DRAFT 2.1 release notes

DRAFT 2.1 release notes

[Serguei Makarov]

These are the current in-progress draft SystemTap 2.1 release notes.
Feedback would be appreciated.

Looking these over, it seems that more NEWS items are needed for
various work, in particular for progress on the dyninst backend.

====

The systemtap team announces release 2.1!

perf event probe improvements, dyninst backend improvements,
error:: and warning:: man pages, experimental regex support,
... TODO ...


= Where to get it

  http://sourceware.org/systemtap/ - our project page
  http://sourceware.org/systemtap/ftp/releases/systemtap-2.1.tar.gz
  http://koji.fedoraproject.org/koji/packageinfo?packageID=615
  git tag release-2.1 (commit ...)

  There have been over ... commits since the last release.
  There have been over ... bugs/features fixed since the last release.


= How to build it

  See the README and NEWS files at
  http://sourceware.org/git/?p=systemtap.git;a=tree
  Further information at http://sourceware.org/systemtap/wiki/


= Systemtap frontend (stap) changes

- EMACS and VIM editor modes for systemtap source files are included / updated.

- The translator now eliminates duplicate tapset files between its
  preferred directory (as configured during the build with --prefix=/
  or specified with the -I /path option), and files it may find under
  $XDG_DATA_DIRS.  This should eliminate a class of conflicts between
  parallel system- and hand-built systemtap installations.

- The translator accepts a --suppress-time-limits option, which defeats
  time-related constraints, to allows probe handlers to run for indefinite
  periods.  It requires the guru mode (-g) flag to work.  Add the earlier
  --suppress-handler-errors flag for a gung-ho "just-keep-going" attitude.

- Some error messages and warnings now refer to additional information
  that is found in man pages.  These are generally named
  error::FOO or warning::BAR (in the 7stap man page section)
  and may be read via
     % man error::FOO
     % man warning::BAR


= Systemtap script language changes

- Perf event probes may now be read on demand. The counter probe is
  defined using the counter-name part: 
  probe perf.type(0).config(0).counter("NAME").  The counter is
  read in a user space probe using @perf("NAME"), e.g.
   process("PROCESS").statement("func@file") {stat <<< @perf("NAME")} 

- Perf event probes may now be bound to a specific task using the
  process-name part:  probe perf.type(0).config(0).process("NAME") { }
  If the probed process name is not specified, then it is inferred
  from the -c CMD argument.  

- The foreach looping construct can now sort aggregate arrays by the user's
  choice of aggregating function.  Previously, @count was implied.  e.g.:
     foreach ([x,y] in array @sum +) { println(@sum(array[x,y])) } 

- Proof of concept support for regular expression matching has been added:
     if ("aqqqqqb" =~ "q*b") { ... }
     if ("abc" !~ "q*b") { ... }

  The eventual aim is to support roughly the same functionality as
  the POSIX Extended Regular Expressions implemented by glibc.
  Currently missing features include extraction of the matched string
  and subexpressions, and named character classes ([:alpha:], [:digit:], &c).


= Systemtap runtime changes

- The dyninst backend's runtime has been improved to allow much more
  concurrency when probing multithreaded processes.


= Systemtap tapset changes

- New tapsets:
  choose_defined.stpm  add @choose_defined() macro

- Added/changed tapsets for dyninst:
  ctime.stp            ctime() returns "<invalid time>" when an error occurs
  context.stp          add support for print_regs()

- Changed tapsets:
  nfs_proc.stp         account for separate nfsv2/3/4 kernel modules
  context.stp          add ppfunc() to parse the function name from pp()

...


= Systemtap sample scripts

- Tracing examples were updated to use ppfunc() instead of probefunc().

- New samples:
  perf.stp          use perf.counter to access performance counters
  threadstacks.stp  override default new-pthread stack sizes
  ltrace.stp        uprobes-based ltrace demo
  block.stp         stapgames Breakout demo using input/output devices
  eater.stp         stapgames Snake demo using input/output devices
  lifegame.stp      stapgames Game of Life demo using input/output devices
  pingpong.stp      stapgames Pong demo using input/output devices

- Changed samples:
  deviceseeks.stp   switch to using ioblock_trace.request probe point

... TODO regex example script ...


