Add results for test-case 2

Signed-off-by: Dylan Aïssi <dylan.aissi@collabora.com>

README.md

@@ -12,7 +12,7 @@ linkcolor: blue
 colorlinks: true
 ---
 
-*[Draft report - 2023-10-30](https://gitlab.apertis.org/daissi/pipewire-evaluation)*
+*[Report - 2023-10-31](https://gitlab.apertis.org/daissi/pipewire-evaluation)*
 
 
 [PipeWire](https://pipewire.org/) is *de facto* the new audio server used by all
@@ -248,13 +248,51 @@ The script `test-case-capture.py` can be used to record sound generated by the
 board sound cards during the test (i.e. with changing capture sample rate of the
 stream redirected to 1 output) to evaluate how the sound is affected.
 
-**TODO**:
-- Fix bug where capture sample rate is not changed when `start_sink("my-sink-B")`
-  is running.
-- Reduce time to applied new property to a device or force the device to be
-  reinitialized with new properties. For now, we have to wait until the device
-  is IDLE otherwise changes (sample rate change) are not applied.
-  Something to improve in PW?
+## Results
+Currently, it's not possible to reconfigure a running device. That means, to
+reconfigure the sample rate input, we have to force the device to be idle. In
+other words, that means both **input and output** of this device must be idle.
+Only forcing the device *input* to be idle is **not enough** to make pipewire
+apply the changes.
+
+During the test, two wav files are generated `test-case-capture1.wav` and
+`test-case-capture2.wav` (both are available in `results/test-case-2/`). The
+first one captures the sound coming from the board USB sound card while the
+second capture the sound coming from the board built-in sound card.
+
+In `test-case-capture1.wav`, we can hear the `test300.wav` tone without any
+artefact and without any interruption. This stream is therefore not affected by
+the reconfiguration of the input of the other sound card.
+
+In the other file `test-case-capture2.wav`, the stream starts with the
+`test300.wav` tone, then at *0:03* we stop hearing the tone to make the device
+idle allowing the reconfiguration of the sample rate at 32000. At *0:08*, we start
+hearing the tone mixed with the `MIC Input` (sample rate at 32000) for 10 secs.
+At *0:18*, we have a new stream break for 5 secs for the reconfiguration of the
+input rate at 44100. Then, the mixed stream (`test300.wav` tone and `MIC Input`)
+restart at *0:18* for 10 secs and so on. The sample rate reconfiguration is
+checked during the test in the `FORMAT` column of the `pw-top` output.
+
+In this case, the output is affected by the reconfiguration because we have to
+wait the device to be idle to make pipewire apply the new configuration. The
+default timeout for switching a device to idle is defined in a wireplumber
+config file (i.e. `/usr/share/wireplumber/main.lua.d/50-alsa-config.lua`) at 5
+secondes (see *session.suspend-timeout-seconds*). It should be possible to
+reduce the timeout to only 1 seconde (setting `0` disables suspend on idle, so
+not the desired behavior), but that will still impact the direct output.
+
+Moreover, the reconfiguration of sample rate was applied only on the input node
+of the sound (i.e. in our example `alsa_input.platform-sound.stereo-fallback`) and
+not on the whole device, but `pw-top`reported during the test that the output node
+(i.e. `alsa_output.platform-sound.stereo-fallback`) was reconfigured in the same
+way. That means both *input* and *output* are affected by the reconfiguration
+on the *input*.
+
+To summarize, the reconfiguration of audio capture impacts the direct output as
+we have to stop the stream to make pipewire apply the changes, but it doesn't
+impact the unrelated output. If feasible, adding a way to force pipewire to
+reconfigure a device without impacting (or at least with by reducing its impact)
+the current streams is probably a way to go.
 
