On Sun, Oct 08, 2017 at 02:37:41PM +0200, Thorsten Leemhuis wrote: > Hi! Find below my second regression report for Linux 4.14. It lists 8 > regressions I'm currently aware of. One regression was fixed since last > weeks report. One was in there that shouldn't have been there. > > == Current regressions == > > "hangs when building e.g. perf" & "Random insta-reboots on AMD Phenom II" > Status: "Mr. Luto better revert the new lazy TLB flushing fun'n'games" > -> "Yeah, working on it. It's not a straightforward revert." > Note: TWIMC: Workaround: wrmsr -a 0xc0010015 0x1000018 > Reported: 2017-09-05 > https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1484723.html > https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1501379.html > https://www.mail-archive.com/linux-kernel@vger.kernel.org/msg1501570.html > Cause: 94b1b03b519b81c494900cb112aa00ed205cc2d9
Uhm, that particular register value in the workaround applies only to a particular CPU, and it sounds dangerous to use it as-is otherwise. On affected families (apparently 15h as well) it's bit 0x00000008 in register 0xc0010015, you need to rdmsr that register then or that bit with whatever value you had before. (And I have no clue about how it works -- merely repeating what the guys say. You'd want to not do so blindly.) Meow! -- ⢀⣴⠾⠻⢶⣦⠀ We domesticated dogs 36000 years ago; together we chased ⣾⠁⢰⠒⠀⣿⡁ animals, hung out and licked or scratched our private parts. ⢿⡄⠘⠷⠚⠋⠀ Cats domesticated us 9500 years ago, and immediately we got ⠈⠳⣄⠀⠀⠀⠀ agriculture, towns then cities. -- whitroth on /.