Avoid a dynamic stack allocation in qjack_process(). Since this
function is a JACK process callback, we are not permitted to malloc()
here, so we allocate a working buffer in qjack_client_init() instead.
The codebase has very few VLAs, and if we can get rid of them all we
can make the compiler error on new additions. This is a defensive
measure against security bugs where an on-stack dynamic allocation
isn't correctly size-checked (e.g. CVE-2021-3527).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-id: 20230818155846.1651287-3-peter.maydell@linaro.org
Avoid a dynamic stack allocation in qjack_client_init(), by using
a g_autofree heap allocation instead.
(We stick with allocate + snprintf() because the JACK API requires
the name to be no more than its maximum size, so g_strdup_printf()
would require an extra truncation step.)
The codebase has very few VLAs, and if we can get rid of them all we
can make the compiler error on new additions. This is a defensive
measure against security bugs where an on-stack dynamic allocation
isn't correctly size-checked (e.g. CVE-2021-3527).
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Francisco Iglesias <frasse.iglesias@gmail.com>
Reviewed-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Message-id: 20230818155846.1651287-2-peter.maydell@linaro.org
Follow PulseAudio backend comment and code, and only implement the
channels QEMU actually supports at this point, and add the same comment
about limits and future mappings. Simplify a bit the code.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230506163735.3481387-13-marcandre.lureau@redhat.com>
D-Bus doesn't support fd-passing on Windows (AF_UNIX doesn't have
SCM_RIGHTS yet, but there are other means to share objects. I have
proposed various solutions upstream, but none seem fitting enough atm).
To make the "-display dbus" work on Windows, implement an alternative
D-Bus interface where all the 'h' (FDs) arguments are replaced with
'ay' (WSASocketW data), and sockets are passed to the other end via
WSADuplicateSocket().
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230606115658.677673-6-marcandre.lureau@redhat.com>
D-Bus on windows doesn't support fd-passing. Let's isolate the
fdlist-related code as a first step, before adding Windows support,
using another mechanism.
Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230606115658.677673-4-marcandre.lureau@redhat.com>
We use the user_ss[] array to hold the user emulation sources,
and the softmmu_ss[] array to hold the system emulation ones.
Hold the latter in the 'system_ss[]' array for parity with user
emulation.
Mechanical change doing:
$ sed -i -e s/softmmu_ss/system_ss/g $(git grep -l softmmu_ss)
Signed-off-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Message-Id: <20230613133347.82210-10-philmd@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
It's already confusing that we have two very similar functions for
wrapping the parse of a 64-bit unsigned value, differing mainly on
whether they permit leading '-'. Adjust the signature of parse_uint()
and parse_uint_full() to be like all of qemu_strto*(): put the result
parameter last, use the same types (uint64_t and unsigned long long
have the same width, but are not always the same type), and mark
endptr const (this latter change only affects the rare caller of
parse_uint). Adjust all callers in the tree.
While at it, note that since cutils.c already includes:
QEMU_BUILD_BUG_ON(sizeof(int64_t) != sizeof(long long));
we are guaranteed that the result of parse_uint* cannot exceed
UINT64_MAX (or the build would have failed), so we can drop
pre-existing dead comparisons in opts-visitor.c that were never false.
Reviewed-by: Hanna Czenczek <hreitz@redhat.com>
Message-Id: <20230522190441.64278-8-eblake@redhat.com>
[eblake: Drop dead code spotted by Markus]
Signed-off-by: Eric Blake <eblake@redhat.com>
This commit adds a new audiodev backend to allow QEMU to use Pipewire as
both an audio sink and source. This backend is available on most systems
Add Pipewire entry points for QEMU Pipewire audio backend
Add wrappers for QEMU Pipewire audio backend in qpw_pcm_ops()
qpw_write function returns the current state of the stream to pwaudio
and Writes some data to the server for playback streams using pipewire
spa_ringbuffer implementation.
qpw_read function returns the current state of the stream to pwaudio and
reads some data from the server for capture streams using pipewire
spa_ringbuffer implementation. These functions qpw_write and qpw_read
are called during playback and capture.
Added some functions that convert pw audio formats to QEMU audio format
and vice versa which would be needed in the pipewire audio sink and
source functions qpw_init_in() & qpw_init_out().
These methods that implement playback and recording will create streams
for playback and capture that will start processing and will result in
the on_process callbacks to be called.
