[Qemu-devel] Fwd: Guest hangs when I do general operation.

[Qemu-devel] Fwd: Guest hangs when I do general operation.

[Amos Kong]

---------- Forwarded message ----------
From: Amos Kong <kongjianjun@gmail.com>
Date: Tue, Sep 7, 2010 at 7:49 AM
Subject: Guest hangs when I do general operation.
To: 王箫 <sirouni@gmail.com>


kvm upstream: 43e413f7db1a4a90671dda0b1d6c1f8cb30673ed KVM: Whitespace
changes to remove differences wrt kvm-updates/2.6.37
qemu upstream: cb93bbdd7db92e50ff5e60a346b23df68acae46b Fix OpenBSD
linker warning

# ./x86_64-softmmu/qemu-system-x86_64 ~/win7-32.qcow2 -m 1024 -vnc :0
-usbdevice tablet -cpu qemu64 -snapshot -enable-kvm -bios pc-bios/bios.bin

Guest hangs when I do general operation.
It's fine when using upstream qemu-kvm.  this problem only occurred
one time when I debug by gdb, after execute 'continue', guest runs
normally.

debug msg of (qemu + kvm)
(gdb) c
Continuing.

Program received signal SIGUSR2, User defined signal 2.
0x00007fc7d4d7bfb3 in select () at ../sysdeps/unix/syscall-template.S:82
82      ../sysdeps/unix/syscall-template.S: No such file or directory.
       in ../sysdeps/unix/syscall-template.S
(gdb) bt
#0  0x00007fc7d4d7bfb3 in select () at ../sysdeps/unix/syscall-template.S:82
#1  0x00000000004270ea in qemu_aio_wait () at aio.c:193
#2  0x0000000000426475 in bdrv_read_em (bs=0x186a340,
sector_num=6343320, buf=0x7fc7c5a9b010 "RCRD(", nb_sectors=104) at
block.c:2432
#3  0x000000000043c437 in qcow_read (bs=0x1838680, start_sect=<value
optimized out>, cluster_offset=<value optimized out>, n_start=<value
optimized out>,
           n_end=<value optimized out>) at block/qcow2-cluster.c:368
#4  copy_sectors (bs=0x1838680, start_sect=<value optimized out>,
cluster_offset=<value optimized out>, n_start=<value optimized out>,
               n_end=<value optimized out>) at block/qcow2-cluster.c:406
#5  0x000000000043c69b in qcow2_alloc_cluster_link_l2 (bs=0x1838680,
m=0x1d5d798) at block/qcow2-cluster.c:689
#6  0x00000000004378d5 in qcow_aio_write_cb (opaque=0x1d5d700, ret=0)
at block/qcow2.c:566
#7  0x0000000000429c5d in posix_aio_process_queue (opaque=<value
optimized out>) at posix-aio-compat.c:459
#8  0x0000000000429d0c in posix_aio_read (opaque=0x183a250) at
posix-aio-compat.c:489
#9  0x000000000051fec6 in main_loop_wait (nonblocking=<value optimized
out>) at /home/devel/qemu/vl.c:1281
#10 0x00000000005209bd in main_loop (argc=0, argv=<value optimized
out>, envp=<value optimized out>) at /home/devel/qemu/vl.c:1332
#11 main (argc=0, argv=<value optimized out>, envp=<value optimized
out>) at /home/devel/qemu/vl.c:2995
---
kvm statistics

 efer_reload                  0       0
 exits                  8714404       0
 fpu_reload              115538       0
 halt_exits               66926       0
 halt_wakeup                  0       0
 host_state_reload      2366344       0
 hypercalls                   0       0
 insn_emulation         1848818       0
 insn_emulation_fail          0       0
 invlpg                  662261       0
 io_exits               1293800       0
 irq_exits               531478       0
 irq_injections          109588       0
 irq_window              114236       0
 largepages                   0       0
 mmio_exits              705388       0
 mmu_cache_miss          355201       0
 mmu_flooded             298554       0
 mmu_pde_zapped           25705       0
 mmu_pte_updated         241815       0
 mmu_pte_write         15701676       0
 mmu_recycled               546       0
 mmu_shadow_zapped       527220       0
 mmu_unsync                4203       0
 nmi_injections               0       0
 nmi_window                   0       0
 pf_fixed               3107522       0
 pf_guest                631148       0
 remote_tlb_flush         31032       0
 request_irq                  0       0
 signal_exits            310597       0
 tlb_flush              2164428       0

Re: [Qemu-devel] Fwd: Guest hangs when I do general operation.