 # Test case 3: Capturing multiple inputs to multiple outputs
 
@@ -422,7 +460,10 @@ built-in sound card, it's a little longer with *1min06* (+6secs) to fully recove
 # Results summary
 - Adding or removing an audio stream in a running output neither impacts that
   output or any other system outputs. See *Test case 1: Results* for more details.
-- TODO. See *Test case 2: Results* for more details.
+- Reconfigurating of audio capture impacts the direct output as we have to make
+  the whole device (input **and** output) idle to make pipewire apply the changes,
+  but it doesn't impact the unrelated output. See *Test case 2: Results* for more
+  details.
 - Starting and/or stopping inputs independently does not cause disruptions in the
   outputs, but mixing streams from different sources gives sound arfects during
   our test. See *Test case 3: Results* for more details.

scripts/test-case-2-decrease-timeout-idle.sh

+#!/bin/sh
+
+# Decrease timeout to suspend device from 5secs to 1sec
+
+sudo mkdir -p /etc/wireplumber/main.lua.d/
+
+sudo cp -a /usr/share/wireplumber/main.lua.d/50-alsa-config.lua \
+                 /etc/wireplumber/main.lua.d/50-alsa-config.lua
+
+sudo sed -i 's/--\["session.suspend-timeout-seconds"\] = 5/--\["session.suspend-timeout-seconds"\] = 1/g' /etc/wireplumber/main.lua.d/50-alsa-config.lua
+
+systemctl restart --user wireplumber.service

scripts/test-case-2-run.py

 #!/usr/bin/env python3
 
+#from pwevaltools import *
+
 from itertools import cycle
 import json
 import select
@@ -30,12 +32,6 @@ def pw_cli_rate(device_id, rate):
     subprocess.run(cmd_pw_cli, shell=True)
 
 
-def pw_cli_reduce_timeout_idle(device_id):
-    cmd_pw_cli = "pw-cli s "+str(device_id)+" Props '{ params: [ \"session.suspend-timeout-seconds\", \"1\" ] }'"
-    print(f".. DBG: {cmd_pw_cli}")
-    subprocess.run(cmd_pw_cli, shell=True)
-
-
 def pw_link(out_pattern, in_pattern):
     cmd_pw_link = ["pw-link",str(out_pattern),str(in_pattern)]
     subprocess.run(cmd_pw_link)
@@ -56,16 +52,20 @@ def pw_metadata(out_pattern, in_pattern):
 def create_loopback_device(name):
     print(f"Creating a virtual device named {name}")
 
-    cmd_pw_loopback = "pw-loopback -m '[FL FR]' --capture-props='media.class=Audio/Sink node.name="+name+"'"
+    cmd_pw_loopback = "pw-loopback -m '[FL FR]'"
+    cmd_pw_loopback += " --capture-props='media.class=Audio/Sink node.name="+name+"'"
+    #cmd_pw_loopback += " --playback-props='node.name="+name+"'"
     subprocess.Popen(cmd_pw_loopback, shell=True)
 
 
 def create_loopback_mic_device(name):
     print(f"Creating a virtual device named {name}")
 
-    cmd_pw_loopback = "pw-loopback -m '[FL FR]' --capture-props='media.class=Audio/Sink node.name="+name+"'"
-    cmd_pw_loopback += " --capture alsa_input.platform-sound.stereo-fallback --playback alsa_output.platform-sound.stereo-fallback"
+    cmd_pw_loopback = "pw-loopback -m '[FL FR]'"
+    cmd_pw_loopback += " --capture-props='media.class=Audio/Sink node.name="+name+" target.object=alsa_input.platform-sound.stereo-fallback'"
+    cmd_pw_loopback += " --playback-props='node.name="+name+" target.object=alsa_output.platform-sound.stereo-fallback'"
     subprocess.Popen(cmd_pw_loopback, shell=True)
+    time.sleep(2) # Wait for the node to be up
 
 
 def start_jackplay(my_sound):
@@ -195,11 +195,13 @@ def start_sink(my_sink_name, my_wav, device_playback, my_id_used=None):
     my_pw_dump = pw_dump("pw-dump-temp-"+my_sink_name+"-4.json")
     pw_dot("pw-dot-temp-"+my_sink_name+"-4.png")
 
+    my_sink_node_group = "output."+my_sink_node_group
+
     # This is required to identify which jackplay are already processsed
-    return my_jackplay_id
+    return my_jackplay_id, my_sink_node_group
 
