Fix: Free variable before strdup() in process_event_node

Fixes coverity #1243040

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Free variables before strdup()

Fixes coverity #1243047 and #1243050

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Add missing free() in spawn_viewer

Fixes coverity #1291948

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Add missing free() in utils_partial_realpath

Fixes coverity #1328401

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in test_create_kernel_event

Found by Coverity:
CID 1243030 (#1 of 1): Buffer not null terminated (BUFFER_SIZE_WARNING)1.
buffer_size_warning: Calling strncpy with a maximum size argument of 256
bytes on destination array ev.name of size 256 bytes might leave the
destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in send_viewer_streams

Found by Coverity:

CID 1243037 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)18. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array
send_stream.path_name of size 4096 bytes might leave the destination
string unterminated.

CID 1243037 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)18. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array
send_stream.channel_name of size 255 bytes might leave the destination
string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in viewer_list_sessions

Found by Coverity:

CID 1243025 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)17. buffer_size_warning: Calling strncpy with a
maximum size argument of 64 bytes on destination array
send_session->hostname of size 64 bytes might leave the destination
string unterminated.

CID 1243025 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)17. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array
send_session->session_name of size 255 bytes might leave the destination
string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in relayd_add_stream

Found by Coverity:

CID 1243017 (#1 of 4): Buffer not null terminated
(BUFFER_SIZE_WARNING)14. buffer_size_warning: Calling strncpy with a
maximum size argument of 264 bytes on destination array msg.channel_name
of size 264 bytes might leave the destination string unterminated.

ID 1243017 (#2 of 4): Buffer not null terminated
(BUFFER_SIZE_WARNING)14. buffer_size_warning: Calling strncpy with a
maximum size argument of 264 bytes on destination array
msg_2_2.channel_name of size 264 bytes might leave the destination
string unterminated.

CID 1243017 (#3 of 4): Buffer not null terminated
(BUFFER_SIZE_WARNING)15. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array msg.pathname of
size 4096 bytes might leave the destination string unterminated.

CID 1243017 (#4 of 4): Buffer not null terminated
(BUFFER_SIZE_WARNING)15. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array
msg_2_2.pathname of size 4096 bytes might leave the destination string
unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in relayd_create_session_2_4

Found by Coverity:
CID 1243024 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)2. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array msg.session_name
of size 255 bytes might leave the destination string unterminated.

CID 1243024 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)3. buffer_size_warning: Calling strncpy with a
maximum size argument of 64 bytes on destination array msg.hostname of
size 64 bytes might leave the destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in session_create

Found by Coverity:

CID 1323138 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)3. buffer_size_warning: Calling strncpy with a
maximum size argument of 64 bytes on destination array session->hostname
of size 64 bytes might leave the destination string unterminated.

CID 1323138 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)3. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array
session->session_name of size 255 bytes might leave the destination
string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in consumer_set_subdir

Found by Coverity:
CID 1243015 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)8. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array
consumer->subdir of size 4096 bytes might leave the destination string
unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in syscall_init_table

Found by Coverity:
CID 1243021 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)25. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array (syscall_table +
index).name of size 255 bytes might leave the destination string
unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in write_pidfile

Found by Coverity:
CID 1243023 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)3. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array pidfile_path of
size 4096 bytes might leave the destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in list_lttng_channels

Found by Coverity:
CID 1243018 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)11. buffer_size_warning: Calling strncpy with a
maximum size argument of 256 bytes on destination array (channels +
i).name of size 256 bytes might leave the destination string
unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in cmd_snapshot_record

Found by Coverity:
CID 1243027 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)20. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array tmp_output.name
of size 255 bytes might leave the destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in output_init

Found by Coverity:

CID 1243028 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)5. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array output->name of
size 255 bytes might leave the destination string unterminated.

CID 1243028 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)10. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array
output->consumer->dst.trace_path of size 4096 bytes might leave the
destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in consumer_set_network_uri

Found by Coverity:
CID 1243029 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)31. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array obj->subdir of
size 4096 bytes might leave the destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in cmd_snapshot_list_outputs

Found by Coverity:

CID 1243031 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)22. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array (list +
idx).ctrl_url of size 4096 bytes might leave the destination string
unterminated.

