Le blog de Jean David TECHER, un Réunionnais à Saint-Priest/Lyon

Les pilotes VirtIO permettent d'améliorer les performances d'une machine virtuelle lorsqu'il s'agit du dual périphérique/pilote. Il s'agit d'une série de pilotes remplaçant les pilotes émulés nativement par qemu. Dans ce billet, une série de vidéos est proposée

• Injection des drivers dans l'iso originelle de Windows 10 Home (Build 1607)
• Présentation du matériel pour Qemu/KVM
• Installation d'une machine virtuelle Windows 10 à partir de l'iso fraîchement modifiée
• Tests de surface

1. Injection des pilotes VirtIO dans l'iso de Windows 10

Je suis parti de l'iso Win10_1607_SingleLang_French_x64.iso. Le script utilisé pour injecter les drivers se trouve ici (avec commentaires en anglais)
Tutoriel - Injection des pilotes VirtIO dans l'iso de Windows 10

2. Présentation du matériel Qemu/KVM

Je présente mon (vieux) matos pour faire mes tests sous Qemu/KVM
Hardware - Présentation du matériel

3. Test de l'iso: création d'une machine virtuelle Windows 10

On teste maintenant l'iso en créant une machine virtuelle. La connexion a lieu à distance depuis VNC. J'en profite au passage pour commenter mon installation software de Qemu/KVM.
Test de l'iso: Création d'un machine virtuelle Windows 10

4. Tests de fin

Comme dit dans la vidéo précédente, on attend le reboot de la machine pour voir si le switch vers l'écran rattaché à la carte graphique a bien eu lieu. On teste ensuite GPU-Z et CPU-Z.
Tests de fin

Youtube video

Referring to ReactOS SVN repository I used revision = 69458 for this VM. As quoted from ReactOS Wiki, sound can work only once :(. But at least I can play one video

Script used for this VM

number=69458	
#number=69461

#version=dbg
version=rel

driver="ac97"
#driver="es1370"

# For ES1370
[ "${driver}" = "es1370" ] && {
	echo "AUDIO DRIVER = ${driver}";
	export QEMU_AUDIO_DRV=alsa;
	export QEMU_AUDIO_TIMER_PERIOD=0;
	export QEMU_AUDIO_DAC_FIXED_CHANNELS=2;
	export QEMU_ALSA_DAC_SIZE_IN_USEC=1;
	export QEMU_ALSA_ADC_SIZE_IN_USEC=1;
	export QEMU_ALSA_ADC_BUFFER_SIZE=2176;
}

# For AC97
[ "${driver}" = "ac97" ] && {
	echo "AUDIO DRIVER = ${driver}";
	export QEMU_AUDIO_DRV=alsa;
	export QEMU_ALSA_DAC_SIZE_IN_USEC=1;
	export QEMU_ALSA_DAC_BUFFER_SIZE=4353 ;
	export QEMU_AUDIO_TIMER_PERIOD=100;  
	export QEMU_AUDIO_DAC_FIXED_CHANNELS=1; 
}



/opt/qemu/2.4.0.1/bin/qemu-system-i386  --enable-kvm  \
	-cpu host -m 2047 -smp  4,sockets=1,cores=4,threads=1 \
	-drive file=/dev/xen-vg/React_OS,if=ide,index=0,media=disk,format=raw,cache=writeback \
        --cdrom /opt/CD_DVD_ISO/ReactOS/bootcd-${number}-${version}.iso  \
	-boot order=cd -vga std -global VGA.vgamem_mb=32 -localtime   \
	-net nic,model=rtl8139 -net bridge,br=br0,helper=/opt/qemu/2.4.0.1/libexec/qemu-bridge-helper \
	-soundhw ${driver}  \
	-writeconfig /root/qemu-2.4.0.1_reactos.conf

I wrote this post so I remember which options I used for Qemu.

Options to build

./configure --prefix=/opt/qemu/2.4.0.1/ --sysconfdir=/opt/qemu/2.4.0.1/etc/ --target-list=arm-softmmu,i386-softmmu,x86_64-softmmu \
--enable-kvm --enable-uuid --enable-virtfs --enable-libnfs --enable-libusb --enable-lzo --enable-bzip2 --enable-libiscsi --enable-attr --enable-libssh2 \
--enable-numa --enable-system --enable-guest-base --enable-guest-agent --audio-drv-list=alsa,pa,sdl --enable-cap-ng --enable-modules --enable-netmap \
--enable-user --enable-linux-user --extra-cflags=-I/opt/Sources/QEMU/netmap/sys/ --enable-tcmalloc --enable-gcov --enable-usb-redir --enable-tcg-interpreter \
--enable-xfsctl --enable-gprof --enable-profiler

Network (bridge mode)

I had to add this paramter to /etc/sysctl.conf file (Allow the host system to forward IP packets)

echo "# Require for QEMU/KVM" >> /etc/sysctl.conf
echo "net.ipv4.ip_forward=1" >> /etc/sysctl.conf

