|
The ps_pstop() function stops the target process identified by ph, while the ps_pcontinue() function allows it to resume.
The libthread_db() function uses ps_pstop() to freeze the target process while it is under inspection. Within the scope of any single call from outside libthread_db to a libthread_db routine, libthread_db
will call ps_pstop(), at most once. If it does, it will call ps_pcontinue() within the scope of the same routine.
The controlling process may already have stopped the target process when it calls libthread_db. In that case, it is not obligated to resume the target process when libthread_db calls ps_pcontinue(). In other words, ps_pstop()
is mandatory, while ps_pcontinue() is advisory. After ps_pstop(), the target process must be stopped; after ps_pcontinue(), the target process may be running.
The ps_lstop() and ps_lcontinue() functions stop and resume a single lightweight process (LWP) within the target process ph.
The ps_lrolltoaddr() function is used to roll an LWP forward out of a critical section when the process is stopped. It is also used to run the libthread_db agent thread on behalf of libthread. The ps_lrolltoaddr() function is always called with the target process stopped, that is, there has been a preceding call to ps_pstop(). The specified LWP must be continued at the address go_addr, or at its current address
if go_addr is NULL. It should then be stopped when its execution reaches stop_addr. This routine does not return until the LWP has stopped at stop_addr.
The ps_kill() function directs the signal signum to the target process for which the handle is ph. It has the same semantics as kill(2).
|
