Fix: sessiond: notification: missing unlock on client skip

Skipping a client must be performed by using the dedicated "skip_client"
label which will unlock the client's lock before continuing the loop
rather than using 'continue' directly.

Currently, a client will remain locked when an hidden trigger emits
a notification to which it is subscribed.

1458230 Missing unlock

May result in deadlock if there is another attempt to acquire the lock.

In notification_client_list_send_evaluation: Missing a release of a lock
on a path (CWE-667)

Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8b69395b91b0ea59ae5e0beadebd9099db623121

Update version to v2.13.0-rc3

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

liblttng-ctl: hide logger_thread_name

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4eb5a86029c6220ad4f48d382ec26126fd82e443

liblttng-ctl: hide MI trigger command variables

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I45eec5bb0fd3353c8f1257b3c94ef08440114b21

Cleanup: rename `get_domain_str()` -> `lttng_domain_type_str()`

Both functions currently exist in the code base and accomplish the same
goal. Let's keep only one of them.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2254b846f0b5bdc883c86d970fde7daffa9e6155

.gitignore: Add hidden trigger test

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iab0fe77c0d4607d5469a7aa57d6bd784d47d8609

Test: unix socket: test credential passing

Since the credential passing over UNIX sockets now makes use of the pid,
the compatiblity wrappers have become more complex as each platform
appears to define its own way of accessing this information.

This new test:
  - creates a named unix socket,
  - forks,
  - gets the parents and child to connect,
  - sends the child's credentials as a data payload and as credentials
    verified by the kernel
  - the parent checks that the two sets of credentials are equal.

This is more of a sanity check for the compatibility wrappers used on
non-Linux platforms.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic0a6213afca7cc95a00617b052e7a145fc88625c

Build fix: retrieve unix socket peer PID on non-unix platforms

The previous attempt at extending the credential retrieval wrapper was
broken and didn't build on FreeBSD, macOS, and cygwin.

A platform-specific way of retrieving the PID of a unix peer is
implemented for FreeBSD (getsockopt using LOCAL_PEERCRED, note that the
cr_pid field is only available from FreeBSD 13 and up),
macOS (getsockopt using LOCAL_PEERPID, macOS 10.8+), and
Solaris (getpeerucreds).

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifcf522c70ee4c2e0799293ae0961f41aebff5056

Fix: sessiond: notification: find_tracer_event_source returns NULL

Due to a bad edit of the original patch (my bad!)
find_tracer_event_source_element always returns NULL.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7febee1d803034a06d5063a2cc9179c4edef4809

Tests: MI: add `diag` statements to test functions

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie56e23a3d0796d1edb07e2fd7cdc259816ac0133

Cleanup: fix comments in `duplicate_{stream,channel}_object()`

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5089d09880d21842bf264f6c30ec7fd5e72b93df

Tests: add hidden trigger visibility test

Add a regression test for the previous commit that verifies that
internal triggers used by the session daemon to implement various
features (automatic session rotations based on their consumed size, in
this instance) are not visible to users of liblttng-ctl.

The test is written in C to use the library directly. This is needed
since the `lttng` client filters-out anonymous triggers and thus, would
not allow us to see those triggers since they are anonymous by default.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I1b8fca648953b8cba49a9888593b3486457d01b2

Fix: sessiond: list-triggers: don't return internal triggers

The session daemon uses triggers internally. For instance, the trigger
and notification subsystem is used to implement the automatic rotation
of sessions based on a size threshold.

Currently, a user of the C API will see those internal triggers if it is
running as the same user as the session daemon. This can be unexpected
by user code that assumes it will be alone in creating triggers.
Moreover, it is possible for external users to unregister those triggers
which would cause bugs.

As the triggers gain more capabilities, it is likely that the session
daemon will keep using them to implement features internally. Thus,
an internal "is_hidden" property is introduced in lttng_trigger.

A "hidden" trigger is a trigger that is not returned by the listings.
It is used to hide triggers that are used internally by the session
daemon so that they can't be listed nor unregistered by external
clients.

