]> sourceware.org Git - systemtap.git/log
systemtap.git
15 years agoFix kprobes.exp.
Ananth N Mavinakayanahalli [Mon, 4 May 2009 10:09:23 +0000 (15:39 +0530)]
Fix kprobes.exp.
Patch from Prerna Saxena.

15 years agoAdd manual for stringat function
Wenji Huang [Sun, 3 May 2009 23:19:32 +0000 (19:19 -0400)]
Add manual for stringat function

15 years agosystemtap.syscall tests do use + and * as metachars.
Mark Wielaard [Sun, 3 May 2009 06:05:43 +0000 (08:05 +0200)]
systemtap.syscall tests do use + and * as metachars.

+ and * are metacharacters, but should always be used
as metacharacters in the expressions, don't escape them.

* testsuite/systemtap.syscall/test.tcl (run_one_test): Remove escaping
  of + and *.

15 years agoDon't recurse into DW_AT_declaration subprograms in iterate_over_labels.
Mark Wielaard [Sun, 3 May 2009 04:49:02 +0000 (06:49 +0200)]
Don't recurse into DW_AT_declaration subprograms in iterate_over_labels.

Newer gcc (4.4) places more DW_TAG_subprograms with DW_AT_declaration
at call sites. iterate_over_labels should only be concerned about
complete declarations.

* tapsets.cxx (dwflpp::iterate_over_labels): Skip DW_TAG_subprogram
  that has attribute DW_AT_declaration.

15 years agoFix funtion -> function typo in alternatives.exp.
Mark Wielaard [Sat, 2 May 2009 15:52:50 +0000 (17:52 +0200)]
Fix funtion -> function typo in alternatives.exp.

* testsuite/systemtap.base/alternatives.exp (local1_script): Fix typo.

15 years agosyscalls testsuite: use different escape sequence
Frank Ch. Eigler [Thu, 30 Apr 2009 21:07:43 +0000 (17:07 -0400)]
syscalls testsuite: use different escape sequence

The syscall testsuite uses embedded "// TEXT" markers in the .c files
to designate expected output, kind of like the "dg-" bunch in the gcc
test suite.  Unfortunately, "//" is not a unique prefix to systemtap
tests, and in particular it can occur in the system headers that will
be picked up with the "gcc -E -C ..." invocation in test.tcl.

So let's switch to "//staptest//".  test.tcl is also modified to
escape a few more mischevious regexp metacharacters that might sneak
past.

15 years agowhitespace tweak in tcl code
Frank Ch. Eigler [Thu, 30 Apr 2009 20:54:04 +0000 (16:54 -0400)]
whitespace tweak in tcl code

15 years agoadd verboser strerrors if posix_spawn failed
Frank Ch. Eigler [Thu, 30 Apr 2009 20:43:41 +0000 (16:43 -0400)]
add verboser strerrors if posix_spawn failed

* buildrun.cxx (run_make_cmd, kernel_built_uprobes,
  copy_uprobes_symbols, run_pass, make_typequery_umod):
  Print strerror after a failed stap_system() if verbose.

15 years agoFixed uaddr tapset function for ppc64 by adding task_pt_regs() definition.
David Smith [Thu, 30 Apr 2009 15:20:41 +0000 (10:20 -0500)]
Fixed uaddr tapset function for ppc64 by adding task_pt_regs() definition.
PR10117 fix.
* tapset/context.stp: Added ppc64 task_pt_regs() definition if it dooesn't
  already exist.

15 years agoPR10007: Avoid probing syscall entry points in the testsuite.
Ananth N Mavinakayanahalli [Thu, 30 Apr 2009 11:30:38 +0000 (17:00 +0530)]
PR10007: Avoid probing syscall entry points in the testsuite.
While there, fix minor issues with the s390x syscall tapset.

15 years agoadded tcp_connections.stp
ddomingo [Thu, 30 Apr 2009 05:56:47 +0000 (15:56 +1000)]
added tcp_connections.stp

15 years agoUse sockaddr instead of timeval in @cast test
Josh Stone [Wed, 29 Apr 2009 21:00:21 +0000 (14:00 -0700)]
Use sockaddr instead of timeval in @cast test

Using timeval had problems on big-endian multi-arch platforms (ppc64),
because the user tv_sec used in the @cast didn't match the kernel tv_sec
used to provide a pointer.  Hopefully reading from a sockaddr should be
more robust, as that type doesn't appear to need any compat wrappers for
multi-archs.