Built a connection to the Pipewire sound system server in the
qpw_audio_init() method.
Signed-off-by: Dorinda Bassey <dbassey@redhat.com>
Reviewed-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230417105654.32328-1-dbassey@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Simplify the resample buffer size calculation.
For audio playback we have
sw->ratio = ((int64_t)sw->hw->info.freq << 32) / sw->info.freq;
samples = ((int64_t)sw->HWBUF.size << 32) / sw->ratio;
This can be simplified to
samples = muldiv64(sw->HWBUF.size, sw->info.freq, sw->hw->info.freq);
For audio recording we have
sw->ratio = ((int64_t)sw->info.freq << 32) / sw->hw->info.freq;
samples = (int64_t)sw->HWBUF.size * sw->ratio >> 32;
This can be simplified to
samples = muldiv64(sw->HWBUF.size, sw->info.freq, sw->hw->info.freq);
With hw = sw->hw this becomes in both cases
samples = muldiv64(HWBUF.size, sw->info.freq, hw->info.freq);
Now that sw->ratio is no longer needed, remove sw->ratio.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-15-vr_qemu@t-online.de>
Upsampling may leave one remaining audio frame in the input
buffer. The emulated audio playback devices are currently
resposible to write this audio frame again in the next write
cycle. Push that task down to audio_pcm_sw_write.
This is another step towards an audio callback interface that
guarantees that when audio frontends are told they can write
n audio frames, they can actually do so.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-13-vr_qemu@t-online.de>
Introduce the new function st_rate_frames_out() to calculate the
exact number of audio output frames the resampling code can
generate from a given number of audio input frames. When upsampling,
this function returns the maximum number of output frames.
This new function replaces the audio_frontend_frames_in()
function, which calculated the average number of output frames
rounded down to the nearest integer. The audio_frontend_frames_in()
function was additionally used to limit the number of output frames
to the resample buffer size. In audio_pcm_sw_read() the variable
resample_buf.size replaces the open coded audio_frontend_frames_in()
function. In audio_run_in() an additional MIN() function is
necessary.
After this patch the audio packet length calculation for audio
recording is exact.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-12-vr_qemu@t-online.de>
The audio_pcm_sw_read() function uses a few very unspecific
variable names. Rename them for better readability.
ret => total_out
total => total_in
size => buf_len
samples => frames_out_max
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-11-vr_qemu@t-online.de>
Replace the resampling loop in audio_pcm_sw_read() with the new
function audio_pcm_sw_resample_in(). Unlike the old resample
loop the new function will try to consume input frames even if
the output buffer is full. This is necessary when downsampling
to avoid reading less audio frames than calculated in advance.
The loop was unrolled to avoid complicated loop control conditions
in this case.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-10-vr_qemu@t-online.de>
Introduce the new function st_rate_frames_in() to calculate the
exact number of audio input frames needed to get a given number
of audio output frames. The exact number of frames depends only
on the difference of opos - ipos and the number of output frames.
When downsampling, this function returns the maximum number of
input frames needed.
This new function replaces the audio_frontend_frames_out() function,
which calculated the average number of input frames rounded down
to the nearest integer. Because audio_frontend_frames_out() also
limited the number of input frames to the size of the resample
buffer, st_rate_frames_in() is not a direct replacement and two
additional MIN() functions are needed. One to prevent resample
buffer overflows and one to limit the available bytes for the audio
frontends.
After this patch the audio packet length calculation for playback is
exact. When upsampling, it's still possible that the audio frontends
can't write the last audio frame. This will be fixed later.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-9-vr_qemu@t-online.de>
The function audio_capture_mix_and_clear() no longer uses
audio_pcm_sw_write() to resample audio frames from one internal
buffer to another. For this reason, the noop_conv() function is
now unused. Remove it.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-8-vr_qemu@t-online.de>
The audio_pcm_sw_write() function is intended to convert a
PCM audio stream to the internal representation, adjust the
volume, and then mix it with the other audio streams with a
possibly changed sample rate in mix_buf. In order for the
audio_capture_mix_and_clear() function to use audio_pcm_sw_write(),
it must bypass the first two tasks of audio_pcm_sw_write().
Since patch "audio: split out the resampling loop in
audio_pcm_sw_write()" this is no longer necessary, because now
the audio_pcm_sw_resample_out() function can be used instead of
audio_pcm_sw_write().