This is a property that can only be set internally by the session
daemon. As such, it is not serialized nor set by a
"create_from_buffer" constructor.

The hidden property is preserved by copies.

Note that notifications originating from an "hidden" trigger will not
be sent to clients that are not within the session daemon's process.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I61b7949075172fcd428289e2eb670d03c19bdf71

unix: receive pid on non-linux platforms

Add a `pid` to the lttng_sock_cred structure definition used on
non-Linux platforms and receive the peer's PID when receiving
credentials.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9c92f6dda6441deca58f9cc85f846f5031cceb6e

Clean-up: sessiond: return an lttng_error_code from list_triggers

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5d44b508a2a5211894c0cc7b6d51a9a03dc8b3f2

notification-thread: remove fd from pollset on LPOLLHUP and friends

When an app dies, it's possible that the notification thread gets an
epoll event (`LPOLLHUP`) that the socket was closed before it gets the
_REMOVE_TRACER_SOURCE command for that source.

In such cases, the notification thread should simply remove the file
descriptor from the pollset and drain the notification on that file
descriptor. It should _not_ remove the _source_element object from the
list.

The removal from the list should only be done when it receives the
_REMOVE_TRACER_SOURCE command.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9525315f9e92d0f6ae5e84e26b83a6b7207dce54

Tests: fix: list triggers: bc missing on system

`bc` is not part of the test suite's dependancies and can be replaced,
in this instance, by a use of `printf`.

This use of `bc` caused a number of failures on the CI's Lava workers.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I1a1b24a23325754c26ebedfdb6b7728378381d97

Clean-up: event-expr: remove unreachable code

1452699 Logically dead code

The indicated dead code may have performed some action; that action will
never occur.

In lttng_event_expr_array_field_element_create: Code can never be
reached because of a logical contradiction (CWE-561)

Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I301e73c8e0cc7b9c4fb889e5bf7ef30d6ecf7d9f

Fix: lttng: remove-trigger: null dereference on MI initialization error

Failures to create an MI writer instance will result in a dereference of
the MI writer when attempting to close the command's output element.

1457842 Dereference after null check

Either the check against null is unnecessary, or there may be a null
pointer dereference.

In cmd_add_trigger: Pointer is checked against null but then
dereferenced anyway (CWE-476)

Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0bc71bf6c83df7d9d938cf93a12d5f6cf6d7ae36

Fix: lttng: list-trigger: leak of error query in query callbacks

1457841 Resource leak

The system resource will not be reclaimed and reused, reducing the
future availability of the resource.

In mi_error_query_trigger_callback: Leak of memory or pointers to system
resources (CWE-404)

Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4e2cde41d77e5299d1758e8c9387b0a1c63efd17

Fix: lttng: add-trigger: null dereference on MI initialization error

Failures to create an MI writer instance will result in a dereference of
the MI writer when attempting to close the command's output element.

1457842 Dereference after null check

Either the check against null is unnecessary, or there may be a null
pointer dereference.

In cmd_add_trigger: Pointer is checked against null but then
dereferenced anyway (CWE-476)

Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I98b844d2f1c7abd43bd42ee472759de57b34484e

lttng: add-trigger: print generated trigger name

Print the generated trigger name when `add-trigger` succeeds.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id858880260513b9a10c4ce5022a95c476e3e32aa

sessiond: generate trigger name: name triggers with the 'trigger' prefix

Generated trigger names currently have the form TN, where N is the
number of generated trigger names over the lifetime of the session
daemon.

The form 'triggerN' seems more in line with autogenerated names such
as channel names (e.g. 'channel0').

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id61cd4716bb4c080d9242853c366e14542f60f7c

Revert "lttng: add-trigger: print generated trigger name"

This reverts commit 8310270a50784aced2af5b21ab23bc7bd9dee47f.

This change is still under review.

Change-Id: If75aa02e2e5daa0bfbcf30bea0a2b54c4aca1fd4
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

lttng: add-trigger: print generated trigger name

Print the generated trigger name when `add-trigger` succeeds. Also,
no message is emited when a trigger is successfully registered as
the command will print an error message if any error occurs.