15 years agoPR 6930: fix flightrec2.exp to handle filesize correctly
Masami Hiramatsu [Tue, 28 Apr 2009 15:18:06 +0000 (11:18 -0400)]
PR 6930: fix flightrec2.exp to handle filesize correctly

* testsuite/systemtap.base/flightrec2.exp: Use stat instead of ls for checking
  file size.
* testsuite/systemtap.base/flightrec2.stp: Increase timer interval for some
  architecture on which the minimum interval is more than 1 ms (e.g. xen)

15 years agoPR10102: tolerate mismatched optional probe
Wenji Huang [Tue, 28 Apr 2009 02:35:05 +0000 (22:35 -0400)]
PR10102: tolerate mismatched optional probe

* elaborate.cxx: Early return for mismatched optional probe.
* testsuite/systemtap.base/optionalprobe.exp: New test case.
* testsuite/systemtap.base/optionalprobe.stp: Ditto.

15 years agonew AUTHOR; fixed my commit --author typo via .mailmap
Frank Ch. Eigler [Mon, 27 Apr 2009 23:21:06 +0000 (19:21 -0400)]
new AUTHOR; fixed my commit --author typo via .mailmap

15 years agotraceio: add human-readable byte-count output
Key Meyer [Mon, 27 Apr 2009 23:12:14 +0000 (19:12 -0400)]
traceio: add human-readable byte-count output

15 years agotraceio sample: tolerate more than a few hundred processes
Key Meyer [Mon, 27 Apr 2009 22:36:32 +0000 (18:36 -0400)]
traceio sample: tolerate more than a few hundred processes

... rather than exiting with MAXACTIONS exceeded

15 years agoMake sure user_string_n() results are always zero terminated strings.
Mark Wielaard [Mon, 27 Apr 2009 22:24:34 +0000 (00:24 +0200)]
Make sure user_string_n() results are always zero terminated strings.

* tapset/conversions.stp (user_string_n2): Explicitly null terminate
  return string.
  (user_string_n_warn): Likewise.

15 years agoHandle more types of types in declaration_resolve
Josh Stone [Mon, 27 Apr 2009 21:38:20 +0000 (14:38 -0700)]
Handle more types of types in declaration_resolve

Instead of just structs and unions, we now also process typedefs, enums,
and base types in iterate_over_globals, so declaration_resolve can find
more variations.  This especially useful to let @casts reference typedef
names.

15 years agoWarn if trying to dereference an enum
Josh Stone [Mon, 27 Apr 2009 21:29:19 +0000 (14:29 -0700)]
Warn if trying to dereference an enum

In translate_components, give a cleaner warning about trying to
dereference an enum type, rather than "unexpected type tag".

15 years agoFixed itrace on RHEL5 (PR10091).
David Smith [Mon, 27 Apr 2009 20:53:21 +0000 (15:53 -0500)]
Fixed itrace on RHEL5 (PR10091).
PR10091 fix.
* runtime/itrace.c: Includes ptrace_compatibility.h.
  (usr_itrace_report_quiesce): Corrected return value for original version
  of utrace.
* runtime/ptrace_compatibility.h: Defines arch_has_single_step() and
  arch_has_block_step() in terms of ARCH_HAS_SINGLE_STEP and
  ARCH_HAS_BLOCK_STEP.

15 years agoPR10081: improve error message for verifying build-id
Wenji Huang [Mon, 27 Apr 2009 10:12:36 +0000 (06:12 -0400)]
PR10081: improve error message for verifying build-id

Output debuginfo file name and build-id if verification
failed.

* runtime/sym.c: Print more info if not matched.

15 years agoPR10099: Extend %M directive to support hexdumping large buffers
Wenji Huang [Mon, 27 Apr 2009 09:38:18 +0000 (05:38 -0400)]
PR10099: Extend %M directive to support hexdumping large buffers

This patch will make %M directive dump the variable width
buffer in hex format instead of returning uint64_t number
as before.

* runtime/vsprintf.c: Modify %M directive.
* stap.1.in: Update description.
* testsuite/systemtap.printf/memory1.stp: Add test case.

15 years agoAdd function of returning the char in given position of string
Wenji Huang [Mon, 27 Apr 2009 08:34:42 +0000 (04:34 -0400)]
Add function of returning the char in given position of string

* tapset/string.stp: New function stringat.
* testsuite/systemtap.printf/char1.exp: Update test case.
* testsuite/systemtap.printf/char1.stp: Ditto.

