Jérémie Galarneau [Wed, 7 Apr 2021 15:54:17 +0000 (11:54 -0400)]
Fix: sessiond: kernel: missing check for unregistration error
1451463 Unused value
An assigned value that is never used may represent unnecessary computation, an incorrect algorithm, or possibly the need for cleanup or refactoring.
In synchronize_tracer_notifier_unregister: A value assigned to a variable is never used. (CWE-563)
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iaea550a2d70730cc4c90f8d1bdcd8cc0a4af6bd8
Jérémie Galarneau [Wed, 7 Apr 2021 15:49:39 +0000 (11:49 -0400)]
Tests: fix: unchecked sscanf return value
1407934 Unchecked return value
If the function returns an error value, the error value may be mistaken for a normal value.
In parse_arguments: Value returned from a function is not checked for errors before being used (CWE-252)
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7733e88ea6af6313379bd8d4ca131cd243cda401
Jérémie Galarneau [Wed, 7 Apr 2021 02:05:26 +0000 (22:05 -0400)]
on-event evaluation: remove trigger name accessor
The trigger name accessor on an on-event evaluation is no longer needed
as the originating trigger is now accessible on a notification.
The tests are adapted to use the name of the trigger as returned by
the lttng_notification received by the client.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I57bd64d7f5ec35f015dfb115ecfadd75806b9a75
Jérémie Galarneau [Wed, 7 Apr 2021 01:10:35 +0000 (21:10 -0400)]
notification: add trigger accessor to lttng_notification
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If154de643f0d3ec0978b96551b3c3873f4b3380b
Jérémie Galarneau [Wed, 7 Apr 2021 00:41:25 +0000 (20:41 -0400)]
notification: transmit originating trigger as part of evaluation
Transmit the full trigger object instead of its condition when sending a
notification to a client. This allows clients to reliably filter on
triggers as, currently, only the name of a trigger is made available to
clients as part of the "on_event" evaluation object (the owner uid is
needed to perform a correct comparison for filtering).
This will also make the trigger accessible for all condition types.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I61d587436177eb34744487a1d3c486d87fe2e1f1
Mathieu Desnoyers [Tue, 6 Apr 2021 18:58:28 +0000 (14:58 -0400)]
Fix: .gitignore: use full paths for test binaries
Using only the binary name in the .gitignore paths yields confusion
for git and search tools relying on this file's content (e.g. ag
silversearcher), because the binary file name may also be part of a
path, e.g. the filename "gen-ust-events" is also part of the path
leading to the .c file implementing this test:
"./tests/utils/testapp/gen-ust-events/gen-ust-events.c", which means
this .c file is ignored, even though it should not be.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I309b268fed8e66b693c1e0670044b40bf90593c4
Mathieu Desnoyers [Tue, 6 Apr 2021 18:26:03 +0000 (14:26 -0400)]
Tests: use TRACEPOINT_INCLUDE
The macro TRACEPOINT_INCLUDE_FILE is deprecated since LTTng-UST 2.1. In
preparation for LTTng-UST 2.13, this compatibility macro is finally
removed.
The tests in lttng-tools need to keep up with times and start using its
replacement: TRACEPOINT_INCLUDE.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id2bb15062f37878de03656dc5a47d9dd35687b41
Jérémie Galarneau [Fri, 2 Apr 2021 17:11:34 +0000 (13:11 -0400)]
Fix: sessiond: ust-app: assertion failure on registration of dead app
While setting up the event notifier group of a newly registered
application (see `ust_app_setup_event_notifier_group()`), communications
with the application can fail (generally because the application exited)
while sending the error counter configuration. This causes the UST
handle to the event notifier group to be released.
However, a second release is performed which results in undefined
behaviour (I have seen it succeed, assert, and crash under different
circumstances) while disposing of the application (see
`delete_ust_app()`).
Set the event notifier group ust handle to NULL to mark it as invalid
and prevent a second release during the clean-up.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ia5623a8d3f57174effa51c278e9654b6adc43289
Jérémie Galarneau [Fri, 2 Apr 2021 13:07:00 +0000 (09:07 -0400)]
common: credentials: hide symbols
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I089502d46bdd0cf34b0721703a08fdd28f61b8f7
Francis Deslauriers [Mon, 8 Mar 2021 21:11:37 +0000 (16:11 -0500)]
trigger: consider name in trigger _is_equal()`
For the moment, this is a breaking change since, before 2.13, users
could not register two identical triggers.
Triggers were previously unnamed, but a name is now generated
automatically. This means that a pre-2.13 user that could expect
a TRIGGER_ALREADY_EXISTS reply when registering a trigger that already
existed will now get "OK".
However, since it doesn't know about trigger names, it could now fail
to unregister a trigger (if the instance used to register the trigger
was not kept around).
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9549980677e2387a7cc0f275233a3b5717fb915e
Francis Deslauriers [Tue, 23 Mar 2021 23:50:30 +0000 (19:50 -0400)]
Fix: sessiond: notification: use after free of trigger object
Background
==========
* Clients can subscribe to certain specific conditions (e.g. buffer
usage) using the `lttng_notification_channel_subscribe()` function.
* This subscription is only useful once a trigger with that condition
AND at least one notify action is registered.
* The sessiond keeps a list for client subscribed to each registered
condition.
* More than one trigger with the same condition may be registered
the sessiond at the same time if they have different actions.
Issue
=====
Currently, when registering a trigger (T1) the sessiond looks if there
is already a client list for the condition of this trigger. If not, the
sessiond links the newly created client list object to that trigger T1
by keeping a pointer to it.