CID 1243031 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)26. buffer_size_warning: Calling strncpy with a
maximum size argument of 255 bytes on destination array (list +
idx).name of size 255 bytes might leave the destination string
unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in list_events

Found by Coverity:
CID 1243022 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)23. buffer_size_warning: Calling strncpy with a
maximum size argument of 256 bytes on destination array (tmp_events +
i).name of size 256 bytes might leave the destination string
unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in disable_event

Found by Coverity:
CID 1243016 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)14. buffer_size_warning: Calling strncpy with a
maximum size argument of 256 bytes on destination array msg.name of size
256 bytes might leave the destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in add_uri_to_consumer

Found by Coverity:

CID 1243038 (#1 of 1): Buffer not null terminated
(BUFFER_SIZE_WARNING)15. buffer_size_warning: Calling strncpy with a
maximum size argument of 4096 bytes on destination array
consumer->dst.trace_path of size 4096 bytes might leave the destination
string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: illegal memory access in _cmd_enable_event

Found by Coverity:

CID 1321742 (#1 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)21. buffer_size_warning: Calling strncpy with a
maximum size argument of 256 bytes on destination array attr->name of
size 256 bytes might leave the destination string unterminated.

CID 1321742 (#2 of 2): Buffer not null terminated
(BUFFER_SIZE_WARNING)22. buffer_size_warning: Calling strncpy with a
maximum size argument of 256 bytes on destination array attr->name of
size 256 bytes might leave the destination string unterminated.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Implement lttng_strncpy safe string copy

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: loading a session prints an error message but the load is successful

Fixes #1013

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Snapshot warning if there are no events in one of the domains

Remove the "no data" check from the UST snapshot code since

1) It would report that no data was recorded in the snapshot even
   if the kernel tracer had recorded events,
2) The kernel consumer does not seem to report such a condition,
   making it hard to perform the check without modifying the
   consumer.

The consumer interfaces could be modified to reliably provide
this information in the future.

However, this warning is not provided in the other tracing
modes, making its usefulness questionable. An empty snapshot is
still a valid trace and users should not always expect a trace
to contain events. Moreover, "warning" that no data was produced
implies an error may have occured and is probably as confusing
as opening an empty trace to the average user.

Fixes #1012

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Snapshot: record: use same datetime for snapshot folder output

Make sure that the ust snapshot is put under the same folder as the
kernel snapshot.

Fixes #970

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Refactor: move the snapshot type check into output type check

Since for both sub-session type the boilerplate is the same
simply move the check for sub-session type presence to the inner
check of type output.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

configure.ac: check for an existing liburcu-cds symbol

Fixes #1001

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

configure.ac: use macro for wrong liburcu error message

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Tests: wording of trace_matches is misleading

trace_matches reports a number of "events found" in a trace while it
actually counts the number of event_name matches.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Typo: catched -> caught

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: free ust channel object after grace period in consumer

[ Backport from master branch
commit b83e03c499 "Fix: free metadata cache after grace period in consumer" ]

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Test fix: ignore stderr at every step of randstring

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Update version to v2.6.2

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: add missing synchronization point for before app test case

Fixes a race where the application could generate all its events before
trace start.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix warning when building Python bindings

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Tests: use configured processor count in tracefile count test

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Tests: use configured processor count in snapshot tests

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Tests: Introduce conf_proc_count()

Tests are using the nproc utility which return the number of
_available_ CPUs. The distinction between online, available and
configured CPUs is subtle, but important.

The number of "online" CPUs can change at runtime as CPUs are
hot-plugged. This could happen during a test and result in
unexpected results.

The number of "configured" CPUs includes any CPU which may be
offline at the time.

The number of "available" CPUs, which is what is returned by the
nproc utility, may differ from both "online" and "configured"
counts. This is the case in containers which are assigned to a
subset of configured CPUs.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Test: prevent the spawning of a daemonized sessiond

Sets the default lttng-sessiond path to /bin/true to prevent the spawning
of a daemonized sessiond. This is necessary since 'lttng create' will spawn
its own sessiond if none is running. It also ensures that 'lttng create'
fails when no sessiond is running.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: do not print error and bug messages when quiet (-q) is present

Conform to the lttng man page

fixes #988

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Build: fallback to AC_CHECK_LIBS when looking for popt and uuid

Not all distro ship .pc so fallback to basic libs searching if necessary.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Reviewed-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

tests/unit: fix object files' location

Referring to *.o files under a .libs/ directory is not recommended
because this belongs to libtool's innards.

Indeed, libtool decides to place the *.o files in an
implementation-specific location:
- PIC *.o files go into a .libs/ directory;
- non-PIC *.o files are generated along side to their corresponding
  source files.

Using PIC objects to build executable is legit, thought it may
introduce some minor overhead at runtime.

However, hard-coding these PIC object files in the Makefile.am to build
executables breaks the build in case of static only build.

In this case, no PIC object files is generated, so the linker will not
found some of the needed objects files.

Changing these dependencies' path fixes the static build, keeping the
shared one ok, though the non-PIC object files are now always built.

Fixes #983.

Fix tested on git master and v2.6 with no change needed.

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

configure.ac: fix static build

For static build, some extra LDFLAGS may be needed.

