Jérémie Galarneau [Wed, 7 Oct 2020 18:10:35 +0000 (14:10 -0400)]
Fix: relayd: failure to read index entry or stream packet after clear
Observed issue
==============
The clear tests occasionally fail with the following babeltrace error
when a live session is stopped following a "clear". Unfortunately, this
problem only seems to occur on certain machines. In my case, I only
managed to reproduce this on the CI's workers.
10-07 12:39:48.333 7679 7679 E PLUGIN/SRC.CTF.LTTNG-LIVE/VIEWER lttng_live_get_stream_bytes@viewer-connection.c:1610 [lttng-live] Received get_data_packet response: error
10-07 12:39:48.333 7679 7679 E PLUGIN/CTF/MSG-ITER request_medium_bytes@msg-iter.c:563 [lttng-live] User function failed: status=ERROR
10-07 12:39:48.333 7679 7679 E PLUGIN/CTF/MSG-ITER ctf_msg_iter_get_next_message@msg-iter.c:2899 [lttng-live] Cannot handle state: msg-it-addr=0x5603c28e2830, state=DSCOPE_TRACE_PACKET_HEADER_BEGIN
10-07 12:39:48.333 7679 7679 E PLUGIN/SRC.CTF.LTTNG-LIVE lttng_live_iterator_next_handle_one_active_data_stream@lttng-live.c:845 [lttng-live] CTF message iterator failed to get next message: msg-iter=0x5603c28e2830, msg-iter-status=ERROR
10-07 12:39:48.333 7679 7679 E PLUGIN/SRC.CTF.LTTNG-LIVE lttng_live_msg_iter_next@lttng-live.c:1665 [lttng-live] Error preparing the next batch of messages: live-iter-status=LTTNG_LIVE_ITERATOR_STATUS_ERROR
10-07 12:39:48.333 7679 7679 W LIB/MSG-ITER bt_message_iterator_next@iterator.c:864 Component input port message iterator's "next" method failed: iter-addr=0x5603c28cb0f0, iter-upstream-comp-name="lttng-live", iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=SOURCE, iter-upstream-comp-class-name="lttng-live", iter-upstream-comp-class-partial-descr="Connect to an LTTng relay daemon", iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
10-07 12:39:48.333 7679 7679 E PLUGIN/FLT.UTILS.MUXER muxer_upstream_msg_iter_next@muxer.c:454 [muxer] Upstream iterator's next method returned an error: status=ERROR
10-07 12:39:48.333 7679 7679 E PLUGIN/FLT.UTILS.MUXER validate_muxer_upstream_msg_iters@muxer.c:991 [muxer] Cannot validate muxer's upstream message iterator wrapper: muxer-msg-iter-addr=0x5603c28dbe70, muxer-upstream-msg-iter-wrap-addr=0x5603c28cd0f0
10-07 12:39:48.333 7679 7679 E PLUGIN/FLT.UTILS.MUXER muxer_msg_iter_next@muxer.c:1415 [muxer] Cannot get next message: comp-addr=0x5603c28dc960, muxer-comp-addr=0x5603c28db0a0, muxer-msg-iter-addr=0x5603c28dbe70, msg-iter-addr=0x5603c28caf80, status=ERROR
10-07 12:39:48.333 7679 7679 W LIB/MSG-ITER bt_message_iterator_next@iterator.c:864 Component input port message iterator's "next" method failed: iter-addr=0x5603c28caf80, iter-upstream-comp-name="muxer", iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=FILTER, iter-upstream-comp-class-name="muxer", iter-upstream-comp-class-partial-descr="Sort messages from multiple inpu", iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
10-07 12:39:48.333 7679 7679 W LIB/GRAPH consume_graph_sink@graph.c:473 Component's "consume" method failed: status=ERROR, comp-addr=0x5603c28dcb60, comp-name="pretty", comp-log-level=WARNING, comp-class-type=SINK, comp-class-name="pretty", comp-class-partial-descr="Pretty-print messages (`text` fo", comp-class-is-frozen=0, comp-class-so-handle-addr=0x5603c28c8140, comp-class-so-handle-path="/home/jenkins/jgalar-debug/build/usr/lib/babeltrace2/plugins/babeltrace-plugin-text.so", comp-input-port-count=1, comp-output-port-count=0
10-07 12:39:48.333 7679 7679 E CLI cmd_run@babeltrace2.c:2548 Graph failed to complete successfully
10-07 12:39:48.333 7679 7679 E PLUGIN/SRC.CTF.LTTNG-LIVE/VIEWER lttng_live_session_detach@viewer-connection.c:1227 [lttng-live] Unknown detach return code 0
ERROR: [Babeltrace CLI] (babeltrace2.c:2548)
Graph failed to complete successfully
CAUSED BY [libbabeltrace2] (graph.c:473)
Component's "consume" method failed: status=ERROR, comp-addr=0x5603c28dcb60,
comp-name="pretty", comp-log-level=WARNING, comp-class-type=SINK,
comp-class-name="pretty", comp-class-partial-descr="Pretty-print messages
(`text` fo", comp-class-is-frozen=0, comp-class-so-handle-addr=0x5603c28c8140,
comp-class-so-handle-path="/home/jenkins/jgalar-debug/build/usr/lib/babeltrace2/plugins/babeltrace-plugin-text.so",
comp-input-port-count=1, comp-output-port-count=0
CAUSED BY [libbabeltrace2] (iterator.c:864)
Component input port message iterator's "next" method failed:
iter-addr=0x5603c28caf80, iter-upstream-comp-name="muxer",
iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=FILTER,
iter-upstream-comp-class-name="muxer",
iter-upstream-comp-class-partial-descr="Sort messages from multiple inpu",
iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
CAUSED BY [muxer: 'filter.utils.muxer'] (muxer.c:991)
Cannot validate muxer's upstream message iterator wrapper:
muxer-msg-iter-addr=0x5603c28dbe70,
muxer-upstream-msg-iter-wrap-addr=0x5603c28cd0f0
CAUSED BY [muxer: 'filter.utils.muxer'] (muxer.c:454)
Upstream iterator's next method returned an error: status=ERROR
CAUSED BY [libbabeltrace2] (iterator.c:864)
Component input port message iterator's "next" method failed:
iter-addr=0x5603c28cb0f0, iter-upstream-comp-name="lttng-live",
iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=SOURCE,
iter-upstream-comp-class-name="lttng-live",
iter-upstream-comp-class-partial-descr="Connect to an LTTng relay daemon",
iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (lttng-live.c:1665)
Error preparing the next batch of messages:
live-iter-status=LTTNG_LIVE_ITERATOR_STATUS_ERROR
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (lttng-live.c:845)
CTF message iterator failed to get next message: msg-iter=0x5603c28e2830,
msg-iter-status=ERROR
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (msg-iter.c:2899)
Cannot handle state: msg-it-addr=0x5603c28e2830,
state=DSCOPE_TRACE_PACKET_HEADER_BEGIN
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (msg-iter.c:563)
User function failed: status=ERROR
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (viewer-connection.c:1610)
Received get_data_packet response: error
This occurs immediately following a 'stop' on the session. As the error
indicates, a request to obtain a data packet fails with a generic
error reply.
Moreover, the following LTTNG_VIEWER_DETACH_SESSION appears to fail
with an invalid status code. This is addressed in a different commit.
Reproducing the test's failure without redirecting the relay daemon's
allows us to see the following errors after the first stop:
PERROR - 14:33:44.
929675253 [25108/25115]: Failed to open fs handle to ust/uid/1001/64-bit/index/chan_0.idx, open() returned: No such file or directory (in fd_tracker_open_fs_handle() at fd-tracker.c:550)
PERROR - 14:33:45.
030037417 [25108/25115]: Failed to open fs handle to ust/uid/1001/64-bit/index/chan_0.idx, open() returned: No such file or directory (in fd_tracker_open_fs_handle() at fd-tracker.c:550)
PERROR - 14:33:45.
130429370 [25108/25115]: Failed to open fs handle to ust/uid/1001/64-bit/index/chan_0.idx, open() returned: No such file or directory (in fd_tracker_open_fs_handle() at fd-tracker.c:550)
PERROR - 14:33:45.
230829447 [25108/25115]: Failed to open fs handle to ust/uid/1001/64-bit/index/chan_0.idx, open() returned: No such file or directory (in fd_tracker_open_fs_handle() at fd-tracker.c:550)
PERROR - 14:33:45.