This means that if another trigger (T2) is registered with the same
condition (but a different name, or different actions) it will reuse the
same client list object and use the pointer to the T1.
This causes problems if T1 is unregistered before T2. In that case, the
pointer to T1 in the client list object is pointing to a deallocated
trigger object.
This issue is not encountered with the current test suite, namely the
`test_notification_multi_app` test case, because triggers with the same
condition also had the same action, so they are considered identical and
are not registered.
This issue was first witnessed when adding a trigger name comparison in
the `lttng_trigger_is_equal()` function.
Fix
===
Change the client list object so that it has its own copy of the
condition and a list of dependent triggers. Each trigger with that
condition has a reference on this client list object.
When unregistering a trigger, the notification thread removes it for the
client list's triggers list and put its reference on the client list
object.
Tests
=====
This commit adds a parameter to the base_client that dictates if the
notify action should be in a group. This is a trick to create triggers
that are not equal but have the same behaviour.
The `test_notification_multi_app` test case is modified to turn on this
option every other trigger registration.
Semi-related cleanups
=====================
* Merge `notification_client_list_create()` and
`publish_notification_client_list()` functions since they are used
together anyway. This removes the need to call
`notification_client_list_put()` at the end of the
`_register_trigger()` function
* Rename `trigger_applies_to_client()` to `condition_applies_to_client()`.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3ebb90a1a64236a440a085e6fc1b82726a0e5af9
Francis Deslauriers [Thu, 5 Nov 2020 19:55:26 +0000 (14:55 -0500)]
sessiond: Extract `{setup,teardown}_tracer_notifier()` functions
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic394b1a50f166afc911b58607aece6ae69422841
Francis Deslauriers [Thu, 5 Nov 2020 17:25:44 +0000 (12:25 -0500)]
Clean-up: sessiond: Extract trigger_ht_element removal function
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4994d1dce7a8fc188b6df6ad72e5cd1081cb0cc7
Francis Deslauriers [Mon, 31 Aug 2020 15:05:41 +0000 (11:05 -0400)]
Tests: event notifier error counters
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9e2a4bab833d9204717029c1d5f594455c2262ce
Francis Deslauriers [Wed, 14 Oct 2020 19:33:46 +0000 (15:33 -0400)]
sessiond: Implement UST event notifier error counter
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I254b2cec39b8c61ad945376043b5b9053de98f6f
Jérémie Galarneau [Thu, 1 Apr 2021 03:38:07 +0000 (23:38 -0400)]
common: document why spawn_viewer() is public
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I09574ebd82bb583db5a7acec2b8f129a7dcd36fc
Jérémie Galarneau [Mon, 1 Mar 2021 20:40:48 +0000 (15:40 -0500)]
Docs: document trigger condition and action ownership in 2.13+
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8a52f166565efc87c6073713ce0f6569ba6d3661
Jérémie Galarneau [Mon, 29 Mar 2021 16:17:12 +0000 (12:17 -0400)]
Fix: sessiond: kernel: close on badfd on initialization error
The various kernel tracer file descriptors are closed during an
initialization error, but are not set to `-1` which causes them to
be closed again when the session daemon performs its clean-up.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I11d3ce7accea51394c34b9ca7ea0445e4ee422fc
Francis Deslauriers [Fri, 26 Mar 2021 21:06:12 +0000 (17:06 -0400)]
Fix: sessiond: kernel error accounting fd still open when unloading modules
Issue
=====
I noticed that some kernel modules were not unloading properly when
shutting down the sessiond. Namely those modules:
lttng_counter_client_percpu_64_modular
lttng_ring_buffer_event_notifier_client
The call to `modprobe -r` for those modules is failing because the
kernel error accounting fd has not been closed.
Fix
===
Only call `modprobe -r` after all kernel resources (file descriptors)
have been released.
Side note
=========
The `modprobe_remove_lttng()` is currently printing that the removal of a
_OPTIONAL module was a success even if it failed. I rework this function
to print a debug message in such situation. I used the
`modprobe_lttng()` as a template for this function.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibb51425ef3deedf6e982084ef879bc83ac823ba3
Francis Deslauriers [Wed, 14 Oct 2020 19:33:19 +0000 (15:33 -0400)]
sessiond: Implement kernel event notifier error counter
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I390113873567a98d7e53db605d48b59a60a1074e
Jérémie Galarneau [Fri, 26 Mar 2021 22:39:30 +0000 (18:39 -0400)]
Fix: on-event condition: don't abort() on invalid event expression type
Protocol errors should not result in crashes. Simply report the
failure to deserialize the event expression.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic5a0be6b248c430fcf1fe487cd734d9bdf5dd36a
Francis Deslauriers [Fri, 26 Mar 2021 18:41:01 +0000 (14:41 -0400)]
Fix: common: index_allocator_get_index_count() returns size
The lttng_index_allocator_get_index_count() should return the number of
allocated indexes. Right now it returns the maximum number of indexes.
I witnessed the following message when running testing the event
notifier error counter feature:
Warning: Destroying index allocator with 4096 slot indexes still in use
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibbcb3f128553df0a4118901ef85ae6fe17ebcecf
Francis Deslauriers [Mon, 7 Dec 2020 21:08:30 +0000 (16:08 -0500)]
on-event: add error counter fields to condition
Modified from proposed change: don't include error counter index in serdes
The error counter index is an internal implementation detail that
doesn't need to be exposed to the client (the accessor is not reacheble
to them anyway).
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibc84d4cc3bcc7e304f4b52bd45f1de010dd8b87b
Mathieu Desnoyers [Thu, 25 Mar 2021 19:18:13 +0000 (15:18 -0400)]
compiler warning cleanup: is_signed_type: compare -1 to 1
Comparing -1 to 0 triggers compiler warnings (gcc -Wtype-limits and
-Wbool-compare) and Coverity warning "Macro compares unsigned to 0".
