1UTRACE_CONTROL(9) utrace core API UTRACE_CONTROL(9)
2
3
4
6 utrace_control - control a thread being traced by a tracing engine
7
9 int utrace_control(struct task_struct * target,
10 struct utrace_engine * engine,
11 enum utrace_resume_action action);
12
14 target
15 thread to affect
16
17 engine
18 attached engine to affect
19
20 action
21
22 enum utrace_resume_action for thread to do
23
25 This is how a tracing engine asks a traced thread to do something. This
26 call is controlled by the action argument, which has the same meaning
27 as the enum utrace_resume_action value returned by event reporting
28 callbacks.
29
30 If target is already dead (target->exit_state nonzero), all actions
31 except UTRACE_DETACH fail with -ESRCH.
32
33 The following sections describe each option for the action argument.
34
36 After this, the engine data structure is no longer accessible, and the
37 thread might be reaped. The thread will start running again if it was
38 stopped and no longer has any attached engines that want it stopped.
39
40 If the report_reap callback may already have begun, this fails with
41 -ESRCH. If the report_death callback may already have begun, this fails
42 with -EALREADY.
43
44 If target is not already stopped, then a callback to this engine might
45 be in progress or about to start on another CPU. If so, then this
46 returns -EINPROGRESS; the detach happens as soon as the pending
47 callback is finished. To synchronize after an -EINPROGRESS return, see
48 utrace_barrier.
49
50 If target is properly stopped before utrace_control is called, then
51 after successful return it's guaranteed that no more callbacks to the
52 engine->ops vector will be made.
53
54 The only exception is SIGKILL (and exec or group-exit by another thread
55 in the group), which can cause asynchronous report_death and/or
56 report_reap callbacks even when UTRACE_STOP was used. (In that event,
57 this fails with -ESRCH or -EALREADY, see above.)
58
60 This asks that target stop running. This returns 0 only if target is
61 already stopped, either for tracing or for job control. Then target
62 will remain stopped until another utrace_control call is made on
63 engine; target can be woken only by SIGKILL (or equivalent, such as
64 exec or termination by another thread in the same thread group).
65
66 This returns -EINPROGRESS if target is not already stopped. Then the
67 effect is like UTRACE_REPORT. A report_quiesce or report_signal
68 callback will be made soon. Your callback can then return UTRACE_STOP
69 to keep target stopped.
70
71 This does not interrupt system calls in progress, including ones that
72 sleep for a long time. For that, use UTRACE_INTERRUPT. To interrupt
73 system calls and then keep target stopped, your report_signal callback
74 can return UTRACE_STOP.
75
77 Just let target continue running normally, reversing the effect of a
78 previous UTRACE_STOP. If another engine is keeping target stopped, then
79 it remains stopped until all engines let it resume. If target was not
80 stopped, this has no effect.
81
83 This is like UTRACE_RESUME, but also ensures that there will be a
84 report_quiesce or report_signal callback made soon. If target had been
85 stopped, then there will be a callback before it resumes running
86 normally. If another engine is keeping target stopped, then there might
87 be no callbacks until all engines let it resume.
88
89 Since this is meaningless unless report_quiesce callbacks will be made,
90 it returns -EINVAL if engine lacks UTRACE_EVENT(QUIESCE).
91
93 This is like UTRACE_REPORT, but ensures that target will make a
94 report_signal callback before it resumes or delivers signals. If target
95 was in a system call or about to enter one, work in progress will be
96 interrupted as if by SIGSTOP. If another engine is keeping target
97 stopped, then there might be no callbacks until all engines let it
98 resume.
99
100 This gives engine an opportunity to introduce a forced signal
101 disposition via its report_signal callback.
102
104 It's invalid to use this unless arch_has_single_step returned true.
105 This is like UTRACE_RESUME, but resumes for one user instruction only.
106
107 Note that passing UTRACE_SINGLESTEP or UTRACE_BLOCKSTEP to
108 utrace_control or returning it from an event callback alone does not
109 necessarily ensure that stepping will be enabled. If there are more
110 callbacks made to any engine before returning to user mode, then the
111 resume action is chosen only by the last set of callbacks. To be sure,
112 enable UTRACE_EVENT(QUIESCE) and look for the report_quiesce callback
113 with a zero event mask, or the report_signal callback with
114 UTRACE_SIGNAL_REPORT.
115
116 Since this is not robust unless report_quiesce callbacks will be made,
117 it returns -EINVAL if engine lacks UTRACE_EVENT(QUIESCE).
118
120 It's invalid to use this unless arch_has_block_step returned true. This
121 is like UTRACE_SINGLESTEP, but resumes for one whole basic block of
122 user instructions.
123
124 Since this is not robust unless report_quiesce callbacks will be made,
125 it returns -EINVAL if engine lacks UTRACE_EVENT(QUIESCE).
126
127 UTRACE_BLOCKSTEP devolves to UTRACE_SINGLESTEP when another tracing
128 engine is using UTRACE_SINGLESTEP at the same time.
129
130
131
132Kernel Hackers Manual 2.6. November 2011 UTRACE_CONTROL(9)