]> sourceware.org Git - systemtap.git/log
systemtap.git
13 years agoCorrect error in 4.3.1. Counting Function Calls Made
William Cohen [Thu, 22 Jul 2010 18:53:15 +0000 (14:53 -0400)]
Correct error in 4.3.1. Counting Function Calls Made

This was reported in Red Hat bz617055.

13 years agoadded a test for the procfs umask functionality, ensuring expected
Lukas Berk [Thu, 22 Jul 2010 16:39:52 +0000 (12:39 -0400)]
added a test for the procfs umask functionality, ensuring expected
permissions match up with what is actually set

13 years agoVersion bumps for the 1.3 release release-1.3
Josh Stone [Wed, 21 Jul 2010 20:43:23 +0000 (13:43 -0700)]
Version bumps for the 1.3 release

13 years agoAdd NEWS for uprobes-on-NOP speedup
Josh Stone [Wed, 21 Jul 2010 20:38:50 +0000 (13:38 -0700)]
Add NEWS for uprobes-on-NOP speedup

13 years agoRemove duped -G NEWS
Josh Stone [Wed, 21 Jul 2010 20:38:11 +0000 (13:38 -0700)]
Remove duped -G NEWS

14 years agoExtend vta-test.exp testcase for PR11825 and add 32-on-64 for fun.
Mark Wielaard [Wed, 21 Jul 2010 20:24:33 +0000 (22:24 +0200)]
Extend vta-test.exp testcase for PR11825 and add 32-on-64 for fun.

* testsuite/systemtap.base/vta-test.c: Provide constants through struct.
* testsuite/systemtap.base/vta-test.exp: Build/Test also for 32-on-64 if
  available.
* testsuite/systemtap.base/vta-test.stp: Take program under probe as @1.

14 years agoPR11825: loc2c: Handle a computed-value piece.
Roland McGrath [Wed, 21 Jul 2010 18:37:14 +0000 (11:37 -0700)]
PR11825: loc2c: Handle a computed-value piece.

* loc2c.c (location_relative): Handle a computed-value piece.

14 years agorhbz#616040 loc2c deref on 32bit address fails.
Mark Wielaard [Wed, 21 Jul 2010 13:59:27 +0000 (15:59 +0200)]
rhbz#616040 loc2c deref on 32bit address fails.

* runtime/loc2c-runtime.h (deref): For __x86_64__ case cast address to
  (unsigned long) everywhere in case we got a 32 bit address of a 32-on-64
  but process.

14 years agofntimes: use @entry for timestamps
Josh Stone [Tue, 20 Jul 2010 23:55:15 +0000 (16:55 -0700)]
fntimes: use @entry for timestamps

14 years agoerrno: use syscall.*'s "name" instead of probefunc()
Josh Stone [Tue, 20 Jul 2010 23:24:32 +0000 (16:24 -0700)]
errno: use syscall.*'s "name" instead of probefunc()

14 years agowait4time: use @entry for timestamps
Josh Stone [Tue, 20 Jul 2010 23:17:04 +0000 (16:17 -0700)]
wait4time: use @entry for timestamps

14 years agosleeptime: use @entry for timestamps
Josh Stone [Tue, 20 Jul 2010 23:06:21 +0000 (16:06 -0700)]
sleeptime: use @entry for timestamps

14 years agoUse @entry and return- in futexes example
Josh Stone [Tue, 20 Jul 2010 22:32:56 +0000 (15:32 -0700)]
Use @entry and return- in futexes example

14 years agoMake sure Dwarf_Ops don't have garbage
Josh Stone [Tue, 20 Jul 2010 22:10:29 +0000 (15:10 -0700)]
Make sure Dwarf_Ops don't have garbage

We synthesize a Dwarf_Op in a couple of places, but we were only setting
the fields we care about, ".atom" and ".number".  The garbage in
".number2" would survive into "DWARF expression:" comment in loc2c,
which made it look unique and subverted caching.

Now we memset-0 the Dwarf_Op before proceeding.

14 years agotcpipstat: use @defined instead of preprocessor
Josh Stone [Tue, 20 Jul 2010 22:09:24 +0000 (15:09 -0700)]
tcpipstat: use @defined instead of preprocessor

14 years agosk_stream_wait_memory spelling fixes
Josh Stone [Tue, 20 Jul 2010 21:18:51 +0000 (14:18 -0700)]
sk_stream_wait_memory spelling fixes

14 years agoUpdate bkl examples with newer syntax
Josh Stone [Tue, 20 Jul 2010 21:06:10 +0000 (14:06 -0700)]
Update bkl examples with newer syntax

* testsuite/systemtap.examples/locks/bkl.stp: Use '!' to alternate
  kernel probepoints between lock_kernel and _lock_kernel (and unlock).
* testsuite/systemtap.examples/locks/bkl_stats.stp: Ditto, and use
  @entry to get timestamps.

