Prev: drivercore/of: Add OF style matching to platform bus
Next: (none)
From: Josh Stone on 21 Jul 2010 20:20
The systemtap team announces release 1.3.

integrated compile-server client, automatic structure
pretty-printing, much faster & better stack backtraces (user and
kernel space), interesting new sample scripts, prototype
debuginfo-less <sys/sdt.h> userspace markers, kernel versions 2.6.9
through 2.6.35-rc4

= Where to get it

http://sourceware.org/systemtap/ - our project page
http://sourceware.org/systemtap/ftp/releases/systemtap-1.3.tar.gz
http://koji.fedoraproject.org/koji/packageinfo?packageID=615
git tag release-1.3 (commit e12e958)

There have been over 500 commits since the last release.
There have been 106 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

- A new integrated compile-server client is now available as part of stap.

o 'stap --use-server ...' is equivalent to 'stap-client ...'
o 'stap --list-servers' is equivalent to 'stap-find-servers'
o 'stap --list-servers=online' is equivalent to 'stap-find-servers --all'
o stap-client and its related tools will soon be deprecated.
o the nss-devel and avahi-devel packages are required for building stap with
the integrated client (checked during configuration).
o nss and avahi are required to run the integrated client.

- The new "--ldd" option automatically adds any additional shared
libraries needed by probed or -d-listed userspace binaries to the -d
list, to enable symbolic backtracing through them. Similarly, the
new "--all-modules" option automatically adds any currently loaded
kernel modules (listed in /proc/modules) to the -d list.

- By default the systemtap-runtime RPM builds now include a shared
library, staplog.so, that allows crash to extract systemtap data from
a vmcore image.

- Backward compatibility flags (--compatible=VERSION, and matching
script preprocessing predicate %( systemtap_v CMP "version" %)
and a deprecation policy are being introduced, in case future
tapset/language changes break valid scripts.

- The <sys/sdt.h> user-space markers no longer default to an implicit
MARKER_NAME_ENABLED() semaphore check for each marker. To check for
enabled markers use a .d declaration file, then:
if (MARKER_NAME_ENABLED()) MARKER_NAME()

- Hyphenated <sys/sdt.h> marker names such as process(...).mark("foo-bar")
are now accepted in scripts. They are mapped to the double-underscore
form ("foo__bar").

- More robust <sys/sdt.h> user-space markers support is included. For
some platforms (x86*, ppc*), this can let systemtap probe the markers
without debuginfo. This implementation also supports preserving
the "provider" name associated with a marker:
probe process("foo").provider("bar").mark("baz") to match
STAP_PROBE<n>(bar, baz <...>)
(Compile with -DSTAP_SDT_V1 to revert to the previous implementation.
Systemtap supports pre-existing or new binaries using them.)

- User space marker arguments no longer use volatile if the version of gcc,
which must be at least 4.5.0, supports richer DWARF debuginfo. Use cflags
-DSTAP_SDT_VOLATILE=volatile or -DSTAP_SDT_VOLATILE= when building
the sys/sdt.h application to override this one way or another.

- New stap option -G VAR=VALUE allows overriding global variables
by passing the settings to staprun as module options.

- The uprobes kernel module now has about half the overhead when probing
NOPs, which is particularly relevant for sdt.h markers.

= Systemtap script language changes

- Iterating with "foreach" can now explicitly save the value for the loop.
foreach(v = [i,j] in array)
printf("array[%d,%s] = %d\n", i, j, v /* array[i,j] */)

- A new operator @entry is available for automatically saving an expression
at entry time for use in a .return probe.
probe foo.return { println(get_cycles() - @entry(get_cycles())) }

- Probe $target variables and @cast() can now use a suffix to print complex
data types as strings. Use a single '$' for a shallow view, or '$$' for a
deeper view that includes nested types. For example, with fs_struct:
$fs$ : "{.users=%i, .lock={...}, .umask=%i,
.in_exec=%i, .root={...}, .pwd={...}}"
$fs$$ : "{.users=%i, .lock={.raw_lock={.lock=%u}}, .umask=%i, .in_exec=%i,
.root={.mnt=%p, .dentry=%p}, .pwd={.mnt=%p, .dentry=%p}}"

- Embedded-C may be used within expressions as values, when in guru mode:
num = %{ LINUX_VERSION_CODE %} // int64_t
name = %{ /* string */ THIS_MODULE->name %} // const char*
printf ("%s %x\n", name, num)
The usual /* pure */, /* unprivileged */, and /* guru */ markers may be used
as with embedded-C functions.

- Probe wildcards can now use '**' to cross the '.' separator.
$ stap -l 'sys**open'
syscall.mq_open
syscall.open

- A new procfs parameter .umask(UMASK) which provides modification of
file permissions using the proper umask value. Default file
permissions for a read probe are 0400, 0200 for a write probe, and
0600 for a file with a read and write probe.

