Forum Discussion
Teams HAL driver update skips coreaudiod restart mediaremoted spins at 100% CPU
Environment: macOS 15.7.7 build 24G720, Apple M3 Max. Teams updated from 26183.1901.4874.5228 to 26198.202.4929.7171 via Microsoft AutoUpdate.
What happens: The delta package installs the MSTeamsAudioDevice component into the system-wide CoreAudio HAL plug-ins folder. The system install log shows the bundle being atomically replaced, after which the postinstall script logs that it is a non-interactive command line installation and therefore will not restart coreaudiod, then exits. Because MAU always runs non-interactively, coreaudiod is never restarted and continues running against the replaced HAL bundle.
Impact: Long-lived processes holding references to the Teams audio device never converge. In my case mediaremoted entered a permanent single-core spin at 107 percent CPU, sustained for days. A sample shows 100 percent of main-thread time inside AudioObjectGetPropertyData and AudioObjectGetPropertyDataSize, with the HALObjectMap CopyObjectByObjectID and ReleaseObject methods accounting for roughly 99 percent of top-of-stack samples. HALPlugIn ObjectHasProperty also emits an os_log call on every probe, so logd burns additional CPU. Teams itself was not running.
Workaround: Force-quit the core audio daemon and then the media remote daemon; both relaunch automatically under launchd.
Suggested fix: Restart coreaudiod from the postinstall regardless of interactivity whenever the HAL bundle payload actually changed, or defer the bundle swap until the next coreaudiod restart.
1 Reply
Your evidence connects the MAU update, replacement of the MSTeamsAudioDevice HAL plug-in, and mediaremoted at 107% CPU. That makes a stale Core Audio state a credible explanation, although the screenshot cannot prove an installer defect or guarantee that restarting coreaudiod is the product fix. For a supported recovery, quit Teams and restart macOS; this reloads Core Audio and dependent daemons without manually killing protected processes. Then install the newest Teams and Microsoft AutoUpdate builds and watch Activity Monitor after the next update. If the spin recurs, collect Teams support files using Help > Collect support files, plus the macOS install log, a process sample of mediaremoted, UTC occurrence time, and before/after Teams versions. Open a Microsoft 365 admin support case and attach that evidence. Keep force-quitting daemons only as a temporary diagnostic workaround; Microsoft must confirm and correct any post-install issue