]> sourceware.org Git - systemtap.git/log
systemtap.git
11 years agoActually deprecate items marked as deprecated in vfs.stp.
David Smith [Fri, 22 Feb 2013 20:34:48 +0000 (14:34 -0600)]
Actually deprecate items marked as deprecated in vfs.stp.

* tapset/linux/vfs.stp: Several items in vfs.stp had a comment saying they
  were deprecated. Acutally make it happen. Deprecate the 'ppos_pos',
  '_dev_minor', and '_dev_major' functions. Deprecate the 'page_index'
  variable in the 'vfs.block_sync_page' and 'vfs.buffer_migrate_page'
  probe aliases. Deprecate the 'write_from' and 'write_upto' variables in
  the '_vfs.block_prepare_write' and '_vfs.block_prepare_write.return'
  probe aliases.
* testsuite/buildok/vfs-embedded.stp: Handle 'ppos_pos' deprecation.
* testsuite/systemtap.examples/io/iodevstats.stp: Switch from deprecated
  '_dev_minor' and '_dev_major' functions to supported 'MINOR' and 'MAJOR'
  functions.
* NEWS: Mention deprecations.

11 years agoFix PR15173 by handling 'origin' change to 'whence'.
David Smith [Fri, 22 Feb 2013 20:14:45 +0000 (14:14 -0600)]
Fix PR15173 by handling 'origin' change to 'whence'.

* tapset/linux/syscalls.stp (syscall.lseek): Handles new kernels where
  'origin' has been renamed to 'whence'.
  (syscall.llseek): Ditto.
* tapset/linux/nfs.stp (nfs.fop.llseek): Deprecated 'origin' and added
  'whence' and 'whence_str'. Updated comment.
* tapset/linux/;vfs.stp (generic.fop.llseek): Ditto.
  (generic.fop.llseek.return): Ditto.
* tapset/linux/nd_syscalls.stp: Updated comment.
* testsuite/buildok/nfs-detailed.stp: Updated.
* testsuite/buildok/vfs-detailed.stp: Ditto.
* NEWS: Mention deprecations.

11 years agoFix PR15171 by updating the inet_get_local_port() tapset function.
David Smith [Thu, 21 Feb 2013 21:53:23 +0000 (15:53 -0600)]
Fix PR15171 by updating the inet_get_local_port() tapset function.

* tapset/linux/inet_sock.stp (inet_get_local_port): Handle kernel changes.

11 years agoKfail fortyseven.stp in testsuite/systemtap.pass1-4/buildok.exp (PR15065).
David Smith [Thu, 21 Feb 2013 21:40:01 +0000 (15:40 -0600)]
Kfail fortyseven.stp in testsuite/systemtap.pass1-4/buildok.exp (PR15065).

11 years agoFixed PR15170 by updating nfsd.stp.
David Smith [Thu, 21 Feb 2013 20:36:53 +0000 (14:36 -0600)]
Fixed PR15170 by updating nfsd.stp.

* tapset/linux/nfsd.stp (nfsd.proc4.write): When setting, 'vlen', if the
  'wr_vlen' field doesn't exist, just return 0.

11 years agopowerpc/aux_syscalls.stp: Define old/deprecated PPC_PTRACE_GET/SET[FP]REGS.
Mark Wielaard [Thu, 21 Feb 2013 13:05:22 +0000 (14:05 +0100)]
powerpc/aux_syscalls.stp: Define old/deprecated PPC_PTRACE_GET/SET[FP]REGS.

kernel commit v3.4-rc5-40-gec34a68 removed them.

Signed-off-by: Mark Wielaard <mjw@redhat.com>
11 years agonsscommon: Include config.h before all else
Josh Stone [Wed, 20 Feb 2013 18:52:56 +0000 (10:52 -0800)]
nsscommon: Include config.h before all else

On 32-bit systems, we need to ensure _FILE_OFFSET_BITS is set to 64
before any system headers are included.

11 years agotestsuite: suppress libloop.c unused-results warning
Frank Ch. Eigler [Wed, 20 Feb 2013 16:12:56 +0000 (11:12 -0500)]
testsuite: suppress libloop.c unused-results warning

11 years agotestsuite: plt.exp should add -Wno-format since it uses printf("%W")
Frank Ch. Eigler [Wed, 20 Feb 2013 15:58:29 +0000 (10:58 -0500)]
testsuite: plt.exp should add -Wno-format since it uses printf("%W")

11 years agotestsuite: move semko/fiftyone.stp to parseko/, where it belongs
Frank Ch. Eigler [Wed, 20 Feb 2013 15:57:59 +0000 (10:57 -0500)]
testsuite: move semko/fiftyone.stp to parseko/, where it belongs

11 years agotranslator: tolerate unavailable System.map
Frank Ch. Eigler [Wed, 20 Feb 2013 02:00:04 +0000 (21:00 -0500)]
translator: tolerate unavailable System.map

On some linux distros, the System.map file is unreadable to unprivileged
users.  systemtap should not overreact to this and assume that the whole
kernel-devel/header world is uninstalled.  Just print a warning and move on.

