Bug 150851
Summary: | general protection fault: 0000 [#1] SMP; native_read_pmc+0x7/0x40 | ||
---|---|---|---|
Product: | Virtualization | Reporter: | JianhongYin (yin-jianhong) |
Component: | kvm | Assignee: | virtualization_kvm |
Status: | NEW --- | ||
Severity: | normal | CC: | asselsm, oenhan, wanpeng.li |
Priority: | P1 | ||
Hardware: | x86-64 | ||
OS: | Linux | ||
Kernel Version: | 4.7.0 | Subsystem: | |
Regression: | No | Bisected commit-id: |
Description
JianhongYin
2016-08-01 04:04:52 UTC
How to reproduce it? I guess it has nothing with virtulization, please give me your CPU information with "cat /proc/cpuinfo | head -n 30" (In reply to Wanpeng Li from comment #1) > How to reproduce it? have not reproduced again yet. (In reply to Huaitong Han from comment #2) > I guess it has nothing with virtulization, please give me your CPU > information with "cat /proc/cpuinfo | head -n 30" The machine is used by others, got it's info from internal manage system: System ------------------------------------------ Host Hypervisor (not virtualized) Vendor IBM Model System x3650 -[7979AC1]- Serial Number KQHTLVV MAC Address 00:1A:64:C7:EC:08 Memory 34029 MB NUMA Nodes 1 CPU ------------------------------------------ Vendor Intel Corp. Model Name Xeon Family 6 Model 23 Stepping 6 Speed 1992.0 Processors 8 Cores 8 Sockets 2 Hyper False Flags fpu fpu_exception wp vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx x86-64 constant_tsc arch_perfmon pebs bts rep_good nopl aperfmperf pni dtes64 monitor ds_cpl vmx est tm2 ssse3 cx16 xtpr pdcm dca sse4_1 lahf_lm dtherm tpr_shadow vnmi flexpriority cpufreq Arch(s) i386 x86_64 You are not being clear if this is falling over in the guest or native, which is it? Can you use gdb to confirm the instruction which is failing? I am no expert but I have built the latest mainline kernel as of today and I believe the failure is in the call to 'rdpmc'. The PCE bit (9th bit) in CR4 appears to be 0 and thus RDPMC can only be used in ring 0. If it is a guest which is failing this might explain the oops. If this is the guest and you are using "-cpu host" you can try "-cpu host,level=9" to disable PMU emulation (see the comments here https://bugs.launchpad.net/qemu/+bug/1037675) RCX: 000000000000001e is suspicious. This seems to be out of range. This looks very much like what is fixed in commit 65ea11ec6a82b1d44aba62b59e9eb20247e57c6e [x86/hweight: Don't clobber %rdi] Can you try with the above applied? |