To take into account this parameter above without rebooting

sudo sysctl -p

Here is my /etc/network/interfaces file

root@amdgaming:~# cat /etc/network/interfaces
# interfaces(5) file used by ifup(8) and ifdown(8)
auto lo
iface lo inet loopback

auto eth0
iface eth0 inet manual


#################
# bridge + dhcp #
#################
auto br0
iface br0 inet dhcp
	bridge_ports eth0
        bridge_stp off
        bridge_fd off
        bridge_maxwait 0

After that

mkdir -pv /opt/qemu/2.4.0.1/etc/qemu/
echo "allow br0" > /opt/qemu/2.4.0.1/etc/qemu/bridge.conf

1. About options

Options are settable through environment variables.

Example:

  export QEMU_AUDIO_DRV=wav
  export QEMU_WAV_PATH=$HOME/tune.wa

v (for csh replace export with setenv in the above)

  qemu ...

2. Audio options

Parameter	Type	Default Value	Description
QEMU_AUDIO_DAC_FIXED_SETTINGS	boolean	1	Use fixed settings for host DAC
QEMU_AUDIO_DAC_FIXED_FREQ	integer	44100	Frequency for fixed host DAC
QEMU_AUDIO_DAC_FIXED_FMT	format	S16	One ofU8 S8 U16 S16 U32 S32 - Format for fixed host DAC
QEMU_AUDIO_DAC_FIXED_CHANNELS	integer	2	Number of channels for fixed DAC (1 - mono,2 - stereo)
QEMU_AUDIO_DAC_VOICES	integer	1	Number of voices for DAC
QEMU_AUDIO_DAC_TRY_POLL	boolean	1	Attempt using poll mode for DAC
QEMU_AUDIO_ADC_FIXED_SETTINGS	boolean	1	Use fixed settings for host ADC
QEMU_AUDIO_ADC_FIXED_FREQ	integer	44100	Frequency for fixed host ADC
QEMU_AUDIO_ADC_FIXED_FMT	format	S16	One of U8 S8 U16 S16 U32 S32 - Format for fixed host ADC
QEMU_AUDIO_ADC_FIXED_CHANNELS	integer	2	Number of channels for fixed ADC (1 - mono , 2 - stereo)
QEMU_AUDIO_ADC_VOICES	integer	1	Number of voices for ADC
QEMU_AUDIO_ADC_TRY_POLL	boolean	1	Attempt using poll mode for ADC
QEMU_AUDIO_TIMER_PERIOD	integer	100	Timer period in HZ (0 - use lowest possible)

3. Available drivers:

3.a) ALSA

Description: ALSA http://www.alsa-project.org

• Theoretically supports many playback voices
• Theoretically supports many capture voices.

Parameter	Type	Default Value	Description
QEMU_ALSA_DAC_SIZE_IN_USEC	boolean	0	DAC period/buffer size in microseconds (otherwise in frames)
QEMU_ALSA_DAC_PERIOD_SIZE	integer	1024	DAC period size (0 to go with system default)
QEMU_ALSA_DAC_BUFFER_SIZE	integer	4096	DAC buffer size (0 to go with system default)
QEMU_ALSA_ADC_SIZE_IN_USEC	boolean	0	ADC period/buffer size in microseconds (otherwise in frames)
QEMU_ALSA_ADC_PERIOD_SIZE	integer	0	ADC period size (0 to go with system default)
QEMU_ALSA_ADC_BUFFER_SIZE	integer	0	ADC buffer size (0 to go with system default)
QEMU_ALSA_THRESHOLD	integer	0	(undocumented)
QEMU_ALSA_DAC_DEV	string	default	DAC device name (for instance dmix)
QEMU_ALSA_ADC_DEV	string	default	ADC device name

3.b) PA (Pulse Audio)

Name: pa Description: http://www.pulseaudio.org/

• Theoretically supports many playback voices
• Theoretically supports many capture voices

Parameter	Type	Default Value	Description
QEMU_PA_SAMPLES	integer	4096	buffer size in samples
QEMU_PA_SERVER	string	(not set)	server address
QEMU_PA_SINK	string	(not set)	sink device name
QEMU_PA_SOURCE	string	(not set)	source device name

3.c) SDL

Name: sdl Description: SDL http://www.libsdl.org

• One playback voice
• Does not support capture

Parameter	Type	Default Value	Description
QEMU_SDL_SAMPLES	integer	1024	Size of SDL buffer in samples

3.d) None

Name: none Description: Timer based audio emulation

• Theoretically supports many playback voices
• Theoretically supports many capture voices

3.e) WAV

Name: wav Description: WAV renderer http://wikipedia.org/wiki/WAV

• One playback voice
• Does not support capture

Parameter	Type	Default Value	Description
QEMU_WAV_FREQUENCY	integer	44100	Frequency
QEMU_WAV_FORMAT	format	S16	(one of: U8 S8 U16 S16 U32 S32) Format
QEMU_WAV_DAC_FIXED_CHANNELS	integer	2	Number of channels (1 - mono, 2 - stereo)
QEMU_WAV_PATH	string	qemu.wav	Path to wave file