= Examples of tested kernel versions

... TODO ...


= Known issues with this release

... TODO ...

... TODO particularly the regex PRs ...


= Contributors for this release

Alexander Lochmann, Daniel Tschan*, David Smith, Frank Ch. Eigler,
Josh Stone, Jóhann B. Guðmundsson*, Lukas Berk, Mark Wielaard,
Masami Hiramatsu, Nathan Scott, Rüdiger Sonderfeld*, Serguei Makarov,
Stan Cox, Tom Tromey, Tomoki Sekiyama*, William Cohen

Special thanks to new contributors, marked with '*' above.


= Bugs fixed for this release <http://sourceware.org/PR#####>

 1887  write emacs/vim editor mode for script language
 6697  need more access_ok() checks in runtime/tapsets
11576  per-process perf events
12022  Allow foreach to order iterations for aggregates on @min @max and @avg operators
12231  reading perfctr events
12443  XDG_DATA_DIRS conflict between preinstalled and hand-built stap
12642  utrace: taskfinder misses events when main thread does not go through at least one quiesce
13022  probe module("*").function("*") does not work on arm
14178  Add a pure-userspace runtime mode
14245  Support debugfs mounted 0700
14390  separate process.**.return tests and mark UNSUPPORTED 
14555  tolerate /proc/sys/kernel/kptr_restrict=2
14571  Make stapdyn multithreaded
14659  task_finder2 breaks ptrace apps
14701  stapdyn needs timer probe support
14707  Support stapdyn -x PID
14709  Support process.function.return in stapdyn
14712  VM_EXECUTABLE went to heaven
14714  stap flag for bigger --suppress-handler-errors hammer
14731  regression in buildok/thirty.stp
14735  stapdyn needs timing ('-t') support
14742  more complete dwz / DW_TAG_partial_unit support
14774  process.statement("*@file:*") probes don't match all functions
14781  kmem_cache conflict leads to panic & badness
14794  autoconf auto-enable dyninst if possible
14795  add stapgames
14804  Kernel panic when running hello world stap script
14820  hang when running semok/nodwf01.stp
14842  Old "-kelf" and "--ignore-dwarf" command line options should be removed
14883  let probefunc() not fail - or fail more informatively, on non-uprobes kernels
14900  collision between tapset locals and user globals: contexts
14929  staperrors man page
14949  cannot use systemtap on linux 3.7 intel x86_64
14957  foreach_aggrsort.exp test hangs on i686 rhel5 system
14997  DW_OP_rot mistranslation
15031  stapdyn needs a _stp_print_binary() runtime function
15059  session.print_error called frequently from listing_mode
15074  pthread_stacks.exp problems

... TODO additional closed PRs ...

Re: DRAFT 2.1 release notes

[Serguei Makarov]

Newer draft of the release notes. Would particularly appreciate
someone looking over the 'Known Issues' section to see if something
should be added / taken away.

====

The systemtap team announces release 2.1!

perf event probe improvements, dyninst backend improvements,
error:: and warning:: man pages, bundled emacs/vim editor modes,
customizable aggregate array sorting, optional time limit suppression,
experimental regex support


= Where to get it

  http://sourceware.org/systemtap/ - our project page
  http://sourceware.org/systemtap/ftp/releases/systemtap-2.1.tar.gz
  http://koji.fedoraproject.org/koji/packageinfo?packageID=615
  git tag release-2.1 (commit ...)

  There have been over ... commits since the last release.
  There have been over ... bugs/features fixed since the last release.


= How to build it

  See the README and NEWS files at
  http://sourceware.org/git/?p=systemtap.git;a=tree
  Further information at http://sourceware.org/systemtap/wiki/


= Systemtap frontend (stap) changes

- EMACS and VIM editor modes for systemtap source files are included / updated.

