Mathieu Desnoyers [Thu, 13 Dec 2018 18:56:35 +0000 (13:56 -0500)]
Fix: run_command_wait() handle partial write
Use lttng_write() to handle partial writes (writing less than the
requested amount of bytes) as well as ret = -1, errno = EINTR.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 12 Dec 2018 22:24:11 +0000 (17:24 -0500)]
Fix: do not repurpose iterator while it is being used
The hash table iteration uses an iterator that needs to stay valid for
the next loop. Using that same iterator variable in a nested lookup
in a different hash table leads to segmentation fault.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 12 Dec 2018 20:11:15 +0000 (15:11 -0500)]
Fix: handle_notification_thread_command: handle partial read
Use lttng_read() to handle partial reads (returning less than the
requested amount of bytes) as well as ret = -1, errno == EINTR.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 12 Dec 2018 20:11:14 +0000 (15:11 -0500)]
Fix: notification thread: free session trigger list on error
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 12 Dec 2018 17:16:44 +0000 (12:16 -0500)]
Fix: notification thread: RCU-safe reclaim of hash table nodes
Nodes that are put in a rculfhash hash table created with the
"auto resize" flag need to beware that a worker thread can access the
hash table nodes as a RCU reader concurrently, and that this worker
thread can modify the hash table content, effectively adding and
removing "bucket" nodes, and changing the size of the hash table
index.
Therefore, even though only a single thread reads and updates the hash
table, a grace period is needed before reclaiming the memory holding
the rculfhash nodes.
Moreover, handle_notification_thread_command_add_channel() misses a
RCU read-side lock around iteration on the triggers hash table. Failure
to hold this read-side lock could cause segmentation faults when
accessing hash table objects if a hash table resize is done by the
worker thread in parallel with iteration over the hash table.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 18 Dec 2018 19:01:08 +0000 (14:01 -0500)]
Fix: error logged on partial recvmsg() in MSG_DONTWAIT
The relay daemon logs a "Resource temporarily unavailable" error
message when the lttcomm_recvmsg_inet_sock() is invoked and
no data is left to be consumed from the lttcomm_sock.
The "recvmsg" socket operation is called in a loop by the relay
daemon to consume the data being received in 64k chunks. If, on
one of those iterations, 0 bytes are available, recvmsg() will
return an error (-1, errno = EAGAIN). This should not be
logged in non-blocking mode.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 14 Dec 2018 02:24:35 +0000 (21:24 -0500)]
Print UTF-8 SI suffix only when allowed by the locale
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 13 Dec 2018 22:30:56 +0000 (17:30 -0500)]
Cleanup: duplicate LDADD of libcommon for utils unit tests
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 13 Dec 2018 22:06:29 +0000 (17:06 -0500)]
Move time utils to their own time.c file
Mathieu Desnoyers [Mon, 19 Nov 2018 21:13:58 +0000 (16:13 -0500)]
Fix: sessiond: don't allocate buffers and files for inactive sessions
When tracing is inactive (before start/after stop), the current behavior
is to track all applications registered as UST data producers and
allocate buffers and files.
However, we guarantee that the trace is readable (invariant) after a
"stop" command has waited for data pending to complete. Unfortunately,
tracking additional applications (and adding their files) after tracing
is stopped (for each pid in per-pid buffers, for new uid in per-uid
buffers) does not respect this guarantee.
Fix this by *not* allocating channels, events, contexts when tracing
is inactive, but rather allocate those lazily just before tracing
starts.
One reason why this was not originally done was to ensure we could
have a fast start command. There are however other ways to achieve this
in the future that will respect the stop invariant guarantees.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Mon, 19 Nov 2018 21:12:17 +0000 (16:12 -0500)]
Cleanup: ust start/stop trace
Move setting/clearing the session "active" state from cmd.c to
cmd_start_trace()/cmd_stop_trace() for better encapsulation of
behavior.
Reduces the amount of code to maintain in the catch-all cmd.c.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Mon, 19 Nov 2018 21:09:28 +0000 (16:09 -0500)]
Fix: relayd: rotation pending off-by-one
We need to compare with <= rather than < in the rotation pending
check on the relay daemon side, similarly to the check done in
the consumer daemon check_stream_rotation_pending().
