Mathieu Desnoyers [Thu, 13 Nov 2014 21:22:33 +0000 (16:22 -0500)]
Fix: define _LGPL_SOURCE in C files
Defining this in headers is a bad practice, since the header is not
necessarily the first in the include list.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Sat, 15 Nov 2014 04:38:36 +0000 (23:38 -0500)]
Fix: HT must not be destroyed with a rcu_read_lock held
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Sat, 15 Nov 2014 04:35:45 +0000 (23:35 -0500)]
Fix: agent events HT should be destroyed from the cleanup thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 13 Nov 2014 21:27:40 +0000 (16:27 -0500)]
Fix: missing rcu_read_lock in trace_ust_destroy_session()
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 13 Nov 2014 21:23:04 +0000 (16:23 -0500)]
Fix: missing rcu_read_lock when calling trace_ust_find_agent()
Added a comment to note that this function is assumed to be
called with the rcu_read_lock held.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 13 Nov 2014 21:09:57 +0000 (16:09 -0500)]
Fix: missing rcu_read_lock in cmd_start_trace()
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 12 Nov 2014 23:18:33 +0000 (18:18 -0500)]
Cleanup: remove "disabled" flag for new release cycle
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 12 Nov 2014 23:18:32 +0000 (18:18 -0500)]
Fix: filter attach vs event enable race
In order to correctly handle the use-case where events are enabled
_after_ trace is started, and _after_ applications are already being
traced, the event should be created in a "disabled" state, so that it
does not trace events until its filter is attached.
This fix needs to be done both in lttng-tools and lttng-ust. In order to
keep ABI compatibility between tools and ust within a stable release
cycle, we introduce a new "disabled" within struct lttng_ust_event
padding (previously zeroed). Newer LTTng-UST checks this flag, and
fallback on the old racy behavior (enabling the event on creation) if it
is unset.
Therefore, old session daemon works with newer lttng-ust of the same
stable release, and vice-versa. However, building lttng-tools requires
an upgraded lttng-ust, which contains the communication protocol with
the new "disabled" field.
This patch should be backported to stable-2.4, stable-2.5, stable-2.6
branches.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 11 Nov 2014 19:36:15 +0000 (14:36 -0500)]
Fix: Mark libxml2 as a mandatory dependency
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Christian Babeux [Fri, 7 Nov 2014 23:18:43 +0000 (18:18 -0500)]
Fix: Warn the user when enabling an event in a new domain
This patch warn the user when trying to enable an event in a new domain
that was not configured prior to the tracing start.
Fixes #826
Signed-off-by: Christian Babeux <christian.babeux@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 7 Nov 2014 22:58:15 +0000 (17:58 -0500)]
Fix: Don't report an error when listing a session with no channels
The LTTng sessiond daemon reports an error when the list_channel
command is invoked on a session which only has a metadata channel.
The error lies in assuming that a session which has an enabled
domains also has (user-visible) channels.
The metadata channel should be listed in a future fix.
Fixes #855
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 7 Nov 2014 21:10:46 +0000 (16:10 -0500)]
Bypass 0 byte allocation when no domains are enabled
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Christian Babeux [Fri, 7 Nov 2014 17:17:02 +0000 (12:17 -0500)]
Fix: Add missing URCU_TLS access to error_log_time
Fixes #849
Signed-off-by: Christian Babeux <christian.babeux@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Nov 2014 22:34:08 +0000 (17:34 -0500)]
Fix: check userspace perf counter name when looking up contexts
create_ust_app_channel_context() looks for a context's existance
in a channel before adding it. However, it only checks for
context types. This is valid for all context types except for
LTTNG_UST_CONTEXT_PERF_THREAD_COUNTER since multiple perf
thread counters may be enabled at the same time.
This fix ensures that the perf counter name is taken into
consideration when checking for a context's presence in a
channel.
Reported-by: Alexander Grigoriev <alexgri@tbricks.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 5 Nov 2014 22:34:14 +0000 (17:34 -0500)]
Update maintainer section of the man pages
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 5 Nov 2014 20:51:29 +0000 (15:51 -0500)]
Update maintainer section of README.md
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 4 Nov 2014 22:59:29 +0000 (17:59 -0500)]
Fix: add README.md to documentation files
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
David Goulet [Fri, 31 Oct 2014 17:23:29 +0000 (13:23 -0400)]
Fix: UST consumer sync all available metadata
In live mode, the sync metadata function was only working on one single
metadata stream of a given session ID. However, we can have multiple
metadata stream for the same session ID thus failing to send the data in
live mode correctly for the other streams.