* session.cxx (systemtap_session::parse_kernel_functions): Tolerate missing
  info.

11 years agostap-report: add some ubuntu-specific data, plus a WARNING for non-root
Frank Ch. Eigler [Wed, 20 Feb 2013 01:52:55 +0000 (20:52 -0500)]
stap-report: add some ubuntu-specific data, plus a WARNING for non-root

11 years agoPR15162: Bump MAX_UNWIND_TABLE_SIZE to 16MB
Josh Stone [Tue, 19 Feb 2013 17:46:14 +0000 (09:46 -0800)]
PR15162: Bump MAX_UNWIND_TABLE_SIZE to 16MB

11 years agoPR15155 partial fix: Added more sendmmsg syscall support.
David Smith [Tue, 19 Feb 2013 17:21:38 +0000 (11:21 -0600)]
PR15155 partial fix: Added more sendmmsg syscall support.

* tapset/linux/syscalls2.stp: Fixed typo in syscall.sendmmsg.
* tapset/linux/nd_syscalls2.stp: Added nd_syscall.sendmmsg and
  nd_syscall.sendmmsg.return probe aliases.
* testsuite/buildok/nd_syscalls2-detailed.stp: Added nd_syscall.sendmmsg
  and nd_syscall.sendmmsg.return tests.
* testsuite/buildok/syscalls2-detailed.stp: Added syscall.sendmmsg and
  syscall.sendmmsg.return tests.

11 years agoPR15162: Don't output anything at all for skipped unwind tables
Josh Stone [Tue, 19 Feb 2013 01:55:46 +0000 (17:55 -0800)]
PR15162: Don't output anything at all for skipped unwind tables

* translate.cxx (dump_unwindsym_cxt_table): New common function to
  output an unwind table.  It checks against MAX_UNWIND_TABLE_SIZE and
  nulls out those that are skipped, so they're truly skipped everywhere.
  (dump_unwindsym_cxt): Use dump_unwindsym_cxt_table for all four unwind
  tables: debug_frame, debug_frame_hdr, eh_frame, and eh_frame_hdr.

11 years agoPR15155 Add syscall.sendmmsg and syscall.sendmmsg.return
Mark Wielaard [Tue, 19 Feb 2013 16:06:37 +0000 (17:06 +0100)]
PR15155 Add syscall.sendmmsg and syscall.sendmmsg.return

11 years agoPR11341: Fix two improper update_visitor assignments
Josh Stone [Sat, 16 Feb 2013 20:26:04 +0000 (12:26 -0800)]
PR11341: Fix two improper update_visitor assignments

Both are cases where a too-specific require<target_symbol> was used,
inferred by the parameter type.  The result is immediately saved as a
generic expression pointer, so require<expression> is appropriate.

I believe this case is such that there was actually no harm caused here
by the formerly blithe static_cast, but it's still nice to see that such
improprieties are now detected.

11 years agoPR11341: Add type safety to update_visitor::require()
Josh Stone [Sat, 16 Feb 2013 03:05:25 +0000 (19:05 -0800)]
PR11341: Add type safety to update_visitor::require()

This uses dynamic_cast<T*> to make a runtime check that the pointer that
was given to update_visitor::provide() is compatible with what the
caller required.  Such a cast can't be used on void*, so the ::stack is
now composed of 'visitable' pointers - a new base class for expressions
and statements.

In the process, 'indexable' is now also brought into the expression
hierarchy.  This looks a little odd since hist_op isn't really an
expression on its own, but otherwise this change untangles a lot of code
trying to deal with hist_op as an outlier, especially visit and print.

* staptree.h (update_visitor::require): Use dynamic_cast<> instead of a
  dangerous static_cast<>, and throw exceptions for bad cases.
  (struct visitable): New base class for expression and statement.
  (struct indexable): Use expression as a base class, and get rid of the
  now-unnecessary visit/print/tok middleman methods.
* staptree.cxx: Use direct indexable->visit/print/tok throughout.
* elaborate.cxx, parse.cxx: Ditto.

11 years agosystemtap.spec: hard-code systemtap instead of most usese of %{name}
Frank Ch. Eigler [Fri, 15 Feb 2013 21:41:19 +0000 (16:41 -0500)]
systemtap.spec: hard-code systemtap instead of most usese of %{name}

... since we'll never change the project name, ever, ever, ever.
(And those special software-collection builds that do change %{name},
we don't want effecting these derived rpm strings, mostly.)

11 years agoStill install systemtap rc.d files when systemd is enabled
Lukas Berk [Fri, 15 Feb 2013 21:44:23 +0000 (16:44 -0500)]
Still install systemtap rc.d files when systemd is enabled

*systemtap.spec: remove systemtap portions from 'with_systemd'
 conditionals so they are still installed even if
 stap-server is using systemd.  They should not be
 combined.

