lttng-tools.git
11 years agoImplement filter expression to bytecode compiler in liblttng-ctl
Mathieu Desnoyers [Tue, 10 Jul 2012 19:15:04 +0000 (15:15 -0400)] 
Implement filter expression to bytecode compiler in liblttng-ctl

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoFix: The session list count should provide unique identifiers
Mathieu Desnoyers [Wed, 11 Jul 2012 22:23:12 +0000 (18:23 -0400)] 
Fix: The session list count should provide unique identifiers

Never decrement it.

Reported-by: Amer Alhalabi <amer.alhalabi@ericsson.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoNetwork streaming support
David Goulet [Wed, 11 Jul 2012 17:20:39 +0000 (13:20 -0400)] 
Network streaming support

This is a huge commit integrating a lot of new feature for network
streaming support with the lttng-relayd previously merged.

Please note that this commit is the initial import of the network
streaming feature meaning that this is NOT stable and SHOULD be
considered experimental hence not suited for production deployment.

API Changes:

lttng_create_session_uri(), lttng_enable_consumer(),
lttng_set_consumer_uri() and lttng_disable_consumer() calls are added.
The lttng command line now supports these new calls by introducing two
new commands and one modified. (Please look at the --help option for
more details).

$ lttng enable-consumer
$ lttng disable-consumer

The "lttng create" command can now take new arguments to specify the
destination URI for network streaming. There is also a new flag which
allows the user to disable the consumer for the session.

At this point, the now old deprecated lttng_create_session() call is
still supported but SHOULD be replaced by lttng_create_session_uri().
More calls will be added to the API in order to facilitate the creation
of lttng URI.

Network streaming:

Both kernel and user space tracing are supported for network streaming.
For now, only IPv4 and IPv6 protocol over TCP is supported. Two network
socket are used where one is for control commands and the second one is
for data transmission. The port are respectively 5342 and 5343.

BUGS: Streaming high throughput traces (e.g lttng enable-event -a -k)
with low bandwitdh available to the relayd, the tracer does not behave
well when the metadata are not emptied enough quickly so data are
missing on the remote target and the tracing is simply not stoppable.
This is a known bug and future iterations will try to fix it.

WARNING: At this stage, there is _NO_ security features meaning no
remote authentication on the relayd, no transport layer encryption or
network secure handshake of some sort. So, uses only for in-vitro
experiment or on _trusted_ networks.

Signed-off-by: Julien Desfossez <julien.desfossez@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoInitial import of the new binary lttng-relayd
Julien Desfossez [Tue, 10 Jul 2012 21:40:27 +0000 (17:40 -0400)] 
Initial import of the new binary lttng-relayd

The lttng-relayd listens on the network and receives traces streamed by
the consumer.

At this first commit, the relayd supports IPv4 and IPv6 over TCP.

The following commits will add the support to use it with the lttng
command line interface, control library, consumer and session daemon.

Signed-off-by: Julien Desfossez <julien.desfossez@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoAdd and change lttcomm socket API
David Goulet [Tue, 10 Jul 2012 21:30:37 +0000 (17:30 -0400)] 
Add and change lttcomm socket API

Add lttcomm_alloc_sock_from_uri(), lttcomm_destroy_sock(),
lttcomm_alloc_copy_sock() and lttcomm_copy_sock(). Also add the relayd
command and lttcomm error message.

Change the create and alloc sock call API arguments.

Fix the fd leak in accept() function.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: mmap write() for large subbuffers and handle EINTR (v2)
Mathieu Desnoyers [Fri, 6 Jul 2012 14:13:05 +0000 (10:13 -0400)] 
Fix: mmap write() for large subbuffers and handle EINTR (v2)

With large subbuffer (packet) size, if write() returns before copying the
entire packet for mmap buffers, the consumerd restarts the write
infinitely, which is not good at all.

This affects both lttng-ust (in default mmap mode) and lttng-kernel (but
only for mmap buffers, which is not the default).

This issue would show up with large subbuffer size.

We need to handle this case, as well as EINTR errors (which need to restart
write).

Also fixing the return value of mmap read functions, which were returning
the amount of data written by the last invocation of write() rather than
the total number of bytes written. splice use had the same issue.

Also now consider a write() that returns more bytes than requested as an
error.

