[Bug tapsets/15219] syscall.exp failures on RHEL5, RHEL6, and rawhide

dsmith at redhat dot com sourceware-bugzilla@sourceware.org
Mon Aug 26 20:44:00 GMT 2013


https://sourceware.org/bugzilla/show_bug.cgi?id=15219

David Smith <dsmith at redhat dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Attachment #7067|0                           |1
        is obsolete|                            |

--- Comment #9 from David Smith <dsmith at redhat dot com> ---
Created attachment 7166
  --> https://sourceware.org/bugzilla/attachment.cgi?id=7166&action=edit
anti-nested syscall patch 2

Here's yet another approach to fixing this one. This one attempts to solve the
nested-syscall problem by defining 2 macros: '@__syscall_enter' (called at each
syscall entry probe) and '@__syscall_exit' (called in each syscall return
probe). It just uses a global array to figure out if we're already in a system
call in this thread. If so, the probe is skipped.

Here are some notes:

- I just modified the syscall tapset enough to get the 32-bit clock syscall
passing, I didn't modify every place that needs changes.

-There are a couple of problems with using a macro library file, listed there.
The best way of fixing these problems might be to combine syscalls.stp and
syscalls2.stp so that the macros could be defined in that file.

- Similar changes would need to be made for the nd_syscall tapset. It would
probably need its own global array.

-- 
You are receiving this mail because:
You are the assignee for the bug.



More information about the Systemtap mailing list