Comparing -1 to 1 instead takes care of silencing those warnings while
keeping the same behavior.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id4aebce6eed214f0ccba2c0289fb3b034808cb64
Jérémie Galarneau [Wed, 24 Mar 2021 23:23:54 +0000 (19:23 -0400)]
common: shm.c: improve logging on various error paths
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I1d68f952b8a01b79aeab7c8d05d98085071a45a7
Francis Deslauriers [Thu, 8 Oct 2020 20:59:57 +0000 (16:59 -0400)]
Cleanup: Move `create_posix_shm()` to common/shm.c
* This function will be used for trigger error accounting.
* This function is renamed to `shm_create_anonymous()` and now takes one
parameter for the owner of the shared memory area. Again this is code
reuse.
* Remove erroneous comment about the `name` parameter for the
`shm_open()` function. It's the absence (and not its presence) of the
'/' at the beginning of the name that makes the behaviour implementation
defined.
* Move the bin/lttng-sessiond/shm.c to common/shm.c so that all shm
related functions are in one location.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I690b38801ffb4772e7c5069c4cccf470b0671f63
Francis Deslauriers [Mon, 24 Aug 2020 21:02:05 +0000 (17:02 -0400)]
common: Add index allocator for error counters
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I677a8f5b774315eadb8e3b2c2f2a5b1aae02133f
Jérémie Galarneau [Wed, 24 Mar 2021 18:21:40 +0000 (14:21 -0400)]
Fix: sessiond: agent: lazy creation of agent on removal
When unregistering a trigger that was previously registered, the
agent corresponding to the trigger's domain will already exist. It is
superfluous to handle the case where it doesn't exist and it would only
hide an internal error.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifc6d14b74586af8423a70b3466e0a2c00491d3c3
Jérémie Galarneau [Wed, 24 Mar 2021 16:17:59 +0000 (12:17 -0400)]
Clean-up: sessiond: make some accesses to conditions const
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I88edd184c06d2f2912331da8e83b624f8ae8bcc2
Francis Deslauriers [Wed, 4 Nov 2020 23:10:52 +0000 (18:10 -0500)]
Extract synchronize tracer notifier functions
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2d92227d93ede0b04bf7314ff819c2b777247262
Francis Deslauriers [Thu, 14 Jan 2021 22:44:49 +0000 (17:44 -0500)]
Cleanup: Extract `ust_app_synchronize_all_channels()` function
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id24eff10cc07c5e061fbb45e871087c17eac9b1d
Mathieu Desnoyers [Tue, 23 Mar 2021 19:01:19 +0000 (15:01 -0400)]
ust-abi: add missing lttng_ust_abi prefixes
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I03f465659cd7b044d1ec10458fa77332d5e9db41
Mathieu Desnoyers [Tue, 23 Mar 2021 16:01:43 +0000 (12:01 -0400)]
clock override: use getter API from LTTng-UST
Currently, the lttng session daemon reimplements its own copy of
struct lttng_trace_clock, which is pretty bad as ABI go.
Use getter functions for each of the clock callback instead. This
refactoring needs to be introduced with a matching commit to implement
the new ABI in LTTng-UST.
Move the implementation of the clock frequency, uuid, name and
description default callbacks (for monotonic clock) to lttng-ust.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If2e671bacacc55d3fb27b6b4d4fc8c16aaffd059
Mathieu Desnoyers [Tue, 23 Mar 2021 15:22:56 +0000 (11:22 -0400)]
Fix: bytecode test should be in UST_FILTERS, not UST_STR_FILTERS
The UST test is in the wrong array, making it always succeed when it
should fail when run with a non-patched UST, because it uses a different
test application which does not contain the targeted fields.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I96f74e26ffdead5e03e069c944f215967b7890f2
Mathieu Desnoyers [Mon, 22 Mar 2021 17:23:45 +0000 (13:23 -0400)]
Tests: test bytecode linker refusal of non-string arrays and sequences
The bytecode linker should refuse loading arrays and sequences of data
which are not strings for the purpose of comparison against other
strings.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id3b14cb6674b0e2e1edaf039fbc5f929708e2451
Jonathan Rajotte [Tue, 15 Dec 2020 19:03:15 +0000 (14:03 -0500)]
Tests: utils.sh: pass arbitrary arguments to start_lttng_sessiond*
Allows for easier arguments passing to lttng-sessiond. This is useful
both for debugging manually and passing specific flag as needed in
testing. Passing "--event-notifier-error-number-of-bucket=X" or "-vvv"
for examples.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I36bd6eaf1c4fc84c3a1e8cfe70a37a46e7130c75
Francis Deslauriers [Tue, 2 Feb 2021 17:00:50 +0000 (12:00 -0500)]
Cleanup: clarify `buffer_reg_channel` and `ust_registry_channel` variables
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3eab9f1c2f53a72209ae95c05a5f9f3669e4ea0f
Jonathan Rajotte [Tue, 2 Mar 2021 20:07:42 +0000 (15:07 -0500)]
Fix: return value of register trigger to the notification thread is not acted upon
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0fd6dea34854b82b6d4c1550c4f84a7c79f9076f
Francis Deslauriers [Mon, 15 Mar 2021 19:31:08 +0000 (15:31 -0400)]
Tests: Cleanup: gen-syscall-events: remove duplicated help line
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id5bc54598f8abf7498f3b81064a83c60379922d6
Francis Deslauriers [Thu, 4 Mar 2021 22:12:57 +0000 (17:12 -0500)]
Tests: Cleanup: add `lttng_{add, remove}_trigger_ok()` bash functions
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I63b1f4d2a72b291f93df4548114014a116824864
Jérémie Galarneau [Mon, 22 Mar 2021 17:02:56 +0000 (13:02 -0400)]
Tests: fix: log level rule: leak of rule after test
1451079 Resource leak
The system resource will not be reclaimed and reused, reducing the future availability of the resource.
