CR for the PR 200286: Shoutcast mp3 stream played through httpfs stops playback after

okay

ManaV wrote:
> Synopsis
> 
> ========
> 
>  
> 
> This CR includes fix for the PR 200286: Shoutcast mp3
stream played 
> through httpfs stops playback after a few seconds,
spins server CPU at 100%
> 
>  
> 
> Quick link to the PR:
> 
> https://bugs.dev.prognet.com/show_bug.cgi?id=20028
6&link=0 
> <https://bugs.dev.prognet.com/show_bug.cgi?id=2
00286&link=0>
> 
>  
> 
> Branch: HEAD
> 
>  
> 
> Suggested Reviewer: Anyone
> 
>  
> 
>  
> 
> Description
> 
> ===========
> 
>  
> 
> While playing the live MP3 (Shoutcast) stream through
httpfsys via RTSP, 
> the stream starts up but playback stops after a few
seconds, and player 
> used to sit on the communicating...
> 
> These are the URL's I used:
> 
>
rtsp://myserver/httpfs/http:%2f/kexp-mp3-1.cac.washington.ed
u:8000/
> 
> rtsp://myserver/httpfs/http:%2f/205.188.215.226:8000/
> 
>  
> 
> The problem was occurring because after some time
server used to get 
> locked inside the LOOP in the function
CRnMp3Fmt::MyCreatePacketObj_hr() 
> and the call for the PacketReady was not happening
which results into 
> server stops sending data packet to the player after
some time. Inside 
> the GetMP3Frame_p when we are not reading at ValidMP3
frame we search 
> for the next valid frame by searching through the
SyncWord. When we find 
> the valid frame after the SearchForSyncWord call, we
search whether this 
> frame contains the meta data or not. If it consists of
meta data then it 
> should only set the m_bMetaPacket and shouldn't modify
the nSyncSize and 
> lScan. Previously on modifying:
> 
> 1. lScan when we need to move to the current frame then
these modified 
> value won't help in progressing to right.
> 
> 2. nSyncSize when search for the next valid frame due
to the addition of 
> m_ulMetaLength to it, it wont be pointing to the next
valid frame and 
> when we return from the loop it will search for the
next SyncWord, and 
> again same thing will happen, and also as the
calculation of next meta 
> position m_ulNextMetaPos is not there in the loop so it
wont be 
> modifying m_ulNextMetaPos which cause the condition
check for the meta 
> data to become true always. In this scenario it used to
stuck in this 
> infinite loop.
> 
>  
> 
> I have fixed this problem by omitting the modification
of the nSyncSize 
> in case of Server Only. I have cross verified it, Now
it is playing the 
> Shoutcast MP3 stream just fine.
> 
>  
> 
> Files Affected
> 
> ==============
> 
>  
> 
> Datatype/mp3/fileformat/mp3ff.cpp
> 
>  
> 
> Testing Performed
> 
> =================
> 
>  
> 
> Unit Tests:
> 
> - None
> 
>  
> 
> Integration Tests:
> 
> - I have tested it. Now it streams the shoutcast MP3
fine and player no 
> more sits on the communicating.
> 
>  
> 
> Leak Tests:
> 
> - None
> 
>  
> 
> Performance Tests:
> 
> - None
> 
>  
> 
>  
> 
> Platforms Tested: win32-i386-vc7
> 
> Build verified: win32-i386-vc7
> 
>  
> 
> QA Hints
> 
> ========
> 
>  
> 
> - None
> 
>  
> 
> Thanks and Regards,
> 
> Manvendra Singh
> 
> 
>
------------------------------------------------------------
------------
> 
> _______________________________________________
> Datatype-dev mailing list
> Datatype-devhelixcommunity.org
> http://lists.helixcommunity.org/mailman/listinfo/da
tatype-dev


_______________________________________________
Datatype-dev mailing list
Datatype-devhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/da
tatype-dev