1 =========================================
2 user_events: User-based Event Tracing
3 =========================================
9 User based trace events allow user processes to create events and trace data
10 that can be viewed via existing tools, such as ftrace and perf.
11 To enable this feature, build your kernel with CONFIG_USER_EVENTS=y.
13 Programs can view status of the events via
14 /sys/kernel/debug/tracing/user_events_status and can both register and write
15 data out via /sys/kernel/debug/tracing/user_events_data.
17 Programs can also use /sys/kernel/debug/tracing/dynamic_events to register and
18 delete user based events via the u: prefix. The format of the command to
19 dynamic_events is the same as the ioctl with the u: prefix applied.
21 Typically programs will register a set of events that they wish to expose to
22 tools that can read trace_events (such as ftrace and perf). The registration
23 process gives back two ints to the program for each event. The first int is
24 the status bit. This describes which bit in little-endian format in the
25 /sys/kernel/debug/tracing/user_events_status file represents this event. The
26 second int is the write index which describes the data when a write() or
27 writev() is called on the /sys/kernel/debug/tracing/user_events_data file.
29 The structures referenced in this document are contained within the
30 /include/uapi/linux/user_events.h file in the source tree.
32 **NOTE:** *Both user_events_status and user_events_data are under the tracefs
33 filesystem and may be mounted at different paths than above.*
37 Registering within a user process is done via ioctl() out to the
38 /sys/kernel/debug/tracing/user_events_data file. The command to issue is
41 This command takes a packed struct user_reg as an argument::
50 The struct user_reg requires two inputs, the first is the size of the structure
51 to ensure forward and backward compatibility. The second is the command string
52 to issue for registering. Upon success two outputs are set, the status bit
55 User based events show up under tracefs like any other event under the
56 subsystem named "user_events". This means tools that wish to attach to the
57 events need to use /sys/kernel/debug/tracing/events/user_events/[name]/enable
58 or perf record -e user_events:[name] when attaching/recording.
60 **NOTE:** *The write_index returned is only valid for the FD that was used*
64 The command string format is as follows::
66 name[:FLAG1[,FLAG2...]] [Field1[;Field2...]]
78 Basic types are supported (__data_loc, u32, u64, int, char, char[20], etc).
79 User programs are encouraged to use clearly sized types like u32.
81 **NOTE:** *Long is not supported since size can vary between user and kernel.*
83 The size is only valid for types that start with a struct prefix.
84 This allows user programs to describe custom structs out to tools, if required.
86 For example, a struct in C that looks like this::
92 Would be represented by the following field::
94 struct mytype myname 20
98 Deleting an event from within a user process is done via ioctl() out to the
99 /sys/kernel/debug/tracing/user_events_data file. The command to issue is
102 This command only requires a single string specifying the event to delete by
103 its name. Delete will only succeed if there are no references left to the
104 event (in both user and kernel space). User programs should use a separate file
105 to request deletes than the one used for registration due to this.
109 When tools attach/record user based events the status of the event is updated
110 in realtime. This allows user programs to only incur the cost of the write() or
111 writev() calls when something is actively attached to the event.
113 User programs call mmap() on /sys/kernel/debug/tracing/user_events_status to
114 check the status for each event that is registered. The bit to check in the
115 file is given back after the register ioctl() via user_reg.status_bit. The bit
116 is always in little-endian format. Programs can check if the bit is set either
117 using a byte-wise index with a mask or a long-wise index with a little-endian
120 Currently the size of user_events_status is a single page, however, custom
121 kernel configurations can change this size to allow more user based events. In
122 all cases the size of the file is a multiple of a page size.
124 For example, if the register ioctl() gives back a status_bit of 3 you would
125 check byte 0 (3 / 8) of the returned mmap data and then AND the result with 8
126 (1 << (3 % 8)) to see if anything is attached to that event.
128 A byte-wise index check is performed as follows::
133 index = status_bit / 8;
134 mask = 1 << (status_bit % 8);
138 if (status_page[index] & mask) {
142 A long-wise index check is performed as follows::
144 #include <asm/bitsperlong.h>
147 #if __BITS_PER_LONG == 64
148 #define endian_swap(x) htole64(x)
150 #define endian_swap(x) htole32(x)
153 long index, mask, *status_page;
155 index = status_bit / __BITS_PER_LONG;
156 mask = 1L << (status_bit % __BITS_PER_LONG);
157 mask = endian_swap(mask);
161 if (status_page[index] & mask) {
165 Administrators can easily check the status of all registered events by reading
166 the user_events_status file directly via a terminal. The output is as follows::
168 Byte:Name [# Comments]
175 For example, on a system that has a single event the output looks like this::
183 If a user enables the user event via ftrace, the output would change to this::
185 1:test # Used by ftrace
191 **NOTE:** *A status bit of 0 will never be returned. This allows user programs
192 to have a bit that can be used on error cases.*
196 After registering an event the same fd that was used to register can be used
197 to write an entry for that event. The write_index returned must be at the start
198 of the data, then the remaining data is treated as the payload of the event.
200 For example, if write_index returned was 1 and I wanted to write out an int
201 payload of the event. Then the data would have to be 8 bytes (2 ints) in size,
202 with the first 4 bytes being equal to 1 and the last 4 bytes being equal to the
203 value I want as the payload.
205 In memory this would look like this::
210 User programs might have well known structs that they wish to use to emit out
211 as payloads. In those cases writev() can be used, with the first vector being
212 the index and the following vector(s) being the actual event payload.
214 For example, if I have a struct like this::
222 It's advised for user programs to do the following::
227 io[0].iov_base = &write_index;
228 io[0].iov_len = sizeof(write_index);
230 io[1].iov_len = sizeof(e);
232 writev(fd, (const struct iovec*)io, 2);
234 **NOTE:** *The write_index is not emitted out into the trace being recorded.*
238 See sample code in samples/user_events.