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-7-vr_qemu@t-online.de>
The audio_pcm_sw_write() function uses a lot of very unspecific
variable names. Rename them for better readability.
ret => total_in
total => total_out
size => buf_len
hwsamples => hw->mix_buf.size
samples => frames_in_max
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-6-vr_qemu@t-online.de>
All call sites of audio_pcm_sw_write() guarantee that sw is not
NULL. Remove the unnecessary NULL check.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-5-vr_qemu@t-online.de>
Replace the resampling loop in audio_pcm_sw_write() with the new
function audio_pcm_sw_resample_out(). Unlike the old resample
loop the new function will try to consume input frames even if
the output buffer is full. This is necessary when downsampling
to avoid reading less audio frames than calculated in advance.
The loop was unrolled to avoid complicated loop control conditions
in this case.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-4-vr_qemu@t-online.de>
Read the maximum possible number of audio frames instead of the
minimum necessary number of frames when the audio stream is
downsampled and the output buffer is limited. This makes the
function symmetrical to upsampling when the input buffer is
limited. The maximum possible number of frames is written here.
With this change it's easier to calculate the exact number of
audio frames the resample function will read or write. These two
functions will be introduced later.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Acked-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-3-vr_qemu@t-online.de>
Change the type of the resample buffer from struct st_sample *
to STSampleBuffer. Also change the name from buf to resample_buf
for better readability.
The new variables resample_buf.size and resample_buf.pos will be
used after the next patches. There is no functional change.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-2-vr_qemu@t-online.de>
Change the type of mix_buf in struct HWVoiceOut and conv_buf
in struct HWVoiceIn from STSampleBuffer * to STSampleBuffer.
However, a buffer pointer is still needed. For this reason in
struct STSampleBuffer samples[] is changed to *buffer.
This is a preparation for the next patch. The next patch will
add this line, which is not possible with the current struct
STSampleBuffer definition.
+ sw->resample_buf.buffer = hw->mix_buf.buffer + rpos2;
There are no functional changes.
Acked-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Message-Id: <20230224190555.7409-1-vr_qemu@t-online.de>
Audio recording with ALSA default settings currently doesn't
work. The debug log shows updates every 0.75s and 1.5s.
audio: Elapsed since last alsa run (running): 0.743030
audio: Elapsed since last alsa run (running): 1.486048
audio: Elapsed since last alsa run (running): 0.743008
audio: Elapsed since last alsa run (running): 1.485878
audio: Elapsed since last alsa run (running): 1.486040
audio: Elapsed since last alsa run (running): 1.485886
The time between updates should be in the 10ms range. Audio
recording with ALSA has the same timing contraints as playback.
Reintroduce the default recording settings and use the same
default settings for recording as for playback.
The term "reintroduce" is correct because commit a93f328177
("alsaaudio: port to -audiodev config") removed the default
settings for recording.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-11-vr_qemu@t-online.de>
The currently used default playback settings in the ALSA audio
backend are a bit unfortunate. With a few emulated audio devices,
audio playback does not work properly. Here is a short part of
the debug log while audio is playing (elapsed time in seconds).
audio: Elapsed since last alsa run (running): 0.046244
audio: Elapsed since last alsa run (running): 0.023137
audio: Elapsed since last alsa run (running): 0.023170
audio: Elapsed since last alsa run (running): 0.023650
audio: Elapsed since last alsa run (running): 0.060802
audio: Elapsed since last alsa run (running): 0.031931
For some audio devices the time of more than 23ms between updates
is too long.
Set the period time to 5.8ms so that the maximum time between
two updates typically does not exceed 11ms. This roughly matches
the 10ms period time when doing playback with the audio timer.
After this patch the debug log looks like this.
audio: Elapsed since last alsa run (running): 0.011919
audio: Elapsed since last alsa run (running): 0.005788
audio: Elapsed since last alsa run (running): 0.005995
audio: Elapsed since last alsa run (running): 0.011069
audio: Elapsed since last alsa run (running): 0.005901
audio: Elapsed since last alsa run (running): 0.006084
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-10-vr_qemu@t-online.de>
Now that the last call site of audio_calloc() was removed, remove
the unused audio_calloc() function.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-9-vr_qemu@t-online.de>
Replace audio_calloc() with the equivalent g_new0().
