[Bug 565302] New: system time runs double speed with PAE kernel
http://bugzilla.novell.com/show_bug.cgi?id=565302 http://bugzilla.novell.com/show_bug.cgi?id=565302#c0 Summary: system time runs double speed with PAE kernel Classification: openSUSE Product: openSUSE 11.1 Version: Final Platform: 32bit OS/Version: Other Status: NEW Severity: Normal Priority: P5 - None Component: Kernel AssignedTo: bnc-team-screening@forge.provo.novell.com ReportedBy: joachim.deguara@amd.com QAContact: qa@suse.de Found By: --- Blocker: --- User-Agent: Mozilla/5.0 (X11; U; Linux i686 (x86_64); en-US; rv:1.9.0.15) Gecko/2009102100 SUSE/3.0.15-0.1.2 Firefox/3.0.15 On some AMD systems (specifically tested on the Dinar) using the *updated* 32bit PAE kernel the system time will run twice as fast. The install kernel 2.6.27.7-9 worked fine but a change in the update 2.6.27.19-3.2 (and further kernels) brok the system time. The workaround currently has been to boot with apic=bigsmp however through a tedious bisect I belive I found the bad change. The following change make the clock run twice as fast Index: kernel-src/include/asm-x86/mach-bigsmp/mach_apic.h =================================================================== --- kernel-src.orig/include/asm-x86/mach-bigsmp/mach_apic.h +++ kernel-src/include/asm-x86/mach-bigsmp/mach_apic.h @@ -131,7 +131,7 @@ static inline unsigned int cpu_mask_to_a int cpu; int apicid; - cpu = first_cpu(cpumask); + cpu = first_cpu(*cpumask); apicid = cpu_to_logical_apicid(cpu); return apicid; } This change is also not in the upstream stable commits to the 2.6.27 kernel. Why was this change added? Can you verify that this change is bad on AMD hardware and if so revert the change? Reproducible: Always Steps to Reproduce: 1. 2. 3. -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
http://bugzilla.novell.com/show_bug.cgi?id=565302
http://bugzilla.novell.com/show_bug.cgi?id=565302#c
Joachim Deguara
http://bugzilla.novell.com/show_bug.cgi?id=565302
http://bugzilla.novell.com/show_bug.cgi?id=565302#c
shuang qiu
http://bugzilla.novell.com/show_bug.cgi?id=565302
http://bugzilla.novell.com/show_bug.cgi?id=565302#c1
Jeff Mahoney
http://bugzilla.novell.com/show_bug.cgi?id=565302
http://bugzilla.novell.com/show_bug.cgi?id=565302#c2
Jiri Slaby
This change is also not in the upstream stable commits to the 2.6.27 kernel. Why was this change added?
Because of 4096 CPUs support patches.
Can you verify that this change is bad on AMD hardware and if so revert the change?
No, the change seems to be correct, as far as I understand your description. first_cpu takes cpumask_t typed variable. cpu_mask_to_apicid takes pointer to that, hence the dereference. You can't even compile the kernel if you revert the commit. Could you specify more precisely what is the problem? Also what's your system? I guess this is SLE 11, not openSUSE 11.1? -- Configure bugmail: http://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.
On some AMD systems (specifically tested on the Dinar) I run into "Time is running twice as fast" issue on the Dinar when booting with
http://bugzilla.novell.com/show_bug.cgi?id=565302
http://bugzilla.novell.com/show_bug.cgi?id=565302#c3
--- Comment #3 from Thomas Renninger
participants (1)
-
bugzilla_noreply@novell.com