In test_log_level_rule_at_least_as_severe_as: 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: Ica923f9f010e51217ca62664faab99cb34999c0b
Jérémie Galarneau [Mon, 22 Mar 2021 17:00:09 +0000 (13:00 -0400)]
Fix: lttng-sessiond: kernel: leak of event notifier rule
1445768 Resource leak
The system resource will not be reclaimed and reused, reducing the future availability of the resource.
In kernel_create_event_notifier_rule: 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: I58862b3b38c7e61806e0047840bf68619273db49
Jérémie Galarneau [Thu, 18 Mar 2021 21:11:37 +0000 (17:11 -0400)]
Tests: add log level rule validation tests
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idabc5420ac2b188d13ed5537be81128976661357
Jérémie Galarneau [Thu, 18 Mar 2021 23:11:32 +0000 (19:11 -0400)]
Fix: event-rule: report log level rules as unsupported for kernel domain
Log levels are not supported for by tracepoints for the kernel domain.
Return `LTTNG_EVENT_RULE_STATUS_UNSUPPORTED` when setting a log level
rule on a kernel-domain tracepoint.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I73b6067ec0a11ae1f83098a6966dc956e1772224
Francis Deslauriers [Thu, 18 Mar 2021 20:01:34 +0000 (16:01 -0400)]
Tests: Fix: list-triggers test case out of date
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9d78c025693fce9dc0ae2177d71d24217a4b1d6a
Francis Deslauriers [Tue, 16 Mar 2021 20:27:23 +0000 (16:27 -0400)]
Tests: add kernel --function basic test case
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I584be60bd7325433aa9c30338ceb652c48806ac6
Francis Deslauriers [Tue, 16 Mar 2021 18:28:08 +0000 (14:28 -0400)]
Rename event rule kretprobe to kernel function
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibbd0cfa74e7d8e5b922debd23fc5f89cdeb14621
Jérémie Galarneau [Wed, 17 Mar 2021 05:10:27 +0000 (01:10 -0400)]
on-event evaluation: introduce on-event evaluation specific status code
Before this change, `lttng_evaluation_on_event_get_captured_values`
returns `LTTNG_EVALUATION_STATUS_INVALID` when an evaluation's
condition has no capture descriptors. That status code is also used to
indicate a handful of "invalid parameter" conditions.
Not having captured values in an evaluation should not be considered an
error. Hence, a status enum that is specific to on-event evaluations is
added to introduce a "NONE" status that is not an error (positive value)
and wouldn't make sense for all evaluations (it could mean an error in
other circumstances).
`LTTNG_EVALUATION_ON_EVENT_STATUS_NONE` is returned when the condition
of an on-event evaluation has no capture descriptors.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic2e3e90317a44c7bb3e95630956565690bffd760
Francis Deslauriers [Thu, 5 Nov 2020 20:22:03 +0000 (15:22 -0500)]
event-notifier: implement `lttng_trigger_needs_tracer_notifier()` function
This function is used to tell if a trigger needs a tracer notifier. This
depends on its condition and its actions.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie3c928ba1c705c32e4c2e4c6b6b22aa8c10837a9
Jonathan Rajotte [Fri, 6 Nov 2020 15:06:23 +0000 (10:06 -0500)]
lttng-ctl: Introduce lttng_log_level_rule
This object corresponds to the criteria a event-rule tracepoint must
respect for an event to be consired under the event-rule tracepoint
criteria.
The lttng tracepoint event rule object now have the following API
function in regards to log level rule:
lttng_event_rule_tracepoint_set_log_level_rule
lttng_event_rule_tracepoint_get_log_level_rule
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iefd361d1b7d85c6803e891166429a2830e2404a2
Jonathan Rajotte [Fri, 6 Nov 2020 15:16:28 +0000 (10:16 -0500)]
Clean-up: sort noinst_PROGRAMS in test/unit/Makefile.am
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I99cb52748495cd7cc364af7ea04ae95a75bb5759
Jonathan Rajotte [Fri, 6 Nov 2020 15:06:36 +0000 (10:06 -0500)]
Clean-up: sort TESTS in test/unit/Makefile.am
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id4d245eef4f0b99038c7ee76d9c783f40d8b849a
Jonathan Rajotte [Thu, 5 Nov 2020 20:03:29 +0000 (15:03 -0500)]
Clean-up: sort files in include/Markefile.am
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idcdd023916084610669472a55164a3d6e0192496
Jonathan Rajotte [Thu, 5 Nov 2020 19:59:58 +0000 (14:59 -0500)]
event-rule: userspace probe: force location on create
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5775184c62249c642af0ca0315103e040cb2502d
Jonathan Rajotte [Thu, 5 Nov 2020 19:04:33 +0000 (14:04 -0500)]
event-rule: userspace probe: rename get/set_name to get/set_event_name
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4a7fbd26b4dfe352100a6fa78b828c1da0fa9928
Jonathan Rajotte [Thu, 5 Nov 2020 17:02:32 +0000 (12:02 -0500)]
event-rule: kernel probe: force location on create
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I485ab3bb32317d97ad46f7de71c72a37afcd64c2
Jonathan Rajotte [Thu, 5 Nov 2020 16:32:35 +0000 (11:32 -0500)]
event-rule: kernel probe: rename set/get_name to set/get_event_name
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I14950a663a4db37fe0f120d8f06c01cae241a871
Jonathan Rajotte [Thu, 5 Nov 2020 16:28:03 +0000 (11:28 -0500)]
event-rule: syscall: set the default pattern to '*'
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8456f7f125b9341d31bc7bc2b8474abbc44b5467
