5 This document describes the tracing infrastructure in QEMU and how to use it
6 for debugging, profiling, and observing execution.
10 1. Build with the 'simple' trace backend:
12 ./configure --trace-backend=simple
15 2. Enable trace events you are interested in:
17 $EDITOR trace-events # remove "disable" from events you want
19 3. Run the virtual machine to produce a trace file:
21 qemu ... # your normal QEMU invocation
23 4. Pretty-print the binary trace file:
25 ./simpletrace.py trace-events trace-*
29 There is a set of static trace events declared in the "trace-events" source
30 file. Each trace event declaration names the event, its arguments, and the
31 format string which can be used for pretty-printing:
33 qemu_malloc(size_t size, void *ptr) "size %zu ptr %p"
34 qemu_free(void *ptr) "ptr %p"
36 The "trace-events" file is processed by the "tracetool" script during build to
37 generate code for the trace events. Trace events are invoked directly from
38 source code like this:
40 #include "trace.h" /* needed for trace event prototype */
42 void *qemu_malloc(size_t size)
45 if (!size && !allow_zero_malloc()) {
48 ptr = oom_check(malloc(size ? size : 1));
49 trace_qemu_malloc(size, ptr); /* <-- trace event */
53 === Declaring trace events ===
55 The "tracetool" script produces the trace.h header file which is included by
56 every source file that uses trace events. Since many source files include
57 trace.h, it uses a minimum of types and other header files included to keep the
58 namespace clean and compile times and dependencies down.
60 Trace events should use types as follows:
62 * Use stdint.h types for fixed-size types. Most offsets and guest memory
63 addresses are best represented with uint32_t or uint64_t. Use fixed-size
64 types over primitive types whose size may change depending on the host
65 (32-bit versus 64-bit) so trace events don't truncate values or break
68 * Use void * for pointers to structs or for arrays. The trace.h header
69 cannot include all user-defined struct declarations and it is therefore
70 necessary to use void * for pointers to structs.
72 Pointers (including char *) cannot be dereferenced easily (or at all) in
73 some trace backends. If pointers are used, ensure they are meaningful by
74 themselves and do not assume the data they point to will be traced. Do
75 not pass in string arguments.
77 * For everything else, use primitive scalar types (char, int, long) with the
78 appropriate signedness.
80 Format strings should reflect the types defined in the trace event. Take
81 special care to use PRId64 and PRIu64 for int64_t and uint64_t types,
82 respectively. This ensures portability between 32- and 64-bit platforms. Note
83 that format strings must begin and end with double quotes. When using
84 portability macros, ensure they are preceded and followed by double quotes:
87 === Hints for adding new trace events ===
89 1. Trace state changes in the code. Interesting points in the code usually
90 involve a state change like starting, stopping, allocating, freeing. State
91 changes are good trace events because they can be used to understand the
92 execution of the system.
94 2. Trace guest operations. Guest I/O accesses like reading device registers
95 are good trace events because they can be used to understand guest
98 3. Use correlator fields so the context of an individual line of trace output
99 can be understood. For example, trace the pointer returned by malloc and
100 used as an argument to free. This way mallocs and frees can be matched up.
101 Trace events with no context are not very useful.
103 4. Name trace events after their function. If there are multiple trace events
104 in one function, append a unique distinguisher at the end of the name.
106 5. Declare trace events with the "disable" keyword. Some trace events can
107 produce a lot of output and users are typically only interested in a subset
108 of trace events. Marking trace events disabled by default saves the user
109 from having to manually disable noisy trace events.
113 The "tracetool" script automates tedious trace event code generation and also
114 keeps the trace event declarations independent of the trace backend. The trace
115 events are not tightly coupled to a specific trace backend, such as LTTng or
116 SystemTap. Support for trace backends can be added by extending the "tracetool"
119 The trace backend is chosen at configure time and only one trace backend can
120 be built into the binary:
122 ./configure --trace-backend=simple
124 For a list of supported trace backends, try ./configure --help or see below.
126 The following subsections describe the supported trace backends.
130 The "nop" backend generates empty trace event functions so that the compiler
131 can optimize out trace events completely. This is the default and imposes no
136 The "stderr" backend sends trace events directly to standard error. This
137 effectively turns trace events into debug printfs.
139 This is the simplest backend and can be used together with existing code that
144 The "simple" backend supports common use cases and comes as part of the QEMU
145 source tree. It may not be as powerful as platform-specific or third-party
146 trace backends but it is portable. This is the recommended trace backend
147 unless you have specific needs for more advanced backends.
149 ==== Monitor commands ====
152 Display the contents of trace buffer. This command dumps the trace buffer
153 with simple formatting. For full pretty-printing, use the simpletrace.py
154 script on a binary trace file.
156 The trace buffer is written into until full. The full trace buffer is
157 flushed and emptied. This means the 'info trace' will display few or no
158 entries if the buffer has just been flushed.
161 View available trace events and their state. State 1 means enabled, state 0
164 * trace-event NAME on|off
165 Enable/disable a given trace event.
167 * trace-file on|off|flush|set <path>
168 Enable/disable/flush the trace file or set the trace file name.
170 ==== Enabling/disabling trace events programmatically ====
172 The st_change_trace_event_state() function can be used to enable or disable trace
173 events at runtime inside QEMU:
177 st_change_trace_event_state("virtio_irq", true); /* enable */
179 st_change_trace_event_state("virtio_irq", false); /* disable */
181 ==== Analyzing trace files ====
183 The "simple" backend produces binary trace files that can be formatted with the
184 simpletrace.py script. The script takes the "trace-events" file and the binary
187 ./simpletrace.py trace-events trace-12345
189 You must ensure that the same "trace-events" file was used to build QEMU,
190 otherwise trace event declarations may have changed and output will not be
193 === LTTng Userspace Tracer ===
195 The "ust" backend uses the LTTng Userspace Tracer library. There are no
196 monitor commands built into QEMU, instead UST utilities should be used to list,
197 enable/disable, and dump traces.
201 The "dtrace" backend uses DTrace sdt probes but has only been tested with
202 SystemTap. When SystemTap support is detected a .stp file with wrapper probes
203 is generated to make use in scripts more convenient. This step can also be
204 performed manually after a build in order to change the binary name in the .stp
207 scripts/tracetool --dtrace --stap \
208 --binary path/to/qemu-binary \
209 --target-type system \
210 --target-arch x86_64 \
211 <trace-events >qemu.stp