Moreover, assigning error = ret after failed splice and write was a
mistake: error is holding the actual error value. ret just holds -1.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix inet return value in the close() fct
David Goulet [Tue, 3 Jul 2012 20:28:05 +0000 (16:28 -0400)] 
Fix inet return value in the close() fct

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: add missing consumer.h to sessiond Makefile.am
Mathieu Desnoyers [Fri, 29 Jun 2012 10:49:21 +0000 (12:49 +0200)] 
Fix: add missing consumer.h to sessiond Makefile.am

Fix make dist.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoFix: test make dist Makefile.am missing files
Mathieu Desnoyers [Fri, 29 Jun 2012 10:43:14 +0000 (12:43 +0200)] 
Fix: test make dist Makefile.am missing files

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoFix: support large files on 32-bit systems
Mathieu Desnoyers [Fri, 29 Jun 2012 10:40:30 +0000 (12:40 +0200)] 
Fix: support large files on 32-bit systems

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoFix: enable event loglevel match function
David Goulet [Fri, 22 Jun 2012 15:02:51 +0000 (11:02 -0400)] 
Fix: enable event loglevel match function

For the loglevel type ALL, the value set in an event inside the session
daemon is -1 where the value received from the client is 0. For this
loglevel type, the loglevel does not matter.

This fixes the commit c0657016bbb98e1769edbaba8c00221f964c7402 done
before version 2.0.2.

(Closes #283 #287)

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoRemove useless arg from kernel consumer send stream
David Goulet [Tue, 19 Jun 2012 19:13:16 +0000 (15:13 -0400)] 
Remove useless arg from kernel consumer send stream

The consumer_data structure was not used when sending kernel channel
streams to the consumer. Now the function uses the consumer_data
structure to get the consumer socket and the sock argument was removed.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: unchecked pointer from getenv() for lttng create
David Goulet [Tue, 19 Jun 2012 15:15:38 +0000 (11:15 -0400)] 
Fix: unchecked pointer from getenv() for lttng create

The lttng create command was not handling the return pointer from
getenv("HOME") and was passing it to a strdup() call.

(close #271)

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: problem in port assignment in uri_parse()
David Goulet [Tue, 19 Jun 2012 14:49:01 +0000 (10:49 -0400)] 
Fix: problem in port assignment in uri_parse()

The functions now puts 0 in the port field of the lttng_uri structure if
NO port are specified. It's now the caller's job to assign a default
port for that case.

Since the uri_parse() function does not know the URI context, a default
port can not be assigned correctly.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoAdd pipe creation to utils facility
David Goulet [Fri, 15 Jun 2012 15:00:11 +0000 (11:00 -0400)] 
Add pipe creation to utils facility

Three new functions are added to create a pipe, create a pipe with
CLOEXEC flag and close a pipe.

Those functions are now used in the session daemon for all pipes and
cleanup.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoAdd network transport layer to sessiond-comm
David Goulet [Thu, 14 Jun 2012 18:48:49 +0000 (14:48 -0400)] 
Add network transport layer to sessiond-comm

Only import the library files in sessiond-comm/ directory for network
communication supporting both TCP and UDP protocols with IPv4 and IPv6.

First layer needed for network streaming and communication with the
upcoming lttng-relayd daemon.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: enable event with different loglevel error
Mathieu Desnoyers [Thu, 14 Jun 2012 14:43:04 +0000 (10:43 -0400)] 
Fix: enable event with different loglevel error

This following example was NOT returning a correct error message.
"event1" is set with a loglevel "TRACE_CRIT". The normal behavior here
is that once enabled, you can not change the loglevel of the enable
event on the tracer side with a second command. It now returns a new
error message like so:

$ lttng enable-event event1 --loglevel-only TRACE-CRIT -u
$ lttng enable-event event1 --loglevel-only TRACE-WARNING -u
[...] Event already enabled with different loglevel

This commit makes the session daemon verify if _both_ the name and
loglevel are the same when enabling an event or else an error is
returned.

Also, the session daemon will continue enabling events and not return an
error is the loglevel does not match event for ust app on the tracer
which returns an EPERM at that stage. This is to address the case where
two applications have the same event name but with different loglevel.

Reported-by: Tan Dung Le Tran <tan.dung.le.tran@ericsson.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoAdd lttng URI support and default network ports
David Goulet [Wed, 13 Jun 2012 21:08:05 +0000 (17:08 -0400)] 
Add lttng URI support and default network ports

This commit simply adds the lttng_uri API/ABI to the upstream branch.
Three functions are provided by common/uri.h and are usable by linking
with libcommon.