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Fri, 16 Nov 2018 18:25:00 +0000 (13:25 -0500)]
Fix: tests: test_crash should start sessions
test_crash expects side-effects of directory creation to happen while
tracing is still stopped. In preparation for changing that behavior,
ensure that tracing is started when those side-effects are expected.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 12 Dec 2018 03:42:19 +0000 (22:42 -0500)]
Fix: missing session reference release on kernel poll update
The iteration performed on all sessions in update_kernel_poll() does
not release the reference taken on the sessions. This causes the
session(s) to be leaked and prevents the shutdown of the session
daemon as it waits for all sessions to be destroyed before completing
its teardown.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 12 Dec 2018 03:27:20 +0000 (22:27 -0500)]
Fix: set client socket permissions after launch of client thread
The client thread is now the owner of the client socket. As the
client socket is now created by the client thread, the socket's
permissions must be set after the launch of the client thread.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 12 Dec 2018 02:26:42 +0000 (21:26 -0500)]
Fix: use assignment-suppression for unused sscanf arguments
This removes the conversion of elements parsed by sscanf() which
are not used anyhow and eliminates a warning on x86 builds
(%lu used on size_t).
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 11 Dec 2018 21:54:20 +0000 (16:54 -0500)]
Fix: report initialization error of app registration thread
The health check tests use the testpoints() in the application
registration thread to force a pthread_exit() or simulate a
catastrophic error within the thread.
The testpoints were moved before the signal that the thread's
initialization was completed by recent changes. This caused the thread
to fail to complete its initialization, causing a deadlock of the
session daemon on launch.
This commit reports initialization errors through the
launch_application_registration_thread() function to the "main" thread
and shuts down the session daemon. It also moves the testpoints after
the thread's initialization to respect the test's intent.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 11 Dec 2018 21:54:01 +0000 (16:54 -0500)]
Fix: report initialization error of client thread
The health check tests use the testpoints() in the client thread
to force a pthread_exit() or simulate a catastrophic error within
the client thread.
The testpoints were moved before the signal that the thread's
initialization was completed by recent changes. This caused the
thread to fail to complete its initialization, causing a deadlock
of the session daemon.
This commit reports initialization errors through the
launch_client_thread() function to the "main" thread and shuts down
the session daemon. It also moves the testpoints after the thread's
initialization to respect the test's intent.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Francis Deslauriers [Wed, 28 Nov 2018 23:04:06 +0000 (18:04 -0500)]
Cleanup: consumer socket creation debug msg always prints fd:-1
Set the fd to -1 _after_ we print the debug message.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Francis Deslauriers [Wed, 21 Nov 2018 00:16:18 +0000 (19:16 -0500)]
Cleanup: remove unused label
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Francis Deslauriers [Wed, 21 Nov 2018 00:15:46 +0000 (19:15 -0500)]
Fix: may be used uninitialized warnings
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Francis Deslauriers [Mon, 26 Nov 2018 16:41:51 +0000 (11:41 -0500)]
userspace-probe: tests: add testcase for unsupported instrumentation
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Francis Deslauriers [Sat, 24 Nov 2018 00:20:45 +0000 (19:20 -0500)]
userspace-probe: Print error on binary not found
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Francis Deslauriers [Sat, 24 Nov 2018 00:03:21 +0000 (19:03 -0500)]
userspace-probe: Print error for unsupported instrumentation mode
This patch adds an error message printed when the user tries to place an
userspace-probe using unsupported probe location descriptions.
Hopefully, this will help users understand why their command failed.
Here are examples of unsupported uses of an address location:
* "elf:/path/to/binary:0x400430"
* "elf:/path/to/binary:
4194364"
Here are examples of unsupported uses of offset from symbol location:
* "elf:/path/to/binary:my_symbol+0x323"
* "elf:/path/to/binary:my_symbol+43"
I expect users to try using the address locations and offset from symbol
locations for ELF instrumentation location because those methods are
available with the --probe option used to instrument the kernel.