This fixes it by simply iterating over all metadata stream for a session
ID and syncing them all.
Signed-off-by: David Goulet <dgoulet@efficios.com>
Nathan Lynch [Tue, 28 Oct 2014 02:00:06 +0000 (21:00 -0500)]
Test: remove bogus argument from prove invocation
An argument of "$2" was added to the prove command line in tests/run.sh
by
68270f0f604e "Mi test: Basic test structure". This is harmless since
run.sh is never called with more than one argument, but it is confusing,
and it breaks patches in OE that add support for running tests on the
target.
Signed-off-by: Nathan Lynch <nathan_lynch@mentor.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Sat, 25 Oct 2014 20:47:31 +0000 (16:47 -0400)]
Fix: typo in pythong test Makefile
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Thu, 23 Oct 2014 14:16:06 +0000 (10:16 -0400)]
Fix: memory leak in libconfig
Fixes Coverity issue
1191755.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 14 Oct 2014 15:15:22 +0000 (11:15 -0400)]
Add Python agent support
Support the new Python agent shipped in liblttng-ust.
This adds the -p, --python option to the list and enable/disable-event
command to control the domain exactly like JUL and LOG4J.
The agent support is for the Python "logging" module.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 7 Oct 2014 19:05:48 +0000 (15:05 -0400)]
Fix: return EINVAL if agent registration fails
The errno value might be 0 thus not returning an error if so. It has
been seen with an unstable python agent code base which means it could
happen in the future if a third part decides to create an agent.
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Mon, 20 Oct 2014 22:50:33 +0000 (18:50 -0400)]
Use lttng-modules ABI version ioctl
Check compatibility between lttng-modules and tools using a version
numbering specifically for the ABI, rather than relying on the major
version of lttng-modules per se. This takes into account that we
sometimes depend on lock-step updates of the toolchain, including
tools and modules.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Mon, 20 Oct 2014 22:50:32 +0000 (18:50 -0400)]
Fix: syscall list ioctl number conflict
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 20 Oct 2014 20:10:19 +0000 (16:10 -0400)]
Update master ChangeLog with 2.5.1 release
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 20 Oct 2014 17:17:42 +0000 (13:17 -0400)]
Update version to v2.6.0-rc1
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 20 Oct 2014 15:21:44 +0000 (11:21 -0400)]
Update man page with LOG4J information
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 20 Oct 2014 14:48:13 +0000 (10:48 -0400)]
Add log4j values to mi XML schema
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Thu, 16 Oct 2014 16:18:00 +0000 (12:18 -0400)]
Fix: add missing headers to makefile for dist tarball
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Thu, 16 Oct 2014 15:16:07 +0000 (11:16 -0400)]
Fix: initialize context list node on creation
This is to avoid a segfault when destroying a kernel context that has
not been added to a channel.
Fixes #845
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 15 Oct 2014 18:29:01 +0000 (14:29 -0400)]
Fix: update man page with log4j option
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 14 Oct 2014 18:43:29 +0000 (14:43 -0400)]
Fix: support log4j loglevel in mi
To achieve this, a couple of functions needed to send the domain of the
event so we can use the right loglevel according to the domain.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 14 Oct 2014 18:16:23 +0000 (14:16 -0400)]
Fix: add log4j domain to session.xsd
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 14 Oct 2014 15:30:59 +0000 (11:30 -0400)]
Fix: add log4j to lttng list help
Signed-off-by: David Goulet <dgoulet@efficios.com>
Jérémie Galarneau [Mon, 6 Oct 2014 23:47:19 +0000 (19:47 -0400)]
Fix: Don't save an empty <contexts> node in kernel channels
An empty <contexts/> node is saved when kernel channels are
saved even when no kernel contexts are enabled.
Note that this is a minor fix as the generated XML remains valid
anyhow.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 6 Oct 2014 23:44:21 +0000 (19:44 -0400)]
Fix: save UST context informations as a event_perf_context_type
This fixes UST perf counters not being recognized by the session
restore code at the linting stage.
Reported-by: Philipe Proulx <philippe.proulx@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
David Goulet [Fri, 3 Oct 2014 18:30:54 +0000 (14:30 -0400)]
Fix: handle kernel event type correctly on enable
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 3 Oct 2014 15:05:13 +0000 (11:05 -0400)]
Test: Add syscall listing
For both available syscall and session event.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 20:10:54 +0000 (16:10 -0400)]
Fix: set hidden attribute to all public mi functions
Reported-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 19:57:34 +0000 (15:57 -0400)]
Fix: set the hidden attribute to a filter function in liblttng-ctl
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:57:36 +0000 (14:57 -0400)]
Fix: handle sysconf possible negative returned value
Fixes Coverity issue
1019947.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:52:33 +0000 (14:52 -0400)]
Fix: channel deref. after NULL check in kernel consumer
Fixes Coverity issue
1040158.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:49:42 +0000 (14:49 -0400)]
Fix: check for kernel session metadata during init
Fixes Coverity issue
1040159.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:40:13 +0000 (14:40 -0400)]
Fix: cmd snapshot record output can't be NULL
The session daemon calls cmd_record_snapshot always with a valid
pointer.
