: kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument

Qemu-devel Applied Luiz Capitulino 6 participants 21 messages May 30, 2013 – Jun 3, 2013 KVM

lore

Share & tools

lore

Luiz Capitulino

author

2013-05-30 16:00 UTC #0 lore raw

The culprit is commit:

commit 235e8982ad393e5611cb892df54881c872eea9e1

Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700

kvm: support using KVM_MEM_READONLY flag for regions

I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().

Paolo Bonzini

maintainer

2013-05-30 16:03 UTC #1 lore raw

Il 30/05/2013 17:46, Luiz Capitulino ha scritto:

Quoted 11 lines

The culprit is commit:
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().

Reproducer?

Paolo

Luiz Capitulino

author

2013-05-30 16:38 UTC #2 lore raw

Quoted 7 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:

Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 6 lines
The culprit is commit:
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700

Quoted 7 lines

Quoted 5 lines
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
Reproducer?

I just try to start a VM (HEAD 87d23f7):

~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/

Jordan Justen

2013-05-30 16:50 UTC #3 lore raw

Quoted 6 lines · Luiz Capitulino

Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:

Quoted 8 lines

Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().

As noted in the code, the first call is for KVM commit 75d61fbc.

I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)

Xiao, any thoughts?

Quoted 8 lines

Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/

Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.

Does the firmware behave as a ROM for you?

-Jordan

Luiz Capitulino

author

2013-05-30 17:04 UTC #4 lore raw

Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:

Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/

Quoted 4 lines

Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?

I think so:

(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)

Is this what you're asking?

Jordan Justen

2013-05-30 17:32 UTC #5 lore raw

Quoted 39 lines · Luiz Capitulino

Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/
Quoted 4 lines
Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?

Quoted 8 lines

I think so:
(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)
Is this what you're asking?

I guess I was meaning ... if you write to an address such as 0xfffffff0, does it update as RAM, or does it retain the original value?

This is easy to test in OVMF at the EFI shell, but I'm not sure how you could easily test it otherwise.

Does the system actually boot for you after the error message?

-Jordan

Luiz Capitulino

author

2013-05-30 17:57 UTC #6 lore raw

Quoted 48 lines · On Thu, 30 May 2013 10:32:36 -0700 Jordan Justen <[email protected]> wrote:

Quoted 46 lines · Luiz Capitulino
Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/
Quoted 4 lines
Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?
I think so:
(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)

Quoted 8 lines

Quoted 1 line
Is this what you're asking?
I guess I was meaning ... if you write to an address such as 0xfffffff0, does it update as RAM, or does it retain the original value?
This is easy to test in OVMF at the EFI shell, but I'm not sure how you could easily test it otherwise.

I could try to hack something, but maybe not today.

Quoted 1 line

Does the system actually boot for you after the error message?

No, I get an abort. That's what kvm_set_phys_mem() does when kvm_set_user_memory_region() fails:

(gdb) bt

#0  0x00007f01f8592ba5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:63
#1  0x00007f01f8594358 in __GI_abort () at abort.c:90
#2  0x00007f01fedb57f1 in kvm_set_phys_mem (section=0x7f01ee6975d0, add=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:692
#3  0x00007f01fedb5bfa in kvm_region_del (listener=0x7f01ff2bc360 <kvm_memory_listener>, section=0x7f01ee6975d0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:795
#4  0x00007f01fedbc92d in address_space_update_topology_pass (as=0x7f01ffadfa60 <address_space_memory>, old_view=..., new_view=..., adding=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:689
#5  0x00007f01fedbd125 in address_space_update_topology (as=0x7f01ffadfa60 <address_space_memory>) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:725
#6  0x00007f01fedbd29c in memory_region_transaction_commit () at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:750
#7  0x00007f01fec1b977 in i440fx_update_memory_mappings (d=0x7f0200511940) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:131
#8  0x00007f01fec1bab3 in i440fx_write_config (dev=0x7f0200511940, address=94, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:154
#9  0x00007f01fec29a9a in pci_host_config_write_common (pci_dev=0x7f0200511940, addr=94, limit=256, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:54