Supporting those location descriptions in the future would require
the validation that the address or the offset from a symbol is indeed at
the boundary of an instruction.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Tue, 27 Nov 2018 19:24:01 +0000 (14:24 -0500)]
Fix: use sys/types.h for ssize_t on Cygwin
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Tue, 27 Nov 2018 19:24:27 +0000 (14:24 -0500)]
Add *.exe to gitignore for Cygwin
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 29 Nov 2018 21:22:07 +0000 (16:22 -0500)]
Revert stubbing of runas functions
All the runas functions were stubbed on builds where the sessiond isn't
built which is the case for all platforms except Linux. This was done
because of 2 new commands that require elf.h which is not present on
MacOSX. However the other commands can be used by the relayd.
Revert this and and only stub the relevant commands when "elf.h" is not
present on the system.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 29 Nov 2018 21:22:06 +0000 (16:22 -0500)]
Revert stubbing of unix socket functions
Instead of stubbing useful UNIX socket functions to work around
Linux-only credential passing, ifdef the relevant parts like it was
already done for other functions.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 29 Nov 2018 21:49:51 +0000 (16:49 -0500)]
Fix: warning 'fd' may be used uninitialized
Initialize fd to invalid '-1' and remove unnecessary file_opened.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 20:49:04 +0000 (15:49 -0500)]
Fix: worker structure is leaked in run_as process
The run_as structure (handle) is allocated and initialized before
the fork() that spawns the run_as process. Currently, that structure
is only cleaned-up on the parent's end.
This fix performs the clean-up on the worker's side as well.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 20:38:14 +0000 (15:38 -0500)]
Fix: ensure the ht clean pipe is empty before processing quit pipe
The ht-cleanup thread does not ensure that all data pending on its
ht_clean_pipe has been read (and processed) before processing the
"quit" event on the quit_pipe.
This causes a number of urcu hash tables to be leaked on exit of
the sessiond even though they have been pushed on the cleanup
pipe.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 19:47:34 +0000 (14:47 -0500)]
Perform the clean-up of application notify sockets in main thread
The notify sockets of applications are owned by the "notify-apps"
thread. If an application exits during the course of the session
daemon's life, the notify thread will take care of cleaning its
associated notify socket.
However, there is no teardown/clean-up code to handle their
clean-up when the daemon is torn down.
This change adds a new step in the sessiond clean-up code that
closes closes all notify sockets present in the
ust_app_by_notify_sock hash table at the time of the sessiond
cleanup.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 19:22:20 +0000 (14:22 -0500)]
Clean-up: remove redundant exit labels from sessiond initialization
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 17:03:48 +0000 (12:03 -0500)]
Make the launch of the application registration thread blocking
Waiting for the application registration thread to be ready to
accept application connection ensures that the application
registration socket is created and being listened-to before the parent
is signalled.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 16:09:32 +0000 (11:09 -0500)]
Stop the application registration thread before orphaned threads
The application registration thread receives new connections from
applications, provides them to the dispatch thread. The dispatch
thread, in turn, forwards the command and notification sockets of
applications (liblttng-ust) to the application management and
application notification threads.
Not shutting down the application registration thread is problematic
since application connections will be accepted but not associated
to an "ust_app" structure as the following threads are no longer
present.
The remaining threads can then be safely torn down as part of the
orphaned threads.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 6 Dec 2018 14:39:02 +0000 (09:39 -0500)]
Rename ust-thread to notify-apps
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 5 Dec 2018 20:00:09 +0000 (15:00 -0500)]
Teardown the notification thread after the sessiond clean-up
The notification thread may receive commands issued through the
call_rcu thread during the destruction of some of the sessiond's
data structure.
This change tears down the notification thread after the clean-up
has occured and the issuance of an RCU barrier. This ensures that
all previously-queued call_rcu work has been performed and that
any ensuing notification thread commands have been queued in return.
It is safe, at that point, to queue a "quit" command in the
notification thread's command queue. The notification thread's
shutdown method will issue the command and wait for its completion
before returning.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 5 Dec 2018 04:19:25 +0000 (23:19 -0500)]
Launch the consumer management thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 4 Dec 2018 22:46:00 +0000 (17:46 -0500)]
Stop sessions before destroying on teardown of session daemon
Stopping sessions ensures that trace data is no longer produced
for a session which will allow pending rotations to complete.
