Bug 186152 – Fix leak of AudioDeviceID array due to an early return in AudioDeviceMac::GetNumberDevices()

• Component: WebRTC
• Status: RESOLVED FIXED

[David Kilzer (:ddkilzer)]

Fix leak of AudioDeviceID array due to an early return in AudioDeviceMac::GetNumberDevices():

Source/ThirdParty/libwebrtc/Source/webrtc/modules/audio_device/mac/audio_device_mac.cc:1586:3: warning: Potential leak of memory pointed to by 'deviceIds'
  WEBRTC_CA_RETURN_ON_ERR(AudioObjectGetPropertyData(kAudioObjectSystemObject,
  ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Source/ThirdParty/libwebrtc/Source/webrtc/modules/audio_device/mac/audio_device_mac.cc:30:7: note: expanded from macro 'WEBRTC_CA_RETURN_ON_ERR'
      logCAMsg(rtc::LS_ERROR, "Error in " #expr, (const char*)&err); \
      ^~~~~~~~

We can use RAII with std::unique_ptr<>() to fix the leak and remove manual release of the memory using this technique:
<https://www.codeproject.com/Articles/820931/Using-std-unique-ptr-RAII-with-malloc-and-free>

[Radar WebKit Bug Importer]

<rdar://problem/40692824>

[David Kilzer (:ddkilzer)]

Created attachment 341679 [details]
Patch v1

[Alex Christensen]

Comment on attachment 341679 [details]
Patch v1

Can't we use std::make_unique?

[David Kilzer (:ddkilzer)]

(In reply to Alex Christensen from comment #3)
> Comment on attachment 341679 [details]
> Patch v1
> 
> Can't we use std::make_unique?

I was aiming to change the code as little as possible, but yes we can!  New patch forthcoming.

[David Kilzer (:ddkilzer)]

Created attachment 341681 [details]
Patch v2

[David Kilzer (:ddkilzer)]

Created attachment 341684 [details]
Patch v3

Remove 'const' from 'numberDevices' since it used in an in-out context in AudioObjectGetPropertyData().

[WebKit Commit Bot]

Comment on attachment 341684 [details]
Patch v3

Clearing flags on attachment: 341684

Committed r232365: <https://trac.webkit.org/changeset/232365>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.