The default network ports are also added for network streaming for both
control and data stream.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: Change the type of enabled in lttng_event to a signed int
Danny Serres [Fri, 8 Jun 2012 14:19:08 +0000 (10:19 -0400)] 
Fix: Change the type of enabled in lttng_event to a signed int

Signed-off-by: Danny Serres <danny.serres@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoCreate kernel-consumer.c/.h and consumer.h
David Goulet [Thu, 7 Jun 2012 17:16:09 +0000 (13:16 -0400)] 
Create kernel-consumer.c/.h and consumer.h

Contains function call to send data to the kernel consumer (for now
session and channel streams).

Nothing new, code has just been moved to simplify main.c and partition
portion of code for clearer and easier understanding.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: move memset in channel_set_attr after NULL check
Danny Serres [Wed, 6 Jun 2012 18:54:30 +0000 (14:54 -0400)] 
Fix: move memset in channel_set_attr after NULL check

Signed-off-by: Danny Serres <danny.serres@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoCreate unix.c/.h for UNIX socket communication
David Goulet [Thu, 7 Jun 2012 16:57:51 +0000 (12:57 -0400)] 
Create unix.c/.h for UNIX socket communication

Simply split some content of sessiond-comm.c/.h to those two files for a
clearer code.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoFix: close all file descriptors when executed as daemon
Mathieu Desnoyers [Mon, 4 Jun 2012 22:08:24 +0000 (18:08 -0400)] 
Fix: close all file descriptors when executed as daemon

Both sessiond and consumerd support the option "-d" to run as daemon. In
some specific cases, e.g. when launched from dpkg installation scripts,
file descriptors 3, 4, 5 are left open and don't seem to have O_CLOEXEC
flag set, so the install script hangs because the sessiond still holds a
reference to them. daemon(3) only closes standard FD 0, 1, 2.

Fix this issue by closing all file descriptors after calling daemon(3).
Note: we make sure no file descriptor is opened before calling daemon(3)
by moving the init_thread_quit_pipe() call after the FD close.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoUpdate lttng-ust header copy for disabled UST compilation
Mathieu Desnoyers [Fri, 1 Jun 2012 21:05:01 +0000 (17:05 -0400)] 
Update lttng-ust header copy for disabled UST compilation

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoFix: ust_app_list_events should return -ENOSYS when no UST
Mathieu Desnoyers [Fri, 1 Jun 2012 21:01:13 +0000 (17:01 -0400)] 
Fix: ust_app_list_events should return -ENOSYS when no UST

Caller thinks that memory is allocated, and incorrectly dereferences a
NULL pointer if we return 0 rather than -ENOSYS.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agocleanup output indentation for field listing
Mathieu Desnoyers [Thu, 31 May 2012 20:14:45 +0000 (16:14 -0400)] 
cleanup output indentation for field listing

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoImplement event fields listing
Mathieu Desnoyers [Wed, 30 May 2012 20:27:39 +0000 (16:27 -0400)] 
Implement event fields listing

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoUpdate lttng-ust-abi.h for event field listing
Mathieu Desnoyers [Fri, 25 May 2012 18:20:35 +0000 (14:20 -0400)] 
Update lttng-ust-abi.h for event field listing

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoFix comments formating in conf.c
Francis Giraldeau [Wed, 30 May 2012 23:50:07 +0000 (01:50 +0200)] 
Fix comments formating in conf.c

Comments should be indented with one space only.

Signed-off-by: Francis Giraldeau <francis.giraldeau@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoUpdate documentation for destroy all option
Francis Giraldeau [Wed, 30 May 2012 23:50:06 +0000 (01:50 +0200)] 
Update documentation for destroy all option

Document the new option -a, --all of lttng destroy

Signed-off-by: Francis Giraldeau <francis.giraldeau@gmail.com>
11 years agoAdd destroy all session option
Francis Giraldeau [Wed, 30 May 2012 23:50:05 +0000 (01:50 +0200)] 
Add destroy all session option

Destroy was limited to one session at a time. The option -a, --all destroys all
registered sessions at once.

* fixed braces

Signed-off-by: Francis Giraldeau <francis.giraldeau@gmail.com>
11 years agoImprove delete of configuration
Francis Giraldeau [Wed, 30 May 2012 23:50:04 +0000 (01:50 +0200)] 
Improve delete of configuration

Adding functions to encapsulate the configuration delete. Test if path to
configuration exists before attempting to remove it.

* fixed braces and goto name

Signed-off-by: Francis Giraldeau <francis.giraldeau@gmail.com>
11 years agoFix: clang llvm warnings
David Goulet [Thu, 31 May 2012 16:21:27 +0000 (12:21 -0400)] 
Fix: clang llvm warnings

Mostly fix unsigned expression and implicit conversion.

URCU still shows warnings but irrelevant.