[Blue Swirl]

On Tue, Sep 7, 2010 at 2:43 AM, Amos Kong <kongjianjun@gmail.com> wrote:
> ---------- Forwarded message ----------
> From: Amos Kong <kongjianjun@gmail.com>
> Date: Tue, Sep 7, 2010 at 7:49 AM
> Subject: Guest hangs when I do general operation.
> To: 王箫 <sirouni@gmail.com>
>
>
> kvm upstream: 43e413f7db1a4a90671dda0b1d6c1f8cb30673ed KVM: Whitespace
> changes to remove differences wrt kvm-updates/2.6.37
> qemu upstream: cb93bbdd7db92e50ff5e60a346b23df68acae46b Fix OpenBSD
> linker warning
>
> # ./x86_64-softmmu/qemu-system-x86_64 ~/win7-32.qcow2 -m 1024 -vnc :0
> -usbdevice tablet -cpu qemu64 -snapshot -enable-kvm -bios pc-bios/bios.bin
>
> Guest hangs when I do general operation.
> It's fine when using upstream qemu-kvm.  this problem only occurred
> one time when I debug by gdb, after execute 'continue', guest runs
> normally.
>
> debug msg of (qemu + kvm)
> (gdb) c
> Continuing.
>
> Program received signal SIGUSR2, User defined signal 2.

This is normal, we use SIGUSR2 for AIO. By default, GDB stops and
prints a message about the signal. The same thing can happen with
SIGALRM.

GDB command 'handle SIGUSR2 nostop noprint pass' lets the program
continue without stopping.

[Qemu-devel] Re: Fwd: Guest hangs when I do general operation.

[Marcelo Tosatti]

On Tue, Sep 07, 2010 at 10:43:38AM +0800, Amos Kong wrote:
> ---------- Forwarded message ----------
> From: Amos Kong <kongjianjun@gmail.com>
> Date: Tue, Sep 7, 2010 at 7:49 AM
> Subject: Guest hangs when I do general operation.
> To: 王箫 <sirouni@gmail.com>
> 
> 
> kvm upstream: 43e413f7db1a4a90671dda0b1d6c1f8cb30673ed KVM: Whitespace
> changes to remove differences wrt kvm-updates/2.6.37
> qemu upstream: cb93bbdd7db92e50ff5e60a346b23df68acae46b Fix OpenBSD
> linker warning
> 
> # ./x86_64-softmmu/qemu-system-x86_64 ~/win7-32.qcow2 -m 1024 -vnc :0
> -usbdevice tablet -cpu qemu64 -snapshot -enable-kvm -bios pc-bios/bios.bin
> 
> Guest hangs when I do general operation.
> It's fine when using upstream qemu-kvm.  this problem only occurred
> one time when I debug by gdb, after execute 'continue', guest runs
> normally.
> 
> debug msg of (qemu + kvm)
> (gdb) c
> Continuing.
> 
> Program received signal SIGUSR2, User defined signal 2.
> 0x00007fc7d4d7bfb3 in select () at ../sysdeps/unix/syscall-template.S:82
> 82      ../sysdeps/unix/syscall-template.S: No such file or directory.
>        in ../sysdeps/unix/syscall-template.S
> (gdb) bt

Try "handle SIGUSR2 nostop" in gdb before executing qemu-kvm.

Re: [Qemu-devel] Fwd: Guest hangs when I do general operation.

[Amos Kong]

On Wed, Sep 8, 2010 at 1:55 AM, Blue Swirl <blauwirbel@gmail.com> wrote:
> On Tue, Sep 7, 2010 at 2:43 AM, Amos Kong <kongjianjun@gmail.com> wrote:
>>
>> kvm upstream: 43e413f7db1a4a90671dda0b1d6c1f8cb30673ed KVM: Whitespace
>> changes to remove differences wrt kvm-updates/2.6.37
>> qemu upstream: cb93bbdd7db92e50ff5e60a346b23df68acae46b Fix OpenBSD
>> linker warning
>>
>> # ./x86_64-softmmu/qemu-system-x86_64 ~/win7-32.qcow2 -m 1024 -vnc :0
>> -usbdevice tablet -cpu qemu64 -snapshot -enable-kvm -bios pc-bios/bios.bin
>>
>> Guest hangs when I do general operation.
>> It's fine when using upstream qemu-kvm.  this problem only occurred
>> one time when I debug by gdb, after execute 'continue', guest runs
>> normally.
>>
>> debug msg of (qemu + kvm)
>> (gdb) c
>> Continuing.
>>
>> Program received signal SIGUSR2, User defined signal 2.
>
> This is normal, we use SIGUSR2 for AIO. By default, GDB stops and
> prints a message about the signal. The same thing can happen with
> SIGALRM.
>
> GDB command 'handle SIGUSR2 nostop noprint pass' lets the program
> continue without stopping.

If ignore this signal in gdb, guest could not break.
Thanks