Josh Stone [Fri, 29 Apr 2011 23:33:35 +0000 (16:33 -0700)]
remote: Normalize the shell used to invoke stapsh
We're using a little bit of shell magic on the remote side to tell
whether a system has stapsh available, but that magic depends on
specific shell functionality. We now explicitly start it with
/bin/bash, so the user's choice of $SHELL doesn't matter.
* remote.cxx (ssh_remote::connect): Wrap the command in /bin/bash -c.
Josh Stone [Fri, 29 Apr 2011 22:48:13 +0000 (15:48 -0700)]
Loop server requests over every remotes' session
Now the list/trust-server options work on every unique session generated
by the requested remote targets. This enables usage such as:
$ stap --remote HOST --list-servers
* main.cxx (passes_0_4): Move the version/session banner to main.
(main): Print the version just once, and the session banner for each
unique session. Perform server actions within the session loop.
Josh Stone [Fri, 29 Apr 2011 22:43:14 +0000 (15:43 -0700)]
Restore signals after removing the tmpdir
This is necessary because passes_0_4_again_with_server() will remove the
tmpdir before trying again on a server. If it is then successful, the
proceeding run should be interruptible as normal.
* main.cxx (remove_temp_dir): Restore signals after removal.
Josh Stone [Thu, 28 Apr 2011 23:35:52 +0000 (16:35 -0700)]
remote: Support ssh on custom ports
* remote.cxx (remote::create): Heuristically disambiguate between URI
scheme:path and SSH host:port.
(ssh_remote::create): Parse an optional :port in the host string.
(ssh_remote::connect): Pass the port to ssh -p.
(ssh_legacy_remote::open_control_master): Pass the port to both
ssh -p and scp -P.
Josh Stone [Wed, 27 Apr 2011 22:23:05 +0000 (15:23 -0700)]
remote: Allow prefixing lines with the host index
When --remote-prefix is given, each line from remote scripts will be
prefixed with "N: ", where N is the index of that host among all the
--remote options on the command line.
* remote.h (class remote): Add an optional prefix string.
* remote.cxx (remote::run): Set the prefix if desired.
(stapsh::handle_poll): When using prefixes, read data line-wise.
* session.cxx (systemtap_session): Add bool use_remote_prefix.
(systemtap_session::parse_cmdline): Set it with --remote-prefix.
Stan Cox [Wed, 27 Apr 2011 02:29:24 +0000 (22:29 -0400)]
tolerate kernel builds with missing .note.gnu.build-id sections
* translate.cxx (dump_unwindsyms): Add back the check "Don't save
build-id if it is located before _stext."
* runtime/sym.c (_stp_module_check): Don't do buildid check for
-DSTP_NO_BUILDID_CHECK.
(_stp_usermodule_check): Likewise.
* buildid.exp: Test -DSTP_NO_BUILDID_CHECK
David Smith [Tue, 26 Apr 2011 18:56:30 +0000 (13:56 -0500)]
Added 'logger.perfile.{LOGFILE}.path' metric in the logger PMDA.
* pcp/src/pmdas/logger/logger.c (logger_fetchCallBack): Add support for
new 'logger.perfile.{LOGFILE}.path' metric, which returns the logfile
pathname.
David Smith [Tue, 26 Apr 2011 15:00:41 +0000 (10:00 -0500)]
Added help text for the dynamic metrics in the logger PMDA.
* pcp/src/pmdas/logger/logger.c (logger_text): New function that provides
help for the dynamic metrics.
(logger_init): Set up dynamic metrics help text.
* pcp/src/pmdas/logger/help: Removed commented out help for the dynamic
metrics.
David Smith [Mon, 25 Apr 2011 17:46:01 +0000 (12:46 -0500)]
Fix setjmp.exp and server tests for RHEL5 systems.
* testsuite/lib/systemtap.exp (start_server): To properly support older
versions of tcl (as on RHEL5 systems), don't use the newer form of
'catch'. Instead, use the global 'errorCode' variable when getting the
exit code of a child process.
* testsuite/systemtap.base/setjmp.exp: Ditto.
David Smith [Thu, 21 Apr 2011 18:45:46 +0000 (13:45 -0500)]
Added more testsuite error handling.
* testsuite/lib/systemtap.exp (start_server): Catch errors when starting a
server.
* testsuite/systemtap.base/buildid.exp: Be sure to call wait after spawn.
David Smith [Tue, 19 Apr 2011 21:41:47 +0000 (16:41 -0500)]
Improved error handling in uprobe probes.