There is also no need to parrot the trigger's name if it was specified
by the user.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9607fbd358298b036bd533834143eb5e9d185cd0

MI: xsd: bump to 4.1

No breaking change were done to the xsd. Only objects related to
triggers, event-rules, actions, condition, and error-query were added.
They do not interfere with the current MI.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ia057c0fbea34f8e5c48cb8d8d307f004acc95a00

Tests: trigger: mi: use utils.sh xsd versions for xml diff

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic1536218b468d300ceb3d16ca160b8a8b891edfc

Tests: utils: regroup xml utils to utils.sh

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idfa0f05d1bde75f4b02c903699281a86494b435f

Tests: MI: {add, list, remove}-trigger

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ica66a759d961cc122c1a1b81ce69fa54b0e78c78

MI: xsd: add objects type definition related to trigger

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If28306f8aaf24890a6d834e9ff69bd00de3da295

MI: xsd: sort output_type

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If2206e4a1c7a54d6d6bc4887c1925b12f035232b

MI: xsd: sort command_string_type

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8df9d69aeaf93050c405ff876ad697efac7c4021

Add pretty_xml utils

This util reads on stdin and outputs an indented/formatted xml.
It is equivalent to "xmllint --format -".

It will be used for MI trigger testing. For testing we will essentially
diff the output of the command against the expected output. While a
nicely formatted multi-line output is not necessary for a machine to
do the diff, the human that will have to debug it will surely appreciate
it.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie1597644941c55ce3e59f7ff16f196ac36325179

Move xml utils from mi subfolder to xml-utils folder

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I268dc544bf4f72f61a701ac3efd0b12488cc2f64

Fix: lttng_triggers count is not equal to the size of the sorted trigger array

Since anonymous triggers can be present in the original lttng_triggers
and that we do not add them to the sorting list, the count to be used
while iterating on the sorted list must be the size of the list itself
and not that of lttng_triggers.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifb1802345199cb20fbb6d401f316be918b8a6443

MI: {add, list, remove} trigger

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie16c5c3a894b921e032a99ed3deda4ed5da17e78

MI: implement all objects related to trigger machine interface

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idb2045135b1ba87853d6214b149afbe27bb7a1ca

Move event-expr-to-bytecode to event-expr

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I74a4b823ae7bbcbb062dbb9a2a0f84785bca287a

Move event-expr from liblttng-ctl to libcommon

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I31c65cd7f63fa4e1c918285b02ab2ab2e82549f6

MI: support double element

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I97411fea238d8b1275028d3d04a6f4f376624001

Fix: rotation client example: leak of handle on error

1452927 Resource leak

The system resource will not be reclaimed and reused, reducing the
future availability of the resource.

In setup_session: Leak of memory or pointers to system
resources (CWE-404)

CID 1452927 (#1 of 1): Resource leak (RESOURCE_LEAK)8. leaked_storage:
Variable chan_handle going out of scope leaks the storage it points to

Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4c215ac4a86f9f70fd5c9d3aa13f944d3d7a2cc7

Silence warnings on GCC 4.8 with -Wmaybe-uninitialized

We still build on SLES12 with GCC 4.8 in which '-Wmaybe-uninitialized'
doesn't seem to be the sharpest tool in the shed. Add explicit
initialization of 'ret' to silence the warnings.

Change-Id: I1f9de535b6be48357735af106ff555ab9eceb730
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

doc/man/common-footer.txt: add missing non-breaking space

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibefd4e7448920f0f346697eea5e1b5d250a93d1f

Rename "tracing session" -> "recording session"

Starting from LTTng 2.13, _tracing_ is defined as attempting to execute
one or more actions when emitting an event, which is very close to the
trigger definition.

To highlight that a tracing session is only about event recording,
rename this concept to _recording session_.

This patch mostly changes the manual pages, although I also updated some
C source and other files which contain user-facing text to use the new
term.

I didn't update logging messages because debugging scripts could still
refer to "tracing sessions".