14 years agoUse actual bytes read/written for iotop
Josh Stone [Tue, 20 Jul 2010 19:54:15 +0000 (12:54 -0700)]
Use actual bytes read/written for iotop

Probe on .return, so we know how many bytes the vfs actually handled.

14 years agoRemove explicit entry probes from iotime example
Josh Stone [Tue, 20 Jul 2010 19:32:47 +0000 (12:32 -0700)]
Remove explicit entry probes from iotime example

For the use of syscall.open/read/write, use .return to do all the work.
This uses @entry to collect timestamps and implicit saving for other
$vars.

Note that $return is used for byte counts instead of the entry $count,
because we should care more about the actual work performed.

14 years agoAdd para-callgraph-verbose example
Josh Stone [Tue, 20 Jul 2010 19:15:08 +0000 (12:15 -0700)]
Add para-callgraph-verbose example

It's like para-callgraph, except using $$parms$$.  The output can get
VERY big, so you probably want to dump it into a file for offline
browsing...

14 years agoUse @defined in examples/general/grapher.stp
Josh Stone [Thu, 15 Jul 2010 22:11:53 +0000 (15:11 -0700)]
Use @defined in examples/general/grapher.stp

14 years agoTweak the systemtap.apps tcl test.
Stan Cox [Tue, 20 Jul 2010 21:03:56 +0000 (17:03 -0400)]
Tweak the systemtap.apps tcl test.

tcl.exp: Bump up timeout value.
stap-tcl.sh: Don't run obj.test as it takes a really long time.

14 years agomention user-space and other recent developments in man pages and tutorial
Frank Ch. Eigler [Tue, 20 Jul 2010 17:02:34 +0000 (13:02 -0400)]
mention user-space and other recent developments in man pages and tutorial

14 years agoNEWS: tweak SDT_V2 blurbage
Frank Ch. Eigler [Tue, 20 Jul 2010 15:14:54 +0000 (11:14 -0400)]
NEWS: tweak SDT_V2 blurbage

14 years agoMake SDT_V2 the default on all architectures.
Stan Cox [Tue, 20 Jul 2010 14:38:24 +0000 (10:38 -0400)]
Make SDT_V2 the default on all architectures.

sdt.h:  Don't special case unsupported STAP_SDT_V2 architectures as STAP_SDT_V1

14 years agoPR11821: sdt.exp should tolerate sdt_v2 dwarf warnings
Frank Ch. Eigler [Tue, 20 Jul 2010 11:29:43 +0000 (07:29 -0400)]
PR11821: sdt.exp should tolerate sdt_v2 dwarf warnings

14 years agoMerge branch 'master' of git://sources.redhat.com/git/systemtap
Frank Ch. Eigler [Tue, 20 Jul 2010 01:26:22 +0000 (21:26 -0400)]
Merge branch 'master' of git://sources.redhat.com/git/systemtap

* 'master' of git://sources.redhat.com/git/systemtap:
  * tapset/syscalls2.stp: Started a pending deprecation of incorrect   'syscall.compat_pselect7a' probe alias name.  Correct name is   'syscall.compat_pselect7'. * testsuite/buildok/syscalls2-detailed.stp: Checks for correct probe alias   name. * NEWS: Mentioned 'syscall.compat_pselect7a' deprecation. * stap.1: Ditto. * HACKING: Updated deprecation description.
  Many small tapset/syscalls2.stp fixups.  Added testcase.
  Fix spurios testcase failure from systemtap.examples/network/socktop.
  added the procfs.umask(UMASK) feature to the NEWS file.

Conflicts:
NEWS

14 years agoPR11815: SDT_V2 asm operand parsing reorg
Frank Ch. Eigler [Tue, 20 Jul 2010 01:18:46 +0000 (21:18 -0400)]
PR11815: SDT_V2 asm operand parsing reorg

* tapsets.cxx (sdt_uprobe_var_expanding_visitor ctor): Use %-adorned
  register names.  Accept other architectures.
  (visit_target_symbol): Rewrite regexp matching / parsing.  Fall back to
  DWARF in case of parse failure.
* testsuite/systemtap.base/cxxclass.exp: Use -w to suppress dwarf warning.
* util.cxx (regexp_match): New function.
* util.h: Declare it.