331223320 [25108/25115]: Failed to open fs handle to ust/uid/1001/64-bit/index/chan_0.idx, open() returned: No such file or directory (in fd_tracker_open_fs_handle() at fd-tracker.c:550)
This is produced with the following back-trace:
(gdb) bt
#0 __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
#1 0x00007ffff69648b1 in __GI_abort () at abort.c:79
#2 0x00005555555b4f1f in fd_tracker_open_fs_handle (tracker=0x55555582c620, directory=0x7fffe8006680,
path=0x7ffff0a25870 "ust/uid/1001/64-bit/index/chan_1.idx", flags=0, mode=0x7ffff0a24508) at fd-tracker.c:550
#3 0x0000555555595c34 in _lttng_trace_chunk_open_fs_handle_locked (chunk=0x7fffe0002130, file_path=0x7ffff0a25870 "ust/uid/1001/64-bit/index/chan_1.idx",
flags=0, mode=432, out_handle=0x7ffff0a24710, expect_no_file=true) at trace-chunk.c:1388
#4 0x0000555555595eef in lttng_trace_chunk_open_fs_handle (chunk=0x7fffe0002130, file_path=0x7ffff0a25870 "ust/uid/1001/64-bit/index/chan_1.idx", flags=0,
mode=432, out_handle=0x7ffff0a24710, expect_no_file=true) at trace-chunk.c:1433
#5 0x00005555555da6c2 in _lttng_index_file_create_from_trace_chunk (chunk=0x7fffe0002130, channel_path=0x7fffe8018c30 "ust/uid/1001/64-bit",
stream_name=0x7fffe8018c10 "chan_1", stream_file_size=0, stream_file_index=0, index_major=1, index_minor=1, unlink_existing_file=false, flags=0,
expect_no_file=true, file=0x7fffe0002270) at index.c:97
#6 0x00005555555dad8a in lttng_index_file_create_from_trace_chunk_read_only (chunk=0x7fffe0002130, channel_path=0x7fffe8018c30 "ust/uid/1001/64-bit",
stream_name=0x7fffe8018c10 "chan_1", stream_file_size=0, stream_file_index=0, index_major=1, index_minor=1, expect_no_file=true, file=0x7fffe0002270)
at index.c:186
#7 0x000055555557640f in try_open_index (vstream=0x7fffe0002250, rstream=0x7fffe8018c50) at live.c:1378
#8 0x0000555555577155 in viewer_get_next_index (conn=0x7fffd4001440) at live.c:1643
#9 0x0000555555579a01 in process_control (recv_hdr=0x7ffff0a27c30, conn=0x7fffd4001440) at live.c:2311
#10 0x000055555557a1db in thread_worker (data=0x0) at live.c:2482
#11 0x00007ffff6d1c6db in start_thread (arg=0x7ffff0a28700) at pthread_create.c:463
#12 0x00007ffff6a45a3f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
That problem is mostly cosmetic in nature (the open can fail
"legitimately") as the PERROR should simply not be printed and is
addressed in a different commit.
This error is also produced after a 'clear' is issued:
PERROR - 14:33:45.
532782268 [25108/25115]: Failed to read from file system handle of viewer stream id 1, offset: 4096: No such file or directory (in viewer_get_packet() at live.c:1849)
Which is produced with the following back-trace:
#0 __GI_raise (sig=sig@entry=6) at ../sysdeps/unix/sysv/linux/raise.c:51
#1 0x00007f53e297c8b1 in __GI_abort () at abort.c:79
#2 0x000055dd77ccef2c in viewer_get_packet (conn=0x7f53c4001100) at live.c:1850
#3 0x000055dd77cd0a15 in process_control (recv_hdr=0x7f53dca3fc30, conn=0x7f53c4001100) at live.c:2315
#4 0x000055dd77cd11db in thread_worker (data=0x0) at live.c:2483
#5 0x00007f53e2d346db in start_thread (arg=0x7f53dca40700) at pthread_create.c:463
#6 0x00007f53e2a5da3f in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
A similar problem occurs, although more rarely, when reading an
index entry in viewer_get_next_index().
Cause
=====
The following situation leads to both failures to get a
packet and failures to get the next index:
- Viewer connects to an existing session,
- Viewer consumes a number of packets, alternating the
GET_NEXT_INDEX and GET_PACKET command,
- The session's streams are rotated to a new trace chunk
(as part of a clear),
- The session is started and stopped, causing new packets
to be produced and received,
- The session is stopped and destroyed, causing the session's
streams to rotate into a "null" trace chunk (no active
trace files),
- Viewer issues GET_NEXT_INDEX or GET_PACKET, but the fact
that a rotation occurred on the receiving end is not detected
as the relay streams' trace chunk are "null".
The crux of the problem is that lttng_trace_chunk_ids_equal() is
bypassed when the current trace chunk of a relay stream is "null".
The rationale for skipping this check is that it is assumed that the
files currently opened by the live server can can still be used even
if the consumer has rotated the corresponding streams into a 'null'
trace chunk, meaning no trace chunk is 'set' for those streams.
This makes sense in one scenario: the session was destroyed and we wish
to allow a connected live client to finish consuming the trace packets
up to the end of the session's lifetime.
Here, the situation is different. The viewer is reading chunk 'A'.
Meanwhile, a rotation occurs into chunk 'B' and packets are received for
chunk 'B'. Then, a rotation to a 'null' chunk (no active chunk) occurs.
In essence, the live server never sees the rotation between chunk 'A'
and 'B', and simply assumes that a rotation from 'A' to 'null' occurred,
as would happen at the end of a session.
In terms of the code, in viewer_get_next_index(), a call to
check_index_status() is performed to determine if an index is available.
The function checks that `index_received_seqcount` is greater than
`index_sent_seqcount`. In that case, it determines that an index must be
available.
Unfortunately, there is no way for the live server to determine that the
remaining indexes are in a chunk that doesn't exist anymore (chunk 'B').
Thus, viewer_get_next_index() attempts to read an index entry from the
current index file and fails.
Solution
========
1) lttng_trace_chunk_ids_equal() is modified to properly handle
'null' trace chunks:
- A null and a non-null trace chunk are not equal,
- Two null trace chunks are equal.
2) Rotation count
A rotation counter is introduced to track the number of rotations
that occurred during a relay stream's lifetime. This counter is
sampled by the matching viewer streams on creation and on rotation
and is used to determine if all rotations were "seen" by the viewer
stream.
Hence, this allows us to handle the special case where a viewer
is consuming the contents of a relay stream that just transitioned
into a 'null' trace chunk (see comments in patch).
The rest of the modifications simply allow the live server to handle
null trace chunks in viewer streams. This fixes another unrelated bug
that I observed while investigating this: sessions that don't have an
active trace chunk are not shown when listing sessions with babeltrace.
To reproduce, simply stop, clear a session, and attempt to list the
sessions of the associated relay daemon.
Known drawbacks
===============
None.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibb3116990e34b7ec3b477f3482d0c0ff1e848d09
Jérémie Galarneau [Tue, 13 Oct 2020 18:55:33 +0000 (14:55 -0400)]
Fix: lttng-ctl: erroneous uses of LTTNG_PACKED
The LTTNG_PACKED macro uses gcc attributes to indicate that a structure
should be packed. Hence, this macro obeys the same rules as the gcc
attribute.
Various mis-uses of the LTTNG_PACKED macros may result in structure not
being packed:
- The LTTNG_PACKED macro should always be placed _before_ an identifier
when a structure is declared in-place.
- Adding LTTNG_PACKED at the definition site has no effect if the
structure was declared elsewhere.
Those mis-uses cause issues when mixing the bitness (32/64) of the
session daemon and liblttng-ctl.
Outstanding issues include the following structures that are not
tagged as LTTNG_PACKED:
- struct lttng_event
- struct lttng_channel
- struct lttng_event_context
Unfortunately, those structures are exposed by the public API and
can't be tagged as being "packed". Doing so would break the ABI
of liblttng-ctl.
These structures should be packed/unpacked explicitly.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I095dc0dffc6bf9e15dc7a7ec797958a5780ef150
Jérémie Galarneau [Fri, 9 Oct 2020 16:04:10 +0000 (12:04 -0400)]
Fix: relayd: live: invalid return code on DETACH_SESSION
Babeltrace 2 reports an invalid return code being returned in reply to a
DETACH_SESSION command.
Reviewing the relevant Babeltrace 2 code, the logging can only be
produced if the reception of the lttng_viewer_detach_session_response
structure succeeds.
This elemininated my first guess that this was caused by the relay
daemon closing the socket before sending the reply. In that case, an
invalid status code of '0' could have been erroneously returned as a
status code since the recv() call on the socket would return 0.
It turns out that on a failure to return a packet, viewer_get_packet()
returns an error status code, but also sends a zero-initialized payload
buffer of the size of the requested packet.
This causes live clients which detach following the error of the
GET_PACKET command to interpret the still-enqueued zero-initialized
buffer as a reply to the DETACH_SESSION command. Since zero is not a
valid status code, it is correctly interpreted as a protocol error.
The reply_size is set to the header's size to only transmit the header
when an error reply is sent.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I69ed74f83404a16353d2bdbaa9f3adcdc2a03892
Jérémie Galarneau [Thu, 8 Oct 2020 22:15:34 +0000 (18:15 -0400)]
Tests: clear: remove test workspace directory
The clear tests only removes its workspace's subdirectory, but
leaves an empty directory behind. Remove the wildcard and remove
the root of the workspace on clean-up.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I551f892af5423c6ed5933beb0c1a13f41a30a26e
Jérémie Galarneau [Mon, 21 Sep 2020 21:24:50 +0000 (17:24 -0400)]
Tests: ns_contexts: discarded events result in test failure
A follow-up change makes all events emited by gen-ust-events
a bit larger, which causes them to no longer fit in the default
channel configuration's buffers.