Fixes Coverity issue
1047269.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:32:12 +0000 (14:32 -0400)]
Fix: incorrect cast in ust consumer assert
Fixes Coverity issue
1068813.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:27:26 +0000 (14:27 -0400)]
Fix: remove useless assert in relayd
Fixes Coverity issue
1127095.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 18:16:07 +0000 (14:16 -0400)]
Fix: bad handling of incoming data in consumer thread
This will fix Coverity issue
1019892.
The issue uncovered a more serious problem. The loop on ready FDs of the
thread was exiting at each branch thus not going on all fd. This is
problematic when the thread quit pipe is triggered and when there is
also at the same time a request for metadata from the consumer since the
metadata request could have been ignored.
This patch makes sure we go through all FDs in the loop when the thread
quit pipe or the metadata fd is triggered.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 17:30:27 +0000 (13:30 -0400)]
Fix: detect size_t overflow in syscall table init
Fixes Coverity issue
1242317.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 17:00:36 +0000 (13:00 -0400)]
Fix: remove non existent header files from Makefile
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 1 Oct 2014 15:30:15 +0000 (11:30 -0400)]
Fix: only initialize syscall table if kernel tracer available
Fixes #844
Signed-off-by: David Goulet <dgoulet@efficios.com>
Philippe Proulx [Wed, 17 Sep 2014 19:33:10 +0000 (15:33 -0400)]
Modernize README with Markdown
This commit also:
* adds a project description at the top
* fixes a few grammar mistakes here and there
* updates the Package contents list
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 30 Sep 2014 17:31:43 +0000 (13:31 -0400)]
Add --list-options to save/load command
Fixes #835
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 30 Sep 2014 16:00:31 +0000 (12:00 -0400)]
Fix: use max between index and nbmem for syscall table realloc
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 30 Sep 2014 15:59:49 +0000 (11:59 -0400)]
Fix: fd leak in inet accept call
Fixes Coverity issue #
1241799.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 29 Sep 2014 19:06:56 +0000 (15:06 -0400)]
Test: add diag to each syscall test
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 29 Sep 2014 14:55:33 +0000 (10:55 -0400)]
Support syscall event in save session
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 22 Sep 2014 18:58:30 +0000 (14:58 -0400)]
Add syscall listing support
This adds two things. First, a global syscall table populated at boot
time that is used to match which syscall is enabled or not using the
lttng kernel get mask call.
The second thing is the ability to list which syscall is enable or not
in a specific session. Keep in mind that for syscalls, NO state is kept
on the session daemon so we have to ask the kernel tracer which syscall
is enabled for a specific channel.
This introduce some changes to the API/ABI. First, an event flag is
added to the lttng_event data structure using 4 bytes of padding for an
enum value that can be ORed together in that field. This is used for now
to know which bitness the event is in case of a syscall event type.
Second, lttng_list_syscalls(...) is added to provide the ability to list
all available syscalls that the user can trace. To use that with the
lttng command line, "--syscall" is added to lttng list.
$ lttng list -k --syscall
The above only lists available syscalls from the kernel tracer.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 29 Sep 2014 13:38:42 +0000 (09:38 -0400)]
Fix: change perror to PERROR in kernel consumer
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Fri, 26 Sep 2014 00:52:27 +0000 (20:52 -0400)]
Fix: syscall test: add missing semicolumns in match
Don't match the system call "openat" for instance.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Fri, 26 Sep 2014 00:49:45 +0000 (20:49 -0400)]
Fix: kernel consumer: issue put_subbuf on error
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Wed, 24 Sep 2014 16:18:19 +0000 (12:18 -0400)]
Test: add disable syscalls when none are enabled
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 24 Sep 2014 01:19:07 +0000 (21:19 -0400)]
Implement tests for kernel syscall tracing feature
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Wed, 24 Sep 2014 01:04:29 +0000 (21:04 -0400)]
Fix: syscall tracing: disable all and missing error handling
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Mathieu Desnoyers [Mon, 15 Sep 2014 19:07:02 +0000 (15:07 -0400)]
Implement kernctl_syscall_mask
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Mon, 15 Sep 2014 18:20:53 +0000 (14:20 -0400)]
Implement syscall mask ioctl
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Fri, 12 Sep 2014 18:22:28 +0000 (14:22 -0400)]
lttng-modules ABI: syscall field "enable"
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Fri, 12 Sep 2014 16:10:29 +0000 (12:10 -0400)]
Add kernctl_syscall_list
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Wed, 10 Sep 2014 20:50:58 +0000 (16:50 -0400)]
Cleanup: Remove LTTNG_ENABLE_ALL_EVENT and LTTNG_DISABLE_ALL_EVENT
Cleanup lttng-ctl to session daemon communication protocol, and cleanup
session daemon code (remove dead code).