The lttng-concepts(7) manual page mentions that the "recording session"
term was "tracing session" before LTTng 2.13.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I620d6b6be9e0f1dac14c0bc5e26094c3b3711c75

doc/man: use double quotes when referring to internal section

This patch adds double quotes to all the manual page internal section
references using their full name. Those references often have the
following AsciiDoc form:

    See the <<id,Full section name>> section below.

With this patch, this would be converted to:

    See the ``<<id,Full section name>>'' section below.

In the rendered manual page, before this patch:

    See the Full section name section below.
            ¯¯¯¯ ¯¯¯¯¯¯¯ ¯¯¯¯
With this patch:

    See the “Full section name” section below.

The purpose of this patch is, thanks to the change in
`doc/man/manpage.xsl`, to remove the italic style for the text of
internal links. Because there's no way to create dynamic internal links
in a manual page, this style causes internal links to look weird when
they're not a full section name, for example:

    Note that the trigger doesn't need to [...]
                  ¯¯¯¯¯¯¯
The HTML rendering of LTTng-tools manual pages can still benefit from
internal links. This patch makes it possible to add more internal links
without degrading the visual style of manual pages when rendered in a
terminal.

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I1a5ef7eab7ff1e66c137e16b51a9c9074e43f583

doc/man: update type/domain options for common event rule spec.

This patch updates manual pages to follow the recent `--type` and
`--domain` option changes of the lttng-add-trigger(1) command, which
accepts the common event rule specification options of
lttng-event-rule(7).

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6064734534e773bf4f03b5f1e849b57134583039

.gitreview: Set default branch to 'stable-2.13'

Change-Id: Ia321edf68795a2a560b38947d5d888536cae08fa
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Fix: use of uninitialised bytes valgrind warning

Issue
=====

Valgrind reports usage of uninitialised stack allocated memory:
  ==2961363== Thread 9 Client manageme:
  ==2961363== Syscall param sendmsg(msg.msg_iov[0]) points to uninitialised byte(s)
  ==2961363==    at 0x521418D: __libc_sendmsg (sendmsg.c:28)
  ==2961363==    by 0x521418D: sendmsg (sendmsg.c:25)
  ==2961363==    by 0x53411B: lttcomm_send_unix_sock (unix.c:294)
  ==2961363==    by 0x48AA8C: send_unix_sock (client.c:896)
  ==2961363==    by 0x484F45: thread_manage_clients (client.c:2865)
  ==2961363==    by 0x480FB4: launch_thread (thread.c:66)
  ==2961363==    by 0x5208608: start_thread (pthread_create.c:477)
  ==2961363==    by 0x5346292: clone (clone.S:95)
  ==2961363==  Address 0x7575389 is 25 bytes inside a block of size 16,384 alloc'd
  ==2961363==    at 0x483DFAF: realloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
  ==2961363==    by 0x4EB618: lttng_dynamic_buffer_set_capacity (dynamic-buffer.c:166)
  ==2961363==    by 0x4EB52C: lttng_dynamic_buffer_append (dynamic-buffer.c:55)
  ==2961363==    by 0x48CBA1: setup_lttng_msg (client.c:125)
  ==2961363==    by 0x48AD70: setup_lttng_msg_no_cmd_header (client.c:860)
  ==2961363==    by 0x489825: process_client_msg (client.c:2253)
  ==2961363==    by 0x484A97: thread_manage_clients (client.c:2807)
  ==2961363==    by 0x480FB4: launch_thread (thread.c:66)
  ==2961363==    by 0x5208608: start_thread (pthread_create.c:477)
  ==2961363==    by 0x5346292: clone (clone.S:95)
  ==2961363==  Uninitialised value was created by a stack allocation
  ==2961363==    at 0x485FE4: process_client_msg (client.c:928)

After some digging, I found that this warning was caused by the padding
of the `struct lttng_session_list_schedules_return` during the
`LTTNG_SESSION_LIST_ROTATION_SCHEDULES` command.

All the fields are of the stack allocated struct are initialised by the
designated initializer but the padding is not.

These padding bytes are reported by Valgrind as being used
uninitialised.

Fix
===

Remove the padding by adding the LTTNG_PACKED attribute to the nested
structs in `struct lttng_session_list_schedules_return`.