#10 0x00007f01fec29ba2 in pci_data_write (s=0x7f0200502a50, addr=2147483742, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:75
#11 0x00007f01fec29d5e in pci_host_data_write (opaque=0x7f02004fa1c0, addr=2, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:128
#12 0x00007f01fedbac6e in memory_region_write_accessor (opaque=0x7f02004fc598, addr=2, value=0x7f01ee697a70, size=1, shift=0, mask=255) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:334 #13 0x00007f01fedbad50 in access_with_adjusted_size (addr=2, value=0x7f01ee697a70, size=1, access_size_min=1, access_size_max=4, access=0x7f01fedbabe9 <memory_region_write_accessor>, opaque=0x7f02004fc598)
at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:364
#14 0x00007f01fedbb1b8 in memory_region_iorange_write (iorange=0x7f0200503550, offset=2, width=1, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:439 #15 0x00007f01fedb33aa in ioport_writeb_thunk (opaque=0x7f0200503550, addr=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:212
#16 0x00007f01fedb2d84 in ioport_write (index=0, address=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:83
#17 0x00007f01fedb3924 in cpu_outb (addr=3326, val=51 '3') at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:289
#18 0x00007f01fedb76c5 in kvm_handle_io (port=3326, data=0x7f01fea6e000, direction=1, size=1, count=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1507 #19 0x00007f01fedb7d59 in kvm_cpu_exec (env=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1659
#20 0x00007f01fed3dc3f in qemu_kvm_cpu_thread_fn (arg=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/cpus.c:759
#21 0x00007f01fcf02d15 in start_thread (arg=0x7f01ee698700) at pthread_create.c:308
#22 0x00007f01f864f48d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:114
(gdb)

Paolo Bonzini

maintainer

2013-05-30 18:06 UTC #7 lore raw

Il 30/05/2013 19:56, Luiz Capitulino ha scritto:

Quoted 53 lines

Quoted 50 lines · On Thu, 30 May 2013 10:32:36 -0700 Jordan Justen <[email protected]> wrote:
Quoted 47 lines · Luiz Capitulino
Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/
Quoted 4 lines
Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?
I think so:
(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)
Is this what you're asking?

Quoted 8 lines

Quoted 6 lines
I guess I was meaning ... if you write to an address such as 0xfffffff0, does it update as RAM, or does it retain the original value?
This is easy to test in OVMF at the EFI shell, but I'm not sure how you could easily test it otherwise.
I could try to hack something, but maybe not today.

Just put a breakpoint on pflash_cfi01_register and see if it is reached.

I cannot reproduce it, but I'm also on 3.8.x. Will look at it tomorrow.

Paolo

Quoted 32 lines

Quoted 1 line
Does the system actually boot for you after the error message?
No, I get an abort. That's what kvm_set_phys_mem() does when kvm_set_user_memory_region() fails:
(gdb) bt
#0  0x00007f01f8592ba5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:63
#1  0x00007f01f8594358 in __GI_abort () at abort.c:90
#2  0x00007f01fedb57f1 in kvm_set_phys_mem (section=0x7f01ee6975d0, add=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:692
#3  0x00007f01fedb5bfa in kvm_region_del (listener=0x7f01ff2bc360 <kvm_memory_listener>, section=0x7f01ee6975d0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:795
#4  0x00007f01fedbc92d in address_space_update_topology_pass (as=0x7f01ffadfa60 <address_space_memory>, old_view=..., new_view=..., adding=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:689
#5  0x00007f01fedbd125 in address_space_update_topology (as=0x7f01ffadfa60 <address_space_memory>) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:725
#6  0x00007f01fedbd29c in memory_region_transaction_commit () at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:750
#7  0x00007f01fec1b977 in i440fx_update_memory_mappings (d=0x7f0200511940) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:131
#8  0x00007f01fec1bab3 in i440fx_write_config (dev=0x7f0200511940, address=94, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:154
#9  0x00007f01fec29a9a in pci_host_config_write_common (pci_dev=0x7f0200511940, addr=94, limit=256, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:54
#10 0x00007f01fec29ba2 in pci_data_write (s=0x7f0200502a50, addr=2147483742, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:75
#11 0x00007f01fec29d5e in pci_host_data_write (opaque=0x7f02004fa1c0, addr=2, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:128
#12 0x00007f01fedbac6e in memory_region_write_accessor (opaque=0x7f02004fc598, addr=2, value=0x7f01ee697a70, size=1, shift=0, mask=255) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:334 #13 0x00007f01fedbad50 in access_with_adjusted_size (addr=2, value=0x7f01ee697a70, size=1, access_size_min=1, access_size_max=4, access=0x7f01fedbabe9 <memory_region_write_accessor>, opaque=0x7f02004fc598)
at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:364
#14 0x00007f01fedbb1b8 in memory_region_iorange_write (iorange=0x7f0200503550, offset=2, width=1, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:439 #15 0x00007f01fedb33aa in ioport_writeb_thunk (opaque=0x7f0200503550, addr=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:212
#16 0x00007f01fedb2d84 in ioport_write (index=0, address=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:83
#17 0x00007f01fedb3924 in cpu_outb (addr=3326, val=51 '3') at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:289
#18 0x00007f01fedb76c5 in kvm_handle_io (port=3326, data=0x7f01fea6e000, direction=1, size=1, count=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1507 #19 0x00007f01fedb7d59 in kvm_cpu_exec (env=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1659
#20 0x00007f01fed3dc3f in qemu_kvm_cpu_thread_fn (arg=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/cpus.c:759
#21 0x00007f01fcf02d15 in start_thread (arg=0x7f01ee698700) at pthread_create.c:308
#22 0x00007f01f864f48d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:114
(gdb)