15 years agoHandle kprobe struct field symbol_name as either char * or const char *.
Mark Wielaard [Sun, 26 Apr 2009 15:17:04 +0000 (17:17 +0200)]
Handle kprobe struct field symbol_name as either char * or const char *.

Older (2.6.18) based kernels defined struct kprobes symbol_name as char *,
but newer (2.6.27) based kernels defined the same field as const char *.

* tapsets.cxx (kprobe_derived_probe_group::emit_module_init):
  Always cast to char * when assigning to kprobes.symbol_name field.

15 years agofix kprobe.* probes so they don't break -p4 if script also has kernel.* probes
Frank Ch. Eigler [Sun, 26 Apr 2009 01:47:54 +0000 (21:47 -0400)]
fix kprobe.* probes so they don't break -p4 if script also has kernel.* probes

* tapsets.cxx (kprobe_derived_probe): Use enter_k[ret]probe>>2<<_probe.
* testsuite/buildok/thirtyone.stp: Test this.

15 years agoAdd grapher executable and doc generated files and dirs to .gitignore files.
Mark Wielaard [Sun, 26 Apr 2009 13:08:35 +0000 (15:08 +0200)]
Add grapher executable and doc generated files and dirs to .gitignore files.

15 years agoAdd two new AUTHORS
Josh Stone [Sat, 25 Apr 2009 17:18:13 +0000 (10:18 -0700)]
Add two new AUTHORS

15 years agoCorrect fd variable name in fadvise64 probe points.
Przemyslaw Pawelczyk [Sat, 25 Apr 2009 15:18:24 +0000 (17:18 +0200)]
Correct fd variable name in fadvise64 probe points.

* tapset/syscalls.stp: Rename fs variable to fd.

15 years agoEnable kprobe.function.return.maxactive
Josh Stone [Sat, 25 Apr 2009 03:06:58 +0000 (20:06 -0700)]
Enable kprobe.function.return.maxactive

The code to emit maxactive was almost there; it just needed to enable
the maxactive probe component and pass the value down.

15 years agoSimplify dwarfless kprobe stmt vs. name logic
Josh Stone [Sat, 25 Apr 2009 02:30:59 +0000 (19:30 -0700)]
Simplify dwarfless kprobe stmt vs. name logic

15 years agoFill out the pp() for dwarfless kprobes
Josh Stone [Sat, 25 Apr 2009 01:37:35 +0000 (18:37 -0700)]
Fill out the pp() for dwarfless kprobes

15 years agoDon't terminate when a dwarfless kprobe fails
Josh Stone [Sat, 25 Apr 2009 01:07:12 +0000 (18:07 -0700)]
Don't terminate when a dwarfless kprobe fails

All of our other kprobe types will let the script continue after a
registration failure, as per PR6749.  The dwarfless kprobes should be
no exception.

Also, the way they were exiting was causing an unclean shutdown, so I
added tests to badkprobe.exp to make sure that cleanup is always allowed
to run.

15 years agoTweak error message in kprobe.function registration
Josh Stone [Fri, 24 Apr 2009 21:28:26 +0000 (14:28 -0700)]
Tweak error message in kprobe.function registration

15 years agoPR5273: uprobes_i386.c instruction table
Jim Keniston [Fri, 24 Apr 2009 20:46:46 +0000 (13:46 -0700)]
PR5273: uprobes_i386.c instruction table
Resynch uprobes1 2-byte-opcode table with uprobes2 x86.  Mostly changes a
bunch of mmx instructions from rejected to accepted.

15 years agoCache and retrieve syscall arguments when needed.
David Smith [Fri, 24 Apr 2009 17:48:29 +0000 (12:48 -0500)]
Cache and retrieve syscall arguments when needed.
* runtime/task_finder.c (__stp_utrace_task_finder_target_syscall_entry):
  New function that caches syscall arguments for mmap(), munmap(), and
  mprotect() for use in __stp_utrace_task_finder_target_syscall_exit().
  (__stp_utrace_task_finder_target_syscall_exit): Uses cached syscall
  argument info when making callbacks.
  (stap_start_task_finder): Initializes map subsytem for storing syscall
  argument data.
* runtime/task_finder_map.c: New file containing functions to save and
  retrieve syscall arguments.