This causes the test to fail occasionnaly when the consumer daemon
fails to consume the packets fast enough to leave room in the
buffers for new events.
The test doesn't need to produce 10,000 events; reducing to
1,000 produced events makes no material difference and works
around the problem.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie87583bb9bb9cdd813f80443231a65164ef67df1
Jérémie Galarneau [Tue, 15 Sep 2020 20:22:14 +0000 (16:22 -0400)]
Fix: PERROR spam when `tracing` group does not exist
The session daemon prints a PERROR on launch when the tracing group does
not exist. This should not occur when the group simply does not exist as
this is not an error. In that case (ESRCH), a DBG statement is
sufficient.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3ade29071a8f4e9fe2eb56bf05ff4150b70fd463
Jérémie Galarneau [Fri, 11 Sep 2020 03:57:54 +0000 (23:57 -0400)]
Build fix: implicit declaration of function 'PERROR' on Solaris
Solaris 10/11 CI builds complain that PERROR is not declared
in memstream.h. This addresses the warning on both platforms.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I09ce7cea216e075d0ef7256f738ea7400d193b6b
Jonathan Rajotte [Thu, 28 Nov 2019 19:19:16 +0000 (14:19 -0500)]
tests: unit: event-rule unit testing
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic25fac00027b9b62c7d9dd060e75228aa7e86c57
Jonathan Rajotte [Wed, 11 Mar 2020 17:58:48 +0000 (13:58 -0400)]
event-rule: introduce event rule tracepoint
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I77495a7b0e359e6f513dfc862dec65b4946d08ba
Jonathan Rajotte [Wed, 11 Mar 2020 17:56:36 +0000 (13:56 -0400)]
event-rule: introduce event rule uprobe
This is the "--userspace-probe" option of the enable-event command line.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I43478cb75bd8fdc28e2927d53e0a9a09f2283dea
Jonathan Rajotte [Wed, 11 Mar 2020 17:55:16 +0000 (13:55 -0400)]
event-rule: introduce event rule syscall
This is the "--syscall" option of the enable-event command line.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9c36cb0440e1f26a3b436782eb437a8ddffbe4fa
Jonathan Rajotte [Wed, 11 Mar 2020 17:52:38 +0000 (13:52 -0400)]
event-rule: introduce event-rule kprobe
This is the "--probe" option of the enable-event command line.
Change-Id: I6d763df53d5e838ea2266d49b1f22bc23a9addb1
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jonathan Rajotte [Wed, 11 Mar 2020 17:50:08 +0000 (13:50 -0400)]
event-rule: lttng_event_rule base object
A lttng_event_rule object is the base object representing an event-rule.
We plan on using the event-rule object for compositing with a new
condition type.
This also paves the way toward further improvement of the lttng_event
related APIs.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5bbfa3b75337f040cf7e565447d4a4af590ed043
Jonathan Rajotte [Mon, 24 Aug 2020 19:50:49 +0000 (15:50 -0400)]
Introduce kernel-probe locations
Kernel probe can be configured by two type of location.
The first one is via address:
lttng_kernel_probe_location_address_create.
The second one is using a symbol name combined with an offset.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Icd280e7a8403c761987472f0a416ef5188a2068d
Jérémie Galarneau [Thu, 10 Sep 2020 16:23:28 +0000 (12:23 -0400)]
userspace-probe: replace explicit null-termination check
Replace explicit null-termination checks by uses of
lttng_buffer_view_contains_string() which provides the same
guarantees and ensures the string pointer is within the view.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If0038a82ad7dfe1ed0e8cdef7870d5e25d62200d
Jérémie Galarneau [Fri, 11 Sep 2020 15:10:44 +0000 (11:10 -0400)]
Revert "userspace-probe: replace explicit null-termination check"
This reverts commit
b9e63e21bd01c0deeaec2195ba912e38460bc038.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I909339dbdcf4a7931ad95c5b4b6af152aa52487c
Jérémie Galarneau [Thu, 10 Sep 2020 16:29:50 +0000 (12:29 -0400)]
Tests: clean-up: remove trailing dot in snapshot test statements
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I95208d67b18ba806f009b9a6b53d5e2cc38b6c74
Jérémie Galarneau [Thu, 10 Sep 2020 16:23:28 +0000 (12:23 -0400)]
userspace-probe: replace explicit null-termination check
Replace explicit null-termination checks by uses of
lttng_buffer_view_contains_string() which provides the same
guarantees and ensures the string pointer is within the view.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If0038a82ad7dfe1ed0e8cdef7870d5e25d62200d
Jérémie Galarneau [Thu, 10 Sep 2020 16:22:59 +0000 (12:22 -0400)]
userspace-probe: log function name on invalid parameter error
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I31bb1e7af0cde85a7ddb4614796cae1da3e5b1b9
Jonathan Rajotte [Fri, 29 Nov 2019 21:12:28 +0000 (16:12 -0500)]
Allow run-as to generate filter bytecode.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3742bb00b7c753b7d256cdff1889a5e90865608b
Michael Jeanson [Wed, 26 Aug 2020 18:30:07 +0000 (14:30 -0400)]
Fix: add missing errno.h in pthread compat
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idf574e6e0c7f535693149f3c88f1784188fb14da
Michael Jeanson [Wed, 12 Aug 2020 21:08:07 +0000 (17:08 -0400)]
Add common util to set thread name
Use the same code to set all the thread names and fail gracefully on
platforms that don't support them.
This sets the minimum requirement for thread names on Linux to Glibc
>= 2.12, which seems reasonable.
Signed-off-by: Michael Jeanson <mjeanson@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I61c0b9adb6c2309fed91b5a1b11ebc5ee2a637ce
Jérémie Galarneau [Fri, 21 Aug 2020 18:28:54 +0000 (14:28 -0400)]
Fix: liblttng-ctl: unchecked return value on buffer append
Allocation failures can cause lttng_dynamic_buffer_append to fail;
its result should always be checked.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id1870b1e19d3451afdd1e992355d83d4028b5723
Jérémie Galarneau [Fri, 21 Aug 2020 18:16:10 +0000 (14:16 -0400)]
Fix: action executor: double work list unlock on error
The action executor executes its queued work items without holding the
work list lock. This can result in a double-unlock when a fatal error
occurs during the processing of a work item.
A check for the "should_quit" flag is added before unlocking since this
is the only other reason for the thread to exit its loop.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8d6b1a3511174abd08848ac3677cdf8a326fa8c5
Jonathan Rajotte [Thu, 28 Nov 2019 21:09:16 +0000 (16:09 -0500)]
Move filter related code to libfilter under libcommon
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I006dcac801ce1ec742de3a03f21a1c5f8b698298
Jérémie Galarneau [Thu, 20 Aug 2020 19:40:10 +0000 (15:40 -0400)]
Clean-up: consumer: consumer_metadata_cache_write is not const-correct
`data` is used as a source argument and can be marked as `const`.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9b2b32fe253e1b89a2605ed9c859a04141b321d5
Jérémie Galarneau [Thu, 20 Aug 2020 19:38:18 +0000 (15:38 -0400)]
Fix: memcpy used on potentially overlapping regions
Caught by reviewing unrelated code, these two uses of memcpy
can operate on overlapping buffers. I checked all other uses
of "raw" memcpy and those appear safe.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I72b1204bc52a92015042adb6a67b022d140f5b4e
Jonathan Rajotte [Tue, 14 Jul 2020 18:56:37 +0000 (14:56 -0400)]
sessiond: notification: use lttng_payload for communications
Allows passing of fds related to object (e.g userspace probes).
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7bb3a91c71016b2939b0e05aca60d57c2da14a20
Jérémie Galarneau [Tue, 18 Aug 2020 20:01:30 +0000 (16:01 -0400)]
Fix: sessiond: client/client_list lock inversion on disconnect
Coverity reports a lock inversion scenario in
handle_notification_thread_client_disconnect() where a client's lock is
held while acquiring the client list lock. This is indeed a problem.
As indicated in the notification_client and notification_client_list
comments, the locking was shoe-horned to make it possible for the action
executor to enqueue notifications in a client's outgoing queue and flush
it.
Since this is the only access pattern that is supported, the client
locking is reworked slightly to only acquire the client lock when
checking the "active" flag, interacting with the outbound communication
state, and sending through a client's socket.
This change makes the client locking regions more narrow which accounts
for the somewhat large number of lines affected.
The updates to the `active` flag on error are moved to the function that
flushes the outbound queue instead of expecting the callers to set it.
This allows the locking to be limited to this function rather than
relying on the callers.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8632d0f7785ec727dabd329bdfba010fd5e4643a
Jérémie Galarneau [Mon, 17 Aug 2020 20:55:39 +0000 (16:55 -0400)]
Fix: sessiond: missing rcu read lock on client in/out events
Users of get_client_from_sock() must hold the RCU read lock
for the duration of the use of the notification_client.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I644e549187ee47c959eeb692e27be111343d8979
Jérémie Galarneau [Fri, 14 Aug 2020 20:59:18 +0000 (16:59 -0400)]
sessiond: enforce user-exclusive session access in session_access_ok
The current session_access_ok logic disallows the access to a session
when:
uid != session->uid && gid != session->gid && uid != 0
This means that any user that is part of the same primary group as the
session's owner can access the session. The primary group is not
necessarily (and most likely) not the `tracing` group. Moreover, the
`tracing` group is not meant to provide shared access to sessions, but
to allow interactions with a root session daemon.