Luiz Capitulino

author

2013-05-30 20:33 UTC #8 lore raw

Quoted 58 lines · On Thu, 30 May 2013 20:05:29 +0200 Paolo Bonzini <[email protected]> wrote:

Il 30/05/2013 19:56, Luiz Capitulino ha scritto:
Quoted 57 lines
Quoted 54 lines · On Thu, 30 May 2013 10:32:36 -0700 Jordan Justen <[email protected]> wrote:
Quoted 39 lines · Luiz Capitulino
Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/
Quoted 4 lines
Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?
Quoted 8 lines
I think so:
(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)
Is this what you're asking?
I guess I was meaning ... if you write to an address such as 0xfffffff0, does it update as RAM, or does it retain the original value?

Quoted 6 lines

Quoted 4 lines
Quoted 2 lines
This is easy to test in OVMF at the EFI shell, but I'm not sure how you could easily test it otherwise.
I could try to hack something, but maybe not today.
Just put a breakpoint on pflash_cfi01_register and see if it is reached.

Reached on a regular boot right?

It's not reached, with or without the offending commit.

Quoted 38 lines

I cannot reproduce it, but I'm also on 3.8.x. Will look at it tomorrow.
Paolo
Quoted 32 lines
Quoted 1 line
Does the system actually boot for you after the error message?
No, I get an abort. That's what kvm_set_phys_mem() does when kvm_set_user_memory_region() fails:
(gdb) bt
#0  0x00007f01f8592ba5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:63
#1  0x00007f01f8594358 in __GI_abort () at abort.c:90
#2  0x00007f01fedb57f1 in kvm_set_phys_mem (section=0x7f01ee6975d0, add=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:692
#3  0x00007f01fedb5bfa in kvm_region_del (listener=0x7f01ff2bc360 <kvm_memory_listener>, section=0x7f01ee6975d0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:795
#4  0x00007f01fedbc92d in address_space_update_topology_pass (as=0x7f01ffadfa60 <address_space_memory>, old_view=..., new_view=..., adding=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:689
#5  0x00007f01fedbd125 in address_space_update_topology (as=0x7f01ffadfa60 <address_space_memory>) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:725
#6  0x00007f01fedbd29c in memory_region_transaction_commit () at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:750
#7  0x00007f01fec1b977 in i440fx_update_memory_mappings (d=0x7f0200511940) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:131
#8  0x00007f01fec1bab3 in i440fx_write_config (dev=0x7f0200511940, address=94, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:154
#9  0x00007f01fec29a9a in pci_host_config_write_common (pci_dev=0x7f0200511940, addr=94, limit=256, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:54
#10 0x00007f01fec29ba2 in pci_data_write (s=0x7f0200502a50, addr=2147483742, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:75
#11 0x00007f01fec29d5e in pci_host_data_write (opaque=0x7f02004fa1c0, addr=2, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:128
#12 0x00007f01fedbac6e in memory_region_write_accessor (opaque=0x7f02004fc598, addr=2, value=0x7f01ee697a70, size=1, shift=0, mask=255) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:334 #13 0x00007f01fedbad50 in access_with_adjusted_size (addr=2, value=0x7f01ee697a70, size=1, access_size_min=1, access_size_max=4, access=0x7f01fedbabe9 <memory_region_write_accessor>, opaque=0x7f02004fc598)
at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:364
#14 0x00007f01fedbb1b8 in memory_region_iorange_write (iorange=0x7f0200503550, offset=2, width=1, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:439 #15 0x00007f01fedb33aa in ioport_writeb_thunk (opaque=0x7f0200503550, addr=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:212
#16 0x00007f01fedb2d84 in ioport_write (index=0, address=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:83
#17 0x00007f01fedb3924 in cpu_outb (addr=3326, val=51 '3') at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:289
#18 0x00007f01fedb76c5 in kvm_handle_io (port=3326, data=0x7f01fea6e000, direction=1, size=1, count=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1507 #19 0x00007f01fedb7d59 in kvm_cpu_exec (env=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1659
#20 0x00007f01fed3dc3f in qemu_kvm_cpu_thread_fn (arg=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/cpus.c:759
#21 0x00007f01fcf02d15 in start_thread (arg=0x7f01ee698700) at pthread_create.c:308
#22 0x00007f01f864f48d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:114
(gdb)

