6 perf-top - System profiling tool.
11 'perf top' [-e <EVENT> | --event=EVENT] [<options>]
15 This command generates and displays a performance counter profile in real time.
22 System-wide collection. (default)
26 Event period to sample.
30 Monitor only on the list of CPUs provided. Multiple CPUs can be provided as a
31 comma-separated list with no space: 0,1. Ranges of CPUs are specified with -: 0-2.
32 Default is to monitor all CPUS.
36 Number of seconds to delay between refreshes.
40 Select the PMU event. Selection can be a symbolic event name
41 (use 'perf list' to list all events) or a raw PMU
42 event (eventsel+umask) in the form of rNNN where NNN is a
43 hexadecimal event descriptor.
47 Display this many functions.
50 --count-filter=<count>::
51 Only display functions with more events than this.
55 Put the counters into a counter group.
59 Profile at this frequency.
63 Child tasks do not inherit counters.
67 Path to vmlinux. Required for annotation functionality.
70 --mmap-pages=<pages>::
71 Number of mmapped data pages.
75 Profile events on existing Process ID (comma separated list).
79 Profile events on existing thread ID (comma separated list).
83 Record events in threads owned by uid. Name or number.
86 --realtime=<priority>::
87 Collect data with this RT SCHED_FIFO priority.
90 --sym-annotate=<symbol>::
94 --hide_kernel_symbols::
103 Dump the symbol table used for profiling.
107 Be more verbose (show counter open errors, etc).
111 Zero history across display updates.
115 Sort by key(s): pid, comm, dso, symbol, parent, srcline, weight, local_weight.
119 Show a column with the number of samples.
121 --show-total-period::
122 Show a column with the sum of periods.
125 Only consider symbols in these dsos.
128 Only consider symbols in these comms.
131 Only consider these symbols.
134 --disassembler-style=:: Set disassembler style for objdump.
137 Interleave source code with assembly code. Enabled by default,
138 disable with --no-source.
141 Show raw instruction encoding of assembly instructions.
143 -G [type,min,order]::
145 Display call chains using type, min percent threshold and order.
147 - flat: single column, linear exposure of call chains.
148 - graph: use a graph tree, displaying absolute overhead rates.
149 - fractal: like graph, but displays relative rates. Each branch of
150 the tree is considered as a new profiled object.
153 - callee: callee based call graph.
154 - caller: inverted caller based call graph.
156 Default: fractal,0.5,callee.
158 --ignore-callees=<regex>::
159 Ignore callees of the function(s) matching the given regex.
160 This has the effect of collecting the callers of each such
161 function into one place in the call-graph tree.
164 Do not show entries which have an overhead under that percent.
167 INTERACTIVE PROMPTING KEYS
168 --------------------------
171 Display refresh delay.
174 Number of entries to display.
177 Event to display when multiple counters are active.
180 Profile display filter (>= hit count).
183 Annotation display filter (>= % of total).
189 Stop annotation, return to full profile display.
192 Toggle event count zeroing across display updates.
197 Pressing any unmapped key displays a menu, and prompts for input.
202 linkperf:perf-stat[1], linkperf:perf-list[1]