For instance:
- the session has uid = 1000, gid = 100
- the current user has uid = 1001, gid = 100
access to the session is granted.
This is way too broad and unexpected from most users as the LTTng
documentation never mentions this "primary group share tracing sessions"
behaviour. The documentation only alludes to the fact that separate
users have "their own set of sessions".
On most distributions, this change will have no impact as `useradd`
creates a new group for every user. Users will never share a primary
group and thus can't control each others' sessions.
However, it is not unusual to have users share a primary group (e.g.
`users`) and set the default umask to `0700`. In that case, there is no
expectation that every user will share files and there would be no
reasonable expectation that they should share all sessions.
For instance, it would be unexpected for one user to tear down the
sessions of other users with a single `lttng destroy -a` command.
If this type of session sharing is desirable to some users, then the
default umask of users could be checked or sessions could be created as
part of a group. However, in doubt, it is preferable to be strict.
This is not marked as a fix since this was most likely deliberate and
the change could, although unlikely, break existing deployment
scenarios.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I98f7ffb29d5f6dcb9d660535c1d3f5a1d1a68293
Jérémie Galarneau [Tue, 11 Feb 2020 04:29:18 +0000 (23:29 -0500)]
sessiond: trigger: run trigger actions through an action executor
The `action executor` interface allows the notification subsystem to enqueue
work items to execute on behalf of a given trigger. This allows the notification
thread to remain responsive even if the actions to execute are blocking (as
through the use of network communication).
Before this commit, the notification subsystem only handled `notify` actions;
handling code for new action types are added as part of the action executor.
The existing `notify` action is now performed through the action executor so
that all actions can be managed in the same way.
This is less efficient than sending the notifications directly, but could be
optimized trivially (if it ever becomes a problem) when:
- the action is a group containing only a `notify` action,
- the action is a `notify` action.
Overview of changes to existing code
===
Managing the new action types requires fairly localized changes to the existing
notification subsystem code. The main code paths that are modified are the sites
where `evaluation` objects are created:
1) on an object state change (session or channel state changes, see
handle_notification_thread_channel_sample and
handle_notification_thread_command_session_rotation),
2) on registration of a trigger (see
handle_notification_thread_command_register_trigger),
3) on subscription to a condition (see client_handle_message_subscription).
To understand the lifetime of most objects involved in a work deferral to the
action executor, see the paragraph in notification-thread-internal.h (line 82)
to understand the relation between clients and client lists.
1) Object state changes
As hinted in the notification_client_list documentation, deferring work on a
state change is straight-forward: a reference is taken on a client list and the
list is provided to the action executor as part of a work item.
Hence, very little changes are made to the the two state-change handling sites
beyond enqueuing a work item rather than directly sending a notification.
2) Subscription to a condition
A notification client can subscribe to a condition before or after a matching
trigger (same condition and containing a notify action) has been registered.
If no matching trigger were registered, no client list exists and there is
nothing to do.
If a matching trigger existed, a client list (which could be empty) will already
exist and the client is simply added to the client list. However, it is
important to evaluate the condition for the client (as the condition could
already be true) and send the notification to that client only and not to all
clients in the list.
Before this change, since everything was done in the same thread, a temporary
list containing only the newly-subscribed client was created on the stack and
the notification was sent/queued immediately. After sending the condition, the
client was removed from the temporary list and added to the "real" client list.
This strategy cannot be used with the action executor as the "temporary" client
list must exist beyond the scope of the function. Moreover, the notification
subsystem assumes that clients are in per-condition client lists and that they
can safely be destroyed when they are not present in any list.
Fortunately, here we know that the action to perform is to `notify` and nothing
else. The enqueuing of the notification is performed "in place" by the
notification thread without deferring to the action executor.
3) Registration of a trigger
When a client subscribes to a condition, the current state of that condition is
immediately evaluated. If the condition is true (for instance, a channel's
buffer are filled beyond X% of their capacity), the action associated with the
trigger is executed right away.
This path requires little changes as a client list is created when a trigger is
registered. Hence, it is possible to use the client list to defer work as is
done in `1`.
4) Trigger registration
Since the `notify` action was the only supported action type, the notification
subsystem always created a client list associated with the new trigger's
condition.
This is changed to only perform the creation (and publication) of the client
list if the trigger's action is (or contains, in the case of a group) a `notify`
action.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I43b54b93c1244591aeff6e0d0fa8076c7b5e0c50
Jérémie Galarneau [Mon, 17 Aug 2020 20:06:16 +0000 (16:06 -0400)]
Revert "Fix: sessiond: erroneous user check logic in session_access_ok"
This reverts commit
4064563ea326f6f26d2c458009beb9ebdb3ba840.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ifbcfd0c18631cfca70b2ea85b16824bb26a2a446
Jérémie Galarneau [Mon, 17 Aug 2020 20:06:14 +0000 (16:06 -0400)]
Revert "sessiond: trigger: run trigger actions through an action executor"
This reverts commit
d1ba29d290281cf72ca3ec7b0222b336c747e925.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I70d57fef86aea94a590720689af751e2554184d0
Jérémie Galarneau [Fri, 14 Aug 2020 20:59:18 +0000 (16:59 -0400)]
Fix: sessiond: erroneous user check logic in session_access_ok
The current session_access_ok logic disallows the access to a session when:
uid != session->uid && gid != session->gid && uid != 0
This means that any user that is part of the same primary group as the session's
owner can access the session. The primary group is not necessarily (and most
likely) not the `tracing` group.
For instance:
- the session has uid = 1000, gid = 100
- the current user has uid = 1001, gid = 100
access to the session is granted.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2e9208286e5508315dae90cb25d34133ca5edcc0
Jérémie Galarneau [Tue, 11 Feb 2020 04:29:18 +0000 (23:29 -0500)]
sessiond: trigger: run trigger actions through an action executor
The `action executor` interface allows the notification subsystem to enqueue
work items to execute on behalf of a given trigger. This allows the notification
thread to remain responsive even if the actions to execute are blocking (as
through the use of network communication).
Before this commit, the notification subsystem only handled `notify` actions;
handling code for new action types are added as part of the action executor.
The existing `notify` action is now performed through the action executor so
that all actions can be managed in the same way.
This is less efficient than sending the notifications directly, but could be
optimized trivially (if it ever becomes a problem) when:
- the action is a group containing only a `notify` action,
- the action is a `notify` action.
Managing the new action types requires fairly localized changes to the existing
notification subsystem code. The main code paths that are modified are the sites
where `evaluation` objects are created:
- on an object state change (session or channel state changes, see
handle_notification_thread_channel_sample and
handle_notification_thread_command_session_rotation),
- on registration of a trigger (see
handle_notification_thread_command_register_trigger),
- on subscription to a condition (see client_handle_message_subscription).
To understand the lifetime of most objects involved in a work deferral to the
action executor, see the paragraph in notification-thread-internal.h (line 82)
to understand the relation between clients and client lists.
Overview of changes
===
Object state changes
Change-Id: I23290e94d98e781992661f0aee88de9986ed274f
---
As hinted in the notification_client_list documentation, defering work on a
state change is straight-forward: a reference is taken on a client list and the
list is provided to the action executor as part of a work item.
Hence, very little changes are made to the the two state-change handling sites
beyond enqueuing a work item rather than directly sending a notification.
Subscription to a condition
---
A notification client can subscribe to a condition before or after a matching
trigger (same condition and containing a notify action) has been registered.
When a client subscribes to a condition, it is a added to a corresponding
"client list"
Registration of a trigger
---
When a client subscribes to a condition, the current state of
that condition is immediately evaluated. If the condition is true
(for instance, a channel's buffer are filled beyond X% of their
capacity),
TODO:
Change-Id: I7f9bc197715c9ca008a4f1fcd4c86e01b6252dce
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Jonathan Rajotte [Tue, 5 May 2020 20:06:58 +0000 (16:06 -0400)]
Fix: notification: deadlock on cmd_queue.lock and client->lock
Observed issue
==============
Deadlock between the notification thread and the action executor thread.
Thread 5 holds cmd_queue.lock and request the client lock.
Thread 6 holds the client lock and request the cmd_queue lock.
Thread 5 have little value in holding the queue lock considering it effectively to a "pop" of the cmd_queue.
Thread 9 is waiting on the cmd_queue lock but does not hold any other
locks and thus not part of the deadlock but is a casualties of this
deadlock and leave a client "hanging".
Other threads are all in their respective waiting state.