Paolo Bonzini

maintainer

2013-05-30 21:24 UTC #9 lore raw

Il 30/05/2013 22:32, Luiz Capitulino ha scritto:

Quoted 64 lines

Quoted 61 lines · On Thu, 30 May 2013 20:05:29 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 19:56, Luiz Capitulino ha scritto:
Quoted 60 lines
Quoted 56 lines · On Thu, 30 May 2013 10:32:36 -0700 Jordan Justen <[email protected]> wrote:
Quoted 39 lines · Luiz Capitulino
Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/
Quoted 4 lines
Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?
Quoted 8 lines
I think so:
(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)
Is this what you're asking?
I guess I was meaning ... if you write to an address such as 0xfffffff0, does it update as RAM, or does it retain the original value?
This is easy to test in OVMF at the EFI shell, but I'm not sure how you could easily test it otherwise.

Quoted 7 lines

Quoted 3 lines
Quoted 1 line
I could try to hack something, but maybe not today.
Just put a breakpoint on pflash_cfi01_register and see if it is reached.
Reached on a regular boot right?
It's not reached, with or without the offending commit.

Thanks. Reproduced with 3.9.4.

Paolo

Quoted 40 lines

Quoted 37 lines
I cannot reproduce it, but I'm also on 3.8.x. Will look at it tomorrow.
Paolo
Quoted 32 lines
Quoted 1 line
Does the system actually boot for you after the error message?
No, I get an abort. That's what kvm_set_phys_mem() does when kvm_set_user_memory_region() fails:
(gdb) bt
#0  0x00007f01f8592ba5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:63
#1  0x00007f01f8594358 in __GI_abort () at abort.c:90
#2  0x00007f01fedb57f1 in kvm_set_phys_mem (section=0x7f01ee6975d0, add=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:692
#3  0x00007f01fedb5bfa in kvm_region_del (listener=0x7f01ff2bc360 <kvm_memory_listener>, section=0x7f01ee6975d0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:795
#4  0x00007f01fedbc92d in address_space_update_topology_pass (as=0x7f01ffadfa60 <address_space_memory>, old_view=..., new_view=..., adding=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:689
#5  0x00007f01fedbd125 in address_space_update_topology (as=0x7f01ffadfa60 <address_space_memory>) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:725
#6  0x00007f01fedbd29c in memory_region_transaction_commit () at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:750
#7  0x00007f01fec1b977 in i440fx_update_memory_mappings (d=0x7f0200511940) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:131
#8  0x00007f01fec1bab3 in i440fx_write_config (dev=0x7f0200511940, address=94, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:154
#9  0x00007f01fec29a9a in pci_host_config_write_common (pci_dev=0x7f0200511940, addr=94, limit=256, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:54
#10 0x00007f01fec29ba2 in pci_data_write (s=0x7f0200502a50, addr=2147483742, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:75
#11 0x00007f01fec29d5e in pci_host_data_write (opaque=0x7f02004fa1c0, addr=2, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:128
#12 0x00007f01fedbac6e in memory_region_write_accessor (opaque=0x7f02004fc598, addr=2, value=0x7f01ee697a70, size=1, shift=0, mask=255) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:334 #13 0x00007f01fedbad50 in access_with_adjusted_size (addr=2, value=0x7f01ee697a70, size=1, access_size_min=1, access_size_max=4, access=0x7f01fedbabe9 <memory_region_write_accessor>, opaque=0x7f02004fc598)
at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:364
#14 0x00007f01fedbb1b8 in memory_region_iorange_write (iorange=0x7f0200503550, offset=2, width=1, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:439 #15 0x00007f01fedb33aa in ioport_writeb_thunk (opaque=0x7f0200503550, addr=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:212
#16 0x00007f01fedb2d84 in ioport_write (index=0, address=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:83
#17 0x00007f01fedb3924 in cpu_outb (addr=3326, val=51 '3') at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:289
#18 0x00007f01fedb76c5 in kvm_handle_io (port=3326, data=0x7f01fea6e000, direction=1, size=1, count=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1507 #19 0x00007f01fedb7d59 in kvm_cpu_exec (env=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1659
#20 0x00007f01fed3dc3f in qemu_kvm_cpu_thread_fn (arg=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/cpus.c:759
#21 0x00007f01fcf02d15 in start_thread (arg=0x7f01ee698700) at pthread_create.c:308
#22 0x00007f01f864f48d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:114
(gdb)