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Tue, 9 Sep 2014 22:56:13 +0000 (18:56 -0400)]
Implement support for lttng-modules syscall filtering
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Sat, 19 Jul 2014 21:16:22 +0000 (17:16 -0400)]
Support lttng-modules syscall filtering in enable-event
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Simon Marchi [Sat, 20 Sep 2014 01:29:23 +0000 (21:29 -0400)]
Fix out-of-tree build
Signed-off-by: Simon Marchi <simon.marchi@polymtl.ca>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Jérémie Galarneau [Tue, 23 Sep 2014 21:46:32 +0000 (17:46 -0400)]
Fix: Propagate socket timeouts on lttcomm_accept_inet_sock()
Ensure that send/recv timeouts are applied to new sockets created
from a lttcomm_accept_inet_sock().
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Mathieu Desnoyers [Mon, 22 Sep 2014 16:39:08 +0000 (12:39 -0400)]
Fix: report UST consumer channel creation error
The session daemon reports no error whatsoever when the consumer daemon
fails to create a channel. We don't want to print errors when failing to
send the channel to the UST application though, because the application
may be concurrently exiting. So only print those errors when receiving
reply from the consumerd.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 15 Sep 2014 19:22:12 +0000 (15:22 -0400)]
Test: Add the new log4j agent
This adds the log4j java agent test.
It also moves the JUL test and build system to use the new UST java
agent jar.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 22 Sep 2014 15:05:06 +0000 (11:05 -0400)]
Fix: memory leak in load session thread
Fixes Coverity issue
1213789.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 22 Sep 2014 15:03:24 +0000 (11:03 -0400)]
Fix: kernel context memory leak on error
Fixes Coverity issue
1223776.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 22 Sep 2014 14:55:52 +0000 (10:55 -0400)]
Fix: possible file descriptor leak in error path
Fixes Coverity issue
1225086.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 22 Sep 2014 14:50:42 +0000 (10:50 -0400)]
Fix: use after free in agent subsystem
Fixes Coverity issue
1230592.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 19 Sep 2014 17:58:31 +0000 (13:58 -0400)]
Add version to agent registration message
Signed-off-by: David Goulet <dgoulet@efficios.com>
Philippe Proulx [Mon, 15 Sep 2014 17:43:56 +0000 (13:43 -0400)]
Fix: ignore error when loading optional kmod
This commit also improves the accuracy of debug messages: a module is
not "successfully" loaded when it's optional and actually not loaded.
Fixes: #837
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Philippe Proulx [Sat, 13 Sep 2014 01:37:18 +0000 (21:37 -0400)]
sessiond: Add --extra-kmod-probes option
This patch adds the --extra-kmod-probes option to lttng-sessiond. The
LTTNG_EXTRA_KMOD_PROBES environment variable may also be used.
The option specifies a list of extra probe kernel modules to be loaded
(and unloaded) by lttng-sessiond. The list is appended to either the
default list or to the user-supplied --kmod-probes list.
This option is especially useful for kernel developers who need the
default LTTng kernel probes plus additional probes in order to
instrument their custom kernel or module. This becomes easy with
--extra-kmod-probes:
lttng-sessiond --extra-kmod-probes=custom_subsys,other
would load all known and available LTTng kernel probes plus
lttng_probe_custom_subsys and lttng_probe_other.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Thu, 18 Sep 2014 19:38:18 +0000 (15:38 -0400)]
Fix: remove break in lttng cmdline session listing
This was causing to only list the first session with a "lttng list"
command.