15 years agogit_version: fall back to git-describe w/o --long, and even git-rev-parse HEAD for...
Frank Ch. Eigler [Fri, 24 Apr 2009 15:43:11 +0000 (11:43 -0400)]
git_version: fall back to git-describe w/o --long, and even git-rev-parse HEAD for old gits

15 years agonew NEWS section
Frank Ch. Eigler [Fri, 24 Apr 2009 15:41:16 +0000 (11:41 -0400)]
new NEWS section

15 years agoPR10082: refactor duplicated code
Kent Sebastian [Fri, 24 Apr 2009 15:26:51 +0000 (11:26 -0400)]
PR10082: refactor duplicated code

15 years agoVersion bumps for 0.9.7 release release-0.9.7
Josh Stone [Thu, 23 Apr 2009 22:11:37 +0000 (15:11 -0700)]
Version bumps for 0.9.7 release

15 years agoi386 uprobes1: enable the 0x0f 0x1_ row of opcodes
Jim Keniston [Thu, 23 Apr 2009 21:19:58 +0000 (14:19 -0700)]
i386 uprobes1: enable the 0x0f 0x1_ row of opcodes
to make sdt.exp pass

15 years agoAvoid a uprobe break setting problem by avoiding 'nop 0' on x86.
Stan Cox [Thu, 23 Apr 2009 20:43:44 +0000 (16:43 -0400)]
Avoid a uprobe break setting problem by avoiding 'nop 0' on x86.

* sdt.h (STAP_NOP): New.
(STAP_PROBE): Use it.

15 years agogit_version: use --long mode to always add the git hash too
Frank Ch. Eigler [Thu, 23 Apr 2009 20:30:29 +0000 (16:30 -0400)]
git_version: use --long mode to always add the git hash too

15 years agoVersion bumps for 0.9.6 release release-0.9.6
Josh Stone [Thu, 23 Apr 2009 19:43:28 +0000 (12:43 -0700)]
Version bumps for 0.9.6 release

15 years agoMake sure values get passed to xmlto.
William Cohen [Thu, 23 Apr 2009 19:37:22 +0000 (15:37 -0400)]
Make sure values get passed to xmlto.

15 years agoMerge branch 'master' of ssh://sources.redhat.com/git/systemtap
William Cohen [Thu, 23 Apr 2009 19:27:31 +0000 (15:27 -0400)]
Merge branch 'master' of ssh://sources.redhat.com/git/systemtap

15 years agoRevert "Make sure that the pool_size and hash_size are passed into xmlto."
William Cohen [Thu, 23 Apr 2009 19:26:32 +0000 (15:26 -0400)]
Revert "Make sure that the pool_size and hash_size are passed into xmlto."

This reverts commit 52bc8b53c4bf03d51edf02d088aa04af809b9005.

15 years agoMake sure that the pool_size and hash_size are passed into xmlto.
William Cohen [Thu, 23 Apr 2009 19:04:05 +0000 (15:04 -0400)]
Make sure that the pool_size and hash_size are passed into xmlto.

15 years agogit_version: emit `git-describe` rather than just sha code, for use in stap -V
Frank Ch. Eigler [Thu, 23 Apr 2009 18:43:00 +0000 (14:43 -0400)]
git_version: emit `git-describe` rather than just sha code, for use in stap -V

15 years agoAdd NEWS entry about EXPERIMENTAL support for user process unwinding.
Mark Wielaard [Thu, 23 Apr 2009 15:37:56 +0000 (17:37 +0200)]
Add NEWS entry about EXPERIMENTAL support for user process unwinding.

15 years agoMerge branch 'master' of ssh://sources.redhat.com/git/systemtap
William Cohen [Thu, 23 Apr 2009 15:13:59 +0000 (11:13 -0400)]
Merge branch 'master' of ssh://sources.redhat.com/git/systemtap

15 years agoBump pool_size and hash_size to build Systemtap Tapset Refence manual.
William Cohen [Thu, 23 Apr 2009 15:12:51 +0000 (11:12 -0400)]
Bump pool_size and hash_size to build Systemtap Tapset Refence manual.

15 years agoMark uaddr tapset function as EXPERIMENTAL.
Mark Wielaard [Thu, 23 Apr 2009 14:06:21 +0000 (16:06 +0200)]
Mark uaddr tapset function as EXPERIMENTAL.

* tapset/context.stp (uaddr): Marked as experimental. Add FIXME for vdso
  tracking bug #10080.

15 years agoTweak kprobes test case
Wenji Huang [Thu, 23 Apr 2009 04:09:12 +0000 (00:09 -0400)]
Tweak kprobes test case

* testsuite/systemtap.base/kprobes.exp: Add expected message.
* testsuite/systemtap.base/kprobes.stp: Ditto.