Thread 9 (Thread 0x7f76f2ffd700 (LWP 240467)):
#0 __lll_lock_wait (futex=futex@entry=0x1ad1308, private=0) at lowlevellock.c:52 [1070/1123]
#1 0x00007f77052c80a3 in __GI___pthread_mutex_lock (mutex=0x1ad1308) at ../nptl/pthread_mutex_lock.c:80
#2 0x00000000004611dd in run_command_wait (handle=0x1ad12f0, cmd=0x7f76f2fe31e0) at notification-thread-commands.c:31
#3 0x000000000046143a in notification_thread_command_unregister_trigger (handle=0x1ad12f0, trigger=0x7f76e4000ef0) at notification-thread-commands.c:148
#4 0x00000000004444af in cmd_unregister_trigger (cmd_ctx=0x7f76e4000d40, sock=68, notification_thread=0x1ad12f0) at cmd.c:4618
#5 0x0000000000483d23 in process_client_msg (cmd_ctx=0x7f76e4000d40, sock=0x7f76f2ffcba4, sock_error=0x7f76f2ffcb90) at client.c:2001
#6 0x000000000047f00b in thread_manage_clients (data=0x1ad1a80) at client.c:2402
#7 0x000000000047b303 in launch_thread (data=0x1ad1af0) at thread.c:66
#8 0x00007f77052c5609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#9 0x00007f77051cc103 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
Thread 6 (Thread 0x7f7700fcf700 (LWP 240464)):
#0 __lll_lock_wait (futex=futex@entry=0x1ad1308, private=0) at lowlevellock.c:52
#1 0x00007f77052c80a3 in __GI___pthread_mutex_lock (mutex=0x1ad1308) at ../nptl/pthread_mutex_lock.c:80
#2 0x0000000000461bf2 in run_command_no_wait (handle=0x1ad12f0, in_cmd=0x7f7700fce340) at notification-thread-commands.c:87
#3 0x0000000000461b93 in notification_thread_client_communication_update (handle=0x1ad12f0, id=1, transmission_status=CLIENT_TRANSMISSION_STATUS_QUEUED) at notification-thread-commands.c:400
#4 0x0000000000497658 in client_handle_transmission_status (client=0x7f76f8004e30, status=CLIENT_TRANSMISSION_STATUS_QUEUED, user_data=0x7f76f8004a00) at action-executor.c:154
#5 0x0000000000467be7 in notification_client_list_send_evaluation (client_list=0x7f76f8004fe0, condition=0x7f76e40041a0, evaluation=0x7f76cc000cc0, trigger_creds=0x7f76e4004288, source_object_creds=0x0, client_report=0x4971a0 <client_ha
ndle_transmission_status>, user_data=0x7f76f8004a00) at notification-thread-events.c:4007
#6 0x00000000004956bb in action_executor_notify_handler (executor=0x7f76f8004a00, work_item=0x7f76f80062d0, action=0x7f76e4004210) at action-executor.c:199
#7 0x00000000004953fd in action_executor_generic_handler (executor=0x7f76f8004a00, work_item=0x7f76f80062d0, action=0x7f76e4004210) at action-executor.c:493
#8 0x0000000000495101 in action_work_item_execute (executor=0x7f76f8004a00, work_item=0x7f76f80062d0) at action-executor.c:506
#9 0x0000000000493ff5 in action_executor_thread (_data=0x7f76f8004a00) at action-executor.c:559
#10 0x000000000047b303 in launch_thread (data=0x7f76f8004aa0) at thread.c:66
#11 0x00007f77052c5609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#12 0x00007f77051cc103 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
Thread 5 (Thread 0x7f77017d0700 (LWP 240463)):
#0 __lll_lock_wait (futex=futex@entry=0x7f76f8004e30, private=0) at lowlevellock.c:52
#1 0x00007f77052c80a3 in __GI___pthread_mutex_lock (mutex=0x7f76f8004e30) at ../nptl/pthread_mutex_lock.c:80
#2 0x0000000000463080 in handle_notification_thread_command (handle=0x1ad12f0, state=0x7f77017cfb00) at notification-thread-events.c:2936
#3 0x000000000045e881 in thread_notification (data=0x1ad12f0) at notification-thread.c:705
#4 0x000000000047b303 in launch_thread (data=0x1ad1420) at thread.c:66
#5 0x00007f77052c5609 in start_thread (arg=<optimized out>) at pthread_create.c:477
#6 0x00007f77051cc103 in clone () at ../sysdeps/unix/sysv/linux/x86_64/clone.S:95
Cause
=====
The action executor holds the client lock across the communication to
prevent simultaneous update to the client state.
The notification thread holds the cmd_queue lock across operation for no
apparent reason (TODO make sure there is no internal add to the queue. if
so we should reacquire the lock only when necessery.)
Solution
========
Reduce the windows for which the cmd_queue lock is held by the
notification thread to only the "pop" action on the queue. As soon as we
have the lock, get the cmd, remove it from the list and release the
lock. This prevent inverted lock acquisition base on the pattern of the
action executor thread.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I91d30c134bc1a128c96058f0e0cdd325808c91bc
Jérémie Galarneau [Thu, 13 Feb 2020 21:50:52 +0000 (16:50 -0500)]
sessiond: notification: add support for async commands
Notification thread commands are currently all blocking. This means
that the emitter of the command uses the notification_thread_command
APIs to (stack) allocate a command context and wait on an lttng_waiter
for a reply.
In preparation for the addition of non-blocking commands, a new type
of asynchroneous commands is introduced. Asynchroneous commands are
"fire-and-forget"; the caller has no way of obtaining the result of
the command.
Asynchroneous command contexts are heap-allocated and their ownership
is transferred to the notification thread. The notification thread
will not attempt to wake-up the emitter and will free() the command
context regardless of the command's result.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6951987dda0262d1500a3b4e1403fb2559a3ff44
Jérémie Galarneau [Thu, 13 Feb 2020 04:34:08 +0000 (23:34 -0500)]
sessiond: notification: refactor: split transmission and poll update
Split the notification transmission logic and its effect on a
notification_client from the logic tied to the management of the
notification thread.
This is to make it possible to send (or queue) notifications from the
notification thread or another thread. If another thread encounters an
error or a full socket buffer, a future mechanism will allow it to
signal the notification thread to update its private state (e.g. poll
mask).
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I35d8943cb11473d82b07a4dbc5a0f093cde25a79
Jérémie Galarneau [Tue, 11 Feb 2020 04:06:19 +0000 (23:06 -0500)]
sessiond: notification: synchronize notification client (and list)
Introduce reference counting to notification_client_list and add
locks to both notification_client and notification_client list.
Of important note, this is in preparation for the introduction of
an action executor thread. The aim of this change is not to make
any part of the notitication sub-system thread-safe in any "general"
sense.
The reference counting and locks are introduced to protect a very
specific usage scenario.
The main thread of the notification subsystem and the action executor
will interact through triggers, evaluations, and client lists.
If the action executor needs to send a notification to a list of
client during the execution of an action group, it obtains the client
list and acquires a reference to it. It then locks the list to iterate
on it, allowing it to send the notification(s) to all subscribed
clients.
Holding the list lock prevents the main thread from disconnecting and
subsequently destroying a client. Holding a reference to the list also
prevents the list from being reclaimed due to a concurent 'unregister
trigger' operation.
No provision for other access scenarios are taken into account.
Squashed fix, otherwise the tests would hang.
Fix: don't hold client lock while handling subscription changes
Holding the client's lock while handling subscription changes causes a
lock inversion between the client list lock and the client lock.
This happens when a client subscribes to a condition that evaluates to
'true' at the time of the subscription. When this happens, a
notification is sent right away and that communication will attempt to
acquire the client lock.
Holding the client lock for such a long period is not needed anyhow
and we can simply protect the communication state when it is actually
modified/used.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7f38f81fa8bc32e5384538acdffab0824862cff2
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I88a9f66d31e3127b2bcc3f04016c364e0b3fe9ce
Jérémie Galarneau [Mon, 10 Feb 2020 23:35:27 +0000 (18:35 -0500)]
sessiond: notification: introduce the notion of 'active' client
Since notification_clients are now accessed from multiple threads, it
is possible for a thread to access a client while it is being
"cleaned-up" following an error.
The 'active' communication flag allows a check to be performed before
any communication is attempted with a client. The communication is
considered 'active' once the handshake has been performed. It is
considered 'inactive' if a fatal protocol error occurs.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If16f83ef12fe9a6ef597ae464867011811389177
Jérémie Galarneau [Sat, 8 Feb 2020 04:21:40 +0000 (23:21 -0500)]
sessiond: notification: maintain an id to notification_client ht
In preparation for the addition of an action execution worker, add a
client_id_ht which will allow the action worker to send commands to
the notification thread referencing clients by 'id' rather than by
their socket.
This is done in order to prevent FD re-use races between the various
threads at play when a communication error occurs on a notification
client socket.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I7b8774c23a4a2a7c41ed8c1652f23d29954a0771
Jonathan Rajotte [Fri, 22 May 2020 14:36:46 +0000 (10:36 -0400)]
Fix: tests: `pgrep -f` flags unrelated process as lttng-sessiond
Observed issue
==============
lttng-sessiond is not started by start_lttng_sessiond_opt and a vim
process is killed on stop_lttng_sessiond_opt.