Reported-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 15 Sep 2014 19:20:51 +0000 (15:20 -0400)]
Fix: use default agent channel name in disable event
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Thu, 11 Sep 2014 18:26:10 +0000 (14:26 -0400)]
Remove enable/disable consumer obsolete command
They've been obsolete for quite some time now so just remove any
reference and file to get rid of it.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Tue, 9 Sep 2014 15:22:50 +0000 (11:22 -0400)]
Add log4j default event and channel name handling
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 10 Sep 2014 20:38:04 +0000 (16:38 -0400)]
Fix: remove an odd no name directory from the test tree
Reported-by: Daniel Thibault <Daniel.Thibault@drdc-rddc.gc.ca>
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Wed, 10 Sep 2014 16:16:48 +0000 (12:16 -0400)]
Fix: move easter egg after arg parsing else it breaks the mi test
Signed-off-by: David Goulet <dgoulet@efficios.com>
Umut Tezduyar Lindskog [Thu, 4 Sep 2014 08:30:53 +0000 (10:30 +0200)]
Load modules through kmod
Instead of forking processes, load modules through libkmod.
This adds an optional package dependency to kmod but can be disabled by
--disable-kmod option in ./configure. If --enable-kmod option is given
but no kmod is found, loading will happen by forking a process.
The startup time has decreased by %36 on our embedded mips product. The
measurement is done by cgroup cpu shares.
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Mon, 8 Sep 2014 18:50:06 +0000 (14:50 -0400)]
Add cscope files to gitignore
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 5 Sep 2014 19:29:23 +0000 (15:29 -0400)]
Fix: list agent event per domain only
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 5 Sep 2014 19:24:58 +0000 (15:24 -0400)]
Fix: typo in lttng list command for log4j
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 5 Sep 2014 16:16:18 +0000 (12:16 -0400)]
Test: add disable-event JUL tests
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 5 Sep 2014 16:13:55 +0000 (12:13 -0400)]
Fix: disable-event for an agent domain
This fixes the disable-event command for an agent that was not working
properly in the case an application was spawned after the command.
Also, for the disable all event command for an agent, the disable
command will try first to disable the '*' event and will disable all
already existing events after. This is to make sure the refcount on the
agent side is synchronized with the amount of enable event prior.
Fixes #831
Signed-off-by: David Goulet <dgoulet@efficios.com>
David Goulet [Fri, 5 Sep 2014 16:12:38 +0000 (12:12 -0400)]
Fix: send disable ALL command code if event is *
$ lttng disable-event -a
and
$ lttng disable-event '*'
Both commands should do the same meaning send the DISABLE_EVENT_ALL
command to the session daemon.
Signed-off-by: David Goulet <dgoulet@efficios.com>
Julien Desfossez [Wed, 27 Aug 2014 17:59:21 +0000 (13:59 -0400)]
Fix: make sure no index is in flight before using inactivity beacons
Since the index is sent in two parts on two separate connections from
the consumer, there can be cases where we receive an inactivity beacon
between the index creation and the data reception.
This fix prevents from using the inactivity beacon if we know a data
index is coming.
Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Jérémie Galarneau [Thu, 28 Aug 2014 16:53:44 +0000 (12:53 -0400)]
Tests: Validate that filters may be used with JUL's logger_name filtering
This test validates that custom filters may be used on JUL events
without interfering with the implicit filtering done on the logger_name
field.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Jérémie Galarneau [Thu, 28 Aug 2014 16:53:43 +0000 (12:53 -0400)]
Fix: Parenthesize previous statement when adding conditions to a filter
Not parenthesizing the clauses in a filter string causes JUL events to be
traced even though they are not enabled when an enable-event command is
issued with a filter and the --loglevel-only option.
For instance,
lttng enable-event -j "my_event" --loglevel-only JUL_CONFIG -f "int_loglevel > 0 || int_loglevel < 0"
results in the following filter being applied:
int_loglevel > 0 || int_loglevel < 0 && logger_name == "my_event" && int_loglevel == 700
The resulting expression will always evaluate to true, regardless of the logger
name.
This fix parenthesizes each clause to ensure they are all independently
evaluated and not unintuitively affected by operator precedence.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Jonathan Rajotte Julien [Thu, 28 Aug 2014 20:51:09 +0000 (16:51 -0400)]
Fix: mi print of probe attributes + validation
Signed-off-by: Jonathan Rajotte Julien <jonathan.r.julien@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Jonathan Rajotte Julien [Wed, 27 Aug 2014 19:46:57 +0000 (15:46 -0400)]
Fix: parse_prob_opts return the actual success of the function
This bug have been triggered by the mi merging and the use of a
command_ret in enable_events functions. Previously, enable_events was
reusing the ret variable for another operation and always replacing ret.
Parse_probe_event returned the last output of sscanf which represent
the number of match and not the success of the operation.
Fixes #830
Signed-off-by: Jonathan Rajotte Julien <jonathan.r.julien@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
This page took 0.042101 seconds and 4 git commands to generate.