itrace.txt 2.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263
  1. i synthesize instructions events
  2. b synthesize branches events (branch misses for Arm SPE)
  3. c synthesize branches events (calls only)
  4. r synthesize branches events (returns only)
  5. x synthesize transactions events
  6. w synthesize ptwrite events
  7. p synthesize power events
  8. o synthesize other events recorded due to the use
  9. of aux-output (refer to perf record)
  10. e synthesize error events
  11. d create a debug log
  12. f synthesize first level cache events
  13. m synthesize last level cache events
  14. t synthesize TLB events
  15. a synthesize remote access events
  16. g synthesize a call chain (use with i or x)
  17. G synthesize a call chain on existing event records
  18. l synthesize last branch entries (use with i or x)
  19. L synthesize last branch entries on existing event records
  20. s skip initial number of events
  21. q quicker (less detailed) decoding
  22. The default is all events i.e. the same as --itrace=ibxwpe,
  23. except for perf script where it is --itrace=ce
  24. In addition, the period (default 100000, except for perf script where it is 1)
  25. for instructions events can be specified in units of:
  26. i instructions
  27. t ticks
  28. ms milliseconds
  29. us microseconds
  30. ns nanoseconds (default)
  31. Also the call chain size (default 16, max. 1024) for instructions or
  32. transactions events can be specified.
  33. Also the number of last branch entries (default 64, max. 1024) for
  34. instructions or transactions events can be specified.
  35. Similar to options g and l, size may also be specified for options G and L.
  36. On x86, note that G and L work poorly when data has been recorded with
  37. large PEBS. Refer linkperf:perf-intel-pt[1] man page for details.
  38. It is also possible to skip events generated (instructions, branches, transactions,
  39. ptwrite, power) at the beginning. This is useful to ignore initialization code.
  40. --itrace=i0nss1000000
  41. skips the first million instructions.
  42. The 'e' option may be followed by flags which affect what errors will or
  43. will not be reported. Each flag must be preceded by either '+' or '-'.
  44. The flags are:
  45. o overflow
  46. l trace data lost
  47. If supported, the 'd' option may be followed by flags which affect what
  48. debug messages will or will not be logged. Each flag must be preceded
  49. by either '+' or '-'. The flags are:
  50. a all perf events
  51. If supported, the 'q' option may be repeated to increase the effect.