- The translator now eliminates duplicate tapset files between its
  preferred directory (as configured during the build with --prefix=/
  or specified with the -I /path option), and files it may find under
  $XDG_DATA_DIRS.  This should eliminate a class of conflicts between
  parallel system- and hand-built systemtap installations.

- The translator accepts a --suppress-time-limits option, which defeats
  time-related constraints, to allows probe handlers to run for indefinite
  periods.  It requires the guru mode (-g) flag to work.  Add the earlier
  --suppress-handler-errors flag for a gung-ho "just-keep-going" attitude.

- Some error messages and warnings now refer to additional information
  that is found in man pages.  These are generally named
  error::FOO or warning::BAR (in the 7stap man page section)
  and may be read via
     % man error::FOO
     % man warning::BAR


= Systemtap script language changes

- Perf event probes may now be read on demand. The counter probe is
  defined using the counter-name part: 
  probe perf.type(0).config(0).counter("NAME").  The counter is
  read in a user space probe using @perf("NAME"), e.g.
   process("PROCESS").statement("func@file") {stat <<< @perf("NAME")} 

- Perf event probes may now be bound to a specific task using the
  process-name part:  probe perf.type(0).config(0).process("NAME") { }
  If the probed process name is not specified, then it is inferred
  from the -c CMD argument.  

- The foreach looping construct can now sort aggregate arrays by the user's
  choice of aggregating function.  Previously, @count was implied.  e.g.:
     foreach ([x,y] in array @sum +) { println(@sum(array[x,y])) } 

- Proof of concept support for regular expression matching has been added:
     if ("aqqqqqb" =~ "q*b") { ... }
     if ("abc" !~ "q*b") { ... }

  The eventual aim is to support roughly the same functionality as
  the POSIX Extended Regular Expressions implemented by glibc.
  Currently missing features include extraction of the matched string
  and subexpressions, and named character classes ([:alpha:], [:digit:], &c).

  Special thanks go to the re2c project, whose public domain code this
  functionality has been based on. For more info on re2c, see:
     http://sourceforge.net/projects/re2c/


= Systemtap runtime changes

- The dyninst backend's runtime has been improved to allow much more
  concurrency when probing multithreaded processes.


= Systemtap tapset changes

- New tapsets:
  choose_defined.stpm  add @choose_defined() macro

- Added/changed tapsets for dyninst:
  ctime.stp            ctime() returns "<invalid time>" when an error occurs
  context.stp          add support for print_regs()

- Changed tapsets:
  nfs_proc.stp         account for separate nfsv2/3/4 kernel modules
  context.stp          add ppfunc() to parse the function name from pp()
  netfilter.stp        fix error in decoding of ar_tha and ar_tpa fields
    
- The folowing tapset variables are deprecated in release 2.1 and will
  be removed in release 2.2:
  - The 'send2queue' variable in the 'signal.send' probe.
  - The 'oldset_addr' and 'regs' variables in the 'signal.handle' probe.

- The following tapset probes are deprecated in release 2.1 and will
  be removed in release 2.2:
  - signal.send.return
  - signal.handle.return


= Systemtap sample scripts

- Tracing examples were updated to use ppfunc() instead of probefunc().

- New samples:
  perf.stp              use perf.counter to access performance counters
  regex.stp             report opened files matching a regex <TODOXXX>
  threadstacks.stp      override default new-pthread stack sizes
  mutex_contention.stp  pthread mutex contention analysis
  ltrace.stp            uprobes-based ltrace demo
  block.stp             stapgames Breakout demo using input/output devices
  eater.stp             stapgames Snake demo using input/output devices
  lifegame.stp          stapgames Game of Life demo using input/output devices
  pingpong.stp          stapgames Pong demo using input/output devices

- Changed samples:
  deviceseeks.stp   switch to using ioblock_trace.request probe point