It also ensure that no data is produced beyond the last rename
of a rotated session's output folder.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 4 Dec 2018 16:50:30 +0000 (11:50 -0500)]
Remove the sessiond "ready" counter mechanism
This commit replaces the sessiond "ready" counter scheme with
the use of the lttng_thread util. The launch of the threads which
need to be active before the sessiond can signal its parents
(when launched in daemon mode) is now blocking. This means that
their associated "launch" functions wait until the threads mark
themselves as ready (through the use of a "ready" semaphore)
before returning and allowing the initialization of the sessiond
to continue.
The threads which expose externally-visible resources (UNIX and
TCP sockets) which must be fully initialized before marking the
session daemon as ready are:
- Health thread,
- Agent thread,
- Client thread.
Previously, the "load session" thread was part of this group.
However, it is no longer necessary to perform the loading of
session configurations in a dedicated thread. The main thread
performs that operation itself. It is safe to do so since it
is performed after the launch of the client thread. The client
thread has to be fully initialized as the session loading code
"impersonates" a client to initialize the loaded sessions.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 3 Dec 2018 00:04:45 +0000 (19:04 -0500)]
Load session configurations from lttng-sessiond's main thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Sun, 2 Dec 2018 22:06:45 +0000 (17:06 -0500)]
Launch the kernel management thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Sun, 2 Dec 2018 21:20:16 +0000 (16:20 -0500)]
Launch agent management thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Sat, 1 Dec 2018 01:04:08 +0000 (20:04 -0500)]
Mark lttng_pipe as const where possible
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 4 Dec 2018 22:37:45 +0000 (17:37 -0500)]
Launch the application notification thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Sat, 1 Dec 2018 00:25:06 +0000 (19:25 -0500)]
Launch the application management thread with lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 30 Nov 2018 21:33:38 +0000 (16:33 -0500)]
Launch application registration thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 30 Nov 2018 20:57:04 +0000 (15:57 -0500)]
Launch the ust registration dispatch thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 30 Nov 2018 19:29:32 +0000 (14:29 -0500)]
Launch the client management thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 30 Nov 2018 17:12:12 +0000 (12:12 -0500)]
Launch the timer thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 28 Nov 2018 21:53:53 +0000 (16:53 -0500)]
Launch the rotation thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 28 Nov 2018 21:12:25 +0000 (16:12 -0500)]
Fix: flush the rotation thread's job queue on exit
The rotation thread's job queue can transitively hold references
to a number of ltt_session objects which will prevent the session
daemon to exit as it waits for all sessions to have completed
their destruction.
This fix ensures that the job queue is flushed when activity
is observed on the quit pipe.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 27 Nov 2018 23:17:16 +0000 (18:17 -0500)]
Stop rotation pending check timer from the rotation thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 27 Nov 2018 21:37:05 +0000 (16:37 -0500)]
Launch the notification thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Tue, 27 Nov 2018 17:39:06 +0000 (12:39 -0500)]
Launch the health management thread using lttng_thread
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 26 Nov 2018 20:15:24 +0000 (15:15 -0500)]
Launch the ht-cleanup thread using lttng_thread util
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 26 Nov 2018 20:09:10 +0000 (15:09 -0500)]
Add a thread utility class and thread list
As part of the re-work of the order of the teardown of the sessiond's
threads, this utility is introduced to track running threads and unify
the mechanisms through which they are launched and shutdown.
This makes it easier to reason about (and track) the order in which
threads are launched and shutdown.
The lttng_thread class allows threads to be implemented by
defining the following methods:
- An entry point ("main" function),
- A shutdown method,
- A clean-up method.
Since the sessiond's threads use a variety of techniques to initiate
their teardown (through an explicit command in a queue, using the
global "quit" pipe, or through a futex + variable), it is more
practical to let them define a shutdown method which notifies the
thread to shutdown than to impose a standard mechanism.
A clean-up method is meant to work around situations where the
ownership of data structures shared between the main thread and
a worker thread can be ambiguous (mostly in error paths).
