rohc team mailing list archive

[Didier Barvaux <didier.barvaux@xxxxxxxxxxxxxxxxxxxx>]

Hi,

> You are right.I was able to solve the problem.
> by doing two things.
> 
> First in d_generic.c packet was being accessed using (uint16*)
> typecast.
>
> The crash was happening here. I changed two byte access.

I'm interested by your change. Could you please give me the function
name and the line you changed and the exact change you made as well ?

 
> another issue i faced after this was feedback, in bidirectional
> decompressor case.
> 
> i stopped that by allocating decompressor with NULL.
> 
> My system is point to point . What are the advantages of bidirectional
> decompressor?

Bidirectional behaviour of the decompressor makes use of feedback
information to warn the ROHC compressor that a problem occurred (lost
or bad packets that lead to compressor/decompressor
mis-synchronization).

When the ROHC compressor receives the feedback information, it sends
information about the compressed stream to the decompressor to
synchronize the decompressor again. This should reduce the number of
ROHC packets the decompressor fails to decompress correctly after a
network problem occurred.

I'm also interested by the root of the problem with the feedback
mechanism. Could you please run your program in gdb, wait for it to
crash and send me the backtrace you get ? If you do not know how to use
gdb, I can give you detailed information to get the backtrace.

Regards,

Didier Barvaux
Viveris Technologies