTC maintainer) to Cc, maybe he can/want to comment. > And, finally, would the suggested change work for other types of > hardware than armel? Yeah, this would also help on other archs that have the RTC drivers built-in. Then the drivers could be changed to modular (again) without the problem you're seeing. Best regards Uwe
Attachment:
signature.asc
Description: PGP signature