The clean-up method is invoked when the last reference to a thread
is released.
While some threads need to be shutdown in a particular order, most of
them can be shutdown in bulk. The lttng_thread utility maintains a
global thread list which allows for a generic path through which
threads can be shutdown using the lttng_thread_list_shutdown_orphans()
function.
The lttng_thread_shutdown() method, in return, allows the user
(most likely the main thread) to explicitly teardown threads which
must be shutdown in a specific order before issuing the bulk
lttng_thread_list_shutdown_orphans() call.
Note that lttng_thread objects are reference counted. The thread
list holds a reference to each thread until it is shutdown. Hence,
it is safe to hold a reference to a thread, invoke its shutdown
method, and then invoke lttng_thread_list_shutdown_orphans().
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 21 Nov 2018 15:41:07 +0000 (10:41 -0500)]
Don't perform an implicit rotation on session stop
Performing a periodic rotation on stop does not allow a user to start
and stop a session within a given chunk.
There is no functionality lost from not performing the rotation
implicitly on stop; the user could decide to explicitly perform the
rotation himself.
The timestamp of the last stop command is sampled to allow the trace
archive chunk to be accurately named should a rotation be performed.
Moreover, this change moves the implicit rotation performed at the
moment of the stop to the moment of the session's destruction. This
ensures that users don't end-up with a partially-named trace archive
chunk folder upon destruction (timestamp_begin-id).
In effect, this will perform an implicit session rotation as part of
the "session destroy" command if the session was rotated at any point
during its lifetime.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 22 Nov 2018 21:07:49 +0000 (16:07 -0500)]
Wait for the destruction of sessions before tearing down main thread
The main thread can no longer assume that it is the last thread to
use the session objects (since they are now ref-counted). Hence,
this commit introduces utils to allow the main thread to wait for
the destruction of all sessions.
The logic of the teardown is as follows:
1) The main thread waits for activity on the quit pipe.
2) Once the main thread unblocks, the main thread waits for the
client thread to stop in order to guarantee that no new sessions
are created.
3) The main thread then waits for the session list to be emptied
4) Once the session list is empty, continue the rest of the teardown
as usual.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Thu, 22 Nov 2018 01:53:43 +0000 (20:53 -0500)]
Fix: rotation_unavailable returned on failure to read time
A "rotation feature unavailable" error code is returned if the
session daemon fails to sample time. The TIME(2) man page has
makes no mention of an error that could happen in this use case
(as time()'s parameter is NULL, EFAULT does not apply).
However, a more general error code is more appropriate should
this call fail.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 21 Nov 2018 23:09:52 +0000 (18:09 -0500)]
Fix: mark rotation pending check timer is never marked as disabled
The rotation pending check timer is launched in "one-shot" mode.
However, it is marked as enabled for the whole duration of the
rotation pending check phase.
This change ensures that the timer is marked as disabled when
the session rotation pending reaches the "completed" state.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 21 Nov 2018 15:30:15 +0000 (10:30 -0500)]
Acquire a reference to a session when a timer is active
The timers associated to a session don't directly
reference a session since, up to recently, session
objects were not reference counted. There was
essentially no mechanism in place to prevent a
session from being destroyed while one of its timers
was active.
For this reason, the session was queried by id on
every execution of its timers. However, this did
not prevent the session from being destroyed; it only
allowed the periodic jobs to handle that condition
gracefully.
This change that a reference to the session is held
at all times by periodic jobs that are "in-flight".
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 21 Nov 2018 00:24:58 +0000 (19:24 -0500)]
Reference count ltt_session objects
The lifetime of ltt_session objects is mishandled in a number
of situations. For instance, if a rotation is launched on a
session and the session is destroyed during the rotation, it
is not possible for the session daemon to identify the completion
of the rotation. This then prevents the session daemon from
renaming the last chunk of a now-destroyed session and from
notifying clients that a new trace archive chunk is available.