Jonathan Rajotte [Thu, 5 Nov 2020 16:24:10 +0000 (11:24 -0500)]
event-rule: tracepoint: set default pattern to '*'
'*' is set on tracepoint creation.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id20b5e3924a11a336fb08ac6ce10f08ad7fcd80a
Jonathan Rajotte [Thu, 5 Nov 2020 02:46:11 +0000 (21:46 -0500)]
Rename files for condition event-rule to on-event
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8bde2f15674be16c0a21c670affff4c4503ed846
Jonathan Rajotte [Thu, 5 Nov 2020 01:48:56 +0000 (20:48 -0500)]
Rename lttng_condition_event_rule to lttng_condition_on_event
Files renaming is following.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Icf74aa7cc84b46b1f23bbe1418b7e1241d8dc081
Jonathan Rajotte [Thu, 5 Nov 2020 00:18:17 +0000 (19:18 -0500)]
Rename uprobe files to userspace-probe
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ide5aeca0d56b8a368fa26bf7cc082098e349d8d9
Jonathan Rajotte [Thu, 5 Nov 2020 00:08:10 +0000 (19:08 -0500)]
Rename uprobe to userspace-probe
Renaming of associated files follows-up.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iabb6f3599efbac25a4c31b892b791b98ed4f5990
Jonathan Rajotte [Wed, 4 Nov 2020 20:45:31 +0000 (15:45 -0500)]
Rename kprobe file to kernel-probe
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I379bd8d88b4fa4a7d5638dc5bee4d7dd86426dc9
Jonathan Rajotte [Wed, 4 Nov 2020 20:23:30 +0000 (15:23 -0500)]
Rename kprobe to kernel_probe
Renaming of the kprobe based file follows.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If9f500234bdb1a97752ffeff8c8549500a5f841a
Jonathan Rajotte [Thu, 7 May 2020 15:08:34 +0000 (11:08 -0400)]
Tests: Implement trigger capture test cases
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I661fd8f72c871c4c38b941e9c9c5f6054383354c
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Wed, 27 May 2020 18:31:13 +0000 (14:31 -0400)]
man: lttng-add-trigger: document --capture option
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id4041c6647ee9b6ea19896d5a926f910fbf1d378
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Simon Marchi [Thu, 21 May 2020 02:42:00 +0000 (22:42 -0400)]
CLI: make list-triggers command print capture expressions
When printing an event rule condition, in the `list-triggers` command,
print the capture expressions associated with that condition.
The result looks like:
- id: T1
condition: event rule hit
rule: allo (type: tracepoint, domain: ust)
captures:
- $ctx.foo
actions:
notify
- id: T2
condition: event rule hit
rule: allo (type: tracepoint, domain: ust)
captures:
- $app.iga:active_clients
actions:
notify
Change-Id: Iaa902369a2df3edf2928a935f7565a7c21acb861
Signed-off-by: Simon Marchi <simon.marchi@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Wed, 13 May 2020 23:02:43 +0000 (19:02 -0400)]
Support capture for kernel tracer
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I26167d259daa8e356a1bb03831ccf6e570b8e305
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Wed, 13 May 2020 20:24:23 +0000 (16:24 -0400)]
lttng: Capture is only supported by tracepoint and syscall event-rules
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idfe07be4e12e0531dfdb778cdb4c011780af7a26
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Wed, 23 Sep 2020 20:14:40 +0000 (16:14 -0400)]
sessiond: notification-thread: Missing action executor status handling
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8666c0728ca140c45d47ac1ecf252bdbf0526e46
Jonathan Rajotte [Wed, 23 Sep 2020 20:14:17 +0000 (16:14 -0400)]
sessiond: notification-thread: Missing domain internal header
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0fbd0cd8bd688b1983430c9036351ca501f5abd7
Jonathan Rajotte [Wed, 23 Sep 2020 20:08:05 +0000 (16:08 -0400)]
sessiond: action-executor: Missing notify header
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6d00f41d1560a0fc73127aca1e37cd9e100f20ee
Philippe Proulx [Fri, 1 May 2020 20:33:26 +0000 (16:33 -0400)]
Make captured field values available to event rule cond. evaluation
liblttng-ctl:
Add the public lttng_evaluation_get_captured_values() function.
This function returns an array event field value which contains
the captured field values as requested by the evaluation's
condition's capture descriptors.
This array event field value contains as many elements as returned
by lttng_condition_event_rule_get_capture_descriptor_count() for the
evaluation's condition. However,
lttng_event_field_value_array_get_element_at_index() could return
`LTTNG_EVENT_FIELD_VALUE_STATUS_UNAVAILABLE` for some or all indexes
if there was no such field to capture at tracing time.
Common:
Internally, an event rule evaluation now contains a captured field
value payload (CFVP) as well as the array event field value to be
returned by lttng_evaluation_get_captured_values().
The CFVP is MessagePack-encoded.
When you create an event rule evaluation with
lttng_evaluation_event_rule_create(), you must pass the CFVP as well
as the `decode_capture_payload` option parameter. If
`decode_capture_payload` is `true`,
lttng_evaluation_event_rule_create() creates the array event field
value from the CFVP thanks to msgpack-c.
lttng_evaluation_event_rule_create() also accepts the event rule
condition now because it relies on the capture descriptors to create
the event field values if `decode_capture_payload` is `true`.