4. Help from Linux shell

Use option -audio-help to get help shown below

root@amdgaming:~# /opt/qemu/2.4.0.1/bin/qemu-system-i386 -audio-help
Audio options:
  QEMU_AUDIO_DAC_FIXED_SETTINGS: boolean, default = 1
    Use fixed settings for host DAC
  QEMU_AUDIO_DAC_FIXED_FREQ: integer, default = 44100
    Frequency for fixed host DAC
  QEMU_AUDIO_DAC_FIXED_FMT: format, default = S16, (one of: U8 S8 U16 S16 U32 S32)
    Format for fixed host DAC
  QEMU_AUDIO_DAC_FIXED_CHANNELS: integer, default = 2
    Number of channels for fixed DAC (1 - mono, 2 - stereo)
  QEMU_AUDIO_DAC_VOICES: integer, default = 1
    Number of voices for DAC
  QEMU_AUDIO_DAC_TRY_POLL: boolean, default = 1
    Attempt using poll mode for DAC
  QEMU_AUDIO_ADC_FIXED_SETTINGS: boolean, default = 1
    Use fixed settings for host ADC
  QEMU_AUDIO_ADC_FIXED_FREQ: integer, default = 44100
    Frequency for fixed host ADC
  QEMU_AUDIO_ADC_FIXED_FMT: format, default = S16, (one of: U8 S8 U16 S16 U32 S32)
    Format for fixed host ADC
  QEMU_AUDIO_ADC_FIXED_CHANNELS: integer, default = 2
    Number of channels for fixed ADC (1 - mono, 2 - stereo)
  QEMU_AUDIO_ADC_VOICES: integer, default = 1
    Number of voices for ADC
  QEMU_AUDIO_ADC_TRY_POLL: boolean, default = 1
    Attempt using poll mode for ADC
  QEMU_AUDIO_TIMER_PERIOD: integer, default = 100
    Timer period in HZ (0 - use lowest possible)

Available drivers:
Name: alsa
Description: ALSA http://www.alsa-project.org
Theoretically supports many playback voices
Theoretically supports many capture voices
Options:
  QEMU_ALSA_DAC_SIZE_IN_USEC: boolean, default = 0
    DAC period/buffer size in microseconds (otherwise in frames)
  QEMU_ALSA_DAC_PERIOD_SIZE: integer, default = 1024
    DAC period size (0 to go with system default)
  QEMU_ALSA_DAC_BUFFER_SIZE: integer, default = 4096
    DAC buffer size (0 to go with system default)
  QEMU_ALSA_ADC_SIZE_IN_USEC: boolean, default = 0
    ADC period/buffer size in microseconds (otherwise in frames)
  QEMU_ALSA_ADC_PERIOD_SIZE: integer, default = 0
    ADC period size (0 to go with system default)
  QEMU_ALSA_ADC_BUFFER_SIZE: integer, default = 0
    ADC buffer size (0 to go with system default)
  QEMU_ALSA_THRESHOLD: integer, default = 0
    (undocumented)
  QEMU_ALSA_DAC_DEV: string, default = default
    DAC device name (for instance dmix)
  QEMU_ALSA_ADC_DEV: string, default = default
    ADC device name

Name: pa
Description: http://www.pulseaudio.org/
Theoretically supports many playback voices
Theoretically supports many capture voices
Options:
  QEMU_PA_SAMPLES: integer, default = 4096
    buffer size in samples
  QEMU_PA_SERVER: string, default = (not set)
    server address
  QEMU_PA_SINK: string, default = (not set)
    sink device name
  QEMU_PA_SOURCE: string, default = (not set)
    source device name

Name: sdl
Description: SDL http://www.libsdl.org
One playback voice
Does not support capture
Options:
  QEMU_SDL_SAMPLES: integer, default = 1024
    Size of SDL buffer in samples

Name: none
Description: Timer based audio emulation
Theoretically supports many playback voices
Theoretically supports many capture voices
No options

Name: wav
Description: WAV renderer http://wikipedia.org/wiki/WAV
One playback voice
Does not support capture
Options:
  QEMU_WAV_FREQUENCY: integer, default = 44100
    Frequency
  QEMU_WAV_FORMAT: format, default = S16, (one of: U8 S8 U16 S16 U32 S32)
    Format
  QEMU_WAV_DAC_FIXED_CHANNELS: integer, default = 2
    Number of channels (1 - mono, 2 - stereo)
  QEMU_WAV_PATH: string, default = qemu.wav
    Path to wave file

Options are settable through environment variables.
Example:
  export QEMU_AUDIO_DRV=wav
  export QEMU_WAV_PATH=$HOME/tune.wav
(for csh replace export with setenv in the above)
  qemu ...