post-portem Debugging in Ubuntu
nio.wiklund at gmail.com
Sat May 12 08:57:36 UTC 2012
On 2012-05-11 15:01, Yotam Rubin wrote:
> Hi guys,
> A big QA challenge is understanding and resolving bugs that occur on
> client machines. More often than not, a bug will be reported and then
> closed, due to a lack of debug information or a failure to reproduce the
> issue. I'd like to refer you to Traces, a tracing infrastructure for C/C++
> Linux applications. Traces enables the tracing of all function calls
> occurring within an application as well as provide an API for traditional,
> explicit tracing. The traces are designed to have small performance
> overhead, so they can be active at all times. Produced trace files are
> atomic in the sense that they contain all the metadata necessary for
> interpreting the trace.
> What do you think about the possibility of adapting Traces into key
> components of the distribution? It could be very useful to have the user
> send a complete trace of a program in the event of a failure.
> Anyway, here's the code:
I think this is a good idea, but I don't know how easy it would be to
run 'traces'. Would I (the end user) have to compile
- the code to be debugged
or would it be more or less automatic after installation?
Please describe the necessary procedure at the end user?
And also, would there be any security implications? In other words,
should it be limited to a testing platform rather than the main computer
for work or private tasks?
More information about the Ubuntu-qa