15 years agoModify the output of kprobes test.
Wenji Huang [Thu, 23 Apr 2009 02:44:05 +0000 (22:44 -0400)]
Modify the output of kprobes test.

* testsuite/systemtap.base/kprobes.stp: Update text.

15 years agoAvoid 64 bit address value on a 32 bit machine.
Stan Cox [Wed, 22 Apr 2009 20:42:42 +0000 (16:42 -0400)]
Avoid 64 bit address value on a 32 bit machine.

* sdt.h (STAP_PROBE_ADDR): New.
(STAP_PROBE_DATA_): Use it.

15 years agoutrace/ia64: Fix syscall_get_set_args_cb() to handle syscalls via syscall()
Masami Hiramatsu [Wed, 22 Apr 2009 17:21:17 +0000 (13:21 -0400)]
utrace/ia64: Fix syscall_get_set_args_cb() to handle syscalls via syscall()

* runtime/syscall.h (syscall_get_set_args_cb): Fix to decode user stack
  collectly in case of syscall(), and check the maximum number of syscall
  arguments.

15 years agoPR 9821: Add a testcase for strftime
Masami Hiramatsu [Wed, 22 Apr 2009 17:18:01 +0000 (13:18 -0400)]
PR 9821: Add a testcase for strftime

* testsuite/systemtap.base/strftime.exp: New test case for strftime.

15 years agoPR 9821: Use genuine strftime in staprun/stapio
Masami Hiramatsu [Wed, 22 Apr 2009 17:16:25 +0000 (13:16 -0400)]
PR 9821: Use genuine strftime in staprun/stapio

* Makefile.am: Add -fno-builtin-strftime to stapio_CFLAGS.
* Makefile.in: Ditto.
* runtime/staprun/common.c (stap_strfloctime): Use strftime(3).
  (parse_args): Remove strftime format limitation message.
* main.cxx (usage): Ditto.
* stap.1.in: Ditto.
* staprun.8.in: Ditto.

15 years agoPR 6930: Add testcases for on-file flight recorder
Masami Hiramatsu [Wed, 22 Apr 2009 17:07:26 +0000 (13:07 -0400)]
PR 6930: Add testcases for on-file flight recorder

* testsuite/systemtap.base/flightrec1.exp: New test case for background mode.
* testsuite/systemtap.base/flightrec2.exp: New test case for file switching.
* testsuite/systemtap.base/flightrec2.stp: Test script for file switching.

15 years agoPR 9941: Add predicate function callback to check for an early abort
Rajan Arora [Wed, 22 Apr 2009 16:10:37 +0000 (12:10 -0400)]
PR 9941: Add predicate function callback to check for an early abort

* tapsets.cxx: Add dwfl_report_offline_predicate to check pending
  interrupts for an early abort.
* tapsets.h: Declare it.
* translate.cxx: Add callback to function call.

15 years agoMake sdt.h big endian aware.
Stan Cox [Wed, 22 Apr 2009 15:34:14 +0000 (11:34 -0400)]
Make sdt.h big endian aware.

* sdt.h: Use .quad instead of .long for .probe section addresses.
ia64 and s390 require 'nop 0' and x86 tolerates it.
* tapsets.cxx (build): Fetch probe_name in a big endian friendly fashion.

15 years agoCorrect sigmon.meta example title and name.
Sunzen Wang [Wed, 22 Apr 2009 11:52:15 +0000 (13:52 +0200)]
Correct sigmon.meta example title and name.

* testsuite/systemtap.examples/process/sigmon.meta: Correct title and name.
* testsuite/systemtap.examples/index.html: Regenerated.
* testsuite/systemtap.examples/index.txt: Likewise.
* testsuite/systemtap.examples/keyword-index.html: Likewise.
* testsuite/systemtap.examples/keyword-index.txt: Likewise.

15 years agoSimplify section size logic.
Mark Wielaard [Wed, 22 Apr 2009 10:53:39 +0000 (12:53 +0200)]
Simplify section size logic.

* translate.cxx (dump_unwindsyms): Just check that dwfl_module_relocations()
  return more than 1 relocation section bases before calling
  dwfl_module_address_section().

15 years ago[tracepoints] Resolve implicit trace_X use
Josh Stone [Wed, 22 Apr 2009 02:57:59 +0000 (19:57 -0700)]
[tracepoints] Resolve implicit trace_X use

Some of the tracepoints are actually being called in inlines in the
common headers (e.g. trace_kmalloc), which is causing errors about
implicit function declarations.  We don't care about ever running the
code in the tracequery module, so I'm just suppressing that error.