Notes
=====

In light of the actual root cause, this is stacktrace is not really
useful.

The realloc call to grow the buffer makes it hard to find what is the
actual uninitialised stack allocation because Valgrind reports the
realloc call as the problematic site.

I was able to track this issue by adding a "consuming" step in the
`lttng_dynamic_buffer_append()` function. This consuming step would sum
all the bytes of the `buf` parameter so as to force Valgrind to check
each byte and not wait until the `sendmsg()` call. This way, I was able
to get a more precise location of the root cause of the issue.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib4a729575e9117cf95716ad25e1417c833f4232b

Fix: build: libcommon fd-tracker dependency is not available

Observed issue
==============

A build configured with:

  ./configure -disable-bin-lttng --disable-bin-lttng-crash --disable-bin-lttng-sessiond --disable-bin-lttng-relayd

Fails at build time with:

  make[3]: *** No rule to make target '../../src/common/fd-tracker/libfd-tracker.la', needed by 'libcommon.la'.  Stop.
  make[3]: *** Waiting for unfinished jobs....
  CC       lttng-elf.lo

Cause
=====

fd-tracker is required by libcommon. This is introduced by commit
8bb66c3cd60938352927ee865759433387324250 [1]

Build of libfd-tracker is disabled at the configure level by
build_lib_fd_tracker which in turn is enabled/disabled by the
--enable/disable-bin-* options.

For the observed issue, the --enable-bin-lttng-consumerd alone does not
enable the build of libfd-tracker.

Solution
========

All dependencies for libcommon are now always built. All bins require
libcommon to be present anyway.

This patch also fix a problem where the examples under the doc are build
even if liblttng-ctl is not built.

Known drawbacks
=========

None.

References
==========

[1]
http://git.lttng.org/?p=lttng-tools.git;a=commit;h=8bb66c3cd60938352927ee865759433387324250

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I94f5d7cdadcb4f8ff9c2617a675659c1f9eb4709

Clean-up: mark lttng_error_query communication header as const

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I166ef90aee0d4d7da9ce1002cbbe2a35eea88757

Add condition-targeting error query

Notifications discarded by the tracers are reported at the level of a
trigger. As those errors are specific to triggers with an "event-rule
matches" condition, they should be reported through a condition-specific
error query.

Note that a condition error query is created from a trigger: there is no
ambiguity since, unlike actions, conditions cannot be nested.

Given the proximity of the final 2.13 release, the code which populated
trigger error query results is simply used to populate the condition
error query results when the condition is of type "event-rule matches".

No trigger-scope errors can be reported for the moment. However, such
error reports will be added in the future.

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie1ac3668142041beb6fd61574ccef506707c55b2

action list: missing renames from previous name "group"

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4373984c2bea96dc67880b1bbb361fb8fbc014ca

Cleanup: ust-app: simplify ust_app_synchronize() error paths

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7fd3636dfb1370ebe224aa2e200189b2fe99002a

Fix: double mutex_unlock() if session is deleted

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9e640396c12496b6d6c191e838dab138679d5f5e

Fix: out of sync lttng_ust_ctl_sigbug_handle() prototype

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3aa3fc711750be320433a2f4d1e7d49d47d71c44

Fix: appending unallocated data from beyond exclusion entries

Issue
=====
If an exclusion string is smaller than the `LTTNG_SYMBOL_NAME_LEN`
integer, the `lttng_dynamic_buffer_append()` call will append
unallocated data to the buffer.