11 years agoloc2c-test lists functions and associated addresses if no address given
Lukas Berk [Fri, 15 Feb 2013 20:04:40 +0000 (15:04 -0500)]
loc2c-test lists functions and associated addresses if no address given

*loc2c-test: if no address is given, list all functions with their
     dwarf_entrypc address along with any
             dwarf_entry_breakpoints

11 years agoregex: Declare the dfa static and its input const
Josh Stone [Thu, 14 Feb 2013 18:00:15 +0000 (10:00 -0800)]
regex: Declare the dfa static and its input const

11 years agoAbstract the synthesizing of deref functions
Josh Stone [Thu, 14 Feb 2013 03:54:13 +0000 (19:54 -0800)]
Abstract the synthesizing of deref functions

Frank first noted the duplication in how we synthesize functions for
derefs, and I found there were four sites doing nearly the same thing.
A quick bit of abstraction relieves this.

* tapsets.cxx (synthetic_embedded_deref_call): New common code to create
  a functiondecl for a code block and the functioncall to use it.
  (dwarf_pretty_print::deref, dwarf_cast_query::handle_query_module,
   tracepoint_var_expanding_visitor::visit_target_symbol_arg,
   dwarf_var_expanding_visitor::visit_target_symbol): Use it.

11 years agoPR15147: clarify comments about _stp_error internal runtime function
Frank Ch. Eigler [Wed, 13 Feb 2013 23:09:02 +0000 (18:09 -0500)]
PR15147: clarify comments about _stp_error internal runtime function

11 years agoPR15148, speculatively drop the PR15123-conflicting assert
Frank Ch. Eigler [Wed, 13 Feb 2013 22:56:52 +0000 (17:56 -0500)]
PR15148, speculatively drop the PR15123-conflicting assert

11 years agoPR15044 partial fix: Raise real errors instead of returning error texts.
David Smith [Wed, 13 Feb 2013 22:28:22 +0000 (16:28 -0600)]
PR15044 partial fix: Raise real errors instead of returning error texts.

* tapset/dyninst/ctime.stp: Instead of returning "cutesy" error texts,
  raise a real error.
* tapset/linux/context-symbols.stp: Ditto.
* tapset/linux/ctime.stp: Ditto.
* testsuite/systemtap.base/ctime.exp: Make test handle "cutesy" error
  tests or real errors.
* testsuite/systemtap.base/ctime.stp: Ditto.
* NEWS: Mention deprecations.

11 years agosystemtap.spec: cover up tracks of time travel
Frank Ch. Eigler [Wed, 13 Feb 2013 19:54:25 +0000 (14:54 -0500)]
systemtap.spec: cover up tracks of time travel

11 years agoversion bumps for SystemTap 2.2 development cycle
Serguei Makarov [Wed, 13 Feb 2013 18:15:59 +0000 (13:15 -0500)]
version bumps for SystemTap 2.2 development cycle

11 years agoMISSING FOR 2.1 -- %changelog entry for systemtap.spec
Serguei Makarov [Wed, 13 Feb 2013 17:47:09 +0000 (12:47 -0500)]
MISSING FOR 2.1 -- %changelog entry for systemtap.spec

11 years agorelease date in NEWS release-2.1
Serguei Makarov [Wed, 13 Feb 2013 16:36:11 +0000 (11:36 -0500)]
release date in NEWS

11 years agomake update-po for release-2.1
Serguei Makarov [Wed, 13 Feb 2013 15:47:08 +0000 (10:47 -0500)]
make update-po for release-2.1

11 years agoPR15101, dummy commit to correct authorship credit
Lukáš Hejtmánek [Wed, 13 Feb 2013 15:24:58 +0000 (10:24 -0500)]
PR15101, dummy commit to correct authorship credit

11 years agotestsuite: weaken pr10854.exp wildcards due to less stable 3.8.0-rc7 kprobes
Frank Ch. Eigler [Wed, 13 Feb 2013 06:28:26 +0000 (01:28 -0500)]
testsuite: weaken pr10854.exp wildcards due to less stable 3.8.0-rc7 kprobes

See also https://bugzilla.redhat.com/show_bug.cgi?id=910649

11 years agotcp tapset: adjust to inet_sock->dport move
Frank Ch. Eigler [Wed, 13 Feb 2013 05:25:52 +0000 (00:25 -0500)]
tcp tapset: adjust to inet_sock->dport move

As per linux commit ce43b03e88, ->inet_dport
was moved to                    ->sk->__sk_common->skc_dport

11 years agotestsuite: use /usr/bin/sleep instead of tcl [after NN] to sleep
Frank Ch. Eigler [Wed, 13 Feb 2013 04:36:34 +0000 (23:36 -0500)]
testsuite: use /usr/bin/sleep instead of tcl [after NN] to sleep

... for the former tolerates jumpy wall-clock time, the latter doesn't.