Cause
=====
We uses "pgrep -f" with the interested pattern to gather the pids that
should be the lttng processes we are interested in. `pgrep -f` yields
false positives since it matches against the complete cmdline including
parent directory of the command and all arguments.
For example, the following will currently match for the sessiond
pattern:
vim src/bin/lttng-sessiond/notification-thread-internal.h
This prevents the launch of sessiond by start_lttng_sessiond_opt and end
up killing the vim process on stop_lttng_sessiond_opt.
Solution
========
To alleviate this, we propose a two stage lookup. The first stage uses
"pgrep -f" yielding potential candidates. The second stage performs
grep on the basename of the first field of the /proc/[pid]/cmdline
for each pid candidates.
The first field of /proc/[pid]/cmdline corresponds to the actual command.
We use the basename to ensure that we do not match on the path to the
executable.
Known drawbacks
=========
None
References
==========
https://review.lttng.org/c/lttng-tools/+/3043
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I479ebad27f4965ae16d4442a6fe58ff3157d76fa
Jérémie Galarneau [Tue, 11 Feb 2020 23:22:13 +0000 (18:22 -0500)]
logging: print human-readable thread names when logging
The lttng_thread interface used by the session daemon uniquely names
all threads. This name can be used to augment the thread's logging
statement with a human-readable name rather than using the pid/tid
tuple used elsewhere.
Additionally, the thread name is set using the pthread API so that it
is visible in GDB and other tools (e.g. htop).
Invocations of pgrep in the test utilities are replaced by 'pgrep -f',
which matches against the process name.
We are not the first to encounter this problem after renaming the main
thread, see
https://github.com/mongodb/mongo/commit/
726cafd713c7333640f8458ec9808ed4f678e3a7#diff-a9003101d1e4a99ac2d43d9b1b839587R122
pgrep uses the name name in /proc/$PID/status which contains the
thread name, not the executable name. In the case of the sessiond,
this is now "Main".
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I73dfe8683b2ea31f7ed0c2ffdfa8332f36e28f9b
Jérémie Galarneau [Tue, 11 Aug 2020 19:58:08 +0000 (15:58 -0400)]
sessiond: clarify the role of notification credentials
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ica7a7370fcc4d34c9af11bcb9e7435e19b29a6f8
Jonathan Rajotte [Tue, 24 Mar 2020 18:08:16 +0000 (14:08 -0400)]
Use lttng_trigger credentials to send evaluation to client
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I49b4b2aeda09d09b7d8630562660dac96f36b3e7
Jonathan Rajotte [Fri, 7 Feb 2020 22:40:46 +0000 (17:40 -0500)]
trigger: introduce refcounting
Will be used for listing and much more use.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic9ee8753257fcbe15fc1321b18582628328c0154
Jonathan Rajotte [Fri, 7 Aug 2020 20:22:36 +0000 (16:22 -0400)]
trigger: use condition and action ref counting to ease internal objects management
Currently a trigger object has multiple relationship types with the action and
condition associated to it.
On the API client side, the trigger does not own the action and condition
associated to it. The user is responsible for managing the lifetime of the
associated action and condition object.
On the sessiond side, triggers created from `lttng_trigger_create_from_payload`
currently require that before calling lttng_trigger_destroy, the action and
condition be fetched and deleted using their respective destructor. This
operation cannot be done inside `lttng_trigger_destroy` since the exposed API is
clear that the trigger does no own the objects.
We can facilitate the lifetime/ownership management of the action and condition
objects using their respective reference counting mechanism.
On a trigger creation, the trigger get references on both object. On destroy,
the trigger put the references on the objects.
From an API client perspective, nothing changes. Even better, it prevents
premature freeing of these objects, since the trigger have references to these
objects.
On the sessiond side, we can now move the actual ownership of the action and
condition object to the trigger object. This is done in
`lttng_trigger_create_from_payload` forcing a put of the local references to the
object, effectively moving the ownership to the trigger object.
Note that the `lttng_trigger_get_{action, condition}` do not `get` a reference
to the object before returning it. This is done to comply with the API that was
introduced back in 2.11 which does expect a client to call lttng_{action,
condition}_destroy on the returned object.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I31d196d00c886fcf42c1b16c55e693949373568b
Jonathan Rajotte [Fri, 7 Aug 2020 19:39:24 +0000 (15:39 -0400)]
condition: introduce reference counting
This will allows easier management of the trigger ownership of its associated
condition and action objects.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I46420b03bd4bf7948bc2d1f44985edfe86c27c61
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: tests: fd-tracker: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I24fde39587495f5f48f900898755e57c10fbf303
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: relayd index: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6a79fcdceebc0959bb66525daac288953e863bb3
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond comm relay: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2c42c7b7574ccec85524a1b0aed33d108725ad4a
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: compat time: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: If28840c388baffd257e61ef1ecad7c1ea3e3fd68
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: kernel consumer: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I55873fe5500e00d725ebaa241f4054a793b15aee
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond ust-app: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4ba505a8206f4ee8838fc3268e3904c496eb3802
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond notification thread: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ieabe7e8a2066b30ef8b413896d03f8ff824ddc99
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond kernel: change spaces to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3b8f476db2f9038fddf3b648dba0fa06f20db5d8
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond kernel: fix include style
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8959283af0cafa94b0939e0025b08b40ff86653e
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond consumer: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I367ffe89e3930b2e84ad8c70b95bd95a968d974c
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idb749331397c2b43ea41a33f09bf35d15c4b8f8a
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond manage-consumer: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic897e04aa96bd58c6e8dd340f8537af86c2dd478
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: relayd: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I4aa7d69ef823e4ea85820cadd23463869f98edf2
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond command: fix include style
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I2f3d695a97d31ca5b098872abc566a8b261cdf9c
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond command: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ib809c788f3bf32360cd68b16972636db3ef4dfd7
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: relayd trace-chunk: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I789edf8a6b9f893c6e3b5f72fb9001f8e68eebd7
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: relayd index: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9fe8b9fa536a89fd04e989cf8b3f5957de88df1e
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: relayd session: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I545728dc95f67113c0f7adf0cbc9fc742e4018c7
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: liblttng-ctl clear: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I423545aa9de0521b861926b29e21b187500e5fd0
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: liblttng-ctl destruction handle: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ie5d8f2fc1e2ad358b287550065f24f8928c8304c
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: mi-lttng: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ia6ebea7f227d05cc9ebb8cf84717cfaf59f3e222
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: time: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I922111fab4667a5af35ed45d9bbdee0bcb477b05
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: consumer-stream: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Idfe9f1202762bd9a026659d61e50112c81641c88
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: consumer metadata cache: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I9abf075960a33bac01379fdc6fc79dca83f84ee2
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: run-as: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Icbf393b779cf66a93b49ecef4279a6db820b4968
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: utils: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I67341c2a849c340fc56cc03dbca9daaf7c9eddb1
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond rotation: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I8a2192da7b69a5a1945e4605167e7d0c91365681
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: sessiond: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I09854bc28773944ad2b70cb7645dcd53e0ee35ab
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: trace-chunk: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0099bf493c8b80aadf7ac8c7b5668c0bb99d27dd
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: consumer: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I688ab6903064cee0ef43b4ee7d395e6a60eb729b
Jérémie Galarneau [Tue, 11 Aug 2020 16:22:19 +0000 (12:22 -0400)]
Clean-up: optional: change space to tabs
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6e08f4cc6198583f3dfd9dd3b02da88244e2aa9d
Jonathan Rajotte [Tue, 24 Mar 2020 16:15:22 +0000 (12:15 -0400)]
trigger: internal: add credentials information
These credentials are for internal use only. We plan on using them for
listing, run-as filter bytecode generation and more.
Introduce LTTNG_OPTIONAL_GET_PTR which is used to assert and return the
pointer for an LTTNG_OPTIONAL underlying element.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I70d7f79918693b52db9b384b56e6476f3449e71d
Jonathan Rajotte [Tue, 24 Mar 2020 16:11:41 +0000 (12:11 -0400)]
credentials: implement is_equal
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I526fd15b8365fd2a77b5aa5250a75d956fd8efc3
Jérémie Galarneau [Fri, 7 Feb 2020 23:25:36 +0000 (18:25 -0500)]
sessiond: clean-up: remove unused `state` function parameter
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id606bc0c0db713902cbab2fafed5c81a05ef5f08
Jérémie Galarneau [Mon, 10 Aug 2020 21:41:28 +0000 (17:41 -0400)]
.gitignore: add test_unix_socket
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ic7df31b45ce2ad9806790c1041506ce96e1a9e8d
Jonathan Rajotte [Tue, 14 Jul 2020 17:00:06 +0000 (13:00 -0400)]
sessiond: comm: add fd count information for lttng-ctl -> sessiond comm
This allows bulk reception of fds incoming from lttng-ctl before doing
any object deserialization.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I16f054be1d8b8d1e7868784c9b4088750b9a3651
Jérémie Galarneau [Fri, 7 Aug 2020 02:05:25 +0000 (22:05 -0400)]
Tests: add unix socket wrapper unit tests
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0db99d77b47326a5e97c39679a053ba4be7f717c
Jérémie Galarneau [Thu, 6 Aug 2020 19:19:54 +0000 (15:19 -0400)]
payload-view: mark lttng_payload_view_get_fd_handle_count as const
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I0f0b1620e04fa5a5ae06b096dc940504a7d7553e
Jérémie Galarneau [Wed, 5 Aug 2020 16:09:36 +0000 (12:09 -0400)]
Fix: uprobe: inequality comparison against NULL
The expression will not have its intended effect.