Fix
===
Use the `exclusion_len` value to copy the actual exclusion and pad the
remaining bytes with zeros.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Simon Marchi <simon.marchi@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I04c6681c28e82de29791541eb490158db9e503d0

Tests: remove leftover temporary files

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie9249820643158c1572c9ce45507d1cba36bf6f7

lttng-disable-channel(1): fix typo

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I37f733bd48d17f2e6de75038918cabd95375e4e2

lttng-concepts(7): remove reference to the section it's in

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I09319ec4849e854d6a4881ab7895cffc551be58a

lttng-concepts(7): fix typo

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ief92d1f4a56f5263a2fed5c70e2b26d660d6d81d

Build fix: build without lttng-ust

Add conditionnal inclusion of ust-sigbus to the tests modified by:

  commit e207fe73b58787c6077ec5d49254e12ff1a82b51
  Author: orbea <orbea@riseup.net>
  Date:   Sat May 29 11:07:41 2021 -0700

    build: Add missing DEFINE_LTTNG_UST_SIGBUS_STATE();

    Fixes:

      liblttng-ust-ctl.so: undefined reference to `lttng_ust_sigbus_state'

Change-Id: If4af7ada6d94ee67603156d151575dcc57baad05
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

build: Add missing DEFINE_LTTNG_UST_SIGBUS_STATE();

Fixes:

  liblttng-ust-ctl.so: undefined reference to `lttng_ust_sigbus_state'

Signed-off-by: orbea <orbea@riseup.net>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I428bdc262168d1701525e024e7580861d2498e2b

build: Pass --no-as-needed directly to the linker

This is not a libtool argument, but rather a linker argument.
GNU libtool will silently ignore arguments it doesn't understand
in many cases while slibtool does not.

Signed-off-by: orbea <orbea@riseup.net>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I55e55f1bb6a5d14d1c81505ceb39beaef84af9e9

build: Use liblttng-sessiond-common.la instead of LIVE

This allow correctly linking live_test with slibtool.

Signed-off-by: orbea <orbea@riseup.net>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie429aa36b817a3d5b22bb4bbe3d21a82c65ffdaa

build: Use liblttng-sessiond-common.la instead of SESSIOND_OBJS

This allows correctly linking test_session and test_ust_data
with slibtool.

Signed-off-by: orbea <orbea@riseup.net>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5c18187e4185218db337ac3548f0d44fb60501f3

build: Add the liblttng-sessiond-comm.la convenience library

This allows correctly linking test_kernel_data with slibtool.

Signed-off-by: orbea <orbea@riseup.net>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib076854dc2a4f0234ad3fb04ea1c3c7c4d4a2b7e

lttng-concepts(7): add missing "commands" word

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib555c944c37b983a40f963ea3597bafeba6c4da2

Tests: crash: remove redundant directory test

The presence of the `$shm_session_path/ust/pid` path is already tested
a few line above. No need to test it again.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If7da3974a6e03817533f650790e0431346401db5

Fix: bump minimal urcu dependency to 0.11

cds_lfht_destroy is used from within RCU read-side critical sections
which is only allowed for urcu >= 0.10 (see userspace-rcu commit
d0ec0ed2f).

Such uses were introduced as part of the 2.11 release (contemporary to
urcu 0.10).

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0b513f58eacd9fc345eacfddcc2662ffa367695a

condition: buffer usage: validation does not check for ratio and bytes threshold

Buffer usage condition do not support having both ratio and bytes
threshold set.

Using `!usage->threshold_ratio.set && !usage->threshold_bytes.set` does
not error out for `threshold_ratio.set == 1` and
`usage->threshold_bytes.set == 1`

0 0 : 1
0 1 : 0
1 0 : 0
1 1 : 0

We want to check for a xnor (`usage->threshold_ratio.set == usage->threshold_bytes.set`)

0 0 : 1
0 1 : 0
1 0 : 0
1 1 : 1

We could also do 2 error check one for not set and one validating that
both are not set.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5bfcf43bcaf9687ddf9fdebe5ace4f15bda28261

Remove LTTNG_EVENT_RULE_TYPE_KERNEL_FUNCTION

It is not supported for 2.13.

Change-Id: If42b18115cc61da9ee17f281390280b21fdc35a6
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

Remove lttng_event_rule_tracepoint

lttng_event_rule_tracepoint is "exploded" in:
   lttng_event_rule_user_tracepoint
   lttng_event_rule_kernel_tracepoint
   lttng_event_rule_jul_logging
   lttng_event_rule_python_logging
   lttng_event_rule_log4j_logging

On the CLI front, the `--domain` option for the add-trigger is removed
since for each event rule type the domain is implicit based on the type.