11 years agoscript examples: regen indexes
Frank Ch. Eigler [Wed, 13 Feb 2013 03:51:32 +0000 (22:51 -0500)]
script examples: regen indexes

11 years agoPR15123: work around broken gcc debuginfo for -mfentry functions
Frank Ch. Eigler [Wed, 13 Feb 2013 02:32:30 +0000 (21:32 -0500)]
PR15123: work around broken gcc debuginfo for -mfentry functions

http://gcc.gnu.org/bugzilla/show_bug.cgi?id=54793 interferes with the
calm zen-like state of probing function parameters in programs built
with CFLAGS=-mfentry, due to a dwarf location-list problem.  This
patch adds a heuristic function dwflpp::pr15123_retry_addr() that
offers callers a post-mfentry-address alternative for locating
variables.  Will they use it, or not?  Join us next week, as our
heroes encounter ... angry mr. dwarf, five steps away.

* dwflpp.cxx (pr15123_retry_addr): New function.
  (translate_location): Call it, in order to enable $var
  and $$parms resolution.
* tapsets.cxx (saveargs): Call it, in order to enable
  stap -L <probe> listings.

11 years agovim: Expand target matching
Josh Stone [Tue, 12 Feb 2013 22:33:49 +0000 (14:33 -0800)]
vim: Expand target matching

- Allow multiple '$' prefix characters for e.g. $$vars
- Allow '$' suffix characters for pretty-printing

11 years agoNEWS: Brag about more stapdyn improvements
Josh Stone [Tue, 12 Feb 2013 22:19:27 +0000 (14:19 -0800)]
NEWS: Brag about more stapdyn improvements

11 years agoAdd missing 'output' fields to couple of systemtap.examples.
Serguei Makarov [Tue, 12 Feb 2013 21:43:06 +0000 (16:43 -0500)]
Add missing 'output' fields to couple of systemtap.examples.

11 years agoRHBZ 906061, PR15112: Fix handling of scope specifiers/ids for IPv6 addresses.
Dave Brolley [Tue, 12 Feb 2013 19:33:26 +0000 (14:33 -0500)]
RHBZ 906061, PR15112: Fix handling of scope specifiers/ids for IPv6 addresses.

- Do not assume a scope specifier of "%eth0" when one is not specified.
- Copy the scope_id to addresses which don't have one when merging server
  information.
- Provide additional diagnostic when a link-local IPv6 address is used
  without a scope specifier.
- Don't reqauire a server specified by address and port to be previously known
  by avahi.
- Add informational messages tracing server resolution at various -v levels.

11 years agoData object (-d) cannot be empty.
Mark Wielaard [Tue, 12 Feb 2013 16:05:07 +0000 (17:05 +0100)]
Data object (-d) cannot be empty.

11 years agostap: Add shorthand option --dyninst for --runtime=dyninst
Josh Stone [Mon, 11 Feb 2013 22:30:48 +0000 (14:30 -0800)]
stap: Add shorthand option --dyninst for --runtime=dyninst

* cmdline.cxx: Define --dyninst as LONG_OPT_RUNTIME_DYNINST.
* session.cxx (systemtap_session::parse_cmdline_runtime): Pull out the
  code to set the runtime option.
  (systemtap_session::parse_cmdline): Use parse_cmdline_runtime for both
  LONG_OPT_RUNTIME and LONG_OPT_RUNTIME_DYNINST.
  (systemtap_session::usage): Document --dyninst.
* man/stap.1: Document --dyninst.

11 years agostapdyn documentation updates
Josh Stone [Mon, 11 Feb 2013 22:09:45 +0000 (14:09 -0800)]
stapdyn documentation updates

* stapdyn/README: Refresh with more current information.
* man/error::pass5.7stap: Add a note about Dyninst vs exceptions.

11 years agostapdyn: Initialize _stp_random_u from /dev/urandom
Josh Stone [Sat, 9 Feb 2013 01:59:53 +0000 (17:59 -0800)]
stapdyn: Initialize _stp_random_u from /dev/urandom

11 years agostapdyn: Validate that probe flags are known
Josh Stone [Fri, 8 Feb 2013 20:58:56 +0000 (12:58 -0800)]
stapdyn: Validate that probe flags are known

If we see unknown flags, we don't want to do the wrong thing with a
probe.  So just warn about it and move on without it.

* runtime/dyninst/stapdyn.h: Introduce STAPDYN_PROBE_ALL_FLAGS combining
  all the currently-known flags.  Also, shift the values of the utracey
  flags to leave some room for more uprobey flags.
* stapdyn/dynprobe.cxx (dynprobe_location::validate): New, make sanity
  check on the probe.  For now, this just means checking that the flags
  are within those we know about.

11 years agoPR11334: last-minute character class tweaks.
Serguei Makarov [Fri, 8 Feb 2013 19:24:22 +0000 (14:24 -0500)]
PR11334: last-minute character class tweaks.

These fix a couple of errors in how small inverted character classes were handled.
e.g. [^a] and such

11 years agostap-report: don't heartlessly destroy incoming $PATH
Frank Ch. Eigler [Fri, 8 Feb 2013 12:32:53 +0000 (07:32 -0500)]
stap-report: don't heartlessly destroy incoming $PATH

11 years agoerror::pass5, mention more permission-related causes for failure
Frank Ch. Eigler [Thu, 7 Feb 2013 20:37:57 +0000 (15:37 -0500)]
error::pass5, mention more permission-related causes for failure

11 years agostapdyn: Make sure interprocess maps share hash seeds
Josh Stone [Fri, 8 Feb 2013 00:43:21 +0000 (16:43 -0800)]
stapdyn: Make sure interprocess maps share hash seeds

We were using a random stap_hash_seed that was initialized separately in
each process.  This meant that different processes would compute
different hashes for the same keys.  Now the hash seed is stored in the
shared memory, so all participating processes can use maps together.