= Examples of tested kernel versions

3.8.0 (x86_64)
3.7.0 (x86_64, i686)
3.6.2 (x86_64)
3.4.0 (armv7l)
2.6.32 (x86_64, i686, ppc64, s390x)
2.6.18 (x86_64, i686, ia64)
2.6.9 (i686, ppc64)


= Known issues with this release

- The regular expression support is still experimental and incomplete:
  + Character class support is somewhat broken, and named character classes are missing. (PR15064)
  + Support for extracting matched strings and substrings is missing. (PR15065)

- Some kernel crashes continue to be reported when a script probes
  broad kernel function wildcards.  (PR2725)

- Some systems mount debugfs with mode 0700, which prevents non-root
  users from connecting to stap modules.  To work around this issue,
  try "chmod 0755 /sys/kernel/debug" as root.  (PR14245)

- The dyninst backend is still a work in progress. See dyninst/README
  and the systemtap/dyninst Bugzilla component
  (http://tinyurl.com/stapdyn-PR-list) if you want all the gory
  details about the state of the feature.


= Contributors for this release

Alexander Lochmann, Daniel Tschan*, David Smith, Frank Ch. Eigler,
Josh Stone, Jóhann B. Guðmundsson*, Lukas Berk, Mark Wielaard,
Masami Hiramatsu, Nathan Scott, Rüdiger Sonderfeld*, Serguei Makarov,
Stan Cox, Tom Tromey, Tomoki Sekiyama*, William Cohen

Special thanks to new contributors, marked with '*' above.


= Bugs fixed for this release <http://sourceware.org/PR#####>

 1887  write emacs/vim editor mode for script language
 6697  need more access_ok() checks in runtime/tapsets
11576  per-process perf events
11955  signal tapsets don't match 
12022  Allow foreach to order iterations for aggregates on @min @max and @avg operators
12231  reading perfctr events
12443  XDG_DATA_DIRS conflict between preinstalled and hand-built stap
12642  utrace: taskfinder misses events when main thread does not go through at least one quiesce
13022  probe module("*").function("*") does not work on arm
14178  Add a pure-userspace runtime mode
14245  Support debugfs mounted 0700
14390  separate process.**.return tests and mark UNSUPPORTED 
14555  tolerate /proc/sys/kernel/kptr_restrict=2
14571  Make stapdyn multithreaded
14659  task_finder2 breaks ptrace apps
14701  stapdyn needs timer probe support
14707  Support stapdyn -x PID
14709  Support process.function.return in stapdyn
14712  VM_EXECUTABLE went to heaven
14714  stap flag for bigger --suppress-handler-errors hammer
14731  regression in buildok/thirty.stp
14735  stapdyn needs timing ('-t') support
14742  more complete dwz / DW_TAG_partial_unit support
14774  process.statement("*@file:*") probes don't match all functions
14781  kmem_cache conflict leads to panic & badness
14794  autoconf auto-enable dyninst if possible
14795  add stapgames
14804  Kernel panic when running hello world stap script
14820  hang when running semok/nodwf01.stp
14842  Old "-kelf" and "--ignore-dwarf" command line options should be removed
14883  let probefunc() not fail - or fail more informatively, on non-uprobes kernels
14900  collision between tapset locals and user globals: contexts
14929  staperrors man page
14949  cannot use systemtap on linux 3.7 intel x86_64
14957  foreach_aggrsort.exp test hangs on i686 rhel5 system
14997  DW_OP_rot mistranslation
15031  stapdyn needs a _stp_print_binary() runtime function
15059  session.print_error called frequently from listing_mode
15074  pthread_stacks.exp problems
15080  The kernel function 'signal_wake_up' is gone
15087  netfilter.arp tapset probes duplicate source/target addresses 

Re: DRAFT 2.1 release notes

[Timo Juhani Lindfors]

Serguei Makarov <smakarov@redhat.com> writes:
> Newer draft of the release notes. Would particularly appreciate
> someone looking over the 'Known Issues' section to see if something
> should be added / taken away.

do you have some planned release date for 2.1? I'm terribly busy but I
did manage to find time to test it against the debian linux 3.7 packages
(currently in debian experimental). The testsuite output is at

http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.sum

but I didn't have much time to compare it against earlier
testsuites. Note that I did the tests under xen.