Jordan Justen

2013-05-30 23:43 UTC #10 lore raw

Quoted 67 lines · Paolo Bonzini

Il 30/05/2013 22:32, Luiz Capitulino ha scritto:
Quoted 66 lines
Quoted 63 lines · On Thu, 30 May 2013 20:05:29 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 19:56, Luiz Capitulino ha scritto:
Quoted 61 lines
Quoted 48 lines · On Thu, 30 May 2013 10:32:36 -0700 Jordan Justen <[email protected]> wrote:
Quoted 46 lines · Luiz Capitulino
Quoted 32 lines · On Thu, 30 May 2013 09:50:10 -0700 Jordan Justen <[email protected]> wrote:
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
Quoted 8 lines
Quoted 1 line
Reproducer?
I just try to start a VM (HEAD 87d23f7):
~/work/virt/ sudo ./qemu-qmp -drive file=disks/test.img,if=virtio,cache=none,aio=native -enable-kvm -m 1G -monitor stdio -cpu host -snapshot QEMU 1.5.50 monitor - type 'help' for more information
(qemu) kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
~/work/virt/
Quoted 4 lines
Sorry. I am working with Linux 3.8.0, and I don't see this. I'll try to update my kernel.
Does the firmware behave as a ROM for you?
I think so:
(qemu) info roms
fw=genroms/kvmvapic.bin size=0x002400 name="kvmvapic.bin" addr=00000000fffe0000 size=0x020000 mem=rom name="bios.bin" (qemu)
Quoted 8 lines
Quoted 1 line
Is this what you're asking?
I guess I was meaning ... if you write to an address such as 0xfffffff0, does it update as RAM, or does it retain the original value?
This is easy to test in OVMF at the EFI shell, but I'm not sure how you could easily test it otherwise.
I could try to hack something, but maybe not today.

Quoted 7 lines

Quoted 5 lines
Quoted 1 line
Just put a breakpoint on pflash_cfi01_register and see if it is reached.
Reached on a regular boot right?
It's not reached, with or without the offending commit.
Thanks. Reproduced with 3.9.4.

I have also reproduced it on 3.9.0. I'll look into it.

Is there a bug filed on this?