`--type` accepts the following:
   kernel
   kernel:tracepoint
   kernel:kprobe
   kernel:uprobe
   syscall
     syscall:entry
     syscall:exit
     syscall:entry+exit
     syscall:*
   kernel:syscall
     kernel:syscall:entry
     kernel:syscall:exit
     kernel:syscall:entry+exit
     kernel:syscall:*
   user
   user:tracepoint
   python
   python:logging
   jul
   jul:logging
   log4j
   log4j:logging
   kprobe
   function

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4a0489f8f2a79a2c967f5d69461bed10f40b8278

Test log level for newly introduced event rule type (*_logging, user_tracepoint)

This is in prevision to lttng_event_rule_tracepoint removal.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iff251d7540a16a2a9561d834fc464ac1a6be4f84

Introduce lttng_event_rule_python_logging

A following patch will remove lttng_event_rule_tracepoint.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I222bd4f2616efd1b0c0e36cfe5a89cf5416b6a88

Introduce lttng_event_rule_log4j_logging

A following patch will remove lttng_event_rule_tracepoint.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie94c172ed9550dd7c3ae9967e353a3d6ac5b5907

Introduce lttng_event_rule_jul_logging

A following patch will remove lttng_event_rule_tracepoint.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I95f40920b73b5deb42c2e92ad50c43dba0b0e436

Introduce lttng_event_rule_user_tracepoint

A following patch will remove lttng_event_rule_tracepoint.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If814c3ba828e4869cde9af4b9d76c8b830f30654

Introduce lttng_event_rule_kernel_tracepoint

A following patch will remove lttng_event_rule_tracepoint.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I614e45be3ef5e41fa081ad24dc1a9dcddb7eb2af

Rename lttng_event_rule_kernel_probe to lttng_event_rule_kernel_kprobe

The add trigger command now accepts `kernel:kprobe` instead of `kernel-
probe`. `kprobe` is still accepted.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifd6753f337e93ea8eb393b8d686596112292c5bb

Rename lttng_event_rule_userspace_probe to lttng_event_rule_kernel_uprobe

The `--type` option for add trigger now only support `kernel:uprobe`.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2c4de8e314d6fba735343d21c04fff366b92cca9

Rename lttng_event_rule_syscall to lttng_event_rule_kernel_syscall

Allow the usage of "kernel:syscall*" as the `--type` argument for add
trigger. A later patch remove the notion of domain.

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib5c87ec937bb0acf9c5133532961dcc31e533884

Rename *emission_site_type to *emission_site

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3027affe23de8674a5a813d18643396e3a1474f1

Rename lttng_event_rule_syscall_(set, get)_pattern to lttng_event_rule_syscall_(set, get)_name_pattern

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ia6d8c55f607cd5af4201a61abaab277994618bb6

Rename *exclusion* to *name_pattern_exclusion*

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic72582dccdc2810dfc07fc66abf776c853d217b6

Rename lttng_event_rule_tracepoint_(set,get)_pattern to lttng_event_rule_tracepoint_(set, get)_name_pattern

Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib018d7ab4fab66842302beef19c0971d704d2d02

Build fix: cygwin: unknown type ssize_t

The build fails on cygwin with the following error:

  In file included from actions/path.c:8:
  ../../include/lttng/action/path-internal.h:31:1: error: unknown type name "ssize_t"; did you mean "_ssize_t"?

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I93aac8ef33cfe19b655d80709c9b95a503815bda

Fix: consumer: unbalanced RCU read-side lock on error

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I22ddc8585af84d07e8f3d19e9df582599352a1b8

lttng-enable-event(1): add usage examples

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6310d753a2fd6db74f4a854a87cda690d3479593

lttng-{enable,disable}-event(1): document default channel limitation

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0b531be8bde9dfd74bb4a28fe70eeef5eeee7f72

doc/man: log level prefixes are not required

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iddd9eaff96ca857b1249421c31b64bec859429c6

