1113846 - Is ALOGV logging enabled in release builds?

Status: RESOLVED INACTIVE

(Core :: Audio/Video: Playback, defect)

Description

[Markus Stange [:mstange]]

Attachment: hang profile

Around 1.5 months ago I caught a Firefox hang with the Activity Monitor sampler. I'm attaching the profile here. Among other things, the profile shows that a media decode thread spends lots of time in __android_log_print, which seems very strange.

I haven't seen this issue since then, but that may be the case because I've forced youtube to use Flash again.

MPEG4Extractor::parseChunk contains calls to a macro called ALOGV, which is defined by /media/libstagefright/system/core/include/log/log.h:

> #ifndef LOG_NDEBUG
> #ifdef NDEBUG
> #define LOG_NDEBUG 1
> #else
> #define LOG_NDEBUG 0
> #endif
> #endif
> 
> // ---------------------------------------------------------------------
> 
> /*
>  * Simplified macro to send a verbose log message using the current LOG_TAG.
>  */
> #ifndef ALOGV
> #if LOG_NDEBUG
> #define ALOGV(...)   ((void)0)
> #else
> #define ALOGV(...) ((void)ALOG(LOG_VERBOSE, LOG_TAG, __VA_ARGS__))
> #endif
> #endif

Is NDEBUG defined in release builds? Because if it is not defined, we end up in the
> #else
> #define ALOGV(...) ((void)ALOG(LOG_VERBOSE, LOG_TAG, __VA_ARGS__))
case, which causes ALOGV to do work. Is the last #if LOG_NDEBUG just reversed?

I'm adding a needinfo to rillian because he was looking into bug 1061395 earlier, which sounds related.

Comment 1

[Ralph Giles (:rillian)]

Sorry I haven't been able to look at this. Will be able to get to it in January.

Comment 2

[Anthony Jones (:ajones, :kentuckyfriedtakahe, :k17e)]

This no longer matters because we only run the Android demuxer when changing streams these days and we're working on replacing it.

Comment 3

[Nils Ohlmeier [:drno]]

Mass closing because of inactivity.
Please feel free to re-open if still relevant.