* runtime/dyninst/common_session_state.h: Add the hash seed to struct
  stp_runtime_session, and provide _stap_hash_seed() to read it.
  Initialize it one time only in stp_session_init().
* runtime/dyninst/runtime.h: Define the former stap_hash_seed to use
  _stap_hash_seed() from shared memory instead.
* testsuite/systemtap.maps/map_hash_interprocess.*: Test the specific
  interprocess, manually-keyed conditions that revealed the bug.

11 years agostapdyn: Seed _stp_random_u with more "precision"
Josh Stone [Thu, 7 Feb 2013 23:42:47 +0000 (15:42 -0800)]
stapdyn: Seed _stp_random_u with more "precision"

Use clock_gettime() rather than time().  It's still not necessarily
great to seed from a time-based source, but at least nanosecond
resolution is a little better.

11 years agoPR14706: Add proto-exec support in stapdyn
Josh Stone [Thu, 7 Feb 2013 21:02:30 +0000 (13:02 -0800)]
PR14706: Add proto-exec support in stapdyn

This adds preliminary support for stapdyn to follow execs, but the code
it instrument the new process is currently blocked out.  I'm seeing
dyninst hang during loadLibrary, waiting for IRPC, but haven't yet
figured out why.

* stapdyn/mutatee.cxx (mutatee_freezer): New helper to automatically
  stop a process and then continue it when leaving that scope.
  (mutatee::begin_callback): Use a mutatee_freezer.
  (mutatee::thread_callback): Use a mutatee_freezer.
  (mutatee::copy_forked_instrumentation): Use two mutatee_freezers.
  (mutatee::instrument_dynprobes): Support a flag to indicate this is
  post-exec, so first run any process.end probes from before exec.
  (mutatee::exec_reset_instrumentation): New, clear all members that are
  invalidated by the process exec.
* stapdyn/mutator.cxx (g_exec_callback): New global callback for execs.
  (mutator::exec_callback): New specific callback for execs.  Clear old
  instrumentation, then instrument it anew (buggy, #if 0 for now).

11 years agoFix typo in sdt_misc.exp to avoid error for unsupported test.
William Cohen [Thu, 7 Feb 2013 18:31:52 +0000 (13:31 -0500)]
Fix typo in sdt_misc.exp to avoid error for unsupported test.

11 years agoClean up mutatee::stop_execution() logic.
David Smith [Thu, 7 Feb 2013 17:52:06 +0000 (11:52 -0600)]
Clean up mutatee::stop_execution() logic.

* stapdyn/mutatee.cxx (mutatee): Removed unneeded 'patch' argument.
  (stop_execution): Remove logic that waited on the process to stop, since
  process->stopExecution() is synchronous.
* stapdyn/mutatee.h: Update prototypes.
* stapdyn/mutator.cxx: Update mutatee constructor calls.

11 years agoPR11334: regex example tweaked to take a parameter
Serguei Makarov [Thu, 7 Feb 2013 15:21:08 +0000 (10:21 -0500)]
PR11334: regex example tweaked to take a parameter

11 years agostap-report: collect all output on stdout, not stderr
Frank Ch. Eigler [Thu, 7 Feb 2013 15:21:31 +0000 (10:21 -0500)]
stap-report: collect all output on stdout, not stderr

11 years agoRemove a stray space in _stp_saved_errno
Josh Stone [Wed, 6 Feb 2013 23:16:33 +0000 (15:16 -0800)]
Remove a stray space in _stp_saved_errno

11 years agoAUTHORS bump
Frank Ch. Eigler [Wed, 6 Feb 2013 22:30:51 +0000 (17:30 -0500)]
AUTHORS bump

11 years agostap-report: add nss+etc to rpm version list
Frank Ch. Eigler [Wed, 6 Feb 2013 21:08:08 +0000 (16:08 -0500)]
stap-report: add nss+etc to rpm version list

11 years agoSave and restore 'errno' in dyninst probes.
David Smith [Wed, 6 Feb 2013 22:18:23 +0000 (16:18 -0600)]
Save and restore 'errno' in dyninst probes.

* tapsets.cxx (common_probe_entryfn_prologue): In dyninst mode, save errno
  at the start of the probe.
  (common_probe_entryfn_epilogue): In dyninst mode, restore errno at the
  end of the probe.

11 years agoWarn about unbalanced attempts to shutdown NSS.
Dave Brolley [Wed, 6 Feb 2013 21:29:44 +0000 (16:29 -0500)]
Warn about unbalanced attempts to shutdown NSS.

Warn about attempts to shut down NSS against a certificate database
which has not been initialized.

