This is the mail archive of the systemtap@sourceware.org mailing list for the systemtap project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

release-0.7-rc1 announcement suggestions


Hi,

Frank tagged the git trunk (specifically  94687c8... fix
variable-name-collision thinko in symbol processing logic) as
release-0.7-rc1. So hopefully by next week we will have the final 0.7
unless major issues are still found/unresolved.

If people could test in particular the issues they had with the
following bugs that are hopefully fixed now:

- bz6429 unwinding/symbol stuff broken on ppc*
         fix committed by fche should help with, but unwinder base not
         there now reliable symbol address, doesn't rely on kernel
- bz6646 SystemTap scripts fail with an Oops on x86
         might be fixed by same fix as above
         which is a duplicate of bz6736
         switch to compiled-in symbol and unwind data

Unfortunately bz6707 oops crashes with 2.6.25 - onoffprobe still fails
for me with the 0.7-rc1 and 2.6.25.9-76.fc9.i686. It seems I am the only
person being able to reproduce this crash. I will try to investigate
again.

If others have other bugs that might be considered showstoppers, please
do speak up.

Attached is the updated 0.7 announcement suggestions based on the new
work done since June and the feedback people gave on the last version.
Please yell and scream if anything is missing or horribly wrong.

Frank, I don't have a writeup of the pr6429-comp-unwindsyms branch merge
that is now part of trunk, could you write up something if you think it
should be metioned?

I only included test results for one system (f8, x86_64, 2.6.24.7). For
some reason a make installcheck now takes 4 hours for me. This seems to
mainly be the syscalls.exp. More results welcome. I'll try to see why
syscalls.exp is taking so long now.

Thanks,

Mark
= Systemtap frontend (stap) changes

- A probe listing mode is available.
  % stap -l vm.*
  vm.brk
  vm.mmap
  vm.munmap
  vm.oom_kill
  vm.pagefault
  vm.write_shared

- Support for limited kernel probing without debuginfo via symbol
  tables -- on i386, x86_64, and powerpc so far.  (See --kelf, --kmap,
  --ignore-vmlinux, and --ignore-dwarf documentation)
  http://sourceware.org/systemtap/wiki/MakeDoWithoutDebugInfo

- The environment variable SYSTEMTAP_DEBUGINFO_PATH now refers to a
  base directory where kernel and module debuginfo when not installed
  in the default location.

For a full overview see the stap(1) manual page.


= Systemtap script language changes

- Globals now accept ; terminators
  global odds, evens;
  global little[10], big[5];

- The vector of script command line arguments is available in a
  tapset-provided global array argv[].  It is indexed 1 ... argc,
  another global.  This can substitute for of preprocessor
  directives @NNN that fail at parse time if there are not
  enough arguments.
      printf("argv: %s %s %s", argv[1], argv[2], argv[3])

For a full overview see the SystemTap Language Reference manual.


= Systemtap probe points changes

- More user-space probe types are added:
  probe process(PID).begin { }
  probe process("PATH").begin { }
  probe process(PID).thread.begin { }
  probe process("PATH").thread.begin { }
  probe process(PID).end { }
  probe process("PATH").end { }
  probe process(PID).thread.end { }
  probe process("PATH").thread.end { }
  probe process(PID).syscall { }
  probe process("PATH").syscall { }
  probe process(PID).syscall.return { }
  probe process("PATH").syscall.return { }

- .statement("func@file+line") probes are now supported to allow
  probing by line-number offset relative to the "{" that begins the
  function.  This allows using the same systemtap script without
  change if source changes are confined to other areas of the .c file.

For a full overview see the stapprobes(5) manual page.


= Standard tapsets functions changes

- Added functions to access different user space data types:
  user_string_n, user_string_n2, user_string_n_warn, user_string_n_quoted,
  user_short, user_short_warn, user_int, user_int_warn, user_long,
  user_long_warn, user_char, user_char_warn.

- Added "cpu register" and "numbered functions argument" support
  functions which help writing "dwarfless" probes.
  register:long (name:string), u_register:long (name:string),
  int_arg:long (n:long), uint_arg:long (n:long), long_arg:long
  (n:long), ulong_arg:long (n:long), longlong_arg:long (n:long),
  ulonglong_arg:long (n:long), pointer_arg:long (n:long), s32_arg:long
  (n:long), u32_arg:long (n:long), s64_arg:long (n:long), u64_arg:long
  (n:long), asmlinkage:unknown (), fastcall:unknown () and
  regparm:unknown (n:long).

- New function returnval:long () for usage with syscall tapset probes
  as counterpart to the returnstr:string (returnp:long) function.