15 years ago[tracepoints] Don't use TRACE_HEADER_MULTI_READ
Josh Stone [Wed, 22 Apr 2009 02:55:47 +0000 (19:55 -0700)]
[tracepoints] Don't use TRACE_HEADER_MULTI_READ

At one point that macro was needed to get all of the tracepoints on the
tip tree, but now it's causing us to get duplicate stapprobe_X
definitions.  AFAICS, we're now getting all tracepoints even without
MULTI_READ, so I'm pulling that workaround out.

15 years agoUse tokenize() for splitting @cast's module list
Josh Stone [Tue, 21 Apr 2009 19:57:37 +0000 (12:57 -0700)]
Use tokenize() for splitting @cast's module list

15 years agoDocument @cast-with-headers
Josh Stone [Tue, 21 Apr 2009 19:34:33 +0000 (12:34 -0700)]
Document @cast-with-headers

15 years agoNB @cast's relative header searching
Josh Stone [Tue, 21 Apr 2009 19:12:43 +0000 (12:12 -0700)]
NB @cast's relative header searching

15 years agoRefine the @cast-with-header syntax
Josh Stone [Tue, 21 Apr 2009 19:08:42 +0000 (12:08 -0700)]
Refine the @cast-with-header syntax

The special syntax to generate a module for type information is now:
- "kernel<path/to/header.h>" to use the kernel's build environment
- "<path/to/header.h>" to use no special build environment, and so use
  gcc's default parameters only (for user mode).

15 years agoSet MAXSTRINGLEN to 133 for uprobes_ustack.exp test on 64 bit.
Mark Wielaard [Tue, 21 Apr 2009 19:33:24 +0000 (21:33 +0200)]
Set MAXSTRINGLEN to 133 for uprobes_ustack.exp test on 64 bit.

Needs extra space since on 64bit the last ubacktrace string is
7 entries * (16 hex + 2 for 0x + 1 space) = 133 chars.
Default MAXSTRINGLEN is 128 chars.

* testsuite/systemtap.base/uprobes_ustack.exp: Add -DMAXSTRINGLEN.

15 years agoMerge branch 'master' of ssh://sourceware.org/git/systemtap
Mark Wielaard [Tue, 21 Apr 2009 19:21:02 +0000 (21:21 +0200)]
Merge branch 'master' of ssh://sourceware.org/git/systemtap

15 years agoRef-count correctly when deranged handler calls do_exit().
Jim Keniston [Tue, 21 Apr 2009 18:57:36 +0000 (11:57 -0700)]
Ref-count correctly when deranged handler calls do_exit().

15 years agodocument user-space probes more fully, mention utrace requirements
Frank Ch. Eigler [Tue, 21 Apr 2009 18:11:18 +0000 (14:11 -0400)]
document user-space probes more fully, mention utrace requirements

15 years agoMerge branch 'user_unwind'
Mark Wielaard [Tue, 21 Apr 2009 18:01:00 +0000 (20:01 +0200)]
Merge branch 'user_unwind'

15 years agoMark ucontext tapset functions EXPERIMENTAL.
Mark Wielaard [Tue, 21 Apr 2009 17:57:02 +0000 (19:57 +0200)]
Mark ucontext tapset functions EXPERIMENTAL.

* tapset/ucontext-symbols.stp (usymname, usymdata, print_ustack):
  mark EXPERIMENTAL.
* tapset/ucontext-unwind.stp (print_ubacktrace, ubacktrace): Likewise.

15 years agoAdd uprobes_ustack testcase and bug fixlet.
Mark Wielaard [Tue, 21 Apr 2009 17:44:55 +0000 (19:44 +0200)]
Add uprobes_ustack testcase and bug fixlet.

* runtime/stack.c (_stp_stack_print): Use _stp_usymbol_print when tsk given.
* testsuite/systemtap.base/uprobes_ustack.exp: New test file.
* testsuite/systemtap.base/uprobes_ustack.stp: Likewise.

15 years agoUpdated utrace tests.
David Smith [Tue, 21 Apr 2009 15:49:15 +0000 (10:49 -0500)]
Updated utrace tests.
2009-04-21  David Smith  <dsmith@redhat.com>

* testsuite/semok/utrace01.stp: New test.
* testsuite/parseko/utrace01.stp: Updated test.
* testsuite/semko/utrace03.stp: Ditto.
* testsuite/semko/utrace04.stp: Ditto.
* testsuite/semko/utrace01.stp: Deleted unneeded test.
* testsuite/semko/utrace08.stp: Ditto.
* testsuite/semko/utrace09.stp: Ditto.
* testsuite/semko/utrace10.stp: Ditto.
* testsuite/semko/utrace11.stp: Ditto.
* testsuite/semko/utrace12.stp: Ditto.
* testsuite/semko/utrace13.stp: Ditto.