11 years agoEnsure that NSS has been initialized before attempting to shut it down.
Jiri Horky [Wed, 6 Feb 2013 21:26:41 +0000 (16:26 -0500)]
Ensure that NSS has been initialized before attempting to shut it down.

11 years agoMerge branch 'master' of ssh://sourceware.org/git/systemtap
David Smith [Wed, 6 Feb 2013 21:07:48 +0000 (15:07 -0600)]
Merge branch 'master' of ssh://sourceware.org/git/systemtap

11 years agoPR14574 partial fix: Add dyninst support to warnings2.exp.
David Smith [Wed, 6 Feb 2013 21:07:24 +0000 (15:07 -0600)]
PR14574 partial fix: Add dyninst support to warnings2.exp.

11 years agoPR11334: add regex.stp to systemtap.examples
Serguei Makarov [Wed, 6 Feb 2013 21:03:42 +0000 (16:03 -0500)]
PR11334: add regex.stp to systemtap.examples

11 years agoPR14574 partial fix: Add dyninst testcase support.
David Smith [Wed, 6 Feb 2013 20:32:21 +0000 (14:32 -0600)]
PR14574 partial fix: Add dyninst testcase support.

* testsuite/systemtap.base/unresolved-struct-typedef.exp: Add dyninst
  support.
* testsuite/systemtap.base/uprobes.exp: Ditto.
* testsuite/systemtap.base/utf_pretty.exp: Ditto.
* testsuite/systemtap.base/utf_user.exp: Ditto.
* testsuite/systemtap.base/utf_user_trunc.exp: Ditto.
* testsuite/systemtap.base/vta-test.exp: Ditto.

11 years agoAccess user strings with 'user_string_*' in systemtap.base/utf_user.stp.
David Smith [Wed, 6 Feb 2013 20:27:52 +0000 (14:27 -0600)]
Access user strings with 'user_string_*' in systemtap.base/utf_user.stp.

11 years agoPR14574 partial fix: Add dyninst testcase support.
David Smith [Wed, 6 Feb 2013 15:28:59 +0000 (09:28 -0600)]
PR14574 partial fix: Add dyninst testcase support.

* testsuite/systemtap.base/rand.exp: Add dyninst support.
* testsuite/systemtap.base/sdt.exp: Ditto.
* testsuite/systemtap.base/sdt_misc.exp: Ditto.
* testsuite/systemtap.base/sdt_va_args.exp: Ditto.
* testsuite/systemtap.base/simple.exp: Ditto.
* testsuite/systemtap.base/stmt_rel_user.exp: Ditto.
* testsuite/systemtap.base/suppress-time-limit.exp: Ditto.
* testsuite/systemtap.base/tri.exp: Ditto.
* testsuite/systemtap.base/trycatch.exp: Ditto.

11 years agoAdd kprobe.function.call, and squash nd_syscall.exit.return
Josh Stone [Tue, 5 Feb 2013 21:17:16 +0000 (13:17 -0800)]
Add kprobe.function.call, and squash nd_syscall.exit.return

* tapsets.cxx (register_standard_tapsets): Bind kprobe .call variants.
* man/stapprobes.3stap: Document this new possibility.
* tapset/linux/nd_syscalls.stp: Use .call to make sure that alias
  suffixes can't tack a .return on nd_syscall.exit and .exit_group.

11 years agoDisregard all pass-2 errors in listing mode
Josh Stone [Tue, 5 Feb 2013 20:34:13 +0000 (12:34 -0800)]
Disregard all pass-2 errors in listing mode

* elaborate.cxx (semantic_pass): In listing mode, the only rc we care
  about is whether any probes were discovered.

11 years agoAdd [nd_]syscall.ustat32.return
Josh Stone [Tue, 5 Feb 2013 19:31:37 +0000 (11:31 -0800)]
Add [nd_]syscall.ustat32.return

This adds the proper .return aliases for the syscall.ustat32 entry
aliases, rather than having them as part of syscall.ustat.return, so
way the entry/return probes are properly balanced.  It also makes sure
that wildcards like syscall.ustat** don't end up in an incomplete
ustat.return probe due to implicit alias suffixes.

I considered wrapping this with version-compatibility checks, but IMO
the former behavior was broken enough that it isn't worth it.

* tapset/linux/syscalls2.stp: Split out syscall.ustat32.return.
* tapset/linux/nd_syscalls2.stp: Split out nd_syscall.ustat32.return.
* testsuite/buildok/syscalls2-detailed.stp: Test ustat32.return.
* testsuite/buildok/nd_syscalls2-detailed.stp: Test ustat32.return.

11 years agoPR11334: use systemtap error reporting for regex functionality
Serguei Makarov [Tue, 5 Feb 2013 21:00:52 +0000 (16:00 -0500)]
PR11334: use systemtap error reporting for regex functionality

11 years agoIndented and cleaned up testsuite/systemtap.base/sdt_misc.exp.
David Smith [Tue, 5 Feb 2013 20:26:04 +0000 (14:26 -0600)]
Indented and cleaned up testsuite/systemtap.base/sdt_misc.exp.