* tapsets.cxx (uprobe_derived_probe_group::emit_module_decls): Make sure
we go through the probe epilogue even in an error in the emitted
enter_uprobe_probe().
In the generated module_exit() function, there exists an infinite loop
that waits for all probe handlers that might still be struggling along
to shut down, before authorizing removal of the probe module. If
something is stuck though, it will stay stuck without any diagnostics
short of a hung stapio/staprun process that's sitting in the kernel.
This patch causes a message to be printk KERN_ERR'd, in case the
shutdown synchronization takes more than a second.
* translate.cxx (emit_module_exit): Print a message if holdon-spinning
for more than a second.
David Smith [Fri, 15 Apr 2011 14:24:27 +0000 (09:24 -0500)]
More testsuite fixes.
* testsuite/systemtap.base/buildid.exp: Check for uprobes, not just
utrace. If we get an error, go ahead and return, since the following
stuff won't work since we've already cleaned up.
* testsuite/systemtap.base/target_set.exp: Better cleanup.
* testsuite/systemtap.base/target_set.stp: Now compiles on s390x and
ppc64 by using @cast().
* testsuite/systemtap.context/backtrace.tcl: If we get an EOF, don't call
exp_contine (which causes a tcl error).
Josh Stone [Tue, 12 Apr 2011 22:06:06 +0000 (15:06 -0700)]
Add STAP_PROBEV for variadic probe arguments
For the moment, this is only created when SDT_USE_VARIADIC is defined,
because we want to support arbitrary compiler flags, and -pedantic tends
to object to variadic macros.
* includes/sys/sdt.h (STAP_PROBEV): New.
* testsuite/systemtap.base/sdt_va_args.c: Use STAP_PROBEV.
* testsuite/systemtap.base/sdt_va_args.exp: Test STAP_PROBEV.
Stan Cox [Tue, 12 Apr 2011 20:25:15 +0000 (16:25 -0400)]
Simplify V3 sdt argument mode handling.
* tapsets.cxx (sdt_uprobe_var_expanding_visitor::visit_target_symbol_arg):
Treat argument extraction as a 64 bit subexpression and handle its sign if needed.
David Smith [Mon, 11 Apr 2011 21:40:31 +0000 (16:40 -0500)]
Updated tapset/vfs.stp for 2.6.39-0.rc2.
* tapset/vfs.stp: Kernel commit 7eaceac removed block_sync_page(), so
those probes were made optional. Kernel commit e64a782 renamed
__remove_from_page_cache() to __delete_from_page_cache(), so add an
alias.
* testsuite/buildok/vfs-detailed.stp: Updated.
David Smith [Mon, 11 Apr 2011 18:45:59 +0000 (13:45 -0500)]
Fixed runtime/procfs.c for 2.6.39-0.rc2.
* runtime/procfs.c (_stp_mkdir_proc_module): Calls kern_path_parent()
instead of path_lookup(). For older kernels kern_path_parent() is
defined as path_lookup().
* buildrun.cxx (compile_pass): Added autoconf-kern-path-parent.c test.
* runtime/autoconf-kern-path-parent.c: New test.
Lukas Berk [Wed, 6 Apr 2011 23:12:17 +0000 (19:12 -0400)]
PR3823: Added gettext functionality to dtrace script
Makefile* - were modified to include dtrace as a gettext file
dtrace.in - script was modified to only enable gettext functionality
if ENABLE_NLS is true
po/POTFILES.in - updated to include dtrace in gettextizing
other po* files - updated as needed
David Smith [Wed, 6 Apr 2011 19:05:33 +0000 (14:05 -0500)]
Fixed BZ692445 by updating regular expression 'all_pass_string'.
* testsuite/lib/stap_run.exp: Make expect happier by removing the '$' at
the end of 'all_pass_string'. The '\r\n' should catch the end of lines
properly.
Lukas Berk [Tue, 5 Apr 2011 15:40:01 +0000 (11:40 -0400)]
PR3823: Added gettext functionality to staprun files.
This was tested on a Fedora 14 x86_64 virtual machine with the
$LANG environment variable set to fr_FR.utf8 with a dummy stapio
usage message. Po files have been updated accordingly.
David Smith [Mon, 4 Apr 2011 16:30:26 +0000 (11:30 -0500)]
Handle logfile paths that end in '|', treat it as a command to read data from.
* pcp/src/pmdas/logger/util.c: New file.