15 years agoAdd ubacktrace(), print_ustack() and print_ubacktrace().
Mark Wielaard [Tue, 21 Apr 2009 15:16:51 +0000 (17:16 +0200)]
Add ubacktrace(), print_ustack() and print_ubacktrace().

* runtime/sym.c (_stp_usymbol_print): New function.
* tapset/ucontext-unwind.stp (print_ubacktrace): New tapset function.
  (ubacktrace): Likewise.
* tapset/ucontext-symbols.stp (print_ustack): Likewise.
* testsuite/buildok/ustack.stp: New test for above three functions.

15 years agoRemove documentation weirdness FIXME in tapset print_stack().
Mark Wielaard [Tue, 21 Apr 2009 14:45:31 +0000 (16:45 +0200)]
Remove documentation weirdness FIXME in tapset print_stack().

Weirdness was fixed in commit d4db5608.

* tapset/context-symbols.stp (print_stack): Remove FIXME.

15 years agoCorrectly handle $syscall in process(PID_OR_PATH).syscall.return probes.
David Smith [Tue, 21 Apr 2009 13:21:34 +0000 (08:21 -0500)]
Correctly handle $syscall in process(PID_OR_PATH).syscall.return probes.
2009-04-21  David Smith  <dsmith@redhat.com>

* tapsets.cxx (utrace_derived_probe::utrace_derived_probe): If
during target-variable-expanding the probe, we added a new block
or probe, add them.
(utrace_var_expanding_visitor::visit_target_symbol_cached): New
function to cache target variables in a generated syscall probe to
use in syscall.return probes.
(utrace_var_expanding_visitor::visit_target_symbol_context): In a
syscall.return probe, you can't access $syscall.  So use
visit_target_symbol_cached() to cache the value for use here.
* testsuite/systemtap.base/utrace_syscall_args.stp: Test use of
$syscall in syscall.return probes.

15 years agoAdd testcases for kprobe.function family
Ananth N Mavinakayanahalli [Tue, 21 Apr 2009 11:49:15 +0000 (17:19 +0530)]
Add testcases for kprobe.function family

15 years agoDocumentation updates for kprobe.function family
Ananth N Mavinakayanahalli [Tue, 21 Apr 2009 11:38:31 +0000 (17:08 +0530)]
Documentation updates for kprobe.function family

15 years agoFrom Prerna Saxena: Add the kprobe.function probe family
Ananth N Mavinakayanahalli [Tue, 21 Apr 2009 11:37:22 +0000 (17:07 +0530)]
From Prerna Saxena: Add the kprobe.function probe family

15 years agoDon't use stack fallback for user space unwinding when using dwarf unwinder.
Mark Wielaard [Tue, 21 Apr 2009 10:06:53 +0000 (12:06 +0200)]
Don't use stack fallback for user space unwinding when using dwarf unwinder.

It never seems to be anywhere near correct.

* runtime/stack-i386.c (__stp_stack_print): Stop unwinding when dwarf
  unwinder fails and we are unwinding user space task.
* runtime/stack-x86_64.c (__stp_stack_print): Likewise.

15 years agoRemove old _stp_ustack bits.
Mark Wielaard [Mon, 20 Apr 2009 21:02:09 +0000 (23:02 +0200)]
Remove old _stp_ustack bits.

This code was never used, nor did it actually work.

* runtime/stack.c (_stp_ustack_print): Removed.

15 years agoMake latencytap.stp compile on i386.
Mark Wielaard [Tue, 21 Apr 2009 09:26:37 +0000 (11:26 +0200)]
Make latencytap.stp compile on i386.

Still needs a .meta file to catch any breakage in the future.

* testsuite/systemtap.examples/profiling/latencytap.stp
  (task_backtrace): Cast task to unsigned long first.

15 years agoSuppress more module-loading errors in @casts
Josh Stone [Tue, 21 Apr 2009 00:53:35 +0000 (17:53 -0700)]
Suppress more module-loading errors in @casts

Commit 462c90c3 did this for user modules, but it's needed for kernel
modules too.