Fix: sessiond: use of uninitialized memory in buffer-usage condition

Issue
=====

valgrind reports the following:
  ==436327== Thread 9 Client manageme:
  ==436327== Syscall param sendmsg(msg.msg_iov[0]) points to uninitialised byte(s)
  ==436327==    at 0x51E418D: __libc_sendmsg (sendmsg.c:28)
  ==436327==    by 0x51E418D: sendmsg (sendmsg.c:25)
  ==436327==    by 0x20D436: lttcomm_send_unix_sock (unix.c:294)
  ==436327==    by 0x186F08: send_unix_sock (client.c:895)
  ==436327==    by 0x18BAFA: thread_manage_clients (client.c:2800)
  ==436327==    by 0x18147D: launch_thread (thread.c:66)
  ==436327==    by 0x51D8608: start_thread (pthread_create.c:477)
  ==436327==    by 0x5314292: clone (clone.S:95)
  ==436327==  Address 0x7540ec1 is 97 bytes inside a block of size 256 alloc'd
  ==436327==    at 0x483DFAF: realloc (in /usr/lib/x86_64-linux-gnu/valgrind/vgpreload_memcheck-amd64-linux.so)
  ==436327==    by 0x1E7556: lttng_dynamic_buffer_set_capacity (dynamic-buffer.c:166)
  ==436327==    by 0x1E72FC: lttng_dynamic_buffer_append (dynamic-buffer.c:55)
  ==436327==    by 0x1DE604: lttng_condition_buffer_usage_serialize (buffer-usage.c:123)
  ==436327==    by 0x1DFF47: lttng_condition_serialize (condition.c:98)
  ==436327==    by 0x20ACD0: lttng_trigger_serialize (trigger.c:328)
  ==436327==    by 0x189EBE: process_client_msg (client.c:2258)
  ==436327==    by 0x18B761: thread_manage_clients (client.c:2742)
  ==436327==    by 0x18147D: launch_thread (thread.c:66)
  ==436327==    by 0x51D8608: start_thread (pthread_create.c:477)
  ==436327==    by 0x5314292: clone (clone.S:95)

This can be reproduce by running the sessiond under valgrind and
launching the following binary to register buffer-usage conditions:
  ./tests/regression/tools/trigger/utils/register-some-triggers test_buffer_usage_conditions

The valgrind report is pointing us toward the `struct
lttng_condition_buffer_usage_comm` stack allocation in the
`lttng_condition_buffer_usage_serialize()` function.

The actual issue is that the threshold_bytes/threshold_ratio will not be
initialized in that structure depending on the way the threshold is
expressed.

Fix
===

Make sure the struct is zeroed at the declaration site.

Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2046f4e0e1d857cfd1a53e0c13aea55f17adada3

lttng-ctl: use lttng_action_path to specify error query actions

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I30a93465231ec963d087b25678b6703a0425ceef

error-query: add lttng_action_path to express the location of an action

Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Icc3056029219cc8f0794b06e7021885e193724be

tests: Move tap-driver.sh out of the autotools aux directory

We have made local modifications to this script, move it to the test
suite directory so it doesn't get overwritten by an updated version from
autotools.

Running 'make check' on current RHEL or Fedora without this patch
results in the following error:

  make[5]: Entering directory '/builddir/build/BUILD/lttng-tools-2.12.4/tests/regression'
  tap-driver.sh: invalid option: '--post-script'

Change-Id: I22823176cc5b98901865c9415250ce8e49384cbd
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>

lttng-enable-channel(1): add usage examples

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I26bcfed2b6d71b0bbc2fa89d33f37925c303c91c

lttng-view(1): add usage examples

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9324b0733ebbaabb824efca4a11330fe3a5c1d0f

lttng-untrack(1): add usage examples

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibafe1662e37110962ae3c7fdef9d450b75f2c0a5

lttng-untrack(1): follow the style of lttng-track(1) for the example

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0c714eb2e504d118cfb3b48bf8e70f33d36d2352

lttng-track(1): add usage examples

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I21a7906626a09b9964fc51dd083f1c2ea5b855c3

lttng-stop(1): add usage examples

Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6794937457d4f34d649b9edb6a755e091a6af67e