* pcp/src/pmdas/logger/util.h: Ditto.
* pcp/src/pmdas/logger/event.c (event_init): Handle opening pipe "paths".
(event_shutdown): New function that stops piped processes and closes all
files.
* pcp/src/pmdas/logger/event.h: Added event_shutdown() prototype.
* pcp/src/pmdas/logger/logger.c (read_config): Moved lstrip/rstrip code to
util.c.
(main): Calls event_shutdown() after pmdaMain() returns to clean
everything up.
* pcp/src/pmdas/logger/GNUmakefile.in: Added util.c and util.h.
* pcp/src/pmdas/logger/Install: Support entering pipe "paths".
Josh Stone [Fri, 1 Apr 2011 20:57:39 +0000 (13:57 -0700)]
Add addr() for retrieving REG_IP
For probes which have CONTEXT->regs, return REG_IP, else return 0.
* tapset/context.stp (addr): New function.
* testsuite/buildok/context-embedded.stp: Test addr() compilation.
* NEWS: Let the town crier make a pronouncement of addr().
William Cohen [Thu, 31 Mar 2011 20:54:09 +0000 (16:54 -0400)]
Correct use of pid for filtering
The blog on http://blog.yufeng.info/archives/1239 mentions that the script
is not properly filtering based on pid. This change corrects the
filtering to use the pid.
David Smith [Thu, 31 Mar 2011 16:23:27 +0000 (11:23 -0500)]
Change the logger PMDA config file lines to be of the form "NAME PATHNAME".
* pcp/src/pmdas/logger/logger.c (valid_pmns_name): New function.
(read_config): Read NAME and PATHNAME from each config file line. Make
sure NAME is a valid PMNS name.
* pcp/src/pmdas/logger/Install: Ask for logfile names along with paths.
* pcp/src/pmdas/logger/event.h: Swap the order of the logfile name and
pathname to mirror the new config file.
* tapset/aux_syscalls.stp (_ptrace_options_str, _ptrace_argstr)
(_ptrace_return_geteventmsg_data): New functions.
* tapset/i386/aux_syscalls.stp: New file.
* tapset/ia64/aux_syscalls.stp: New file.
* tapset/nd_syscalls2.stp (nd_syscall.ptrace): Set argstr using
_ptrace_argstr.
(nd_syscall.ptrace.return): Set geteventmsg_data and arch_prctl_addr.
* tapset/powerpc/aux_syscalls.stp: New file.
* tapset/s390/aux_syscalls.stp: New file.
* tapset/syscalls2.stp (syscall.ptrace): Set argstr using
_ptrace_argstr.
(syscall.ptrace.return): Set geteventmsg_data and arch_prctl_addr.
* tapset/x86_64/aux_syscalls.stp: New file.
* testsuite/buildok/aux_syscalls-embedded.stp (begin): Call
_ptrace_argstr, _ptrace_return_geteventmsg_data and
_ptrace_return_arch_prctl_addr.
Stan Cox [Wed, 30 Mar 2011 15:14:43 +0000 (11:14 -0400)]
Add more syntax for shared library probing.
* dwflpp.cxx (iterate_over_libraries): New. Get library list via
ld-linux.so.
* dwflpp.h (iterate_over_libraries): Likewise.
* tapsets.cxx (base_query): If this is a library then save the
executable path.
(query_module): For a library, call query_library_callback via
iterate_over_libraries.
(query_library_callback): New. Calls query_library.
(query_library): Substitute the library name into the chain then
derive_probes.
William Cohen [Tue, 29 Mar 2011 17:57:08 +0000 (13:57 -0400)]
Remove unneeded initializer
The variable is already static so will be initialized to 0. Removed the
unneeded initializer and elminate a warning message from the compiling the
code.
Frank Ch. Eigler [Tue, 29 Mar 2011 13:33:53 +0000 (09:33 -0400)]
PR12612: compatibility with old staprun/stapio
commit 3abb860f changed the staprun<->runtime message format for
the STP_RELOCATE message, without regard for backward compatibility
for older/mismatching stapruns. That leads to kernel errors and
no kernel probes and such nonsense. Let's change the runtime to
interoperate both with old staprun and with the current one.
* runtime/transport/symbols.c (_stp_do_relocation): Also accept
systemtap-1.3 level relocation message, if the incoming message
size suggests.
David Smith [Mon, 28 Mar 2011 21:04:46 +0000 (16:04 -0500)]
The logger PMDA supports multiple clients and multiple logfiles.