- New function user_mode:long () that returns 1 if the probe point
  occurred in user-mode.

For a full overview see the stapfuncs(5) manual page.


= Standard tapset global variable changes

- NULL is now a new global variable (defined as global NULL = 0).

For a full overview see the stapvars(5) manual page.


= Systemtap examples changes

- examples are now installed by default by make install under
  <prefix>/doc/systemtap*/examples
  The examples are now explicitly run during make check and
  make installcheck to make sure they are always up to date
  (they can be found under testsuite/systemtap.examples).

- New examples:

 * io_submit.stp - When a reschedule occurs during an AIO io_submit
   call, accumulate the traceback in a histogram. When the script
   exits prints out a sorted list from most common to least common
   backtrace.

 * traceio.stp - Every second print out the top ten executables
   sorted in descending order based on cumulative I/O traffic
   observed.

 * iotop.stp - Every five seconds print out the top ten executables
   generating I/O traffic during that interval sorted in descending
   order.

 * disktop.stp - Get the status of reading/writing disk every 5
   seconds, output top ten entries during that period.

 * sigkill.stp - The script traces any SIGKILL signals. When that
   SIGKILL signal is sent to a process, the script prints out the
   signal name, the desination executable and process ID, the
   executable name user ID that sent the signal.

 * thread-times.stp - Sets up time-based sampling. Every five seconds
   it prints out a sorted list with the top twenty processes with
   samples broken down into percentage total time spent in user-space
   and kernel-space.

 * functioncallcount.stp - Takes one argument, a list of functions to
   probe. The script will run and count the number of times that each
   of the functions on the list is called. On exit the script will
   print a sorted list from most frequently to least frequently called
   function.

 * para-callgraph.stp - Takes two arguments: the first argument is the
   function to starts/stops the per thread call graph traces and the
   second argument is the list of functions to generate trace
   information on. The script prints out a timestap for the thread,
   the function name and pid, followed by entry or exit symboly and
   function name.

 * traceio2.stp - Print out the executable name and process number as
   reads and writes to the specified device occur.

 * sleepingBeauties.stp - Monitors time threads spend waiting for IO
   operations (in "D" state) in the wait_for_completion function. If a
   thread spends over 10ms wall-clock time waiting, information is
   printed out describing the thread number and executable name. When
   slow the wait_for_completion function complete, backtraces for the
   long duration calls are printed out.

 * graphs.stp - Tracks the disk and CPU utilization. The resulting
   output of the script can be piped into gnuplot to generate a graph
   of disk and CPU USE.

For more information about examples see also the stapex(5) manual page.


= Miscellaneous changes

- Much faster batch unregistering of (thousands of) kprobes implemented
  (needs linux 2.6.25).

- Scripts relying on probe point wildcards like "syscall.*" that expand
  to distinct kprobes are processed significantly faster than before.

- Shiny new Vim highlighting (see vim/syntax/stap.vim).

- stap now warns about variables read but not set ("read-only
  variables"), which are usually due to script typos.

- Configure with --enable-dejazilla for optional automatic test result
  reporting.

- Configure now accepts --enable-staticdw, which alters the link
  settings so they work with static libdw.


= The following people helped with code contributions for this release:

  Ananth N Mavinakayanahalli, Dave Brolley, David Smith, Eugene Teo,
  Frank Ch. Eigler, James Bottomley, Jim Keniston, Josh Stone,
  Mark Wielaard, Martin Hunt, Masami Hiramatsu, Shaohua Li, Srikar
  Dronamraju, Srinivasa DS, Stan Cox, Theodore Ts'o, Tim Moore,
  Wenji Huang, William Cohen and Zhaolei


= Know issues with this release:

  - http://sourceware.org/bugzilla/show_bug.cgi?id=6707
    oops crashes with 2.6.25 - onoffprobe
    Feedback appreciated since this is a very hard problem to reproduce.