14 years ago* tapset/syscalls2.stp: Started a pending deprecation of incorrect
David Smith [Mon, 19 Jul 2010 21:44:20 +0000 (16:44 -0500)]
* tapset/syscalls2.stp: Started a pending deprecation of incorrect
  'syscall.compat_pselect7a' probe alias name.  Correct name is
  'syscall.compat_pselect7'.
* testsuite/buildok/syscalls2-detailed.stp: Checks for correct probe alias
  name.
* NEWS: Mentioned 'syscall.compat_pselect7a' deprecation.
* stap.1: Ditto.
* HACKING: Updated deprecation description.

14 years agoMany small tapset/syscalls2.stp fixups. Added testcase.
David Smith [Mon, 19 Jul 2010 21:19:49 +0000 (16:19 -0500)]
Many small tapset/syscalls2.stp fixups.  Added testcase.

* tapset/syscalls2.stp (syscall.compat_sys_recvmsg): Make sure it supports
  the same variables as the non-compat probe.
  (syscall.compat_sys_semctl): Ditto.
  (syscall.compat_sys_sendmsg): Ditto.
  (syscall.sched_setaffinity): At some point, x86_64 kernels couldn't
  find $len, so a special version of this probe was written for it.  Now
  just uses @defined().
  (syscall.compat_signalfd): Prefers compat_sys_signalfd4 over
  compat_sys_signalfd.
  (syscall.ustat32): Merged the 2 probes into 1.
* testsuite/buildok/syscalls2-detailed.stp: New testcase.

14 years agoFix spurios testcase failure from systemtap.examples/network/socktop.
David Smith [Mon, 19 Jul 2010 17:49:37 +0000 (12:49 -0500)]
Fix spurios testcase failure from systemtap.examples/network/socktop.

* testsuite/systemtap.examples/network/socktop: Make sure the exit status
  is from stap, not from the cleanup commands.

14 years agoadded the procfs.umask(UMASK) feature to the NEWS file.
Lukas Berk [Mon, 19 Jul 2010 15:47:37 +0000 (11:47 -0400)]
added the procfs.umask(UMASK) feature to the NEWS file.

14 years agoMerge branch 'master' of git://sources.redhat.com/git/systemtap
Frank Ch. Eigler [Mon, 19 Jul 2010 14:31:36 +0000 (10:31 -0400)]
Merge branch 'master' of git://sources.redhat.com/git/systemtap

* 'master' of git://sources.redhat.com/git/systemtap:
  PR 11729 adding a umask parameter to modify the permissions of the file

14 years agoPR 11729 adding a umask parameter to modify the permissions of
Lukas Berk [Fri, 16 Jul 2010 19:17:46 +0000 (15:17 -0400)]
PR 11729 adding a umask parameter to modify the permissions of
the file

14 years agotweak -G documentation wording
Frank Ch. Eigler [Thu, 15 Jul 2010 15:52:21 +0000 (11:52 -0400)]
tweak -G documentation wording

14 years agoMinor timestamp.stp/timestap_gtod.stp/ucontext.stp buildok test improvements.
David Smith [Wed, 14 Jul 2010 15:33:01 +0000 (10:33 -0500)]
Minor timestamp.stp/timestap_gtod.stp/ucontext.stp buildok test improvements.

* testsuite/buildok/timestamp-embedded.stp: Renamed from
  buildok/timestamp.stp.  Only covers function from timestamp.stp.
* testsuite/buildok/timestamp_gtod-embedded.stp: New testcase.
* testsuite/buildok/ucontext-embedded.stp: Renamed from ucontext.stp

14 years agoMerge branch 'master' of ssh://sources.redhat.com/git/systemtap
Rayson Ho [Wed, 14 Jul 2010 05:39:05 +0000 (01:39 -0400)]
Merge branch 'master' of ssh://sources.redhat.com/git/systemtap

14 years agoValiduate -G & -m user input w/ assert_regexp_match().
Rayson Ho [Wed, 14 Jul 2010 05:37:13 +0000 (01:37 -0400)]
Validuate -G & -m user input w/ assert_regexp_match().

14 years agoPR 11373: Cleanup residual stap_*.ko modules when shutting down the compile-server.
Dave Brolley [Tue, 13 Jul 2010 19:48:13 +0000 (15:48 -0400)]
PR 11373: Cleanup residual stap_*.ko modules when shutting down the compile-server.

14 years agoPR 11274: server.exp: Add tests.
Dave Brolley [Tue, 13 Jul 2010 19:18:05 +0000 (15:18 -0400)]
PR 11274: server.exp: Add tests.