-Jordan

Quoted 41 lines

Quoted 40 lines
Quoted 37 lines
I cannot reproduce it, but I'm also on 3.8.x. Will look at it tomorrow.
Paolo
Quoted 32 lines
Quoted 1 line
Does the system actually boot for you after the error message?
No, I get an abort. That's what kvm_set_phys_mem() does when kvm_set_user_memory_region() fails:
(gdb) bt
#0  0x00007f01f8592ba5 in __GI_raise (sig=sig@entry=6) at ../nptl/sysdeps/unix/sysv/linux/raise.c:63
#1  0x00007f01f8594358 in __GI_abort () at abort.c:90
#2  0x00007f01fedb57f1 in kvm_set_phys_mem (section=0x7f01ee6975d0, add=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:692
#3  0x00007f01fedb5bfa in kvm_region_del (listener=0x7f01ff2bc360 <kvm_memory_listener>, section=0x7f01ee6975d0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:795
#4  0x00007f01fedbc92d in address_space_update_topology_pass (as=0x7f01ffadfa60 <address_space_memory>, old_view=..., new_view=..., adding=false) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:689
#5  0x00007f01fedbd125 in address_space_update_topology (as=0x7f01ffadfa60 <address_space_memory>) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:725
#6  0x00007f01fedbd29c in memory_region_transaction_commit () at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:750
#7  0x00007f01fec1b977 in i440fx_update_memory_mappings (d=0x7f0200511940) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:131
#8  0x00007f01fec1bab3 in i440fx_write_config (dev=0x7f0200511940, address=94, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci-host/piix.c:154
#9  0x00007f01fec29a9a in pci_host_config_write_common (pci_dev=0x7f0200511940, addr=94, limit=256, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:54
#10 0x00007f01fec29ba2 in pci_data_write (s=0x7f0200502a50, addr=2147483742, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:75
#11 0x00007f01fec29d5e in pci_host_data_write (opaque=0x7f02004fa1c0, addr=2, val=51, len=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/hw/pci/pci_host.c:128
#12 0x00007f01fedbac6e in memory_region_write_accessor (opaque=0x7f02004fc598, addr=2, value=0x7f01ee697a70, size=1, shift=0, mask=255) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:334 #13 0x00007f01fedbad50 in access_with_adjusted_size (addr=2, value=0x7f01ee697a70, size=1, access_size_min=1, access_size_max=4, access=0x7f01fedbabe9 <memory_region_write_accessor>, opaque=0x7f02004fc598)
at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:364
#14 0x00007f01fedbb1b8 in memory_region_iorange_write (iorange=0x7f0200503550, offset=2, width=1, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/memory.c:439 #15 0x00007f01fedb33aa in ioport_writeb_thunk (opaque=0x7f0200503550, addr=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:212
#16 0x00007f01fedb2d84 in ioport_write (index=0, address=3326, data=51) at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:83
#17 0x00007f01fedb3924 in cpu_outb (addr=3326, val=51 '3') at /home/lcapitulino/work/src/upstream/qmp-unstable/ioport.c:289
#18 0x00007f01fedb76c5 in kvm_handle_io (port=3326, data=0x7f01fea6e000, direction=1, size=1, count=1) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1507 #19 0x00007f01fedb7d59 in kvm_cpu_exec (env=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/kvm-all.c:1659
#20 0x00007f01fed3dc3f in qemu_kvm_cpu_thread_fn (arg=0x7f02004e24e0) at /home/lcapitulino/work/src/upstream/qmp-unstable/cpus.c:759
#21 0x00007f01fcf02d15 in start_thread (arg=0x7f01ee698700) at pthread_create.c:308
#22 0x00007f01f864f48d in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:114
(gdb)

Xiao Guangrong

2013-05-31 06:51 UTC #11 lore raw

In reply to Jordan Justen

Quoted 16 lines · Jordan Justen

Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().

Quoted 6 lines

As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?

I have reproduced this bug, it seems that the bug is caused by double free the memslot. After these change, it can boot the guest now.

kvm-all.c +2 -1

index 8e7bbf8..405480e 100644
@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Jordan Justen

2013-05-31 07:14 UTC #12 lore raw

Quoted 19 lines · Xiao Guangrong

Quoted 18 lines · Jordan Justen
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.

Quoted 7 lines

Quoted 4 lines
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
I have reproduced this bug, it seems that the bug is caused by double free the memslot. After these change, it can boot the guest now.

I think that I might have found that the flash memory had an issue with transitioning from flash to rom mode with code like this on the 3.8 kernel. I'll test it on 3.8 & 3.9.