When an event rule evaluation is serialized:
Only its CFVP is serialized (as is).
When an event rule evaluation is deserialized:
lttng_evaluation_event_rule_create() is called with the
`decode_capture_payload` parameter set to `true`.
Session daemon:
action_executor_notify_handler() calls
lttng_evaluation_event_rule_create() with the
`decode_capture_payload` parameter set to `false` to send the
evaluation to the clients with the CFVP.
This CFVP directly comes from the tracer.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib2ae39dae005ad555eb8309e0bfd53f746d5e8dd
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Mon, 8 Mar 2021 20:59:01 +0000 (15:59 -0500)]
notification: fetch capture payload on notification reception
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id52749b59969276b5390f4494df48940662d5d8c
Mathieu Desnoyers [Mon, 15 Mar 2021 19:58:13 +0000 (15:58 -0400)]
Fix: set_ust_capture: missing ABI rename
struct lttng_ust_object_data has become struct
lttng_ust_abi_object_data. It gives a compiler warning.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ia29eac5903b43a15c02df7204cf5344226d30381
Jonathan Rajotte [Mon, 15 Mar 2021 15:25:07 +0000 (11:25 -0400)]
tests: perf: UNHALTED_REFERENCE_CYCLES might not be actionable on a host
This patch does NOT address the root problem and only addresses the
validation of the context to be added during the test suite.
Observed issue
==============
The system_tests jobs for master hangs on the perf event test suites.
Cause
=====
The hang is caused by a cleanup problem (reference counting of the trace
chunk on session destroy/rotation) when the activation of a context
fails on a ust app channel.
This patch does NOT address the root problem and only addresses the
validation of the context to be added during the test suite. In all
cases we need to handle when a context fail, but for this test we need
to validate that the context can be added and skip the tests as
necessary based on the host.
The perf tests depend on the presence and accessibility of the
UNHALTED_REFERENCE_CYCLES PMU counter. This test suite was previously
run "manually" and since it required the presence and access to that
PMU. Since that the perf test suite is now run on `make check` when
libpfm is present, we need to automate the discovery of
UNHALTED_REFERENCE_CYCLES and validate that we can access it.
There are three major scenarios were we want to skip the tests.
1) UNHALTED_REFERENCE_CYCLES is simply not present in the PMU sets for
that hosts.
2) UNHALTED_REFERENCE_CYCLES is present in the PMU sets but not
actionable. This can happen on qemu guests.
3) UNHALTED_REFERENCE_CYCLES is present but not accessible. This can be
happen if the `/proc/sys/kernel/perf_event_paranoid` prevents the usage
of the PMU.
Solution
========
Two problems were found with `find_event.c`.
1) It took the first event matching the passed name even if it was in a
PMU not supported by the host. In our use case it worked since the only
platform that does not use `r300` is not currently in our testing set.
-> = PMU set currently choosen
-* = The correct PMU set
e.g:
-> Intel Core
r300
Intel Atom
r300
Intel Nehalem
r300
Intel Nehalem EX
r300
Intel X86 architectural PMU
r13c
...
-* Intel Skylake
r300
On my system only the following are "detected" as per libpfm example
found here [1].
[18, ix86arch, "Intel X86 architectural PMU"]
[51, perf, "perf_events generic PMU"]
[110, rapl, "Intel RAPL"]
[114, perf_raw, "perf_events raw PMU"]
[200, skl, "Intel Skylake"]
Hence the `skl` PMU set should be used.
2) libpfm does not perform any validation as to if the event is actually
usable or not.
To fix those two problems, we use pfm_get_os_event_encoding and
perf_event_open.
pfm_get_os_event_encoding [2] is responsible for performing the query
across valid PMU sets and encoding it to the perf struct format.
perf_event_open is used to validate that the event can be used. It tests
the availability on the running host and the accessibility of the PMU.
Based on the result of `find_event` the tests are skipped or failed as
necessary.
Known drawbacks
========
The only drawback is that the tests, albeit having libpfm as a
dependency, are not guarantee to run on all hosts. There is not much we
can do here. We can only validate that it is indeed run on our CI, most
probably using lava hardware-based workers.
References
==========
[1] https://sourceforge.net/p/perfmon2/libpfm4/ci/
288483932c3eb83202b0d8762aa0ed8534982c3f/tree/examples/check_events.c
[2] https://man7.org/linux/man-pages/man3/pfm_get_os_event_encoding.3.html
[3] https://man7.org/linux/man-pages/man2/perf_event_open.2.html
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iea7794dc28d019953930992a2237a1b606368d1f
Jonathan Rajotte [Mon, 15 Mar 2021 17:13:37 +0000 (13:13 -0400)]
Fix: tests: missing LOG_DRIVER and LOG_DRIVER_FLAGS
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I808bb2bb80824a574850cc3015cb24eec7b3ff81
Mathieu Desnoyers [Fri, 12 Mar 2021 21:52:42 +0000 (16:52 -0500)]
Adapt to lttng-ust ust-abi.h naming prefix update
LTTng-UST updates all its ust-abi.h symbols and defines to use the
lttng_ust_abi_ and LTTNG_UST_ABI_ prefixes. Adapt lttng-tools
accordingly.