 
-def capture_mic_reconf(my_input_mic):
+def capture_mic_reconf(my_input_mic, node_to_close):
     my_pw_dump = pw_dump("pw-dump-temp-mic-0.json")
     my_capture_id = None
     for pw_obj in my_pw_dump:
@@ -214,18 +216,12 @@ def capture_mic_reconf(my_input_mic):
         my_capture_id = pw_obj["id"]
         break
 
-    # Reduce timeout for setting a device IDLE, this is requied since we will
-    # change device's properties (sample rate), but those are applied only when
-    # device is IDLE.
-    #pw_cli_reduce_timeout_idle(my_capture_id)
-    # Doesn't work...
-
     # Due to hardware limitation we cannot directly capture at 8K or 16K Hz
     # Only 32000, 44100 and 480000 are supported
     # So, instead we will switch between 32K and 48K to capture "MIC Input"
     my_supported_sample_rates = [32000, 44100, 48000]
     my_loopback_mic_name = "my_loopback_mic"
-    create_loopback_device(my_loopback_mic_name)
+    create_loopback_mic_device(my_loopback_mic_name)
     pw_dot("pw-dot-temp-mic-0.png")
 
     print("\nEverything is ready for capturing MIC input!")
@@ -236,20 +232,30 @@ def capture_mic_reconf(my_input_mic):
 
     # Change sample rate every 10 secs
     for my_srate in cycle(my_supported_sample_rates):
+      pw_link_delete(my_input_mic, my_loopback_mic_name)
+      pw_link_delete(node_to_close, my_sound_cards[1])
+      pw_link_delete(my_loopback_mic_name+":output_FL", my_sound_cards[1]+":playback_FL")
+      pw_link_delete(my_loopback_mic_name+":output_FR", my_sound_cards[1]+":playback_FR")
+
+      # Need to wait until device is idle otherwise reconfigure is not applied,
+      # Something to improve in pipewire? adding an option to force a device
+      # reconfig without having to wait idle?
+      time.sleep(5)
+
       print(f"Setting capture sample rate to {my_srate}")
-      # !!! FIXME !!! changing srate (i.e. pw_cli_rate) doesn't work when `start_sink("my-sink-B")` is running
       pw_cli_rate(my_capture_id, my_srate)
-      time.sleep(5)
-      # Need to wait until device is IDLE otherwise reconfigure isn't applied,
-      # TODO: this part needs to be improved in pw? (force a device to be reinitilizaed if we change properties?)
+
       pw_link(my_input_mic, my_loopback_mic_name)
+      pw_link(node_to_close, my_sound_cards[1])
+      pw_link(my_loopback_mic_name+":output_FL", my_sound_cards[1]+":playback_FL")
+      pw_link(my_loopback_mic_name+":output_FR", my_sound_cards[1]+":playback_FR")
+
       # pw_dot("pw-dot-temp-mic-X.png")
       print(f"Please check sample rate with pw-link") # For DEBUG
       print(f"... sample rate for {my_input_mic} should be {my_srate}") # For DEBUG
       a, b, c = select.select([sys.stdin], [], [], 10)
       if (a):
         break
-      pw_link_delete(my_input_mic, my_loopback_mic_name)
 
 
 if __name__ == "__main__":
@@ -264,13 +270,13 @@ if __name__ == "__main__":
     my_input_mic = "alsa_input.platform-sound.stereo-fallback"
 
     # Make sound with USB sound card
-    my_jackplay_id_used_A = start_sink("my-sink-A", "test300.wav", my_sound_cards[0])
+    my_jackplay_id_used_A, my_jackplay_node_group_A = start_sink("my-sink-A", "test300.wav", my_sound_cards[0])
 
     # Make sound with onboard card
-    my_jackplay_id_used_B = start_sink("my-sink-B", "test300.wav", my_sound_cards[1], my_jackplay_id_used_A)
+    my_jackplay_id_used_B, my_jackplay_node_group_B = start_sink("my-sink-B", "test300.wav", my_sound_cards[1], my_jackplay_id_used_A)
 
     # Capture MIC INPUT on board with audio input reconfiguration
-    capture_mic_reconf(my_input_mic)
+    capture_mic_reconf(my_input_mic, my_jackplay_node_group_B)
 
     # End of test case 2
     print("Press any key to stop the playback...")