09:14 <sonal> hi, i want to use string.h functions in stap script, how to use it using guru mode

10:52 <sonal> can we club system tap with Python?

11:25 <sanoj> I am trying to write an embedded C routine to iterate through certain structures in my user space code. The structures need header file like sys/types.h . The issue I am facing is user space headers are conflicting with kernel headers.

11:26 <sanoj> How do I selove this. eg: /usr/include/stdint.h:40:19: error: conflicting types for ‘int64_t’ . include/linux/types.h:113:17: note: previous declaration of ‘int64_t’ was here

12:21 <fche> sonal, userspace string.h are not appropriate for use from with kernelspace modules; might have more luck with kernel string libraries

12:22 <fche> sanoj, how are you getting at that sys/types.h file? have you tried @cast() to generate debuginfo from it, and using script language to iterate, vs. embedded-c with those problematic #include's ?

12:22 <fche> sanoj, you might also find stap --runtime=dyninst worth a try - that runs pure userspace, so no kernel header conflicts are in play

12:22 <sonal> fche, can we club system tap with python files?

16:03 <sanoj> fche, I was using the embedded C to include headers from of my project and examine its structures, Using --runtime=dyninst did help alleviate the collision with headers in kernel name space.

16:06 <sanoj> however ther are more collision probably between some of the internal variables in generated code and the ones from my project. For instance (list_head , THIS had defintion in both places). So Iam still getting redefinition errors. Is there anyway to isolate the name space or have the generated code compiled into different compilation units and link them