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifde8b73604c84ee37e9123f5dd912e7a806deb38
Depends-on: lttng-ust: I1273be9615846304266206769700e799bde4e51f
Jonathan Rajotte [Thu, 16 Apr 2020 20:31:16 +0000 (16:31 -0400)]
Use bytecode seqnum to force the evaluation ordering of capture bytecode
This ensure that the resulting payload is ordered as expected.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I74c87cdc7098030e05b060f287561b4177739cc1
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Thu, 9 Apr 2020 01:40:41 +0000 (21:40 -0400)]
sessiond: ust-app: set capture bytecode on event notifier on creation
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I82d2dd27da92c4dc1587b22628f8e1a8097e06c2
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Tue, 28 Apr 2020 14:37:55 +0000 (10:37 -0400)]
sessiond: ust-app: add utils to add a capture bytecode to a ust object
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If5b5954cadfafbf851e3fc3a9e37c76c3840a4a3
Depends-on: lttng-ust: I8423c510bf6af2f9bf85256e8d6f931d36f7054b
Jonathan Rajotte [Thu, 14 May 2020 16:44:38 +0000 (12:44 -0400)]
UST abi: add `struct lttng_ust_capture_bytecode`
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I512bb3c5e0232f70b860d33c72d61176303146bc
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Jérémie Galarneau [Fri, 12 Mar 2021 19:54:52 +0000 (14:54 -0500)]
Fix: sessiond: off-by-one poll check when draining an event notifier
When an event source is removed (on the death of an application), the
notification thread "drains" any remaining notifications from the
event notifier pipe.
In doing so, it creates a new poll set containing the event source to
check if messages are left in the event notification pipe.
The invocation of `LTTNG_POLL_GETEV(&events, 1)` means to check the
events pending for the first (and only) fd in the poll set. This check
is off by one since `0` should be used.
For some reason, this worked everywhere except when using a 32-bit
userland on a 64-bit kernel (on x86_64).
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6f274fdd7c80d5676fd48ae20a14adb3cc010142
Philippe Proulx [Fri, 1 May 2020 20:31:01 +0000 (16:31 -0400)]
lttng-ctl: add event field value API
This patch adds a new read-only event field value API to liblttng-ctl.
It also adds the internal functions to create such objects.
As of this patch, an event field value is one of:
* An unsigned integer (64-bit).
* A signed integer (64-bit).
* An unsigned enumeration (64-bit).
* A signed enumeration (64-bit).
* A real number (double precision).
* A string (UTF-8).
* An array of zero or more event field values.
The first purpose of this API will be to access the captured field
values of a notification's trigger condition evaluation.
An enumeration event field value conceptually is an integer event field
value, therefore lttng_event_field_value_unsigned_int_get_value() works
for an unsigned enumeration and
lttng_event_field_value_signed_int_get_value() works for a signed
enumeration.
The compound event field values (array only, as of this patch) can
return the `LTTNG_EVENT_FIELD_VALUE_STATUS_UNAVAILABLE` status when you
access a contained event field value. This indicates that there is no
field at this location. For the captured field value use case, it means
the event did not have any field at that position considering the
condition's capture descriptors. Internally, call
lttng_event_field_value_array_append_unavailable() to append an
unavailable event field value to an array field value.
The lttng_event_field_value_array_get_element_at_index() function can
also return an event field value of which the type, as returned by
lttng_event_field_value_get_type(), is
`LTTNG_EVENT_FIELD_VALUE_TYPE_UNKNOWN`. This is different from the
`LTTNG_EVENT_FIELD_VALUE_STATUS_UNAVAILABLE` status: the event field
value exists (which is why there's a valid
`struct lttng_event_field_value *`), but its type is unknown by this
version of the library. This makes it possible to add new event field
value types in the future while remaining forward compatible.
Note that enumeration labels will not visible to clients since the
kernel tracer does not make their content accessible to the session
daemon.
Signed-off-by: Philippe Proulx <eeppeliteloop@gmail.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I04d327d9f50ff6c6e2ad1f5a1e56c3af25c92c15
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Francis Deslauriers [Thu, 11 Mar 2021 16:18:40 +0000 (11:18 -0500)]
Cleanup: sessiond: fix comments to match with behavior
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib8b7826d9b86df84eae048a7607ecc5a93aa5f56
Francis Deslauriers [Thu, 11 Mar 2021 16:06:53 +0000 (11:06 -0500)]
Fix: sessiond: return _OK on _SET_SESSION_SHM_PATH command success
Issue
=====
When running the `test_crash` tests, I witness the following warning:
Warning: Command returned an invalid status code, returning unknown error: command type = 40, ret = 0
The `cmd_set_session_shm_path()` function is returning 0 on success
instead of the expected LTTNG_OK which leads to this warning being
printed.
Fix
===
Return LTTNG_OK.
Side note
=========
I added the string version of the command name in the warning printing
for easier debugging.
Signed-off-by: Francis Deslauriers <francis.deslauriers@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I79fe41afd49a1c701e760e415b72aabdc9b25f6c
Jonathan Rajotte [Thu, 9 Apr 2020 00:03:06 +0000 (20:03 -0400)]
Generate capture bytecode on trigger reception
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibf86096bd17be078cfc5e8bcd82347797389d448
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Jonathan Rajotte [Tue, 3 Nov 2020 23:04:34 +0000 (18:04 -0500)]
Implement lttng_condition_event_rule_get_capture_bytecode_at_index
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0478c3c1a5866f924fa8a5b49fdc61570a802989
Jonathan Rajotte [Wed, 8 Apr 2020 21:51:52 +0000 (17:51 -0400)]
condition: implement lttng_condition_event_rule_generate_capture_descriptor_bytecode
Generate the bytecode related to the tracer notification capture
feature.