14 years agoEnsure that server.exp test names have the "with server" suffix.
Dave Brolley [Tue, 13 Jul 2010 19:09:49 +0000 (15:09 -0400)]
Ensure that server.exp test names have the "with server" suffix.

14 years agovalidate input for -G
Rayson Ho [Tue, 13 Jul 2010 19:00:46 +0000 (15:00 -0400)]
validate input for -G

14 years agoAdd tests for PR11808 and PR11809
Josh Stone [Tue, 13 Jul 2010 18:10:32 +0000 (11:10 -0700)]
Add tests for PR11808 and PR11809

* testsuite/semok/pr11808.stp: Make sure we don't abort().
* testsuite/semok/pr11809.stp: Make sure we don't underflow.

14 years agoAllow pretty-printing with partial availability
Josh Stone [Tue, 13 Jul 2010 17:25:48 +0000 (10:25 -0700)]
Allow pretty-printing with partial availability

Especially with local structs, DWARF may only have parts of a variable
available at runtime.  This change lets us print '?' for inaccessible
parts, so we can still print the parts we do have.

* tapsets.cxx (dwarf_pretty_print::push_deref): New, try a deref and
  push a '?' instead if it fails.
  (dwarf_pretty_print::recurse): Use push_deref instead of deref.
  (dwarf_pretty_print::recurse_base): Ditto.
  (dwarf_pretty_print::recurse_pointer): Ditto.
  (dwarf_pretty_print::print_chars): Ditto.

14 years agoPR 11274: csclient.cxx: Revert previous fix. Pass "--" before passing script arguments.
Dave Brolley [Tue, 13 Jul 2010 16:54:10 +0000 (12:54 -0400)]
PR 11274: csclient.cxx: Revert previous fix. Pass "--" before passing script arguments.

14 years agoPR 11274: csclient.cxx: DOn't pass script arguments to the server.
Dave Brolley [Tue, 13 Jul 2010 15:25:53 +0000 (11:25 -0400)]
PR 11274: csclient.cxx: DOn't pass script arguments to the server.

14 years agofix commit misnames
Frank Ch. Eigler [Tue, 13 Jul 2010 12:15:01 +0000 (08:15 -0400)]
fix commit misnames

14 years agosome more git hints
Frank Ch. Eigler [Tue, 13 Jul 2010 12:12:03 +0000 (08:12 -0400)]
some more git hints

14 years agoNew stap -G option to support passing global variables to the kernel module
rho [Tue, 13 Jul 2010 06:27:39 +0000 (02:27 -0400)]
New stap -G option to support passing global variables to the kernel module
(Bugzilla Bug 5642)

14 years agotest case for stap -G
rho [Tue, 13 Jul 2010 06:09:26 +0000 (02:09 -0400)]
test case for stap -G

14 years agoPR11809: loc2c: handle empty pieces
Roland McGrath [Tue, 13 Jul 2010 02:40:15 +0000 (19:40 -0700)]
PR11809: loc2c: handle empty pieces

* loc2c.c (struct location): Add new type loc_unavailable.
(translate): Generate that for an empty DW_OP_piece.
(location_relative): Handle it.
(c_translate_location, c_translate_addressof): Likewise.
(c_translate_array, emit_base_fetch, emit_base_store): Likewise.

14 years agoPR11808: loc2c fix for fields of noncontiguous location
Roland McGrath [Tue, 13 Jul 2010 02:22:21 +0000 (19:22 -0700)]
PR11808: loc2c fix for fields of noncontiguous location

* loc2c.c (location_relative): Chain onto loc_noncontiguous head so piece
can be found from head of original list.

14 years agoloc2c-test cleanup
Roland McGrath [Tue, 13 Jul 2010 02:20:48 +0000 (19:20 -0700)]
loc2c-test cleanup

* loc2c-test.c (handle_fields): Clean up #if code.

14 years agoloc2c-test: Support "return" testing
Roland McGrath [Tue, 13 Jul 2010 01:55:53 +0000 (18:55 -0700)]
loc2c-test: Support "return" testing

* loc2c-test.c (handle_fields): New function, broken out of ...
(handle_variable): ... here.  Call it.
(main): Handle special "return" pseudo-variable.

14 years agoAnchor bz10078.exp's regexp patterns
Josh Stone [Mon, 12 Jul 2010 22:05:40 +0000 (15:05 -0700)]
Anchor bz10078.exp's regexp patterns

If we are unlucky in buffering, then the test output could be read
into expect as a single buffer.  Then one of our regexps could match the
end of the buffer and consume the whole thing, even if we also wanted to
match something earlier in the buffer.