In lttng_userspace_probe_location_function_serialize: Comparing a
pointer against NULL using an operator such as < or >=.
`binary_fd` is now a fd_handle instance rather than a "raw" fd. All
instances of `binary_fd` are renamed to `binary_fd_handle` to prevent
such errors in the future.
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I57f1a3d5a01029084889a183881fac6f1fe9b6d9
Jonathan Rajotte [Tue, 7 Jul 2020 23:50:06 +0000 (19:50 -0400)]
Fix: action: lttng_action_group_add_action returns status invalid
IS_GROUP_ACTION macro was performed on the action to be added instead
of the action representing the group, `group` in this function.
This always resulted in error on call to lttng_action_group_add_action.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Iecf133fb2165686d1e8abd2225a8f90df22c41dc
Jérémie Galarneau [Tue, 23 Jun 2020 22:06:53 +0000 (18:06 -0400)]
Fix: evaluation: dereference before NULL check in create_from_payload
An evaluation payload view is created from the view passed to
lttng_evaluation_create_from_payload. Since a view contains a const
copy of the _fds array, it must be initialized as the declaration site.
However, src_view is checked for NULL after the initalization. Coverity
rightfully warns that:
1429799 Dereference before null check
There may be a null pointer dereference, or else the comparison against
null is unnecessary.
In lttng_evaluation_create_from_payload: All paths that lead to this
null pointer comparison already dereference the pointer
earlier (CWE-476)
This is not reachable right now, but it is fixed to silence the warning
and prevent future mistakes.
Reported-by: Coverity Scan
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3ceace4117ff54265b1f0cf6a4c638aec95f2879
Jérémie Galarneau [Thu, 30 Jul 2020 03:13:16 +0000 (23:13 -0400)]
Tests: rework tracefile_count test to meet the tracefile count limit
The original test simply fires a test application and checks that file
count limits are not exceeded.
The test is reworked to produce events until the total stream size
stabilizes. Then the number of files is checked against the expected
limit.
Some shellcheck warnings are fixed as part of this change.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id9ca3e16c05898c4009b4fb8fcf7882fdcb9dcf1
Jérémie Galarneau [Fri, 31 Jul 2020 15:20:40 +0000 (11:20 -0400)]
Fix: extraneous empty/inactive flush on rotation out of a trace chunk
Observed issue
==============
A test (tests/regression/tools/tracefile-limits/test_tracefile_count)
occasionally fails on ppc64. The trace validation steps in the fails in
the case where the trace file count limit is set to 1.
Examining the resulting trace shows that the last packet of data
produced by the test application appears to be missing.
The test case enables a channel in "overwrite" mode. Normally, this
would guarantee that the last data produced will always be available in
the resulting trace.
Cause
=====
An empty/inactive flush is performed when rotating "out" of a non-null
trace chunk to ensure that the trace chunk contained at least one
packet.
Looking at the test's resulting trace and by following the consumerd
logs, we see that the test application runs on one CPU for most of its
lifetime. The stream file is repeatedly replaced to make room for the
latest data.
Eventually, the application is migrated to another CPU. A number of
packets are written to this new stream. The session is then stopped
which causes an active flush to occur to close the current packet of
all streams (see `ust_app_stop_trace_all`).
Then, when the session is destroyed, an empty/inactive flush is
performed to ensure that at least one packet was produced in the current
trace chunk [1].
At the moment of writing this empty packet, the consumer daemon sees
that there is not enough space left in the stream file to honour the
trace file size restriction. It thus overwrites the file resulting in
the loss of the last events to replace them with the empty "end of
chunk" packet that occupies a single page.
While the problem is not specific to PowerPC 64, it has a lot more
chances to occur there as pages are typically configured to be of 64kb
length. Due to current implementation limitations, empty packets have
a size of one page.
In other words, 4kb pages typically fit in the space left in the file,
causing the problem to not be easily reproducible on x64.
Note that while the file size limit is specified as "3 * PAGE_SIZE" in
the test, it is rounded-up to 512kb to accomodate at least one
sub-buffer.
Solution
========
[1] This empty/inactive flush is no longer necessary since
f96af312b
as an "open packet" (which performs an empty/inactive flush) is
performed when a stream enters a non-null trace chunk. There is no
concern that a trace chunk will be left empty unless this initial flush
fails (see patch comments and work-around).
The empty flush that was performed for data streams is converted
into an active flush under most circumstances; the packet is simply
closed.
Known drawbacks
===============
None.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I5602b7ab8318374f75060489cf9c27af4e058805
Jérémie Galarneau [Thu, 30 Jul 2020 16:14:21 +0000 (12:14 -0400)]
Fix: relayd: double unlock on viewer stream creation error
viewer_stream_create must be called with the relay stream's
lock held since
9edaf114d. A call to pthread_mutex_unlock
was forgotten in the error path of viewer_stream_create resulting
in a double-unlock in some error scenarios.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I613f67440669af52f625b7bdc4e9ae6ce75956ea
Jérémie Galarneau [Tue, 28 Jul 2020 15:29:18 +0000 (11:29 -0400)]
Fix: relayd: live connection fails to open file during clear
Issue observed
==============
A `session clear` test occasionaly fails on the CI (very rarely, and
more often on PowerPC executors for some reason) with babeltrace
reporting that the live connection was closed by the remote side:
PASS: tools/clear/test_ust 276 - Waiting for live viewers on url: net://localhost
07-21 16:17:07.058 23855 23855 E PLUGIN/SRC.CTF.LTTNG-LIVE/VIEWER lttng_live_recv@viewer-connection.c:198 [lttng-live] Remote side has closed connection
07-21 16:17:07.058 23855 23855 E PLUGIN/SRC.CTF.LTTNG-LIVE/VIEWER lttng_live_session_get_new_streams@viewer-connection.c:1701 [lttng-live] Error receiving get new streams reply
07-21 16:17:07.058 23855 23855 E PLUGIN/SRC.CTF.LTTNG-LIVE lttng_live_msg_iter_next@lttng-live.c:1665 [lttng-live] Error preparing the next batch of messages: live-iter-status=LTTNG_LIVE_ITERATOR_STATUS_ERROR
07-21 16:17:07.058 23855 23855 W LIB/MSG-ITER bt_message_iterator_next@iterator.c:864 Component input port message iterator's "next" method failed: iter-addr=0x1014d6a8, iter-upstream-comp-name="lttng-live", iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=SOURCE, iter-upstream-comp-class-name="lttng-live", iter-upstream-comp-class-partial-descr="Connect to an LTTng relay daemon", iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
07-21 16:17:07.059 23855 23855 E PLUGIN/FLT.UTILS.MUXER muxer_upstream_msg_iter_next@muxer.c:454 [muxer] Upstream iterator's next method returned an error: status=ERROR
07-21 16:17:07.059 23855 23855 E PLUGIN/FLT.UTILS.MUXER validate_muxer_upstream_msg_iters@muxer.c:991 [muxer] Cannot validate muxer's upstream message iterator wrapper: muxer-msg-iter-addr=0x1014d668, muxer-upstream-msg-iter-wrap-addr=0x1014e210
07-21 16:17:07.059 23855 23855 E PLUGIN/FLT.UTILS.MUXER muxer_msg_iter_next@muxer.c:1415 [muxer] Cannot get next message: comp-addr=0x1014cca8, muxer-comp-addr=0x1015afd8, muxer-msg-iter-addr=0x1014d668, msg-iter-addr=0x1014d598, status=ERROR
07-21 16:17:07.059 23855 23855 W LIB/MSG-ITER bt_message_iterator_next@iterator.c:864 Component input port message iterator's "next" method failed: iter-addr=0x1014d598, iter-upstream-comp-name="muxer", iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=FILTER, iter-upstream-comp-class-name="muxer", iter-upstream-comp-class-partial-descr="Sort messages from multiple inpu", iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
07-21 16:17:07.059 23855 23855 W LIB/GRAPH consume_graph_sink@graph.c:473 Component's "consume" method failed: status=ERROR, comp-addr=0x1014d128, comp-name="pretty", comp-log-level=WARNING, comp-class-type=SINK, comp-class-name="pretty", comp-class-partial-descr="Pretty-print messages (`text` fo", comp-class-is-frozen=1, comp-class-so-handle-addr=0x10159dd8, comp-class-so-handle-path="/home/jenkins/workspace/lttng-tools_stable-2.12_portbuild/arch/powerpc/babeltrace_version/stable-2.0/build/std/conf/agents/liburcu_version/stable-0.12/test_type/base/deps/build/lib/babeltrace2/plugins/babeltrace-plugin-text.so", comp-input-port-count=1, comp-output-port-count=0
07-21 16:17:07.059 23855 23855 E CLI cmd_run@babeltrace2.c:2548 Graph failed to complete successfully