This change only introduces reference counting of the ltt_session
objects; it does not change any behavior with regards to the
lifetime of the sessions themselved. Follow-up commits introduce
those changes.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 16 Nov 2018 22:27:09 +0000 (17:27 -0500)]
Fix: session info reference is not put on unhandled condition type
This commit is only part of the stable-2.11 branch since the bug
results from a bad merge from master.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 15 Nov 2018 17:24:48 +0000 (12:24 -0500)]
Fix: typo 'occured' -> 'occurred'
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 15 Nov 2018 17:24:47 +0000 (12:24 -0500)]
Fix: typo 'retrive' -> 'retrieve'
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 15 Nov 2018 17:16:37 +0000 (12:16 -0500)]
Fix: typo 'convet' -> 'convert'
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Michael Jeanson [Thu, 15 Nov 2018 17:16:36 +0000 (12:16 -0500)]
Fix: typo 'leat' -> 'least'
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Tue, 13 Nov 2018 17:12:21 +0000 (12:12 -0500)]
Fix: max_t/min_t macros are missing cast on input
The semantic expected from max_t and min_t is to perform the max/min
comparison in the type provided as first parameter.
Cast the input parameters to the proper type before comparing them,
rather than after. There is no more need to cast the result of the
expression now that both inputs are cast to the right type.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Tue, 13 Nov 2018 17:12:20 +0000 (12:12 -0500)]
Fix: Connect timeout arithmetic in inet/inet6 (v4)
The nanoseconds part of the timespec struct time_a is not always
bigger than time_b since it wraps around each second.
Use 64-bit arithmetic to compute the difference.
Merge/move duplicated code into utils.c.
This function is really doing two things. Split it into
timespec_to_ms() and timespec_abs_diff().
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 16 Nov 2018 21:46:36 +0000 (16:46 -0500)]
Fix: cmd_snapshot_record can return unexpected error codes
cmd_snapshot_record() does not correctly handle the various error
reporting conventions used by the functions it uses. In certain cases,
invalid error codes (not part of enum lttng_error_code) can bubble up
to the command handler which results in erroneous error messages being
reported by the client.
This change harmonizes the error handling and changes a number of
functions used to explicitly return 'enum lttng_error_code' instead
of 'int'.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Fri, 16 Nov 2018 18:55:57 +0000 (13:55 -0500)]
Docs: document why a negative socket may be passed
The fix in
4821263c4 checks for a negative socket. While the
commit message explains the reason for the check, it is not
obvious why this check is needed when looking at the code
itself.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 13 Nov 2018 17:52:04 +0000 (12:52 -0500)]
lttng-enable-event.1.txt: fix SDT acronym, add missing "are supported"
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Tue, 13 Nov 2018 17:48:52 +0000 (12:48 -0500)]
lttng-enable-rotation.1.txt: --timer option should be fairly precise
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 9 Nov 2018 17:01:37 +0000 (12:01 -0500)]
doc/man: remove AUTHORS section
Many people have contributed to LTTng-tools and the original authors are
not necessarily the top ones now. For instance, GitHub reports this top
5 as of this patch:
1. jgalar 952 commits 45,889 ++ 13,959 --
2. compudj 880 commits 56,579 ++ 28,095 --
3. PSRCode 303 commits 25,553 ++ 10,909 --
4. eepp 199 commits 12,119 ++ 6,461 --
5. jdesfossez 174 commits 32,546 ++ 3,381 --
Also, man-pages(7), the conventions for writing Linux man pages, says:
> Use of an AUTHORS section is strongly discouraged.
Let's just keep the project's copyrights and some special thanks without
selecting specific authors to list on each man page.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 9 Nov 2018 16:46:38 +0000 (11:46 -0500)]
lttng-enable-event.1.txt: uprobe: update style and content
* Better inline styles for option variables.
* Option descriptions moved to the OPTIONS section.
* Expand SDT acronym.
* Add note indicating that SDT probes surrounded by reference counts
are not supported.
* Use "application" and "library" instead of "executable".