Using PKG_CHECK_MODULES instead of AC_CHECK_LIB for librairy detection
allows to get all these flags. Then, the LIBS variable can be extended
with everything that is needed.

So, use PKG_CHECK_MODULES for popt and uuid detection; which both depend
on libintl.

This changes fixes build failures triggered with Buildroot, e.g.:
  http://autobuild.buildroot.net/results/0f1/0f1e015a0c5a5ac2beeb5011d31a1e0058a32a0d/build-end.log

Signed-off-by: Samuel Martin <s.martin49@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: close indexes when rotating the trace files in mmap mode

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: close indexes when rotating the trace files in splice mode

The consumer needs to close the old index file when doing a file
rotation before opening a new one.
The relay does not have this problem (handled with refcounts).

Signed-off-by: Julien Desfossez <jdesfossez@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: use liburcu flavor header instead of local copy

The layout of struct rcu_flavor_struct has changed in newer URCU
versions. There is no need to keep a local copy of this structure. Use
the library header instead.

This ensures that the caller of _cds_lfht_new(), which passes a flavor
pointer with the library layout defined by liburcu, has the same layout
expected internally within the local copy of rculfhash.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: rculfhash should not include specific rcu flavor

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix live timer calculation error

There is an calculation error for live timer. Variable
switch_timer_interval is based on microsecond, and it is not
right to assign switch_timer_interval mod 1000000 to var tv_nsec
which is based on nanosecond.

Signed-off-by: Mikael Beckius <mikael.beckius@windriver.com>
Signed-off-by: Jianchuan Wang <jianchuan.wang@windriver.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Update version to v2.6.1

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Revert "Fix: add missing test file to EXTRA_DIST"

This reverts commit fb7d878577570de1470d891757abad9f078ebeec.

Fix: regression tests

Fix racy session/relayd wait-after-kill scheme.
Fix racy live test where application may not have generated events yet
when we attach to the live trace.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: tests: stderr not correctly redirected

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Help: add -j -l -p option to help string

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Only log app handle release on failure

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: prevent dangling pointer in utils_partial_realpath

Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: rcu_read_unlock without parentheses has no side-effect

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: add missing test file to EXTRA_DIST

Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: disable all ust events

Handle * event correctly.

Fixes #936

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: sessiond: disable: match app event by name

The use of a simple lookup and match on event name is insufficient
to identify the corresponding ust app event.

Fixes #914

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: dereferencing null index pointer

Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: memleak in utils_partial_realpath

Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: error on no/multiple domain options

Fixes: #927
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd: handle consumerd crashes without leak

We can be clever about indexes partially received in cases where we
received the data socket part, but not the control socket part: since
we're currently closing the stream on behalf of the control socket, we
*know* there won't be any more control information for this socket.
Therefore, we can destroy all indexes for which we have received only
the file descriptor (from data socket). This takes care of consumerd
crashes between sending the data and control information for a packet.
Since those are sent in that order, we take care of consumerd crashes.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: LPOLLHUP and LPOLLERR when there is still data in pipe/socket

The event mask returned by poll/epoll is a bitwise mask made of all the
events observed. On bidirectional sockets, there are cases where
combinations of LPOLLHUP/LPOLLERR and LPOLLIN/LPOLLPRI can be raised at
the same time.

Currently the overall behavior in sessiond, consumerd and relayd is to
handle LPOLLHUP or LPOLLERR immediately, whether or not there is still
data to read in the socket. Unfortunately, this behavior may discard the
last information made available on the pipe or socket.

Audit all uses of LPOLLHUP and LPOLLERR on sockets on which we expect
data to ensure that we deal with LPOLLIN or LPOLLPRI, and catch the
hangup when read or recvmsg returns 0. Keep the LPOLLHUP and LPOLLERR
handling, but only when LPOLLIN is not raised, just in case some
unforeseen error happens when sending the reply.

This is one correct case where we can handle LPOLLHUP and LPOLLERR
directly without caring about LPOLLIN: sockets where we are expected to
write and then read the reply (e.g. command sockets). It is then OK
for a dedicated thread to watch for LPOLLHUP and LPOLLERR.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Remove dot after enable-event message

The other domains' enable event confirmation messages don't have
a trailing dot. Removing this one for consistency.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: don't print the default channel name when enabling agent events

Enabling an event in the python domain erroneously reported the
channel as being the default `channel0`. Instead, don't report the
channel name when enabling an event in an agent domain.

Fixes: #910
Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: fail gracefully on --exclude on unsupported domains

Trying to use event name exclusions on unsupported domains other than
kernel (i.e. log4j, jul, and python) would hang the client. Instead,
report the error appropriately.

Fixes: #909
Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: initialize live_timer to 0 for snapshot session

The live timer was being initialized to -1 for snapshot sessions,
instead of the expected default value of 0 used elsewhere in the code.

