GSoC Project | Basic Support for Trace Compass

Sebastian Huber sebastian.huber at
Mon Jul 15 06:01:48 UTC 2019

On 15/07/2019 07:46, Ravindra Kumar Meena wrote:
> *Plan of the week:*
> Last week packet.header part from both client-side and metadata was 
> completed. I will continue to work on adding packet.context etc.
> I got the response from lltng about stream_instance_id.

Ok, did you double check that the stream_instance_id is equal to the 
cpu_id in an example LTTNG file?

> I have made changes accordingly.
> I have moved the generation of binary stream files directly to 
> ./misc/record/ctf/. See above commit.

Why have you changed this? The tool should output the files in the 
current directory. Please change it back to the original version.

> Blocker:
> The packet.context is needed at the starting of the binary file(just 
> after packet.header). The packet.context has timestamp_end which can be 
> obtained after processing last record item but print_item() is already 
> writing the record_item in the binary file. The same case is with 
> packet_size and content_size. These two values can be obtained after 
> processing the last record item in the file but we need at the starting 
> of a binary file.
> We need packet.context just after packet.header in the binary file. So 
> as of now, I have just added cpu_id in packet.context on both 
> client-side and metadata.
> It has no warnings.
> I have one approach for it. We can try the buffer approach for it. Just 
> write the file packet.header and packet.context in a new binary file and 
> append the old file content in a new file.

Please have a look at the fseek() function:

Sebastian Huber, embedded brains GmbH

Address : Dornierstr. 4, D-82178 Puchheim, Germany
Phone   : +49 89 189 47 41-16
Fax     : +49 89 189 47 41-09
E-Mail  : sebastian.huber at
PGP     : Public key available on request.

Diese Nachricht ist keine geschäftliche Mitteilung im Sinne des EHUG.

More information about the devel mailing list