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Reviewed-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 9 Nov 2018 15:28:49 +0000 (10:28 -0500)]
doc/man: add links between man pages about session rotation
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 9 Nov 2018 00:09:26 +0000 (19:09 -0500)]
lttng-enable-event.1.txt: document dynamic user space probes
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Fri, 9 Nov 2018 00:08:48 +0000 (19:08 -0500)]
lttng-{enable,disable}-rotation.1.txt: both types can be specified
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 8 Nov 2018 21:37:27 +0000 (16:37 -0500)]
lttng-disable-rotation.1.txt: remove LIMITATIONS section
`lttng disable-rotation` only unsets what has been previously set by
`lttng enable-rotation`. The real limitations are for
`lttng enable-rotation` and `lttng rotate`.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 8 Nov 2018 21:36:57 +0000 (16:36 -0500)]
lttng-enable-rotation.1.txt: add that SIZE/PERIOD are approximations
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Philippe Proulx [Thu, 8 Nov 2018 21:36:29 +0000 (16:36 -0500)]
lttng-rotate.1.txt: lttng-relayd(8) also has --output option
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 20:34:49 +0000 (15:34 -0500)]
Fix: sessiond: ust_app_rotate_session error handling
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 21:46:05 +0000 (16:46 -0500)]
Fix: sessiond: snapshot: handle consumer return codes
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 21:35:52 +0000 (16:35 -0500)]
Fix: consumer: snapshot error return code
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 21:28:24 +0000 (16:28 -0500)]
Fix: sessiond: consumer.c: rotation error handling
Return more specific error codes in the various error paths
of the session rotation commands.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 21:11:48 +0000 (16:11 -0500)]
Fix: consumer: rotation error return codes
The return codes of rotation commands from consumerd are not
in sync with reality. Some are simply copy-pasted from old code.
Add new return codes to describe each error situation, and split the
"channel lookup" error from other errors so sessiond can distinguish
between an error caused by an exiting application (per-pid buffers)
(LTTCOMM_CONSUMERD_CHAN_NOT_FOUND) and an actual error while performing
the command.
Move the channel lookup into the rotation caller, because it is used by
two sub-functions: we don't want to lookup to succeed for the first and
then fail for the second.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 22:49:21 +0000 (17:49 -0500)]
Fix: create_channel_per_pid: remove channel on error
The notification system of the session daemon tracks the lifetime
of per-pid buffers (in effect, channel keys) using two hooks:
- the successful completion of create_channel_per_pid()
- the deletion of the channel from an application's registry
Multiple error paths in create_channel_per_pid() can leave a channel
in an app's registry without notifying the notification system of its
existence.
When the channel is finally cleaned-up, the channel deletion hook will
inform the notification system that the channel should be deleted,
resulting in an error as that channel was previously unknown.
This fix ensures we remove the application's channel from its registry
on error (with the notify parameter set to 'false'), thus ensuring
that the notification system never tracks the incomplete channel.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 22:46:26 +0000 (17:46 -0500)]
Fix: channel errors on local stress-tests
On local stress-tests with per-pid buffers, error paths triggered
when trying to create a channel on an application already going away
end up tiggering the channel destruction with "-1" socket parameter.
This function attempts to send a notification of channel destruction
(erroneously) and attempts to sum counters into a channel object which
may not have been populated yet.
Therefore, don't perform those side-effects when invoked from error path
caused by channel creation error.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Mathieu Desnoyers [Wed, 14 Nov 2018 22:45:38 +0000 (17:45 -0500)]
Fix: channel_ust_create: remove channel on agent error
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 22:30:17 +0000 (17:30 -0500)]
Fix: cmd_rotate_session() returns unexpected error codes
Certain paths in cmd_rotate_session() result in its return value not
obeying the convention for commands: return LTTNG_OK on success, and a
negative LTTNG_ERR_* code on error.
This patch separates the use of 'int ret' from a separate 'enum
lttng_error_code' value to ensure 'ret' values never bubble-up to the
caller.
Note that this patch assumes that ust_app_rotate_session() returns an
lttng_error_code, which is not the case right now (upcoming patch).
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 22:07:48 +0000 (17:07 -0500)]
Fix: session rotation logging statement references snapshot
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 22:02:04 +0000 (17:02 -0500)]
Clean-up: typo in logging message Totation -> Rotation
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 21:52:12 +0000 (16:52 -0500)]
Fix: rotation error may leave session in "ONGOING" state
The errors that can occur during the launch of a rotation may
leave a session's rotation state in the "ONGOING" state.