Fixes #879

Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: correct mismatched function signatures

The extern declaration of `_lttng_create_session_ext` in `create.c`
had a superfluous `live_timer` parameter not present in the actual
function definition in `lttng_ctl.c`. The -1 value with which it was
called was therefore unused.

Signed-off-by: Antoine Busque <abusque@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: consumer signal handling race

If a signal comes in after ctx has been destroyed, it will try to use a
closed file descriptor.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: list_ust_events(): dangling pointer

Fixes #908

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Buggy string comparison in ust registry ht_match_event

The second strncmp compares the first "strlen(event->signature) != 0"
characters of the event signatures because of a missing parenthesis.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Overwrite of ret in relay_recv_metadata

relay_recv_metadata() interchangeably uses ret and size_ret.
This causes ret to take various (positive) values in case
of success, most often corresponding to the size of the metadata
padding which was written during the call.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Wait for in-flight data before closing a stream

A stream's closing conditions are evaluated in three places:
    1) When a close command is received
    2) When the control connection owning it is closed
    3) The stream has received all of its data following
       a close request.

These checks are performed in try_stream_close().

A known downside of this approach is that a stream will never
be closed if it has not received all of its data.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: unpublish stream on close

Fixes a race where data connection can still add indexes after close,
preventing graceful teardown of the stream.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Tests: Fix flaky live test client

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: Announce empty streams on live attach

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd: file rotation and live read

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relay: viewer_get_next_index handle null vstream

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd: make viewer streams consider metadata sent

The metadata stream does not use prev seq, and is therefore not sent to
viewers if we depend on prev seq. Use the metadata_received field
instead to achieve the same purpose: if a viewer try to attach to a
session that has not received metadata yet, it will get and error
(metadata stream cannot be found when attaching).

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: don't expose empty streams

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd: don't check new metadata on get packet

We only care about this when we get the next index. There may be new
metadata that appears between get next index and get packet, but it
should not matter.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd: don't check for new streams in get packet

Only needed in get next index.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: ask new streams HUP

Test closed streams for content to check for HUP.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: reply error if get packet vstream fails

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd reply error to client if cannot find viewer stream

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: relayd reply with error if cannot find metadata

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: ust-app: protect app socket protocol with lock

Many threads can access the application socket (cmd handling thread and
application handling thread) concurrently. Therefore, we need to protect
it with a mutex.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
 Conflicts:
src/bin/lttng-sessiond/ust-app.c

Fix: add missing rcu_barrier at end of sessiond main

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
 Conflicts:
src/bin/lttng-sessiond/main.c

Fix: add missing rcu_barrier at end of consumer main

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
 Conflicts:
src/bin/lttng-consumerd/lttng-consumerd.c

Fix: app cmd leak on sessiond exit

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
 Conflicts:
src/bin/lttng-sessiond/main.c

Fix: relayd live don't send incomplete stream list

Sample the "closed" flag with session lock held.
Also, if the session connection is closed while we attach to it, reply
with HUP, because there are risks of attaching to an incomplete session
(e.g. the metadata stream could be already closed).

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: consumer timer misses RCU thread registration

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: sessiond consumer thread should register as RCU thread

Fixes RCU race where objects are accessed by this thread under RCU
read-side lock after free. Since this thread is not a registered RCU
reader, the read-side lock has no effect.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: don't chain RCU free

We only do a single rcu_barrier() on teardown of sessiond. Therefore, if
we chain call_rcu, they may not all be executed before exit.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: sessiond vs consumerd push/get metadata deadlock

We need to unlock the registry while we push metadata to break a
circular dependency between the consumerd metadata lock and the sessiond
registry lock. Indeed, pushing metadata to the consumerd awaits that it
gets pushed all the way to relayd, but doing so requires grabbing the
metadata lock. If a concurrent metadata request is being performed by
consumerd, this can try to grab the registry lock on the sessiond while
holding the metadata lock on the consumer daemon. Those push and pull
schemes are performed on two different bidirectionnal communication
sockets.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: sessiond vs consumerd push/get metadata deadlock

We need to unlock the registry while we push metadata to break a
circular dependency between the consumerd metadata lock and the sessiond
registry lock. Indeed, pushing metadata to the consumerd awaits that it
gets pushed all the way to relayd, but doing so requires grabbing the
metadata lock. If a concurrent metadata request is being performed by
consumerd, this can try to grab the registry lock on the sessiond while
holding the metadata lock on the consumer daemon. Those push and pull
schemes are performed on two different bidirectionnal communication
sockets.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: streamline ret/errno of run_as()

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
 Conflicts:
src/bin/lttng-sessiond/ust-registry.c
src/common/runas.c
src/common/ust-consumer/ust-consumer.c

Fix: Double unlock on error path

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>