= Systemtap tapset changes

- It is now possible in some situations to use print_ubacktrace() to
get a user space stack trace from a kernel probe point. e.g. for
user backtraces when there is a pagefault:
$ stap -d /bin/sort --ldd -e 'probe vm.pagefault {
if (pid() == target()) {
printf("pagefault @0x%x\n", address); print_ubacktrace();
} }' -c /bin/sort
[...]
pagefault @0x7fea0595fa70
0x000000384f07f958 : __GI_strcmp+0x12b8/0x1440 [libc-2.12.so]
0x000000384f02824e : __gconv_lookup_cache+0xee/0x5a0 [libc-2.12.so]
0x000000384f021092 : __gconv_find_transform+0x92/0x2cf [libc-2.12.so]
0x000000384f094896 : __wcsmbs_load_conv+0x106/0x2b0 [libc-2.12.so]
0x000000384f08bd90 : mbrtowc+0x1b0/0x1c0 [libc-2.12.so]
0x0000000000404199 : ismbblank+0x39/0x90 [sort]
0x0000000000404a4f : inittables_mb+0xef/0x290 [sort]
0x0000000000406934 : main+0x174/0x2510 [sort]
0x000000384f01ec5d : __libc_start_main+0xfd/0x1d0 [libc-2.12.so]
0x0000000000402509 : _start+0x29/0x2c [sort]
[...]

- New tapset functions to get a string representation of a stack trace:
sprint_[u]backtrace() and sprint_[u]stack().

- New tapset function to get the module (shared library) name for a
user space address umodname:string(long). The module name will now
also be in the output of usymdata() and in backtrace addresses even
when they were not given with -d at the command line.

- Kernel backtraces are now much faster (replaced a linear search
with a binary search).

- A new family of set_kernel_* functions make it easier for gurus to write
new values at arbitrary memory addresses.

- The tapset alias 'syscall.compat_pselect7a' was misnamed. It should
have been 'syscall.compat_pselect7' (without the trailing 'a').
Starting in release 1.4, the old name will be deprecated.

= New script examples

- profiling/errno.stp Show which processes and system calls
return errors most frequently
- io/deviceseeks.stp Histograms of seek behavior for each device
- process/noptrace.stp Disable ptrace(2) from hierarchies of processes
- general/sizeof.stp Print the size of a C type
- profiling/linetimes.stp Show time spent on each line if a function
- process/cycle_thief.stp Track IRQs and other processes stealing
cycles from a task
- general/para-callgraph-verbose.stp Verbose callgraph tracing

= Contributors for this release

Adrien Kunysz*, Charley Wang, Dave Brolley, David Smith,
Don Domingo, Frank Ch. Eigler, Josh Stone, Lars R. Damerow*,
Lukas Berk*, Mark Wielaard, Przemysław Pawełczyk, Rayson Ho*,
Roland McGrath, Srikar Dronamraju, Stan Cox, Tom Callaway*,
Tony Jones*, Wenji Huang, William Cohen

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

Thanks also to David Smith for compiling these notes.

= Examples of tested kernel versions

2.6.9 (el4/i686)
2.6.18 (el5/ia64/i686/x86_64)
2.6.32 (el6/i686/x86_64)
2.6.32 (f12/i686/x86_64)
2.6.33 (f13/i686/x86_64)
2.6.35-rc4 (rawhide/x86_64)

= Known issues with this release

- When using the systemtap client and server udp port 5353 must be
open in your firewall in order for the clent to find servers using
avahi-browse.

- The stap-server may not be sufficiently robust to operate on an open
untrusted network.