This means that clients polling for the rotation's state (or using the
notification channel) will never see the rotation enter the ERROR
or COMPLETED states, resulting in a hang.
This change introduces session_reset_rotation_state() which
implements the logic needed to reset a session's rotation state
and populate the result of the last rotation.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 21:00:15 +0000 (16:00 -0500)]
Fix: session_list lock must be held to launch a rotation
The timer.h functions are documented as needing the session_list
lock to be held. Since timers are setup during the launch of a
rotation, the session_list lock must be taken while a job is
executed by the rotation thread.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 20:43:49 +0000 (15:43 -0500)]
Add an ASSERT_LOCKED(lock) macro
This macro validates that a given lock is taken by using
pthread_mutex_trylock().
The use of this macro is disouraged in hot paths for performance
reasons. It is meant to catch errors in "cold" code path (e.g. error
handling) where the additional safety it provides has no material
effect on performance.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Wed, 14 Nov 2018 20:32:24 +0000 (15:32 -0500)]
Clean-up: move global sessiond symbols out of main.o
The global symbols defined in main.c prevent the use of numerous
functions from various other files. For instance, using timer.h
utils from session.o forces the redefinition of symbols that
are defined in main.o in the session unit tests.
This is a problem that has often occured in the past, more or less
forcing the unit tests to link against all sessiond objects. Moving
the global symbols to their own files (thread-utils.c,
process-utils.c, globals.c, and ready.c) does not solve this problem,
but it allows the unit tests to link against their resulting
objects and forego the stub-ing of symbols.
In the future, we could consider moving most the session daemon
to an internal library which could be re-used by the unit tests.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 12 Nov 2018 21:52:26 +0000 (16:52 -0500)]
Fix: split index and data file rotation logic
Issue
---
There is no guarantee that index and data positions (sequence numbers)
match at a given time as both are received on different sockets.
Currently, the relay determines whether or not it should rotate a
stream's data and index files based on the lowest sequence number
of both the data and index file.
When the data connection "lags" behind the control connection, index
entries received on the control connection that belong in the "next"
chunk are written to the "previous" chunk's index file.
While there is logic to ensure that trace data received after the
current rotation position is copied to the "previous" chunk when the
rotation is performed, the same is not true for the index file
rotation.
This is fine as it should not be needed. The consumerd <-> relayd
protocol ensures that a rotation command issued on a stream will
provide a rotation position that is either:
- the stream's current position,
- a position that will be reached in the future.
Then, the control connection will receive index entries that are
either <= to the rotation target sequence number.
Therefore, it is correct to check if an index file should be rotated
everytime an index entry is flushed and change the index output file
when the rotation position is reached.
Again, this is not true for the data connection as it would be
possible to receive a rotation command with a position that is
before the current data sequence number.
Solution
---
This change splits the logic to evaluate the moment at which a
rotation should be performed (changing the actual file to which data
is written) so that both index and data files are rotated
independently.
Hence, when an index is flushed, the relay will always evaluate
whether or not it should switch the index file to its new destination
(i.e. perform the rotation of the index).
The "data" reception logic remains mostly unchanged, except that
the trace sequence number is no longer used. Only the data stream's
position is considered to evaluate whether the data stream file
should be rotated.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 12 Nov 2018 22:48:53 +0000 (17:48 -0500)]
Fix: trace_archive_id is not sent in add_stream command
The trace_archive_id argument of relayd_add_stream() is not
forwarded to the "2.11" implementation of the command. This means
that all streams on the relayd are created in the "chunk_id == 0"
EPOCH.
This may cause remote rotations to never complete when a stream
is created while a rotation is ongoing. In such a case, the
stream's chunk_id would be 0, but it would not have a rotation
sequence number. This would cause the rotation completion check
to always return that the rotation is "ongoing".
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jérémie Galarneau [Mon, 12 Nov 2018 22:26:46 +0000 (17:26 -0500)]
Fix: condition unsubscription error leaves session locked
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
This page took 0.05833 seconds and 4 git commands to generate.