https://bugzilla.novell.com/show_bug.cgi?id=663675
https://bugzilla.novell.com/show_bug.cgi?id=663675#c7
--- Comment #7 from Marius Tomaschewski
Fam. 5/6 are 32bit CPUs. I suspect the hypervisor is confused by a mix of 32bit and 64bit guests and does invalid CPUID caching?
I think, we have a combination of two bugs here and should split the two bug reports again and fix the libgmp to not to call abort() or make unreliable assumptions bases on the cpu family. I've checked the cpu family on some hosts and I think, the cpu family flag in /proc/cpuinfo is not reliable to call abort() because of it: 64bit CPUs: AMD Athlon(tm) 64 X2 Dual Core Processor 4600+ cpu family : 15 AMD Athlon(tm) Dual Core Processor 5050e cpu family : 15 AMD Turion(tm) 64 Mobile Technology MT-34 cpu family : 15 Intel(R) Core(TM)2 Duo CPU E7200 @ 2.53GHz cpu family : 6 Intel(R) Core(TM)2 Duo CPU E8400 @ 3.00GHz cpu family : 6 Intel(R) Core(TM) i7 CPU 965 @ 3.20GHz cpu family : 6 Intel(R) Xeon(R) CPU X5376 @ 2.80GHz cpu family : 6 32bit CPUs: Intel(R) Xeon(TM) CPU 2.40GHz cpu family : 15 Pentium III (Coppermine) cpu family : 6 So as you see, Family 6 is either definitively not a 32bit CPU and family 15 is also not definitively a 64 bit CPU. I guess, this has to do with the detection in arch/x86/kernel/cpu/common.c, cpu_detect() (and another places in this file), where it is using cpuid() and makes some calculations ... -- Configure bugmail: https://bugzilla.novell.com/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are on the CC list for the bug.