11 years agoGet syscall tapsets working on ppc64 RHEL4 by providing missing defines.
David Smith [Tue, 5 Feb 2013 18:58:12 +0000 (12:58 -0600)]
Get syscall tapsets working on ppc64 RHEL4 by providing missing defines.

* tapset/linux/powerpc/aux_syscalls.stp: Add missing PTRACE_* defines.

11 years agoAs part of PR11955, update the signal-handle.stp buildok testcase.
David Smith [Mon, 4 Feb 2013 18:54:22 +0000 (12:54 -0600)]
As part of PR11955, update the signal-handle.stp buildok testcase.

* testsuite/buildok/signal-handle.stp: Handle 'oldset_addr' and 'regs'
  deprecation.

11 years agoFix up documention comment for signal.send.return so docproc can parse it.
William Cohen [Mon, 4 Feb 2013 15:45:09 +0000 (10:45 -0500)]
Fix up documention comment for signal.send.return so docproc can parse it.

11 years agore2c: don't implement strndup
Frank Ch. Eigler [Fri, 1 Feb 2013 20:51:59 +0000 (15:51 -0500)]
re2c: don't implement strndup

Linux has had it forever; drop the conditional and fallback
implementation.

11 years agoperf.counter probe warnings.
Stan Cox [Fri, 1 Feb 2013 19:58:21 +0000 (14:58 -0500)]
perf.counter probe warnings.

* tapset-perfmon.cxx (statement_counter): New.
  (perf_builder::build):  Add nonempty block warning. Add if..next to block.

11 years agoSmall RHEL4 compile fix.
David Smith [Fri, 1 Feb 2013 17:12:32 +0000 (11:12 -0600)]
Small RHEL4 compile fix.

11 years agoFixed PR11955 by updating the signal tapset.
David Smith [Fri, 1 Feb 2013 17:01:40 +0000 (11:01 -0600)]
Fixed PR11955 by updating the signal tapset.

* tapset/linux/signal.stp (signal.send): Now uses the 'signal_generate'
  tracepoint where available. Deprecated the 'send2queue' variable.
  (signal.send.return): Deprecated.
  (signal.handle): Now uses the 'signal_deliver' tracepoint where
  available. Deprecated the 'oldset_addr' and 'regs' variables.
  (signal.handle.return): Deprecated.
* testsuite/buildok/signal-detailed.stp: Updated with 'signal.send',
  'signal.send.return', and 'signal.handle.return' deprecations.
* NEWS: Added deprecation notes.

11 years agostap.1: stats arrays do support %; drop comment about its absence
Frank Ch. Eigler [Fri, 1 Feb 2013 16:06:12 +0000 (11:06 -0500)]
stap.1: stats arrays do support %; drop comment about its absence

11 years agostap.1 man page: reword/remention array % (wrapping) decoration
Frank Ch. Eigler [Fri, 1 Feb 2013 15:56:32 +0000 (10:56 -0500)]
stap.1 man page: reword/remention array % (wrapping) decoration

11 years ago(PR6971) Mark the nd_syscall.exp tests as kfails on ia64.
David Smith [Thu, 31 Jan 2013 22:36:35 +0000 (16:36 -0600)]
(PR6971) Mark the nd_syscall.exp tests as kfails on ia64.

* testsuite/systemtap.syscall/nd_syscall.exp (test_procedure): Make the
  tests as kfails for ia64.

11 years agoPR11576: simplify runtime *perf_counters[] widget
Frank Ch. Eigler [Thu, 31 Jan 2013 17:29:36 +0000 (12:29 -0500)]
PR11576: simplify runtime *perf_counters[] widget

Take away unnecessary indirection; *long[] -> *long.

11 years agoNEWS: don't forget re2c project acknowledgement
Serguei Makarov [Thu, 31 Jan 2013 16:09:35 +0000 (11:09 -0500)]
NEWS: don't forget re2c project acknowledgement

11 years agoPR14574 partial: Let stapdyn tests run despite the kernel
Josh Stone [Thu, 31 Jan 2013 02:52:09 +0000 (18:52 -0800)]
PR14574 partial: Let stapdyn tests run despite the kernel

Some tests are guarded by [uprobes_p] or [uretprobes_p] before they're
allowed to run, but --runtime=dyninst cares not for such features.
Rework these tests so the kernel features are more tightly considered
only right before running in kernel mode.

11 years agoPR14705: Modify SDT semaphores in stapdyn
Josh Stone [Wed, 30 Jan 2013 20:20:41 +0000 (12:20 -0800)]
PR14705: Modify SDT semaphores in stapdyn

For SDT probes which are guarded by a semaphore, stapdyn needs to update
the value to show when the probes are active.

