09:20 <ukleinek> hmm, does arm64 use arch/arm/tools/syscall.tbl ?

11:01 <arnd> ukleinek: I had a patch to make it do that at some point, but that never got merged, and it still uses arch/arm64/include/asm/unistd32.h instead

11:12 <ukleinek> arnd: ah, thanks

11:14 <arnd> fwiw, if anyone wants to volunteer converting include/uapi/asm-generic/unistd.h and arch/arm64/include/asm/unistd32.h to the syscall.tbl format, I can help them figure out how to do it, it's been on my list for a long time, but I never quite around to doing it

14:53 <abelloni> arnd: WTF is this amd-pmc doing?

14:54 <arnd> abelloni: I was wondering if that was your reaction

14:54 <arnd> I have no idea, besides the build regression

14:54 <abelloni> I replied to the original patch but seriously...

14:55 <abelloni> I don't get how x86 still works when I see patches like that

15:11 <geertu> The x86 people are mad

15:14 <broonie> cute enterprise nonsense hacks

15:24 <arnd> https://www.irccloud.com/pastebin/nSKaTtWL/

15:24 <arnd> abelloni: if you want revenge ^

15:29 <abelloni> arnd: note that I'm not handling CONFIG_RTC_SYSTOHC_DEVICE anymore anyway, all the code has been moved out of drivers/rtc

15:30 <arnd> ah, I had not noticed that, I assumed it's still next to the Kconfig option

15:30 <abelloni> I guess we should also move the config option out :)

15:34 <abelloni> I convinced tglx to get maintainership of that

15:34 <abelloni> as this is NTP specific

15:38 <geertu> arnd: That's gonna break multi-platform kernels including AMC_PMC support?

15:38 <geertu> s/AMC/AMD/

15:40 <arnd> geertu: I think it will still set the time through update_persistent_clock64()/mach_set_rtc_mmss() in arch/x86/kernel/rtc.c

15:46 <geertu> arnd: no idea, timekeeping is like black magic to me ;-)

15:47 <arnd> geertu: I cleaned up that crud for m68k and a few other architectures at some point, didn't quite get it done for i386

15:47 <geertu> My poor Amiga has seen way too many e2fsck runs due to time (not) being set correctly during boot-up.

15:52 <arnd> abelloni: actually now that arch/x86/platform/intel-mid/intel_mid_vrtc.c is gone, we can probably remove the x86 version of update_persistent_clock64(): one calls the same thing as RTC_SYSTOHC does, and the other two (kvm and xen) are nops

15:52 <abelloni> I'd love to get rid of them :)

15:56 <abelloni> I think mach_set_rtc_mmss is just internal to x86 now

16:05 <arnd> looks like that was already the case. The only thing that has changed is that it's the only non-empty x86_platform.set_wallclock() now.

16:05 <arnd> For VM guests, there are still custom .get_wallclock() callbacks though

16:08 <arnd> which is an annoying inconsistency, but understandable: those platforms don't even provide an emulated RTC and don't really need a driver for one either, but they do know the exact time during boot