0001 i synthesize instructions events
0002 b synthesize branches events (branch misses for Arm SPE)
0003 c synthesize branches events (calls only)
0004 r synthesize branches events (returns only)
0005 x synthesize transactions events
0006 w synthesize ptwrite events
0007 p synthesize power events (incl. PSB events for Intel PT)
0008 o synthesize other events recorded due to the use
0009 of aux-output (refer to perf record)
0010 I synthesize interrupt or similar (asynchronous) events
0011 (e.g. Intel PT Event Trace)
0012 e synthesize error events
0013 d create a debug log
0014 f synthesize first level cache events
0015 m synthesize last level cache events
0016 M synthesize memory events
0017 t synthesize TLB events
0018 a synthesize remote access events
0019 g synthesize a call chain (use with i or x)
0020 G synthesize a call chain on existing event records
0021 l synthesize last branch entries (use with i or x)
0022 L synthesize last branch entries on existing event records
0023 s skip initial number of events
0024 q quicker (less detailed) decoding
0025 A approximate IPC
0026 Z prefer to ignore timestamps (so-called "timeless" decoding)
0027
0028 The default is all events i.e. the same as --itrace=ibxwpe,
0029 except for perf script where it is --itrace=ce
0030
0031 In addition, the period (default 100000, except for perf script where it is 1)
0032 for instructions events can be specified in units of:
0033
0034 i instructions
0035 t ticks
0036 ms milliseconds
0037 us microseconds
0038 ns nanoseconds (default)
0039
0040 Also the call chain size (default 16, max. 1024) for instructions or
0041 transactions events can be specified.
0042
0043 Also the number of last branch entries (default 64, max. 1024) for
0044 instructions or transactions events can be specified.
0045
0046 Similar to options g and l, size may also be specified for options G and L.
0047 On x86, note that G and L work poorly when data has been recorded with
0048 large PEBS. Refer linkperf:perf-intel-pt[1] man page for details.
0049
0050 It is also possible to skip events generated (instructions, branches, transactions,
0051 ptwrite, power) at the beginning. This is useful to ignore initialization code.
0052
0053 --itrace=i0nss1000000
0054
0055 skips the first million instructions.
0056
0057 The 'e' option may be followed by flags which affect what errors will or
0058 will not be reported. Each flag must be preceded by either '+' or '-'.
0059 The flags are:
0060 o overflow
0061 l trace data lost
0062
0063 If supported, the 'd' option may be followed by flags which affect what
0064 debug messages will or will not be logged. Each flag must be preceded
0065 by either '+' or '-'. The flags are:
0066 a all perf events
0067 o output to stdout
0068
0069 If supported, the 'q' option may be repeated to increase the effect.