ERROR: [Babeltrace CLI] (babeltrace2.c:2548)
Graph failed to complete successfully
CAUSED BY [libbabeltrace2] (graph.c:473)
Component's "consume" method failed: status=ERROR, comp-addr=0x1014d128,
comp-name="pretty", comp-log-level=WARNING, comp-class-type=SINK,
comp-class-name="pretty", comp-class-partial-descr="Pretty-print messages
(`text` fo", comp-class-is-frozen=1, comp-class-so-handle-addr=0x10159dd8,
comp-class-so-handle-path="/home/jenkins/workspace/lttng-tools_stable-2.12_portbuild/arch/powerpc/babeltrace_version/stable-2.0/build/std/conf/agents/liburcu_version/stable-0.12/test_type/base/deps/build/lib/babeltrace2/plugins/babeltrace-plugin-text.so",
comp-input-port-count=1, comp-output-port-count=0
CAUSED BY [libbabeltrace2] (iterator.c:864)
Component input port message iterator's "next" method failed:
iter-addr=0x1014d598, iter-upstream-comp-name="muxer",
iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=FILTER,
iter-upstream-comp-class-name="muxer",
iter-upstream-comp-class-partial-descr="Sort messages from multiple inpu",
iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
CAUSED BY [muxer: 'filter.utils.muxer'] (muxer.c:991)
Cannot validate muxer's upstream message iterator wrapper:
muxer-msg-iter-addr=0x1014d668, muxer-upstream-msg-iter-wrap-addr=0x1014e210
CAUSED BY [muxer: 'filter.utils.muxer'] (muxer.c:454)
Upstream iterator's next method returned an error: status=ERROR
CAUSED BY [libbabeltrace2] (iterator.c:864)
Component input port message iterator's "next" method failed:
iter-addr=0x1014d6a8, iter-upstream-comp-name="lttng-live",
iter-upstream-comp-log-level=WARNING, iter-upstream-comp-class-type=SOURCE,
iter-upstream-comp-class-name="lttng-live",
iter-upstream-comp-class-partial-descr="Connect to an LTTng relay daemon",
iter-upstream-port-type=OUTPUT, iter-upstream-port-name="out", status=ERROR
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (lttng-live.c:1665)
Error preparing the next batch of messages:
live-iter-status=LTTNG_LIVE_ITERATOR_STATUS_ERROR
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (viewer-connection.c:1701)
Error receiving get new streams reply
CAUSED BY [lttng-live: 'source.ctf.lttng-live'] (viewer-connection.c:198)
Remote side has closed connection
ok 277 - Clear session J7WXjh7fmMleTCfE
PASS: tools/clear/test_ust 277 - Clear session J7WXjh7fmMleTCfE
ok 278 - Clear session J7WXjh7fmMleTCfE
PASS: tools/clear/test_ust 278 - Clear session J7WXjh7fmMleTCfE
ok 279 - Stop lttng tracing for session J7WXjh7fmMleTCfE
PASS: tools/clear/test_ust 279 - Stop lttng tracing for session J7WXjh7fmMleTCfE
ok 280 - Destroy session J7WXjh7fmMleTCfE
PASS: tools/clear/test_ust 280 - Destroy session J7WXjh7fmMleTCfE
# Wait for viewer to exit
not ok 281 - Babeltrace succeeds
Cause
=====
Looking at the relay daemon logs, it appears that the live client
requests an enumeration of the available streams while a rotation is
ongoing (clear).
Ultimately, this results in the relay daemon attempting to open a
non-existing file:
PERROR - 16:33:59.
242388809 [734380/734387]: Failed to open fs handle to ust/uid/1000/64-bit/chan_0, open() returned: No such file or directory (in fd_tracker_open_fs_handle() at fd-tracker.c:550)
The logs indicate that this file existed at some point. However, it
was unlinked and its newest instance was created in a trace chunk
named "20200727T163359-0400-1".
This chunk name is a temporary name used until the original trace
chunk can be unlinked (cleared) and the newest can be moved in its
place.
The file is being opened as part of the creation of a viewer stream
when make_viewer_stream() fails to find it. This implies that, somehow,
an outdated trace chunk is being used to open the viewer stream's file.
The reason why is that make_viewer_stream is called with the
viewer session's current trace chunk. During a rotation, the use of the
viewer session's current trace chunk is touchy due to the way the
switch-over to a new chunk is handled.
How viewer session/stream trace chunks are rotated
--------------------------------------------------
The viewer polls the relay daemon for new data/metadata to consume using
the `GET_NEXT_INDEX` and `GET_METADATA` commands. Both commands will
indicate to the viewer that it should try again later if a rotation is
ongoing on the "side" of the relay session.
When a rotation is not ongoing, the relay compares the `id` of the
target viewer stream's trace chunk with the relay session's current
trace chunk. If those `id`s don't match, the viewer session's current
trace chunk is then updated to a copy of the relay session's trace
chunk. The viewer stream's files are then closed and re-opened in the
context of the viewer session's now-updated trace chunk.
While the live protocol allows `GET_NEXT_INDEX` and `GET_METADATA` to
communicate to the viewer that it should "retry" later, there is no such
provisions made for the paths that lead to the creation of the viewer
streams. This means viewer streams must be created even if a rotation is
ongoing.
Solution
========
If a rotation is ongoing at the moment of the creation of the viewer
streams, we wish to use a copy of the trace chunk being used by the
relay stream. This way, we can be sure that the streams' files exist.
It is okay for viewer streams to hold references to different copies of
the trace chunks since no user-visible actions are performed when the
reference to those chunks is released. This is different from relay
streams where the detection of the completion of a rotation is done when
the last relay stream releases its reference to a specific trace chunk
instance.
Known drawbacks
===============
None beyond a slight increase (temporary until the next rotation) in the
number of FDs used when a client connects during a session rotation.
Note
====
Since make_viewer_streams now acts on relay and viewer counterparts of
the session and stream objects, the various variables are prefixed with
`relay` and `viewer` to make the code easier to understand.
The locking period of the relay stream is extended to most of the
iteration in make_viewer_stream() rather than only in
viewer_stream_create(). As a result, callers of viewer_stream_create()
must hold the relay stream's lock.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I3b9c52a57835134712eadba0f68f6a21e922356b
Jonathan Rajotte [Tue, 28 Jul 2020 14:26:02 +0000 (10:26 -0400)]
Fix: sessiond: unchecked return value
From Coverity:
CID
1431048 (#1 of 1): Unchecked return value (CHECKED_RETURN)
1. check_return: Calling lttng_dynamic_buffer_set_size without
checking return value (as is done elsewhere 32 out of 40 times).
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Id15632e3932052cf7dce5d57a08ac6efc84fc92f
Jonathan Rajotte [Tue, 28 Jul 2020 14:22:28 +0000 (10:22 -0400)]
Fix: common: unchecked return value
From Coverity:
CID
1431050 (#1 of 1): Unchecked return value (CHECKED_RETURN)
1. check_return: Calling lttng_dynamic_buffer_set_size without
checking return value (as is done elsewhere 32 out of 40 times).
Solution
========
Since we are inside a the clear operation for the payload, ignore the
return value.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I45c1054de7252aab4b77f330102ff4f489f20a6d
Jonathan Rajotte [Tue, 28 Jul 2020 14:19:30 +0000 (10:19 -0400)]
Fix: common: improper use of negative return
From Coverity:
CID
1431053 (#1 of 2): Improper use of negative value (NEGATIVE_RETURNS)
5. negative_returns: fd_count is passed to a parameter that cannot be
negative.
Solution
========
Check return value for fd_count and goto error if negative.
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ibdb2f065f0ebe51efae9125630a17730386395ac
Jonathan Rajotte [Tue, 28 Jul 2020 14:15:04 +0000 (10:15 -0400)]
Fix: sessiond: unchecked return value
From coverity:
CID
1431054 (#1 of 1): Unchecked return value (CHECKED_RETURN)
1. check_return: Calling lttng_dynamic_buffer_set_size without
checking return value (as is done elsewhere 32 out of 40 times).
Signed-off-by: Jonathan Rajotte <jonathan.rajotte-julien@efficios.com>
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I6b5b18d90f194c8081b91b3fbb80dccd29ba19ca
Jérémie Galarneau [Mon, 27 Jul 2020 22:14:30 +0000 (18:14 -0400)]
Fix: relayd: wrong stream type used in DBG statement
get_next_index() is never invoked for metadata streams; the
debug statement was erroneously copy-pasted from the get_metadata()
function.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: Ia57dac7f102eddc78e24781905b390b4fa54bd80
Jérémie Galarneau [Fri, 24 Jul 2020 17:48:00 +0000 (13:48 -0400)]
sessiond: client: use common payload send fds util
Use lttcomm_send_payload_view_fds_unix_sock instead of an open-coded
version of the same functionality.
Signed-off-by: Jérémie Galarneau <jeremie.galarneau@efficios.com>
Change-Id: I36627423287a059e8b2ab1cc5cb42ca0db53e144
This page took 0.058323 seconds and 4 git commands to generate.