Hi, >> Decreasing the realtime clock across the y2k38 threshold is one reliable way >> to reproduce. >> Allegedly this can also happen just by running ntpd, I have not managed to >> reproduce that other >> than booting with rtc at >2038 and then running ntp.
> Isn't it the expected behavior. Here is what I see for powermac: >... gives us a time in the documented 1904..2040 year range. I decided not This (or some related issue involving monotonic time going wonky) surfaced on a system running 4.5 kernel with sane clock values running ntpd. It may not be limited to setting the clock beyond 2038, that is just a convenient way to reproduce something visibly wrong. If the "ntpd-related monotonic" issue is separate from what I reported, I did not find anything in the commit history that looks likely to have fixed that, so I think that is still a thing. Another way of looking at the issue is that when you set the clock beyond 2038 (or the mystery ntpd thing happens), vdso and syscall versions of clock_gettime start reporting wildly different values for CLOCK_MONOTONIC. If someone manages to somehow interlace calls to those two assuming they are consistent, crazy bugs will ensue... Regards, Jakub