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

= Bugs fixed for this release

1155 inline function parameters
2168 incomplete translation of histogram bucket foreach
2228 benchmark suite
3672 Support formatted dump of struct $pointers
3833 nfs_proc.stp and LKET/nfs_proc.stp tapset files broken
4396 script to dump hung task's kernel stack
4529 tapset coverage option
4978 need better printf() documentation
5151 rpc-all-probes.stp fails on 2.6.23-rc9 kernel
5621 RFE: support for a general buffer kread()
5949 instruction/block tracing for userspace
6582 register() needs pt_regs check
6731 improve -l (listing mode) to list probe/alias variables
6739 global alias cache too slow
6819 stapprobes(5) - syntactically (in)valid probepoints
6826 shorthand or variable for probed executable's path
6898 marker $$args / $$parms
6902 optimize use of elfutils offline module search
7030 signal tapset may be referring to inline functions.
7070 registration error (rc -22), module probes, missing reloc info
9958 Provide a way for C tapset functions to assert guru mode
9998 Update tapset/i686/registers.stp per pt_regs change
10023 Examples showing kernelpoint use
10025 wildcards lose when embedded in conditional %(%) expressions
10054 improve $expr->foo->bar->zoo error messages
10065 Create modules for Hardware Breakpoint
10080 track vdso for process symbols/backtrace
10115 Regression with inet_sock @cast from 0.9 to 0.9.7 on 2.6.9 kernel
10228 Add more vma-tracking for user space symbol/backtraces
10245 Allow "-" in static marker names
10327 follow symbol aliases for debuginfo search
10393 sdt.exp utrace tests fail on 2.6.31 rawhide kernel
10599 Always print module/map names for kernel/executable addresses
10665 backtrace() is too slow
10830 new pp() variant for source-level probe point name
10913 bad/misleading translator error message
10943 Support generic @entry saving in .return probes
10978 Getting the size of a data type
11095 Support "provider" for mark() probes
11222 ABRT interface
11266 var_expanding_visitor barfs on non-assignment lvalues
11274 stap-client doesn't understand "--"
11340 directly-accessed tracepoint args are surviving target_symbols
11343 backward compatibility flags/conditionals
11373 let stap-client use cache
11399 Allow ** in matching probe points
11425 probefunc() broken when used with kprobe.function
11427 systemtap 1.2 does not compile on sparc/sparc64
11443 Add a "list unprivileged" option to stap
11446 Server installed by systemtap-server rpm should be locally trusted
11455 use XDG_DATA_DIRS for tapset path searching
11457 unresolved *mib_filter_key functions
11473 Support optional "VALUE =" in foreach loops
11474 iotop example is very slow in pass-2
11480 bad kfail for rpc-all-probes.stp test
11498 don't have a cow with oversize debug_frame
11508 automate -d option for shared library searching
11514 nd_syscall.exp takes 10 times longer than syscall.exp testcase
11553 need better message for return probe on inline function
11556 @cast() with array indexing on pointers
11566 embedded-C expressions
11573 uprobe_get_pc, sayonara
11575 TRYLOCKDELAY too short
11590 optimized pre/postfix operators throw away initial assignment
11592 sdt.h should have a 'disable-semaphores' override option
11598 @defined() should allow the address-of operator "&"
11599 Support data pointer in tracepoint callbacks
11609 -d ALL_LOADED_MODULES option
11617 Gluing string literals doesn't work with token substitution
11641 FAIL: fib (19 1) backtrace from uretprobe fails
11649 workarounds for incomplete kernel include/linux/events/*.h decls
11661 recvmmsg missing from syscall tapset
11663 switch to embedded-c-expressions for kernel-marker tapset functions
11664 try-catch squashes normal exits from next/return
11665 umodname tapset function
11672 utrace_report_syscall_exit crash
11678 stap -d /bin/prog --ldd doesn't pick up ld.so
11680 Systemtap build broken on PPC64
11684 eliminate duplication between language reference guide and tapset docs
11690 Semaphores are broken under STAP_SDT_V2
11698 stap fails to figure out powerpc function aliases
11706 unautoconfiscation of most stap*.in man pages
11707 on ppc, all nd_syscall probes fail
11708 SDT on i686 is register-poor
11710 on s390x, all nd_syscall probes fail
11713 listing of available $variables should include $
11719 on ppc, the cmd_parse.exp and preprocessor.exp test fail
11721 the usymbols.exp testcase shouldn't touch '__user' memory directly
11722 on ppc, usymbols.exp should be probing syscall.rt_sigaction32
11723 transok/buildko.stp fails incorrectly
11725 on RHEL5 ppc, systemtap.printf/bin6.exp fails
11726 on ppc, systemtap.printf/memory1.exp fails with compile errors
11727 the systemtap.server/server.exp testcase reports invalid failures
11729 option for procfs files to specify permissions
11752 on ppc, the signal.handle_signal tapset alias needs to be updated
11761 stap -L 'kernel.mark("*")' only lists $arg1..$argN variables
11762 the systemtap.base/flightrec*.exp tests randomly fail
11771 on ppc, the syscall.pipe tapset alias fails when accessing pipe0/pipe1
11775 ensure all tapsets are referenced in tapsets.tmpl
11778 sprint_ustack & sprint_stack
11785 some $foo$ pretty-printing bugs
11808 loc2c unfinished chain from a struct $return
11809 loc2c vs empty piece
11815 SDT_V2 sdt.h broken for modern gcc on x86-64, cxxclass.exp
11821 SDT_V2 dwarfless ambiguities with powerpc // -mno-regnames
11825 The piece we want is part of a computed value!

= Test results on various systems

- After running "sudo make installcheck" from the test suite, on a
suitably equipped machine (kernel debugging data and other stuff
installed), you should see 1400-1600 passes and a small handful of
failures. On a machine without kernel utrace support, you should
see 800-1000 passes.

*** Hark! A gopher's spiffy crankshaft! ***
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo(a)vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
 | 
Pages: 1
Prev: drivercore/of: Add OF style matching to platform bus
Next: (none)