AudioDeviceManager: Avoid dereferencing null channel pointers - #1708
Draft
emezeske wants to merge 1 commit into
Draft
AudioDeviceManager: Avoid dereferencing null channel pointers#1708emezeske wants to merge 1 commit into
emezeske wants to merge 1 commit into
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
A device that passes null pointers for channels it was not opened with crashes on the audio thread, in AudioDeviceManager::LevelMeter::updateLevel, which dereferences them unconditionally. AudioIODeviceCallback documents that those pointers may be null. The ALSA case in #1706 is one way to get there.
CallbackMaxSizeEnforcer can produce the same situation on its own. It sizes its stored channel vectors from the device's active channel bitsets when audioDeviceAboutToStart runs, but fills in only the number of channels the device actually passes to the callback, and then forwards the vectors' full size. If a device's bitsets and channel counts disagree, the tail of those vectors still holds the null pointers that resize() default constructed. The assertions catch it in a debug build, but a release build passes the nulls straight through to the inner callback. This forwards the counts that were really filled in instead.