The solution is to anchor the patterns to the beginning of the buffer
with '^', so we're sure to match everything we can.

14 years agoBetter catch server errors in the testsuite
Josh Stone [Mon, 12 Jul 2010 21:51:10 +0000 (14:51 -0700)]
Better catch server errors in the testsuite

If "avahi-browse ... | grep ..." doesn't find anything, like when a
firewall blocks the way, we should catch that gracefully.  Use tcl's
"catch" instead of "set" for this.

* testsuite/lib/systemtap.exp (setup_server): Catch avahi-browse|grep
  when it falls.

14 years agoMinor task_time.stp and tcpmib.stp buildok test improvements.
David Smith [Mon, 12 Jul 2010 20:28:26 +0000 (15:28 -0500)]
Minor task_time.stp and tcpmib.stp buildok test improvements.

* testsuite/buildok/task_time-embedded.stp: Added missing function.
* testsuite/buildok/tcpmib-all-probes.stp: Moved function tests to
  tcpmib-embedded.stp.
* testsuite/buildok/tcpmib-embedded.stp: New test.

14 years agoPR 11727: Synchronize expected failures of buildok.exp and server.exp.
Dave Brolley [Mon, 12 Jul 2010 20:21:18 +0000 (16:21 -0400)]
PR 11727: Synchronize expected failures of buildok.exp and server.exp.

FWIW, I did not rename testsuite/systemtap.pass1-4/buildok.exp to
testsuite/lib/stap_buildok.exp, but that's the way git sees it.

14 years agoAdd all the weirdness to compat-%p
Josh Stone [Mon, 12 Jul 2010 20:01:51 +0000 (13:01 -0700)]
Add all the weirdness to compat-%p

I thought the old %p behavior was as simple as %#.8x or %#.16x, but it
was even weirder than that, as the ptr.exp test reveals.  The compat
path is now able to pass that test, and a new test makes sure of the new
code behaves as expected too.

* runtime/vsprintf.c (_stp_vsnprintf): compat-%p is better preserved.
* testsuite/systemtap.printf/ptr.exp -> ptr-1.2.exp: Keep as a compat
  test, and add a check that %.Np precision is ignored.
* testsuite/systemtap.printf/ptr-1.3.exp: New, make sure that %p really
  does act like %#x now, with various field_width.precision combos.

14 years agoMerge branch 'master' of ssh://sources.redhat.com/git/systemtap
William Cohen [Mon, 12 Jul 2010 14:57:38 +0000 (10:57 -0400)]
Merge branch 'master' of ssh://sources.redhat.com/git/systemtap

14 years agoRemove the unneeded default filter functions from tests.
William Cohen [Mon, 12 Jul 2010 14:56:34 +0000 (10:56 -0400)]
Remove the unneeded default filter functions from tests.

14 years agomention git pushurl
Frank Ch. Eigler [Mon, 12 Jul 2010 14:47:39 +0000 (10:47 -0400)]
mention git pushurl

14 years agoAdd default filter functions for ipmib, tcpmib, and linuxmib tapset
William Cohen [Mon, 12 Jul 2010 14:25:30 +0000 (10:25 -0400)]
Add default filter functions for ipmib, tcpmib, and linuxmib tapset

The ipmib, tcpmib, and linuxmib tapsets use filter functions. There
needs to be defaults for each of the filter functions so -L and -l work.

14 years agoFix usymbols.exp test for 32-on-64 (again). PR11722.
Mark Wielaard [Fri, 9 Jul 2010 22:26:37 +0000 (00:26 +0200)]
Fix usymbols.exp test for 32-on-64 (again). PR11722.

Bug introduced by commit 82ea0c. Add a comment why we need the same
"chopping" of addr in both _stp_umod_lookup and _stp_kallsyms_lookup.

* runtime/sym.c (_stp_kallsyms_lookup): Detect long used as 32bit address.

14 years agoKeep track of name, not dentry, in vma tracker.
Mark Wielaard [Fri, 9 Jul 2010 21:33:48 +0000 (23:33 +0200)]
Keep track of name, not dentry, in vma tracker.

Now even the vdso has a module name.

