NetBSD-Bugs archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: kern/54229: audio in firefox stops playing

The following reply was made to PR kern/54229; it has been noted by GNATS.

From: Tetsuya Isaki <>
Subject: Re: kern/54229: audio in firefox stops playing
Date: Mon, 27 May 2019 22:17:48 +0900

 > 1. I play audio in cmus and open youtube tab
 > 2. pause cmus while youtube is loading
 > 3. firefox will refuse to play any audio until it is killed.
 Firefox reopens sound device on the page (video) transition on
 youtube.  And firefox (media/libcubeb/src/cubeb_oss.c) uses
 "/dev/sound" that some parameters (including pause) are sticky.
 Therefore, if someone pauses audio, firefox is affected.  And it
 also happens on previous audio driver.
  1. boot any of
     * NetBSD 8 or
     * -current before merging isaki-audio2 (<8.99.39) or
     * -current after merging isaki-audio2 (>=8.99.39).
     (I think NetBSD7 is the same but I haven't confirmed)
  2. audioctl -d /dev/sound -w play.pause=1
  3. execute firefox and play something on youtube.
     -> firefox cannot play video.
 Unless firefox expects that it stops by external operation, I think
 that using /dev/sound in firefox is a bug, although I'm not sure
 about pkgsrc/www/firefox history.
 --- www/firefox/patches/patch-aa.ORIG	2018-11-12 21:50:52.000000000 +0900
 +++ www/firefox/patches/patch-aa	2019-05-27 15:19:05.000000000 +0900
 @@ -60,7 +60,7 @@
  +    dnl Assume NetBSD implementation over SunAudio
  +    AC_CHECK_LIB(ossaudio, _oss_ioctl,
  +         MOZ_OSS_LIBS="$MOZ_OSS_LIBS -lossaudio"])
 There is another theme to consider.
 On NetBSD7 and earlier, /dev/sound maintained the last /dev/audio or
 /dev/sound's parameters.
 On NetBSD8, /dev/sound maintained the last /dev/sound's parameters.
 On >=8.99.39(isaki-audio2), /dev/sound maintains the last /dev/audio
 or /dev/sound's parameters (it's the same as NetBSD7).
 firefox uses /dev/sound and cmus uses /dev/audio.  So this problem
 was found easier after merge.
 I don't know why NetBSD7 and 8 behave differently.
 I personally think /dev/sound's such sticky behavior is the good old
 days and is not suitable for modern system...
 Tetsuya Isaki < />

Home | Main Index | Thread Index | Old Index