Bug#802926: linux-image-4.2.0-1-amd64: KVM hangs with 100% cpu on 4.2
This still affects linux-image-4.2.0-1-amd644.2.5-1 but it is fixed with linux-image-4.3.0-rc7-amd64 4.3~rc7-1~exp1
Bug#802926: linux-image-4.2.0-1-amd64: KVM hangs with 100% cpu on 4.2
It seems that real mode virtualisation on Nehalem has regressed in 4.2: On Sun, 2015-10-25 at 10:08 +0100, Stefan Fritsch wrote: [...] > I cannot use KVM with 4.2, qemu loops with 100% CPU during seabios > initialization. Booting with the latest linux-image-4.1.0-2-amd64 fixes > the issue. [...] > kvm tracing shows: > > ... > qemu-system-x86-3219 [007] 1090.728418: kvm_set_irq: gsi 12 level 0 > source 0 > qemu-system-x86-3219 [007] 1090.728418: kvm_pic_set_irq: chip 1 pin 4 > (edge|masked) > qemu-system-x86-3219 [007] 1090.728419: kvm_ioapic_set_irq: pin 12 > dst 0 vec=0 (Fixed|physical|edge|masked) > qemu-system-x86-3219 [007] 1090.728419: kvm_set_irq: gsi 1 level 0 > source 0 > qemu-system-x86-3219 [007] 1090.728420: kvm_pic_set_irq: chip 0 pin 1 > (edge|masked) > qemu-system-x86-3219 [007] 1090.728420: kvm_ioapic_set_irq: pin 1 dst > 0 vec=0 (Fixed|physical|edge|masked) > qemu-system-x86-3219 [007] 1090.728420: kvm_set_irq: gsi 12 level 0 > source 0 > qemu-system-x86-3219 [007] 1090.728421: kvm_pic_set_irq: chip 1 pin 4 > (edge|masked) > qemu-system-x86-3219 [007] 1090.728421: kvm_ioapic_set_irq: pin 12 > dst 0 vec=0 (Fixed|physical|edge|masked) > qemu-system-x86-3219 [007] 1090.728723: kvm_set_irq: gsi 0 level 0 > source 0 > qemu-system-x86-3219 [007] 1090.728724: kvm_pic_set_irq: chip 0 pin 0 > (edge) > qemu-system-x86-3219 [007] 1090.728725: kvm_ioapic_set_irq: pin 2 dst > 0 vec=0 (Fixed|physical|edge|masked) > qemu-system-x86-3219 [007] 1090.728725: kvm_set_irq: gsi 0 level 0 > source 0 > qemu-system-x86-3219 [007] 1090.728725: kvm_pic_set_irq: chip 0 pin 0 > (edge) > qemu-system-x86-3219 [007] 1090.728725: kvm_ioapic_set_irq: pin 2 dst > 0 vec=0 (Fixed|physical|edge|masked) > qemu-system-x86-3219 [007] 1090.728726: kvm_set_irq: gsi 0 level 0 > source 0 > qemu-system-x86-3219 [007] 1090.728726: kvm_pic_set_irq: chip 0 pin 0 > (edge) > qemu-system-x86-3219 [007] 1090.728726: kvm_ioapic_set_irq: pin 2 dst > 0 vec=0 (Fixed|physical|edge|masked) > qemu-system-x86-3221 [000] d... 1090.729926: kvm_write_tsc_offset: vcpu=0 > prev=18446740943986499809 next=18446740943986499809 > qemu-system-x86-3221 [000] 1090.729927: kvm_track_tsc: vcpu_id 0 > masterclock 1 offsetmatched 0 nr_online 1 hostclock tsc > qemu-system-x86-3221 [000] 1090.730004: kvm_update_master_clock: > masterclock 1 hostclock tsc offsetmatched 1 > qemu-system-x86-3221 [000] d... 1090.730010: kvm_entry: vcpu 0 > qemu-system-x86-3221 [000] 1090.730013: kvm_emulate_insn: > :fff0:ea 5b e0 00 f0 (real) > qemu-system-x86-3221 [000] d... 1090.730016: kvm_entry: vcpu 0 > qemu-system-x86-3221 [000] d... 1090.730017: kvm_exit: reason > EPT_VIOLATION rip 0xe05b info 81 0 > qemu-system-x86-3221 [000] 1090.730018: kvm_page_fault: address > feffc000 error_code 81 > qemu-system-x86-3221 [000] d... 1090.730032: kvm_entry: vcpu 0 > qemu-system-x86-3221 [000] d... 1090.730034: kvm_exit: reason > EXCEPTION_NMI rip 0xe05b info 0 8b08 > qemu-system-x86-3221 [000] 1090.730035: kvm_inj_exception: #DF (0x0) > qemu-system-x86-3221 [000] d... 1090.730040: kvm_entry: vcpu 0 > qemu-system-x86-3221 [000] d... 1090.730041: kvm_exit: reason > EXCEPTION_NMI rip 0xfea5 info 0 8b08 > qemu-system-x86-3221 [000] 1090.730041: kvm_inj_exception: #DF (0x0) > qemu-system-x86-3221 [000] d... 1090.730043: kvm_entry: vcpu 0 > ... [...] > I am rather surprised that there is no bug report about this yet, so I > guess it only happens on some hardware. Mine is a Core i7-860. Ben. -- Ben Hutchings Beware of bugs in the above code; I have only proved it correct, not tried it. - Donald Knuth signature.asc Description: This is a digitally signed message part
Bug#802926: linux-image-4.2.0-1-amd64: KVM hangs with 100% cpu on 4.2
Package: src:linux Version: 4.2.3-2 Severity: important Hi, I cannot use KVM with 4.2, qemu loops with 100% CPU during seabios initialization. Booting with the latest linux-image-4.1.0-2-amd64 fixes the issue. The simplest reproducer is qemu-system-x86_64 --enable-kvm but it happens with all kinds of qemu command lines, directly or via libvirt. kvm tracing shows: ... qemu-system-x86-3219 [007] 1090.728418: kvm_set_irq: gsi 12 level 0 source 0 qemu-system-x86-3219 [007] 1090.728418: kvm_pic_set_irq: chip 1 pin 4 (edge|masked) qemu-system-x86-3219 [007] 1090.728419: kvm_ioapic_set_irq: pin 12 dst 0 vec=0 (Fixed|physical|edge|masked) qemu-system-x86-3219 [007] 1090.728419: kvm_set_irq: gsi 1 level 0 source 0 qemu-system-x86-3219 [007] 1090.728420: kvm_pic_set_irq: chip 0 pin 1 (edge|masked) qemu-system-x86-3219 [007] 1090.728420: kvm_ioapic_set_irq: pin 1 dst 0 vec=0 (Fixed|physical|edge|masked) qemu-system-x86-3219 [007] 1090.728420: kvm_set_irq: gsi 12 level 0 source 0 qemu-system-x86-3219 [007] 1090.728421: kvm_pic_set_irq: chip 1 pin 4 (edge|masked) qemu-system-x86-3219 [007] 1090.728421: kvm_ioapic_set_irq: pin 12 dst 0 vec=0 (Fixed|physical|edge|masked) qemu-system-x86-3219 [007] 1090.728723: kvm_set_irq: gsi 0 level 0 source 0 qemu-system-x86-3219 [007] 1090.728724: kvm_pic_set_irq: chip 0 pin 0 (edge) qemu-system-x86-3219 [007] 1090.728725: kvm_ioapic_set_irq: pin 2 dst 0 vec=0 (Fixed|physical|edge|masked) qemu-system-x86-3219 [007] 1090.728725: kvm_set_irq: gsi 0 level 0 source 0 qemu-system-x86-3219 [007] 1090.728725: kvm_pic_set_irq: chip 0 pin 0 (edge) qemu-system-x86-3219 [007] 1090.728725: kvm_ioapic_set_irq: pin 2 dst 0 vec=0 (Fixed|physical|edge|masked) qemu-system-x86-3219 [007] 1090.728726: kvm_set_irq: gsi 0 level 0 source 0 qemu-system-x86-3219 [007] 1090.728726: kvm_pic_set_irq: chip 0 pin 0 (edge) qemu-system-x86-3219 [007] 1090.728726: kvm_ioapic_set_irq: pin 2 dst 0 vec=0 (Fixed|physical|edge|masked) qemu-system-x86-3221 [000] d... 1090.729926: kvm_write_tsc_offset: vcpu=0 prev=18446740943986499809 next=18446740943986499809 qemu-system-x86-3221 [000] 1090.729927: kvm_track_tsc: vcpu_id 0 masterclock 1 offsetmatched 0 nr_online 1 hostclock tsc qemu-system-x86-3221 [000] 1090.730004: kvm_update_master_clock: masterclock 1 hostclock tsc offsetmatched 1 qemu-system-x86-3221 [000] d... 1090.730010: kvm_entry: vcpu 0 qemu-system-x86-3221 [000] 1090.730013: kvm_emulate_insn: :fff0:ea 5b e0 00 f0 (real) qemu-system-x86-3221 [000] d... 1090.730016: kvm_entry: vcpu 0 qemu-system-x86-3221 [000] d... 1090.730017: kvm_exit: reason EPT_VIOLATION rip 0xe05b info 81 0 qemu-system-x86-3221 [000] 1090.730018: kvm_page_fault: address feffc000 error_code 81 qemu-system-x86-3221 [000] d... 1090.730032: kvm_entry: vcpu 0 qemu-system-x86-3221 [000] d... 1090.730034: kvm_exit: reason EXCEPTION_NMI rip 0xe05b info 0 8b08 qemu-system-x86-3221 [000] 1090.730035: kvm_inj_exception: #DF (0x0) qemu-system-x86-3221 [000] d... 1090.730040: kvm_entry: vcpu 0 qemu-system-x86-3221 [000] d... 1090.730041: kvm_exit: reason EXCEPTION_NMI rip 0xfea5 info 0 8b08 qemu-system-x86-3221 [000] 1090.730041: kvm_inj_exception: #DF (0x0) qemu-system-x86-3221 [000] d... 1090.730043: kvm_entry: vcpu 0 ... >From there, the last 3 lines are repeated forever. The trace complete from the >beginning is available at http://www.sfritsch.de/~stf/kvm.trace.xz I am rather surprised that there is no bug report about this yet, so I guess it only happens on some hardware. Mine is a Core i7-860. -- Package-specific info: ** Version: Linux version 4.2.0-1-amd64 (debian-ker...@lists.debian.org) (gcc version 4.9.3 (Debian 4.9.3-4) ) #1 SMP Debian 4.2.3-2 (2015-10-14) ** Command line: BOOT_IMAGE=/boot/vmlinuz-4.2.0-1-amd64 root=UUID=d73302da-e64e-4b2f-bfa4-8bf2dc4df11d ro ** Not tainted ** Kernel log: [4.877300] iTCO_vendor_support: vendor-support=0 [4.879419] input: HDA Intel MID Rear Mic as /devices/pci:00/:00:1b.0/sound/card0/input9 [4.879573] input: HDA Intel MID Front Mic as /devices/pci:00/:00:1b.0/sound/card0/input10 [4.879726] input: HDA Intel MID Line as /devices/pci:00/:00:1b.0/sound/card0/input11 [4.879875] input: HDA Intel MID Line Out Front as /devices/pci:00/:00:1b.0/sound/card0/input12 [4.880221] input: HDA Intel MID Line Out Surround as /devices/pci:00/:00:1b.0/sound/card0/input13 [4.882180] input: HDA Intel MID Line Out CLFE as /devices/pci:00/:00:1b.0/sound/card0/input14 [4.882341] nouveau [ DEVICE][:01:00.0] BOOT0 : 0x0a8280a2 [4.882347] input: HDA Intel MID Line Out Side as /devices/pci:00/:00:1b.0/sound/card0/input15 [4.882468] nouveau [