* runtime/task_finder_vma.c (__stp_tf_vma_entry): Remove dentry, add name.
  (stap_add_vma_map_info): Take and store name argument, not dentry.
  (stap_find_vma_map_info): Likewise.
  (stap_find_vma_map_info_user): Likewise.
* runtime/sym.c (_stp_umod_lookup): Lookup name directly with
  stap_find_vma_map_info.
* runtime/vma.c (_stp_vma_match_vdso): Pass "vdso" as name.
  (_stp_vma_mmap_cb): Extract and store name from dentry.

14 years agoImprovements to the scsi tapset and build testcases.
David Smith [Fri, 9 Jul 2010 19:19:48 +0000 (14:19 -0500)]
Improvements to the scsi tapset and build testcases.

* tapset/scsi.stp: Added missing convenience variable documentation.
* testsuite/buildok/scsi-detailed.stp: Renamed from buildok/scsi.stp and
  updated with the rest of the convenience variables.
* testsuite/buildok/scsi-embedded.stp: Updated with the rest of the
  functions.

14 years agoAdded build testcases for random.stp, string.stp, and system.stp.
David Smith [Fri, 9 Jul 2010 19:05:51 +0000 (14:05 -0500)]
Added build testcases for random.stp, string.stp, and system.stp.

* testsuite/buildok/random-embedded.stp: New testcase.
* testsuite/buildok/string-embedded.stp: Ditto.
* testsuite/buildok/system-embedded.stp: Ditto.

14 years agoClean up testsuite/systemtap.examples/network/socktop.ko left over after 'make check'.
Dave Brolley [Fri, 9 Jul 2010 18:39:17 +0000 (14:39 -0400)]
Clean up testsuite/systemtap.examples/network/socktop.ko left over after 'make check'.

14 years agoPR10080 Add vdso tracking (x86).
Mark Wielaard [Fri, 9 Jul 2010 16:02:33 +0000 (18:02 +0200)]
PR10080 Add vdso tracking (x86).

The translator will now pick up all the possible vdso.so files (luckily
they are small). At runtime when the task finder notifies us of a new
process we will check against the build id an add the vdso to the vma map.

* translate.cxx (find_vdso): New function.
  (add_unwindsym_vdso): Likewise.
  (emit_symbol_data): If vma tracker enabled call add_unwindsym_vdso.
* buildrun.cxx (compile_pass): Check STAPCONF_MM_CONTEXT_VDSO.
* runtime/autoconf-mm-context-vdso.c: New test.
* runtime/vma.c (_stp_vma_match_vdso): New function.
  (_stp_vma_exec_cb): Call _stp_vma_match_vdso when register_p && process_p.

14 years agoRemove symtab boolean field from session. staprun symtab loading is gone.
Mark Wielaard [Fri, 9 Jul 2010 10:54:52 +0000 (12:54 +0200)]
Remove symtab boolean field from session. staprun symtab loading is gone.

* session.h (systemtap_session): Remove symtab.
* session.cxx (systemtap_session::systemtap_session): Remove symtab init.

14 years ago%p really does take an int64_t argument
Josh Stone [Fri, 9 Jul 2010 01:38:44 +0000 (18:38 -0700)]
%p really does take an int64_t argument

(fixes a regression from commit 1881ea92)

14 years agoPR11785: Fix pretty-printing for tracepoint base types
Josh Stone [Fri, 9 Jul 2010 00:09:20 +0000 (17:09 -0700)]
PR11785: Fix pretty-printing for tracepoint base types

We need slightly special code to deal with tracepoint base types, which
should not be dereferenced at all, but just used as-is.

* tapsets.cxx (dwarf_pretty_print): Add deref_p, which is usually true,
  but is false for tracepoints if "pointer" is really a direct value.
  (dwarf_pretty_print::deref): Copy the value directly if !deref_p.
  (dwarf_pretty_print::recurse_pointer): Detect void pointees more
  reliably, e.g. in case of a const node with no type -> const void.
  (dwarf_cast_query::handle_query_module): @cast always has deref_p.
  (tracepoint_var_expanding_visitor::visit_target_symbol_arg): Permit
  pretty-printing on !isptr args, and use that to set deref_p.
  (tracepoint_derived_probe::resolve_tracepoint_arg_type): Check
  pointers more carefully for a pointee type, again for const void.

14 years agoInclude session::compatible in the script hash
Josh Stone [Fri, 9 Jul 2010 00:07:58 +0000 (17:07 -0700)]
Include session::compatible in the script hash

Now that the compatibility version has runtime-level effects, we need to
include it in the script hash.

* hash.cxx (find_script_hash): Add session::compatible.

14 years agoAdjust systemtap.example checks for the current behavior of "stap -l ..."
William Cohen [Thu, 8 Jul 2010 22:50:15 +0000 (18:50 -0400)]
Adjust systemtap.example checks for the current behavior of "stap -l ..."