-Jordan

Quoted 16 lines

kvm-all.c +2 -1

index 8e7bbf8..405480e 100644
@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Jordan Justen

2013-05-31 08:01 UTC #13 lore raw

Quoted 24 lines · Jordan Justen

Quoted 22 lines · Xiao Guangrong
Quoted 21 lines · Jordan Justen
Quoted 6 lines · Luiz Capitulino
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)

Quoted 8 lines

Quoted 4 lines
Quoted 1 line
Xiao, any thoughts?
I have reproduced this bug, it seems that the bug is caused by double free the memslot. After these change, it can boot the guest now.
I think that I might have found that the flash memory had an issue with transitioning from flash to rom mode with code like this on the 3.8 kernel. I'll test it on 3.8 & 3.9.

Yes, the issue still occurs on 3.9, and even the "double free" doesn't work on 3.9 like it did on 3.8.

This issue is a bit of a corner case for flash, but still important. I hope maybe Xiao might have time to help me understand it. (I'll email Xiao off-list about this.)

But, for now, I think Xiao's patch will fix things for the non-flash case, and will not totally break flash support either.

-Jordan

Paolo Bonzini

maintainer

2013-05-31 08:23 UTC #14 lore raw

In reply to Xiao Guangrong

Il 31/05/2013 08:51, Xiao Guangrong ha scritto:

Quoted 36 lines · Jordan Justen

Quoted 16 lines · Luiz Capitulino
Quoted 6 lines
Quoted 3 lines · On Thu, 30 May 2013 18:03:04 +0200 Paolo Bonzini <[email protected]> wrote:
Il 30/05/2013 17:46, Luiz Capitulino ha scritto:
Quoted 2 lines
The culprit is commit:
Quoted 8 lines
Quoted 8 lines
Quoted 8 lines
commit 235e8982ad393e5611cb892df54881c872eea9e1
Author: Jordan Justen <[email protected]>
Date:   Wed May 29 01:27:26 2013 -0700
kvm: support using KVM_MEM_READONLY flag for regions
I'm running 3.9.2-200.fc18, btw. And, error checking is missing on the first call to kvm_vm_ioctl().
Quoted 6 lines
As noted in the code, the first call is for KVM commit 75d61fbc.
I'm not sure we want to fail if an error occurs when making that call. (I'm pretty sure we don't want to in fact.)
Xiao, any thoughts?
I have reproduced this bug, it seems that the bug is caused by double free the memslot. After these change, it can boot the guest now.
kvm-all.c +0 -0
index 8e7bbf8..405480e 100644

kvm-all.c static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot) · line 213

Quoted 9 lines

kvm-all.c +2 -1

@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Thanks, can you submit it for uq/master? Please Cc [email protected] too, and use [PATCH uq/master] as the prefix.

Paolo

Xiao Guangrong

2013-05-31 08:52 UTC #15 lore raw

PATCH 1/1 fix double free the memslot in kvm_set_phys_mem

Luiz Capitulino reported that guest refused to boot and qemu complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument

It is caused by commit 235e8982ad that did double free for the memslot so that the second one raises the -EINVAL error

Fix it by reset memory size only if it is needed

Reported-by Luiz Capitulino <[email protected]>

Signed-off-by Xiao Guangrong <[email protected]>

---
 kvm-all.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

kvm-all.c +2 -1

index 8e7bbf8..405480e 100644
@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Signature · 1 line

1.7.7.6

Paolo Bonzini

maintainer Reviewed-by on v1 1/1

2013-05-31 12:27 UTC #16 lore raw

Re: [PATCH uq/master] fix double free the memslot in kvm_set_phys_mem

Il 31/05/2013 10:52, Xiao Guangrong ha scritto:

Quoted 21 lines

Luiz Capitulino reported that guest refused to boot and qemu complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
It is caused by commit 235e8982ad that did double free for the memslot so that the second one raises the -EINVAL error
Fix it by reset memory size only if it is needed
Reported-by Luiz Capitulino <[email protected]>
Signed-off-by Xiao Guangrong <[email protected]>
---
 kvm-all.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
kvm-all.c +0 -0
index 8e7bbf8..405480e 100644

kvm-all.c static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot) · line 213

Quoted 9 lines

kvm-all.c +2 -1

@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Reviewed-by Paolo Bonzini <[email protected]>

