kvm_stat.txt 2.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124
  1. kvm_stat(1)
  2. ===========
  3. NAME
  4. ----
  5. kvm_stat - Report KVM kernel module event counters
  6. SYNOPSIS
  7. --------
  8. [verse]
  9. 'kvm_stat' [OPTION]...
  10. DESCRIPTION
  11. -----------
  12. kvm_stat prints counts of KVM kernel module trace events. These events signify
  13. state transitions such as guest mode entry and exit.
  14. This tool is useful for observing guest behavior from the host perspective.
  15. Often conclusions about performance or buggy behavior can be drawn from the
  16. output.
  17. While running in regular mode, use any of the keys listed in section
  18. 'Interactive Commands' below.
  19. Use batch and logging modes for scripting purposes.
  20. The set of KVM kernel module trace events may be specific to the kernel version
  21. or architecture. It is best to check the KVM kernel module source code for the
  22. meaning of events.
  23. INTERACTIVE COMMANDS
  24. --------------------
  25. [horizontal]
  26. *b*:: toggle events by guests (debugfs only, honors filters)
  27. *c*:: clear filter
  28. *f*:: filter by regular expression
  29. :: *Note*: Child events pull in their parents, and parents' stats summarize
  30. all child events, not just the filtered ones
  31. *g*:: filter by guest name/PID
  32. *h*:: display interactive commands reference
  33. *o*:: toggle sorting order (Total vs CurAvg/s)
  34. *p*:: filter by guest name/PID
  35. *q*:: quit
  36. *r*:: reset stats
  37. *s*:: set delay between refreshs
  38. *x*:: toggle reporting of stats for child trace events
  39. :: *Note*: The stats for the parents summarize the respective child trace
  40. events
  41. Press any other key to refresh statistics immediately.
  42. OPTIONS
  43. -------
  44. -1::
  45. --once::
  46. --batch::
  47. run in batch mode for one second
  48. -c::
  49. --csv::
  50. log in csv format. Requires option -l/--log or -L/--log-to-file.
  51. When used with option -L/--log-to-file, the header is only ever
  52. written to start of file to preserve the format.
  53. -d::
  54. --debugfs::
  55. retrieve statistics from debugfs
  56. -f<fields>::
  57. --fields=<fields>::
  58. fields to display (regex), "-f help" for a list of available events
  59. -g<guest>::
  60. --guest=<guest_name>::
  61. limit statistics to one virtual machine (guest name)
  62. -h::
  63. --help::
  64. show help message
  65. -i::
  66. --debugfs-include-past::
  67. include all available data on past events for debugfs
  68. -l::
  69. --log::
  70. run in logging mode (like vmstat)
  71. -L<file>::
  72. --log-to-file=<file>::
  73. like -l/--log, but logging to a file. Appends to existing files.
  74. -p<pid>::
  75. --pid=<pid>::
  76. limit statistics to one virtual machine (pid)
  77. -s::
  78. --set-delay::
  79. set delay between refreshs (value range: 0.1-25.5 secs)
  80. -t::
  81. --tracepoints::
  82. retrieve statistics from tracepoints
  83. -z::
  84. --skip-zero-records::
  85. omit records with all zeros in logging mode
  86. SEE ALSO
  87. --------
  88. 'perf'(1), 'trace-cmd'(1)
  89. AUTHOR
  90. ------
  91. Stefan Hajnoczi <[email protected]>