* pcp/src/pmdas/logger/logger.c: Changed to support dynamic metrics for
logfile data, which enables multiple client/multiple logfile support.
* pcp/src/pmdas/logger/event.c: Changes to support multiple clients and
multiple logfiles. Abstracted the interface a bit.
* pcp/src/pmdas/logger/event.h: Moved 'struct LogfileData' definition here
from logger.c. Removed event_cleanup() declaration.
* pcp/src/pmdas/logger/help: Updated.
* pcp/src/pmdas/logger/pmns: Make 'perfile' the start of the dynamic
metrics.
* pcp/src/pmdas/logger/GNUmakefile.in: Make logger.o depend on domain.h.
Frank Ch. Eigler [Sun, 27 Mar 2011 03:59:24 +0000 (23:59 -0400)]
sdt_v3: avoid misleading warning on complex operands
Roland reported seeing the "Downgrading SDT_V2 probe argument to dwarf ..."
warning message on an ambitious SDT_V3 object file. The message is
not applicable, since there is no dwarf emitted at all for SDT_V3.
* tapsets.cxx (sdt_uprobe_var_expanding_visitor::visit_target_symbol_arg):
Remove SDT_V2 language for SDT_V3 probe.
Josh Stone [Thu, 24 Mar 2011 18:18:45 +0000 (11:18 -0700)]
Find SDT semaphores even in RWX mappings
Some S390 environments don't have noexec mappings for data sections, so
the .probes section with SDT semaphores is mapped RWX rather than RW-.
Our logic which checks vm flags needs to accommodate this.
* runtime/uprobes-common.c (stap_uprobe_mmap_found): Deal with mappings
that are both executable and writable, so semaphores can be found.
Josh Stone [Tue, 22 Mar 2011 03:01:39 +0000 (20:01 -0700)]
Add a workaround for old systems without ppoll
A coarse approximation of ppoll is added using sigprocmask and poll, as
suggested in the ppoll manpage, but it doesn't carry the same atomic
guarantee. I've forced it to have a non-infinite polling timeout, so we
won't block forever.
NB: This only matters on the client connecting to a remote ssh host.
* configure.ac: Check for the function ppoll.
* configure, config.in: Regenerate.
* remote.cxx (ppoll): Fill in for missing ppoll, with a healthy comment
explaining why I think this implementation is Good Enoughâ„¢.
Josh Stone [Mon, 21 Mar 2011 22:30:32 +0000 (15:30 -0700)]
remote: Try the legacy ssh code if stapsh isn't found
The ssh remote command is restructured so that if stapsh isn't found in
the remote PATH, it will exit 127 immediately. On the local side, when
we fail to read the stapsh uname response, we can check the exit status
to know stapsh wasn't found. Then we can try again with the old ssh
code with a ControlMaster. However, I'm limiting the fallback to only
when there's a single remote, as signals and cleanup aren't really
handled as well. (That's the point of stapsh...)
* remote.cxx (ssh_remote::create): New static function which tries a
self-instantiation first, then if that fails to find stapsh, it tries
again using ssh_legacy_remote instead.
(ssh_remote::connect): Renamed from remote. Now using a "type -P"
trick to silently determine if stapsh is present, by exit code alone.
(ssh_legacy_remote): Un-#ifdef'ed the old ssh_remote code, to get new
life as our legacy access.
(remote::create): Get ssh through ssh_remote::create.
Josh Stone [Mon, 21 Mar 2011 22:22:09 +0000 (15:22 -0700)]
nsscommon: Heed what fgets returns
My F14 gcc rightfully warned that a return value of fgets was unused.
It may return NULL on error or EOF, in which case we should free our
memory and return NULL to NSS as well. I also tweaked the newline stomp
to tolerate missing newline, e.g. with oversized input, although that
case it not likely going to work anyway.
* nsscommon.c (nssPasswordCallback): Check fgets' return for success.
Dave Brolley [Mon, 21 Mar 2011 20:25:31 +0000 (16:25 -0400)]
PR 12101: stap-server should use proper cert passwords.
Automatically regenerate certificate databases containing a password file.
Otherwise automatically started servers will fail because they won't know
the password.
Dave Brolley [Mon, 21 Mar 2011 18:06:06 +0000 (14:06 -0400)]
PR 12101: stap-server should use proper cert passwords
Step 1: The server's certificate database now uses no password by default
(useful for full automation).
Preliminary support for password prompting for databases created
with a password.