Luiz Capitulino

author Tested-by on v1 1/1

2013-05-31 12:39 UTC #17 lore raw

Re: [PATCH uq/master] fix double free the memslot in kvm_set_phys_mem

In reply to Xiao Guangrong

Quoted 11 lines · On Fri, 31 May 2013 16:52:18 +0800 Xiao Guangrong…

Luiz Capitulino reported that guest refused to boot and qemu complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
It is caused by commit 235e8982ad that did double free for the memslot so that the second one raises the -EINVAL error
Fix it by reset memory size only if it is needed
Reported-by Luiz Capitulino <[email protected]>
Signed-off-by Xiao Guangrong <[email protected]>

Tested-by Luiz Capitulino <[email protected]>

Thanks Xiao for the fix, and thanks everyone for debugging this issue!

Quoted 18 lines

---
 kvm-all.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

kvm-all.c +2 -1

index 8e7bbf8..405480e 100644
@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Richard W.M. Jones

Tested-by on v1 1/1

2013-06-02 17:36 UTC #18 lore raw

Re: [Qemu-devel] [PATCH uq/master] fix double free the memslot in kvm_set_phys_mem

In reply to Xiao Guangrong

Quoted 20 lines · Xiao Guangrong

Luiz Capitulino reported that guest refused to boot and qemu complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
It is caused by commit 235e8982ad that did double free for the memslot so that the second one raises the -EINVAL error
Fix it by reset memory size only if it is needed
Reported-by Luiz Capitulino <[email protected]>
Signed-off-by Xiao Guangrong <[email protected]>
---
 kvm-all.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
kvm-all.c +0 -0
index 8e7bbf8..405480e 100644

kvm-all.c static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot) · line 213

Quoted 9 lines

kvm-all.c +2 -1

@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Tested and fixes it for me too.

Tested-by Richard W.M. Jones <[email protected]>

Rich.

Signature · 3 lines

Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
libguestfs lets you edit virtual machines.  Supports shell scripting,
bindings from many languages.  http://libguestfs.org

Jordan Justen

Reviewed-by on v1 1/1

2013-06-02 22:08 UTC #19 lore raw

Re: [PATCH uq/master] fix double free the memslot in kvm_set_phys_mem

In reply to Xiao Guangrong

Reviewed-by Jordan Justen <[email protected]>

Quoted 32 lines · Xiao Guangrong

Luiz Capitulino reported that guest refused to boot and qemu complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
It is caused by commit 235e8982ad that did double free for the memslot so that the second one raises the -EINVAL error
Fix it by reset memory size only if it is needed
Reported-by Luiz Capitulino <[email protected]>
Signed-off-by Xiao Guangrong <[email protected]>
---
 kvm-all.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)
kvm-all.c +2 -2
index 8e7bbf8..405480e 100644
@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;
--
1.7.7.6

Gleb Natapov

Applied

2013-06-03 06:57 UTC #20 lore raw

Re: [PATCH uq/master] fix double free the memslot in kvm_set_phys_mem

In reply to Xiao Guangrong

Quoted 11 lines · Xiao Guangrong

Luiz Capitulino reported that guest refused to boot and qemu complained with:
kvm_set_phys_mem: error unregistering overlapping slot: Invalid argument
It is caused by commit 235e8982ad that did double free for the memslot so that the second one raises the -EINVAL error
Fix it by reset memory size only if it is needed
Reported-by Luiz Capitulino <[email protected]>
Signed-off-by Xiao Guangrong <[email protected]>

Thanks, applied.

Quoted 20 lines

---
 kvm-all.c |    3 ++-
 1 files changed, 2 insertions(+), 1 deletions(-)

kvm-all.c +2 -1

index 8e7bbf8..405480e 100644
@@ -206,7 +206,8 @@ static int kvm_set_user_memory_region(KVMState *s, KVMSlot *slot)
     if (s->migration_log) {
         mem.flags |= KVM_MEM_LOG_DIRTY_PAGES;
     }
-    if (mem.flags & KVM_MEM_READONLY) {
+
+    if (slot->memory_size && mem.flags & KVM_MEM_READONLY) {
         /* Set the slot size to 0 before setting the slot to the desired
          * value. This is needed based on KVM commit 75d61fbc. */
         mem.memory_size = 0;

Signature · 1 line

1.7.7.6

--
Gleb.