The -l option only return an error if no matches are found. A number of
the tests would falsely pass because some of the probe points would match.
The tests need to check each probe point individually to avoid missing a
needed probe point.

14 years agoAdd _STP_REGS_USER_FLAG to regflags for hardware based perf event probes.
Mark Wielaard [Thu, 8 Jul 2010 21:55:15 +0000 (23:55 +0200)]
Add _STP_REGS_USER_FLAG to regflags for hardware based perf event probes.

* tapset-perfmon.cxx (perf_derived_probe_group::emit_module_decls): Check
  type of event and whether interrupted in user mode to set regflags.

14 years agoRemove (commented out) perfmon probes section from langref.
Mark Wielaard [Thu, 8 Jul 2010 21:45:57 +0000 (23:45 +0200)]
Remove (commented out) perfmon probes section from langref.

it's gone baby, gone

* doc/langref.text (Probes to monitor performance): Removed.

14 years agoRemove kfailed perfmon01.stp buildok test. Perfmon support is no longer.
Mark Wielaard [Thu, 8 Jul 2010 21:41:21 +0000 (23:41 +0200)]
Remove kfailed perfmon01.stp buildok test. Perfmon support is no longer.

* testsuite/buildok/perfmon01.stp: Removed.
* testsuite/systemtap.pass1-4/buildok.exp: Remove kfail for above test.
* testsuite/systemtap.server/server.exp: Likewise.

14 years agoRemove need_vma_tracker from systemtap_session, tracked through task_finder.
Mark Wielaard [Thu, 8 Jul 2010 21:25:38 +0000 (23:25 +0200)]
Remove need_vma_tracker from systemtap_session, tracked through task_finder.

See task_finder.h enable_vma_tracker() and vma_tracker_enabled().

* session.h (systemtap_session): Remove unused need_vma_tracker field.

14 years agoperfmon support is no more, remove the old tapset.
Mark Wielaard [Thu, 8 Jul 2010 21:24:52 +0000 (23:24 +0200)]
perfmon support is no more, remove the old tapset.

* tapset/perfmon.stp: Removed.

14 years agoMerge branch 'master' of ssh://sourceware.org/git/systemtap
Lukas Berk [Thu, 8 Jul 2010 21:28:47 +0000 (17:28 -0400)]
Merge branch 'master' of ssh://sourceware.org/git/systemtap

14 years agoupdated a few comments from before and changed the flow slightly
Lukas Berk [Thu, 8 Jul 2010 21:27:42 +0000 (17:27 -0400)]
updated a few comments from before and changed the flow slightly
from input from others

14 years agoPR11785: Limit pretty-printing to 32 args
Josh Stone [Thu, 8 Jul 2010 20:41:47 +0000 (13:41 -0700)]
PR11785: Limit pretty-printing to 32 args

As per PR10750, we can't have an unbounded number of printf arguments.
This change now prevents pretty-printing from generating code that would
be rejected in pass-3.  It's still likely that MAXSTRINGLEN will cause
truncation, but users are free to increase that limit if needed.

* tapsets.cxx (dwarf_pretty_print::recurse_array): Break the loop if we
  already have reached 32 args in the print_format.
  (dwarf_pretty_print::recurse_struct_members): Ditto.

14 years agoRemove the default 0-padding from %p
Josh Stone [Thu, 8 Jul 2010 18:58:14 +0000 (11:58 -0700)]
Remove the default 0-padding from %p

It is now equivalent to %#x by default, but the old behavior can be
accessed with --compatible less than 1.3.

* runtime/vsprintf.c (_stp_vsnprintf): Modify %p formatting.

14 years agoExpose session::compatible to the runtime
Josh Stone [Thu, 8 Jul 2010 18:53:01 +0000 (11:53 -0700)]
Expose session::compatible to the runtime

I'm exposing it as (MAJ<<8)+MIN, so it requires us to be a little
more strict about what form session::compatible can take.

* translate.cxx (translate_pass): Add STAP_VERSION(a,b) to canonicalize
  versions, and define STAP_COMPAT_VERSION from session::compatible.

14 years agocleanup merge artifact
Frank Ch. Eigler [Thu, 8 Jul 2010 21:14:38 +0000 (17:14 -0400)]
cleanup merge artifact

14 years agoAddressing bug 11775 which is addressing the need to audit
Lukas Berk [Thu, 8 Jul 2010 21:02:04 +0000 (17:02 -0400)]
Addressing bug 11775 which is addressing the need to audit
which tapsets are documented in tapset.tmpl and any tapsets
that may have been removed without removing the documentation.