15 years agoMerge branch 'cast_header'
Josh Stone [Tue, 21 Apr 2009 00:25:35 +0000 (17:25 -0700)]
Merge branch 'cast_header'

15 years agoAdd tests for @cast-generated modules
Josh Stone [Tue, 21 Apr 2009 00:18:32 +0000 (17:18 -0700)]
Add tests for @cast-generated modules

15 years agoCleanup typequery user modules from the cache too
Josh Stone [Mon, 20 Apr 2009 23:58:37 +0000 (16:58 -0700)]
Cleanup typequery user modules from the cache too

15 years agoPR10055: generate dummy modules w/ types for @cast
Josh Stone [Mon, 20 Apr 2009 23:45:26 +0000 (16:45 -0700)]
PR10055: generate dummy modules w/ types for @cast

The module field in @cast can now also be "kmod<path/to/header.h>" or
"umod<path/to/header.h>" to generate a kernel or user module which
includes the specified header.  The appropriate compiler flags are used
to save all possible type debuginfo from the header.

15 years agoSuppress module-loading errors in @casts
Josh Stone [Mon, 20 Apr 2009 23:43:15 +0000 (16:43 -0700)]
Suppress module-loading errors in @casts

If a @cast encounters a module that it can't load, it should just go on
to the next module instead of throwing an exception.  If there is no
next module, we'll get a better "type not found" exception anyway.

15 years agoAdd a function to hash typequery modules
Josh Stone [Mon, 20 Apr 2009 23:41:52 +0000 (16:41 -0700)]
Add a function to hash typequery modules

15 years agoPR10078: uretprobes on functions returning structs/unions
Jim Keniston [Mon, 20 Apr 2009 23:39:26 +0000 (16:39 -0700)]
PR10078: uretprobes on functions returning structs/unions
arch_predict_sp_at_ret() for x86_32 now accommodates ret $4.
Added bz10078 regression test.

15 years agoPass task from tapset, through stack and unwind functions for lookup.
Mark Wielaard [Mon, 20 Apr 2009 20:43:37 +0000 (22:43 +0200)]
Pass task from tapset, through stack and unwind functions for lookup.

* runtime/stack-arm.c (__stp_stack_print): Take struct task_struct *.
* runtime/stack-ia64.c (__stp_stack_print): Likewise.
* runtime/stack-ppc64.c (__stp_stack_print): Likewise.
* runtime/stack-s390.c (__stp_stack_print): Likewise.
* runtime/stack-i386.c (__stp_stack_print): Likewise.
  And add check and pass to unwind() and _stp_func_print().
* runtime/stack-x86_64.c: Likewise.
* runtime/stack.c *_stp_stack_print): Take and pass on task_struct.
  (_stp_stack_snprint): Likewise.
* runtime/unwind.c (unwind): Take and use task_struct for
  _stp_mod_sec_lookup().
* tapset/context-unwind.stp (print_backtrace): Pass NULL to
  _stp_stack_print().
  (backtrace): Pass NULL to _stp_stack_snprint().

15 years agoAdd comment to try -d kernel -d systemtap_test_module1 on backtrace.tcl test.
Mark Wielaard [Mon, 20 Apr 2009 20:42:56 +0000 (22:42 +0200)]
Add comment to try -d kernel -d systemtap_test_module1 on backtrace.tcl test.

15 years agoMerge branch 'master' of ssh://sources.redhat.com/git/systemtap
David Smith [Mon, 20 Apr 2009 18:55:45 +0000 (13:55 -0500)]
Merge branch 'master' of ssh://sources.redhat.com/git/systemtap

15 years agoUses upstream ia64 syscall functions.
David Smith [Mon, 20 Apr 2009 18:54:51 +0000 (13:54 -0500)]
Uses upstream ia64 syscall functions.
2009-04-20  David Smith  <dsmith@redhat.com>

* runtime/syscall.h (syscall_get_nr): Uses upstream version of
syscall_get_nr() for ia64.
(syscall_get_arguments): Ditto.
(in_syscall): New ia64-only function from upstream.
(syscall_get_set_args_cb): Ditto.
(ia64_syscall_get_set_arguments): Ditto.
* runtime/task_finder.c (stap_register_task_finder_target):
Removed ia64 register cache.

15 years agoHave appropriate types and cast to allow compilation on RHEL-5.
William Cohen [Mon, 20 Apr 2009 18:45:24 +0000 (14:45 -0400)]
Have appropriate types and cast to allow compilation on RHEL-5.

This page took 0.062006 seconds and 5 git commands to generate.