Re: DRAFT 2.1 release notes

[Josh Stone]

On 02/06/2013 01:42 PM, Timo Juhani Lindfors wrote:
> Serguei Makarov <smakarov@redhat.com> writes:
>> Newer draft of the release notes. Would particularly appreciate
>> someone looking over the 'Known Issues' section to see if something
>> should be added / taken away.
> 
> do you have some planned release date for 2.1? I'm terribly busy but I
> did manage to find time to test it against the debian linux 3.7 packages
> (currently in debian experimental). The testsuite output is at
> 
> http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.sum
> 
> but I didn't have much time to compare it against earlier
> testsuites. Note that I did the tests under xen.

We're hoping to release very soon, perhaps in the next few days.

Your results look very bad - even the first add.exp fails!  Can you also
share the full systemtap.log to see what happened there?

Re: DRAFT 2.1 release notes

[Timo Juhani Lindfors]

Josh Stone <jistone@redhat.com> writes:
> Your results look very bad - even the first add.exp fails!  Can you also
> share the full systemtap.log to see what happened there?

Yeah sorry about that, I forgot to add setuid bit for staprun and
thought make installcheck would use sudo automatically bit turns out it
only uses it for some tests and not all. I reran the testsuite after
adding setuid, the results are here:

http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.sum
http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.log


            === systemtap Summary ===

# of expected passes              927
# of unexpected failures          598
# of unexpected successes         3
# of expected failures            275
# of known failures               7
# of untested testcases           267
# of unsupported tests            2

Re: DRAFT 2.1 release notes

[Frank Ch. Eigler]

Timo Juhani Lindfors <timo.lindfors@iki.fi> writes:

> [...]
> Yeah sorry about that, I forgot to add setuid bit for staprun
> [...]
> http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.sum
> http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.log
> [...]
> # of expected passes              927
> # of unexpected failures          598

Those are still far worse then they should be; the .log file shows
similar permission problems.  Try running, ** as root **
        # make installcheck

- FChE

Re: DRAFT 2.1 release notes

[Josh Stone]

On 02/07/2013 11:47 PM, Timo Juhani Lindfors wrote:
> Josh Stone <jistone@redhat.com> writes:
>> Your results look very bad - even the first add.exp fails!  Can you also
>> share the full systemtap.log to see what happened there?
> 
> Yeah sorry about that, I forgot to add setuid bit for staprun and
> thought make installcheck would use sudo automatically bit turns out it
> only uses it for some tests and not all. I reran the testsuite after
> adding setuid, the results are here:

You still have this bit in the log:

> ERROR: The effective user ID of staprun must be set to the root user.
>   Check permissions on staprun and ensure it is a setuid root program.

Just setting the setuid bit makes the process run as the owner of the
file, so it must also be owned by root to be able to load modules.

You can also run the entire installcheck as root, like Frank suggests,
but if you have staprun now as setuid non-root, it will effectively be
defeated by *dropping* permission.

Re: DRAFT 2.1 release notes

[Timo Juhani Lindfors]

Josh Stone <jistone@redhat.com> writes:
> You can also run the entire installcheck as root, like Frank suggests,
> but if you have staprun now as setuid non-root, it will effectively be
> defeated by *dropping* permission.

Yeah sorry for wasting your time. After sudo make installcheck the
results look bit better:

             === systemtap Summary ===

# of expected passes               1627
# of unexpected failures           86
# of unexpected successes          3
# of expected failures             278
# of known failures                7
# of untested testcases            276
# of unsupported tests             2


http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.sum
http://lindi.iki.fi/lindi/systemtap/testsuite_43e1259c0fea19e3348d2adbfe5a65e9abde30ca.log