Re: [Qemu-devel] [PATCH 0/3] QEMU logging improvements

[Pavel Butsykin <pbutsykin@odin.com>]

On 15.10.2015 17:49, Kashyap Chamarthy wrote:
> On Thu, Oct 15, 2015 at 10:30:21AM +0300, Denis V. Lunev wrote:
>> The following is done:
>> - QMP/HMP events are now logged
>
> I applied this patch series locally, and tried to test it the below way:
>
> Launch a minimal QEMU instance:
>
>    $ /home/kashyapc/build/build-qemu-upstream/x86_64-softmmu/qemu-system-x86_64 \
>     -nographic                      \
>     -nodefconfig                    \
>     -nodefaults                     \
>     -m 2048                         \
>     -cpu SandyBridge                \
>     -device virtio-scsi-pci,id=scsi \
>     -device virtio-serial-pci       \
>     -drive file=./cirros.qcow2,if=virtio,cache=writeback \
>     -serial stdio \
>     -D ./output-of-qemu.txt \
>     -qmp unix:./qmp-sock,server
>
> And, issued QMP commands via `qmp-shell`:
>
>    $ ./qmp-shell /export/qmp-sock
>    Welcome to the QMP low-level shell!
>    Connected to QEMU 2.4.50
>    (QEMU)
>    (QEMU) query-status
>    {"return": {"status": "running", "singlestep": false, "running": true}}
>    (QEMU)
>    (QEMU) blockdev-snapshot-internal-sync device=virtio0 name=snapshot1
>    {"return": {}}
>
> As we can see, the QMP commands gets successfully invoked, but nothing
> gets logged to the output-of-qemu.txt file.  Maybe because everything is
> logged onto my stdout of qmp-shell?
>
> A couple of related questions:
>
>    - Is the way I supplied '-D' option correct?  If so, why the
>      output-of-qemu.txt file does not get created?  _Should_ it?
>
>    - If I `touch output-of-qemu.txt`, then re-try issuing QMP commmands
>      from 'qmp-shell', still, nothing gets logged to it.  Probably
>      because only stderr is supposed to get logged to '-D logfile'?
>
> Do I have to turn on any other special knob to take advantage of
> LOG_CMD/LOG_TIME flags mentioned in your patch[1]?
>
> I'm pretty sure I'm missing something. . .
>
> [1] http://lists.nongnu.org/archive/html/qemu-devel/2015-10/msg03316.html
>
>> - timestamp is added to the log message
>> - arguments of qemu_log/qemu_log_mask are not calculated now if the log
>>    is disabled
>>
>> These patches are sent in the hope that it is not too late for QEMU 2.5.
>>
>> Logging subsystem in QEMU looks rudimentary to me and should be definitely
>> improved, f.e. log on stalled NFS could ruin the system completely. Anyway,
>> it is too late for this at the moment.
>>
>> Signed-off-by: Denis V. Lunev <den@openvz.org>
>> CC: Pavel Butsykin <pbutsykin@virtuozzo.com>
>> CC: Markus Armbruster <armbru@redhat.com>
>> CC: Luiz Capitulino <lcapitulino@redhat.com>
>> CC: Eric Blake <eblake@redhat.com>
>> CC: Peter Maydell <peter.maydell@linaro.org>
>>
>>
>

You forgot to specify the mask: -d cmd,time

qemu$ qemu-system-x86_64 -h|grep log
...
-d item1,...    enable logging of specified items (use '-d help' for a 
list of log items)
-D logfile      output log to logfile (default stderr)

So it should work:
     $ 
/home/kashyapc/build/build-qemu-upstream/x86_64-softmmu/qemu-system-x86_64 \
      -nographic                      \
      -nodefconfig                    \
      -nodefaults                     \
      -m 2048                         \
      -cpu SandyBridge                \
      -device virtio-scsi-pci,id=scsi \
      -device virtio-serial-pci       \
      -drive file=./cirros.qcow2,if=virtio,cache=writeback \
      -serial stdio \
      -D ./output-of-qemu.txt \
      -d cmd,time \
      -qmp unix:./qmp-sock,server