1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
|
INITIALIZATION
init_module() is defined in runtime.h. It returns _stp_transport_init().
int _stp_transport_init(void)
{
/* allocates buffers, creates /proc/systemtap/stuff */
if (_stp_register_procfs() < 0)
return -1;
/* starts up the work queue */
schedule_delayed_work(&stp_exit, STP_WORK_TIMER);
/* always succeeds */
return 0;
}
*** Module is now loaded ***
In staprun, librelay.c, init_stp() forks any commands specified with
the "-c" parameter. Then it sends STP_TRANSPORT_INFO message with
the buffer size (if it was specified on the staprun command line) and
the pid from the fork (or 0 otherwise).
In procfs.c, _stp_proc_write_cmd() receives the message. It calls
_stp_transport_open().
_stp_transport_open() initializes relayfs if necessary. It returns a
STP_TRANSPORT_INFO message with the transport_mode
(relayfs or procfs-only), and buffer information.
staprun receives the STP_TRANSPORT_INFO message in stp_main_loop(),
It initializes relayfs (if necessary) and replies with an
STP_START message.
***Communications are now fully established.***
_stp_proc_write_cmd() receives the STP_START. It calls _stp_handle_start().
_stp_handle_start() calls probe_start(), which the translator generates.
It allocates memory and registers probes. When that returns,
an atomic variable is set indicating probe_start finished. An STP_START
message is returned to staprun with the return value from probe_start()
staprun receives the STP_START message in stp_main_loop(). If the return value
is < 0 (indicating probe_start failed to register probes, etc) staprun
cleans up and exits.
*** Probes running. Everything Up. ***
SHUTDOWN AND UNLOADING
There are 3 ways to initiate shutdown.
1. staprun can initiate it. This happens when staprun receives a ^C or
a child specified with "-c" exits.
2. Something can call _stp_exit().
3. rmmod. This can happen when staprun dies for an unexpected reason
and the module is still loaded. So the user does an rmmod. Also, when
STP_START fails, staprun simply rmmods the module.
For #1, staprun sends an STP_EXIT message.
_stp_proc_write_cmd() receives the STP_EXIT. It sets _stp_exit_flag
and returns.
The next time _stp_work_queue() runs, it will notice that
_stp_exit_flag is set. It checks to see if probe_start() finished
because we don't want to start exiting until then. It will cancel itself,
and call _stp_cleanup_and_exit(0). That will call probe_exit() which
unregisters the probes. Then it sends an STP_EXIT(0) message to
staprun. staprun gets the message and calls cleanup_and_exit(0)
For #2, _stp_exit() sets _stp_exit_flag.
For #3, cleanup_module() calls _stp_transport_close().
_stp_transport_close() cancels the work_queue, then calls
_stp_cleanup_and_exit(1). The "1" is passed as an argument to
the STP_EXIT message. Staprun calls cleanup_and_exit(1), which cleans
up everything on its side, but does not do an "rmmod". Finally,
_stp_unregister_procfs() to remove the procfs stuff.
|