= Bugs closed for this release:

#1165 implement .relative(nnn)
#1194 check operation of "-o" (output direction) flag
#2071 Probes on ISR with probes on task thread's prehandler crash the
      system
#2600 should optimize away assignments in other contexts
#2949 need "probe listing" command line option
#3051 print_backtrace() prints most of stack on a single line on ppc64
#3542 request new batch registration/unregistration API
#4014 improve command line option processing
#4904 hard to use statement probes for line-by-line tracing
#5001 ctime() vs _stp_ctime() duplication
#5101 stap abort in loc2c.c
#5102 improve code generation of error-free constructs
#5106 Remove old map and histogram formatting code
#5189 Support a '*' dynamic field width specifier in printf
#5231 Setting correct argument for bio_endio in ioblock.stp
#5528 more user_* data conversions needed
#5636 $builddir/SNAPSHOT should get recomputed at make check time
#5643 profile syscall tests to improve translator speed
#5645 Too many sub-buffers generated with -s option
#5648 Unaligned access to memory in _stp_vsnprintf() on ia64
#5716 staprun/stapio setuid/capability simplification
#5759 probes in sched.c on preempt kernel cause reboot
#5770 Seen lot of "eof" errors on systemtap-20080216 snapshot
#5891 Insertion of kprobe module crashes the system on 2.6.25-rc3-mm1,
      if CONFIG_DEBUG_RODATA is set on x86 system.
#5897 derived_probe derivation chain tweak
#5928 CFLAGS vs. EXTRA_CFLAGS in generated kbuild makefile
#5955 declaration of global variables with semicolon gives semantic error
#5956 Usage of "NULL" in systemtap script throws non-apparent error
      messages
#5963 testsuite/systemtap.maps/pmap_agg_overflow.stp crashes on
      2.6.25-0.121.rc5.git4.fc9
#5975 Systemtap build error with -DNDEBUG
#5980 set dejagnu snapshot from stap -V as fallback
#6008 support larger on-memory buffer
#6393 tag build with more precise git commit code
#6400 New runtime unwinder code does not make use of locally built elfutils
#6410 New unwind code causes errors on ia64
#6416 SystemTap should recognize and disble probes on __devinit/__devexit
      functions
#6417 Wildcard probe hangs system
#6429 unwinding/symbol stuff broken on ppc*
#6432 use 2.6.26 probe_kernel_{read,write}
#6451 staprun shutdown deadlock
#6454 stap -l cut off the format of marker-based probe point
#6455 marker probe: Format mismatch error for markings defined with
      MARK_NOARGS
#6466 more elision needed to support iffy tapsets
#6469 duplicate warning elimination
#6470 export command line arguments or run-time script parsing
#6474 build setuid staprun with -fpie
#6481 timers: stp_time_timer_callback calls mod_timer with interrupts
      disabled
#6492 stap -l outputs so many WARNINGS.
#6499 BUG with systemtap's utrace support
#6500 Systemtap's utrace support needs to follow new threads
#6509 syscall.get*id probes broken
#6520 Systemtap 20080503 and 20080510 snapshot fails on s390
#6524 ctime() on bad values hangs system
#6529 error() from nested function can be lost
#6534 utsname() becoming unusable
#6538 warn about read-only variables
#6550 add --enable-dejazilla
#6563 sys.stp fails on latest systemtap-20080524 snapshot
#6588 syscall.exit[_group].return should define name, retstr
#6601 systemtap with --kelf option looks for data symbol to probe a function
      on powerpc systemtap
#6644 Impure arguments prevent eliding pure function calls
#6703 tapsets.cxx does not build on gcc 4.10
#6732 task_struct parent field removed from RHEL-5 and fedora 9 kernels
#6736 switch to compiled-in symbol and unwind data


= Test results on various systems:

kernel location: /usr/lib/debug/lib/modules/2.6.24.7-92.fc8/vmlinux
kernel version: 2.6.24.7-92.fc8
systemtap location: /usr/local/systemtap/bin/stap
systemtap version: version 0.7/0.131 git branch master, commit 94687c8d
Test Run By root on Fri Jul 11 12:27:33 2008
Native configuration is x86_64-redhat-linux-gnu

Host: Linux dijkstra.wildebeest.org 2.6.24.7-92.fc8 #1 SMP Wed May 7 16:26:02 EDT 2008 x86_64 x86_64 x86_64 GNU/Linux
Snapshot: version 0.7/0.131 git branch master, commit 94687c8d
Distro: Fedora release 8 (Werewolf)

FAIL: debugpath-good (eof)
FAIL: PROCFS shutdown (eof)
FAIL: backtrace of yyy_func4 (1)
FAIL: print_stack of yyy_func4 (1)
FAIL: 64-bit access
FAIL: 64-bit chmod
FAIL: 64-bit forkwait
FAIL: 64-bit link
FAIL: 64-bit timer
FAIL: 32-bit access
FAIL: 32-bit chmod
FAIL: 32-bit forkwait
FAIL: 32-bit link
FAIL: 32-bit rt_signal
FAIL: 32-bit timer

# of expected passes            565
# of unexpected failures        15
# of expected failures          197
# of unknown successes          1
# of known failures             6
# of untested testcases         20
# of unsupported tests          2


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]