14 years agoRename --server to --use-server and --server-status to --list-servers.
Dave Brolley [Thu, 8 Jul 2010 20:45:53 +0000 (16:45 -0400)]
Rename --server to --use-server and --server-status to --list-servers.

14 years agoFix ppc register handling for usdt.
Stan Cox [Thu, 8 Jul 2010 20:32:07 +0000 (16:32 -0400)]
Fix ppc register handling for usdt.

tapsets.cxx (sdt_uprobe_var_expanding_visitor): Special case power reg handling.

14 years agoAdd NEWS entries for new/better/faster backtraces.
Mark Wielaard [Thu, 8 Jul 2010 19:12:53 +0000 (21:12 +0200)]
Add NEWS entries for new/better/faster backtraces.

14 years agoFixed PR11661 by adding support for recvmmsg syscall in syscall tapset.
David Smith [Thu, 8 Jul 2010 15:39:56 +0000 (10:39 -0500)]
Fixed PR11661 by adding support for recvmmsg syscall in syscall tapset.

* tapset/syscalls2.stp: Added syscall.recvmmsg and syscall.compat_recvmmsg
  syscall support.

14 years agorhbz608768: revert sdt.h to "g" asm constraint
Frank Ch. Eigler [Thu, 8 Jul 2010 15:31:36 +0000 (11:31 -0400)]
rhbz608768: revert sdt.h to "g" asm constraint

14 years agoAllow scheduler.cpu_on probe point to be an inline function
William Cohen [Thu, 8 Jul 2010 14:26:18 +0000 (10:26 -0400)]
Allow scheduler.cpu_on probe point to be an inline function

This fix allow the the graphs.stp and chng_cpu.stp examples to run on RHEL-5
kernels, which inline the finish_task_switch() function.

14 years agorpm spec: Add license docs and PD license tag to -sdt-devel subrpm
Frank Ch. Eigler [Wed, 7 Jul 2010 23:03:27 +0000 (19:03 -0400)]
rpm spec: Add license docs and PD license tag to -sdt-devel subrpm

14 years agoSwitch over to version 2 sdt.h.
Stan Cox [Wed, 7 Jul 2010 21:45:33 +0000 (17:45 -0400)]
Switch over to version 2 sdt.h.

* dtrace.in (def_append):  Make STAP_SDT_V2 the default.
* sdt.h:  Likewise.
* sdt_misc.exp: Likewise.
* tapsets.cxx (sdt_uprobe_var_expanding_visitor):  Add need_debug_info.
(sdt_query::handle_query_module): Use dwarf for v2 if a variable ref needs it.

14 years agoTake X86_UNIREGS into account in runtime/unwind/x86_64.h.
Mark Wielaard [Wed, 7 Jul 2010 21:21:23 +0000 (23:21 +0200)]
Take X86_UNIREGS into account in runtime/unwind/x86_64.h.

* runtime/unwind/x86_64.h (arch_unw_init_frame_info): Copy correct register
  names based on STAPCONF_X86_UNIREGS.

14 years agoDocument the --server and --server-status options to stap.
Dave Brolley [Wed, 7 Jul 2010 21:30:28 +0000 (17:30 -0400)]
Document the --server and --server-status options to stap.

14 years agoAdd NEWS entry for new stap options --server and --server-status.
Dave Brolley [Wed, 7 Jul 2010 20:55:04 +0000 (16:55 -0400)]
Add NEWS entry for new stap options --server and --server-status.

14 years agoCorrect expected output of passes when using a server.
Dave Brolley [Wed, 7 Jul 2010 16:58:27 +0000 (12:58 -0400)]
Correct expected output of passes when using a server.

14 years agoFixed semok/entry03.stp on RHEL5 and added more rpc embedded function tests.
David Smith [Wed, 7 Jul 2010 16:43:16 +0000 (11:43 -0500)]
Fixed semok/entry03.stp on RHEL5 and added more rpc embedded function tests.

* testsuite/semok/entry03.stp: Fixed for RHEL5 systems.
* testsuite/buildok/rpc-embedded.stp: Added the rest of the rpc.stp
  embedded functions to the test.

14 years agoRefresh network/socktop and incorporate it into testing framework.
William Cohen [Wed, 7 Jul 2010 16:36:08 +0000 (12:36 -0400)]
Refresh network/socktop and incorporate it into testing framework.

This page took 0.059208 seconds and 5 git commands to generate.