* stapdyn/mutatee.h (mutatee::semaphores): New vector of variables in
  this mutatee which represent active semaphores.
* stapdyn/mutatee.cxx (mutatee::update_semaphores): New, used to
  increment or decrement the value of semaphores in this mutatee.
  (mutatee::instrument_dynprobe_target): When a probe has a semaphore,
  create a variable representation of it.
  (mutatee::instrument_object_dynprobes): Increment semaphores in bulk.
  (mutatee::copy_forked_instrumentation): Copy semaphore variables too.
  (mutatee::remove_instrumentation): Decrement semaphores in bulk.
* tapsets.cxx (sdt_query::handle_query_module): Calculate the difference
  in load addresses between .text and .data as semaphore_load_offset.
  (sdt_query::record_semaphore): Subtract the semaphore_load_offset for
  dyninst to get a file-based offset of the semaphore.

11 years agonetfilter tapset: fix arp ar_tha & ar_tpa fields
Frank Ch. Eigler [Thu, 31 Jan 2013 00:20:43 +0000 (19:20 -0500)]
netfilter tapset: fix arp ar_tha & ar_tpa fields

There was a typo in decoding the arp packets that duplicated source &
target addresses.

11 years agoAdd better cleanup to a setjmp.exp and statement.exp.
David Smith [Wed, 30 Jan 2013 21:11:48 +0000 (15:11 -0600)]
Add better cleanup to a setjmp.exp and statement.exp.

* testsuite/systemtap.base/setjmp.exp: Cleanup test programs. Allow the
  user to override this using the '-v' vebose flag.
* testsuite/systemtap.base/statement.exp: Ditto.

11 years agotestuite: comment typo fix
Frank Ch. Eigler [Wed, 30 Jan 2013 16:39:09 +0000 (11:39 -0500)]
testuite: comment typo fix

11 years agoPR15080 fixed by handling signal_wake_up_state() replacing signal_wake_up().
David Smith [Wed, 30 Jan 2013 16:29:29 +0000 (10:29 -0600)]
PR15080 fixed by handling signal_wake_up_state() replacing signal_wake_up().

* runtime/stp_utrace.c (stp_signal_wake_up): New function.
  (utrace_init): Lookup the address of signal_wake_up_state().
  (utrace_control): Call stp_signal_wake_up().
* buildrun.cxx (compile_pass): Add export test for 'signal_wake_up_state'.
* runtime/linux/runtime.h: Added 'kallsyms_signal_wake_up_state' declaration.
* tapset/linux/signal.stp: Handle signal_wake_up_state() replacing
  signal_wake_up() in newer kernels for the 'signal.wakeup' probe alias.

11 years agoerror::pass4 man page: menton STAP_ARG* changes
Frank Ch. Eigler [Wed, 30 Jan 2013 15:59:07 +0000 (10:59 -0500)]
error::pass4 man page: menton STAP_ARG* changes

11 years agotestsuite: conditionalize client-server tests on feature: NSS
Frank Ch. Eigler [Wed, 30 Jan 2013 14:41:34 +0000 (09:41 -0500)]
testsuite: conditionalize client-server tests on feature: NSS

* testsuite/lib/systemtap.exp (nss_p): New test proc.
  (dyninst_p): Switch to globals instead of $env().
  (setup_server): Use it.
* testsuite/systemtap.server/*.exp: Do an [nss_p] test early on.

11 years agotestsuite: fix semko/badregex01.stp so it passes parseok
Frank Ch. Eigler [Wed, 30 Jan 2013 00:11:32 +0000 (19:11 -0500)]
testsuite: fix semko/badregex01.stp so it passes parseok

11 years agotestsuite: pass -w to tracepoints.exp pass-5 test
Frank Ch. Eigler [Tue, 29 Jan 2013 23:59:17 +0000 (18:59 -0500)]
testsuite: pass -w to tracepoints.exp pass-5 test

... since it may print warnings that we don't want to see

11 years agotestsuite: tweak bz10078 test case to use noinline functions
Frank Ch. Eigler [Tue, 29 Jan 2013 22:03:42 +0000 (17:03 -0500)]
testsuite: tweak bz10078 test case to use noinline functions

11 years agotestsuite: fix systemtap.examples/check.exp to properly CHDIR back & forth
Frank Ch. Eigler [Tue, 29 Jan 2013 21:25:03 +0000 (16:25 -0500)]
testsuite: fix systemtap.examples/check.exp to properly CHDIR back & forth

A failed test_check fragment (such as from virtualization/*.meta)
could result in loss of the original pwd=$builddir/testsuite, because
the loop neglected to restore the old pwd ($curdir) before
"continue"'ing.  We now save & restore $pwd outside the loop in
addition to the intra-loop cwd twitchery.

11 years agoexamples index regen
Frank Ch. Eigler [Tue, 29 Jan 2013 21:24:54 +0000 (16:24 -0500)]
examples index regen

11 years agoPR12231: use literal_string parsers/types in @perf() parameters
Frank Ch. Eigler [Tue, 29 Jan 2013 17:10:25 +0000 (12:10 -0500)]
PR12231: use literal_string parsers/types in @perf() parameters

Type-casting is so passe.

11 years agore2c: fix some possibly-uninit-variable compiler warnings
Frank Ch. Eigler [Tue, 29 Jan 2013 17:09:12 +0000 (12:09 -0500)]
re2c: fix some possibly-uninit-variable compiler warnings

This page took 0.065693 seconds and 5 git commands to generate.