With a n_structs argument >= 1, g_new0() never returns NULL.
Also remove the unnecessary NULL checks.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-8-vr_qemu@t-online.de>
Use g_malloc0() as a direct replacement for audio_calloc().
Since the type of the parameter n_bytes of the function g_malloc0()
is unsigned, the type of the variables voice_size_out and
voice_size_in has been changed to size_t. This means that the
function argument no longer has to be checked for negative values.
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Daniel P. Berrangé <berrange@redhat.com>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-7-vr_qemu@t-online.de>
Replace audio_calloc() with the equivalent g_new0().
The value of the g_new0() argument count is >= 1, which means
g_new0() will never return NULL. Also remove the unnecessary
NULL check.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-6-vr_qemu@t-online.de>
Replace audio_calloc() with the equivalent g_new0().
With a n_structs argument of 1, g_new0() never returns NULL.
Also remove the unnecessary NULL checks.
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-5-vr_qemu@t-online.de>
Remove the unused #define AUDIO_STRINGIFY. It was last used before
commit 470bcabd8f ("audio: Replace AUDIO_FUNC with __func__").
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-4-vr_qemu@t-online.de>
Use a consistent friendly name for the HWVoiceOut and HWVoiceIn
structures.
Reviewed-by: Thomas Huth <thuth@redhat.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Message-Id: <20230121094735.11644-3-vr_qemu@t-online.de>
Let the audio_pcm_create_voice_pair_* functions handle error
reporting. This avoids an additional error message in case
the guest selected an unimplemented sample rate.
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-2-vr_qemu@t-online.de>
Some emulated audio devices allow guests to select very low
sample rates that the audio subsystem doesn't support. The lowest
supported sample rate depends on the audio backend used and in
most cases can be changed with various -audiodev arguments. Until
now, the audio_bug function emits an error message similar to the
following error message
A bug was just triggered in audio_calloc
Save all your work and restart without audio
I am sorry
Context:
audio_pcm_sw_alloc_resources_out passed invalid arguments to
audio_calloc
nmemb=0 size=16 (len=0)
audio: Could not allocate buffer for `ac97.po' (0 samples)
and the audio subsystem continues without sound for the affected
device.
The fact that the selected sample rate is not supported is not a
guest error. Instead of displaying an error message, the missing
audio support is now logged. Simply continuing without sound is
correct, since the audio stream won't transport anything
reasonable at such high resample ratios anyway.
The AUD_open_* functions return NULL like before. The opened
audio device will not be registered in the audio subsystem and
consequently the audio frontend callback functions will not be
called. The AUD_read and AUD_write functions return early in this
case. This is necessary because, for example, the Sound Blaster 16
emulation calls AUD_write from the DMA callback function.
Acked-by: Christian Schoenebeck <qemu_oss@crudebyte.com>
Signed-off-by: Volker Rümelin <vr_qemu@t-online.de>
Reviewed-by: Marc-André Lureau <marcandre.lureau@redhat.com>
Message-Id: <20230121094735.11644-1-vr_qemu@t-online.de>
This commit was created with scripts/clean-includes.
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Reviewed-by: Michael S. Tsirkin <mst@redhat.com>
Reviewed-by: Juan Quintela <quintela@redhat.com>
Message-Id: <20230202133830.2152150-18-armbru@redhat.com>
This moves these commands from MAINTAINERS sections "Human
Monitor (HMP)" and "QMP" to "Overall Audio backends".
Signed-off-by: Markus Armbruster <armbru@redhat.com>
Message-Id: <20230124121946.1139465-3-armbru@redhat.com>
Currently the -audiodev accepts any audiodev type regardless of what is
built in to QEMU. An error only occurs later at runtime when a sound
device tries to use the audio backend.
With this change QEMU will immediately reject -audiodev args that are
not compiled into the binary. The QMP schema will also be introspectable
to identify what is compiled in.
This also helps to avoid compiling code that is not required in the
binary. Note: When building the audiodevs as modules, the patch only
compiles out code for modules that we don't build at all.
Signed-off-by: Daniel P. Berrangé <berrange@redhat.com>
[thuth: Rebase, take sndio and dbus devices into account]
Message-Id: <20230123083957.20349-3-thuth@redhat.com>
Signed-off-by: Thomas Huth <thuth@redhat.com>