The bytecode is stored alongside the lttng_event_expr in an internal
lttng_capture_descriptor object.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4cfbdd9ab91328a5540ad19049e056fe3ce7f5ba
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Jonathan Rajotte [Wed, 29 Apr 2020 19:36:00 +0000 (15:36 -0400)]
condition: implement lttng_condition_event_rule_get_internal_capture_descriptor_at_index
The lttng_capture_descriptor object is only exposed internally. A
getter is necessary to allow more control on the
lttng_capture_descriptor object. This will be reused to set the bytecode
later in the patchset.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7fdfdc9d2ac1920bcf1a098352dfd105949e1062
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Jérémie Galarneau [Thu, 11 Mar 2021 20:26:01 +0000 (15:26 -0500)]
Fix: lttng: add-trigger: payload capture fields specified as contexts
ir_op_load_expr_to_event_expr() re-uses the same `load_expr_op` variable
to traverse the IR. A check for the type of the original load expression
is performed after `load_expr_op->next` is assigned to load_expr_op
which is expected to have remained the same.
The original event_expr child type is sampled at the start of the
function and re-used as needed.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I14ebcb9088c5107f3133ff5b43c65c3fc6101077
Jérémie Galarneau [Tue, 9 Mar 2021 21:39:11 +0000 (16:39 -0500)]
Fix: event-expr-to-bytecode: use after free in logging on error
1448709 Use after free
This could cause an immediate crash or incorrect values might be read
subsequently resulting in incorrect computations.
In event_expr_to_bytecode_recursive: A pointer to freed memory is
dereferenced, used as a function argument, or otherwise used (CWE-416)
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifde0e21dbf6bf08be07b38f9c38896bdce973132
Jérémie Galarneau [Tue, 9 Mar 2021 21:23:37 +0000 (16:23 -0500)]
msgpack: silence uninitialized msg_pack_object warning
Present in the upstream msgpack code. I couldn't find a problematic use,
but this will silence Coverity and is cleaner.
Nothing should access 'via' when the type of a msg_pack_object is NIL.
Coverity report:
1448708 Uninitialized scalar variable
The variable will contain an arbitrary value left from earlier
computations.
In template_callback_root: Use of an uninitialized variable (CWE-457)
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4fe63b0279ca7c3f4c34ee8762ec724a731894c5
Jérémie Galarneau [Tue, 9 Mar 2021 21:21:11 +0000 (16:21 -0500)]
Fix: lttng: add-trigger: erroneous null check on location return
1447042 Dereference before null check
There may be a null pointer dereference, or else the comparison against
null is unnecessary.
In parse_kernel_probe_opts: All paths that lead to this null pointer
comparison already dereference the pointer earlier (CWE-476)
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifcaf1b1854ea7177084ef5433f5a1b386381676a
Jonathan Rajotte [Wed, 8 Apr 2020 14:45:40 +0000 (10:45 -0400)]
condition: introduce lttng_capture_descriptor struct
The capture_index corresponds to the index at which the capture is
expected on reception of the trigger payload from the tracer.
This field will be populated on reception of the trigger on the sessiond
side and is internal use only.
This is necessary due to the fact that a single condition can have
multiple capture actions associated to it with capture descriptor
duplicate. We plan on forming an ordered set of these capture
descriptors and configure only the set for the enabler. This ensures
that we capture only what is required on the tracer side. On reception,
each capture descriptor will know the index at which the payload field
it corresponds to is situated.
In retrospect this is purely an optimization to prevent the tracer from
doing unnecessary work.
Initially this complexity was planned to be contained to the
lttng-sessiond, but since we now transfer the msgpack payload all the
way to liblttngctl we now have to carry this index all the way to
liblttng-ctl.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie3309511347818f88c61c750723f21c73d82f80b
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Jérémie Galarneau [Fri, 25 Sep 2020 17:08:51 +0000 (13:08 -0400)]
sessiond: Add msgpack-c 3.3.0 to the tree
Add a copy of the msgpack-c v3.3.0 release to the tree. Only the files
of the 'C' version of the library are included and the only
modifications are to add the 'vendor/' prefix to the includes to make
sure we don't interact with a system installed version of msgpack and
the addition of lttng-config.h to glue in our endianess compat code.
Currently, the project is built against the system's libmsgpack-c.
However, this isn't viable for release since the library has had
breaking changes (sometimes without bumping the major version) every
year since its initial release.
Given the variety of distribution versions we have to support, I don't
feel comfortable commiting to support all possible versions of this
library.
Nonetheless, this appears to be the most robust C msgpack implementation
and it is permissively licensed, so it is simply easier to add a copy to
the tree.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Change-Id: Ic3d3fedff9a24ca81ba66cc0c970ba0f8f0cba9a
Jonathan Rajotte [Wed, 8 Apr 2020 14:43:55 +0000 (10:43 -0400)]
ust-app: pass down trigger object instead of event-rule
This is necessary since the conditions can contain capture descriptor
related to the event rule. The trigger object allows for access to both
conditions and actions.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I65452340145278b52d897c9d53c402b22b8941d8
Depends-on: lttng-ust: I5a800fc92e588c2a6a0e26282b0ad5f31c044479
Jérémie Galarneau [Mon, 8 Mar 2021 22:41:37 +0000 (17:41 -0500)]
Clean-up: bytecode: harmonize storage-class specifier placement
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2dd85b25caa9e53d897dc773c1a250ee10a70baf
Jérémie Galarneau [Mon, 8 Mar 2021 22:40:45 +0000 (17:40 -0500)]
Clean-up: bytecode: hide file-local bytecode_reserve util
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic68ab476ed16bc2f327d5967438fcb9febe2ead1
This page took 0.046762 seconds and 4 git commands to generate.