debugging mms stream failure?

[dbrashear]

something recently changed with a stream I listen to and now VLC won't open it. other tools still do. i've looked at what's up with TCPdump and nothing obvious is happening. the stream simply stops being acked (tcp-wise) as if we have stopped bothering to service it.

i'm perfectly happy to get my hands dirty, just looking for a hint as to what i might pursue to get some verbosity from the asf stream support so perhaps it will tell me where to look.

[Jean-Baptiste Kempf]

Did you look at the verbose logs from VLC?

[dbrashear]

running in --verbose=2 from a command line I get nothing but I'm betting you mean something else?

fwiw it's mms://wmc1.den.liquidcompass.net/WWCDFM

[dbrashear]

[0x1030078a0] access_mms access error: invalid chunk FATAL (0x44a4)
[0x1030078a0] access_mms access error: EOF!

if i remove the error i can get skippy audio. still looking at *why*

[dbrashear]

Something's weird with GetWLE apparently; in the cases where the type is 0x44A4 i set it to 0x4424 and add 0x80 to the size, the stream plays fine.

So what does this mean?

[dbrashear]

This patch fixes it, but i don't know if it's "right"
--- /Users/shadow/Source/vlc/modules/access/mms/mmsh.c 2011-01-11 02:51:15.000000000 -0500
+++ modules/access/mms/mmsh.c 2011-01-16 20:13:15.000000000 -0500
@@ -948,6 +910,10 @@

p_ck->i_type = GetWLE( p_sys->buffer);
p_ck->i_size = GetWLE( p_sys->buffer + 2);
+ if (p_ck->i_type == 0x44A4) {
+ p_ck->i_type = 0x4424;
+ p_ck->i_size += 0x80;
+ }

restsize = p_ck->i_size;
if( restsize > 8 )

[Jean-Baptiste Kempf]

Do you have documentation about this?

[dbrashear]

None. The sum of my knowledge is Windows Media Player plays the stream, Silverlight plays the stream, mplayer plays the stream, VLC opens a connection and does not play the stream, and the above appears to be the difference. The stream URL is there and open, so it should be evident, at least, that this is not a problem of my own making. It's 100% reproducible. The server there claims to be Server: Rex/11.0.5721.5265.

[dbrashear]

Oh, as far as why those numbers? Because 0x44A4 is 0x80 more than 0x4424, and my suspicion on seeing that was one bit in the type is actually part of the size. But I haven't been able to find anything which documents that.