Signed-off-by: David Goulet <dgoulet@efficios.com>
11 years agoAdd CodingStyle to tarball
Mathieu Desnoyers [Wed, 30 May 2012 13:53:42 +0000 (09:53 -0400)] 
Add CodingStyle to tarball

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoAdd coding style document
Mathieu Desnoyers [Wed, 30 May 2012 13:10:57 +0000 (09:10 -0400)] 
Add coding style document

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agocompat mman.h: add cygwin support
Mathieu Desnoyers [Thu, 24 May 2012 17:46:40 +0000 (13:46 -0400)] 
compat mman.h: add cygwin support

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agocompat socket.h: add cygwin support
Mathieu Desnoyers [Thu, 24 May 2012 17:46:21 +0000 (13:46 -0400)] 
compat socket.h: add cygwin support

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agokernel-ctl: cygwin compatibility compile fix
Mathieu Desnoyers [Thu, 24 May 2012 16:07:21 +0000 (12:07 -0400)] 
kernel-ctl: cygwin compatibility compile fix

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agopoll compat wrapper: add cygwin support
Mathieu Desnoyers [Thu, 24 May 2012 16:01:47 +0000 (12:01 -0400)] 
poll compat wrapper: add cygwin support

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoUpdate fcntl cygwin support
Mathieu Desnoyers [Thu, 24 May 2012 15:59:38 +0000 (11:59 -0400)] 
Update fcntl cygwin support

cygwin has posix_fadvise

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agocygwin support: only freebsd needs to typedef loff_t
Mathieu Desnoyers [Thu, 24 May 2012 15:49:55 +0000 (11:49 -0400)] 
cygwin support: only freebsd needs to typedef loff_t

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agofix: add missing parenthesis
Mathieu Desnoyers [Thu, 24 May 2012 15:01:03 +0000 (11:01 -0400)] 
fix: add missing parenthesis

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agocompat fcntl.h: add cygwin support
Mathieu Desnoyers [Thu, 24 May 2012 15:00:11 +0000 (11:00 -0400)] 
compat fcntl.h: add cygwin support

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
11 years agoMove futex.c/.h to libcommon
David Goulet [Tue, 8 May 2012 19:47:53 +0000 (15:47 -0400)] 
Move futex.c/.h to libcommon

This is done so the future lttng-relayd and other possible binaries
could use it.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoUpdate version to 2.0.1 stable
David Goulet [Thu, 19 Apr 2012 15:12:09 +0000 (11:12 -0400)] 
Update version to 2.0.1 stable

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: test script for kernel module validation
David Goulet [Thu, 19 Apr 2012 15:03:38 +0000 (11:03 -0400)] 
Fix: test script for kernel module validation

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: report error to client on consumerd error
David Goulet [Thu, 19 Apr 2012 14:58:42 +0000 (10:58 -0400)] 
Fix: report error to client on consumerd error

Adds a consumer daemon state flag that allows us to know the state of
the consumer during client command processing.

Please refer to the comment in lttng-sessiond/main.c above the
definition of those flags for more information and a possible race
condition issue.

This is a quick fix for bug #137 that will be merged in 2.0-stable
branch.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: replace 'modprobe -l' by lsmod in kernel test
David Goulet [Mon, 16 Apr 2012 15:37:49 +0000 (11:37 -0400)] 
Fix: replace 'modprobe -l' by lsmod in kernel test

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: add missing ht_destroy at consumer teardown
Mathieu Desnoyers [Thu, 12 Apr 2012 15:45:36 +0000 (11:45 -0400)] 
Fix: add missing ht_destroy at consumer teardown

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoRemove useless test and fix message output
David Goulet [Mon, 16 Apr 2012 15:19:15 +0000 (11:19 -0400)] 
Remove useless test and fix message output

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoWait for all apps in high-throughput test
David Goulet [Thu, 12 Apr 2012 18:54:28 +0000 (14:54 -0400)] 
Wait for all apps in high-throughput test

Don't wait an arbitrary amount of time for application to end.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoAdd multi-session test for UST
David Goulet [Thu, 12 Apr 2012 18:22:55 +0000 (14:22 -0400)] 
Add multi-session test for UST

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: consumer fd recv thread should write into non-blocking pipe
Mathieu Desnoyers [Thu, 12 Apr 2012 01:57:31 +0000 (21:57 -0400)] 
Fix: consumer fd recv thread should write into non-blocking pipe

Writing into a blocking pipe will cause the writer thread to block on
the poll fds thread when the pipe is full. Given that we would like to
batch stream array reallocation as much as possible, this wakeup should
not block.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix: work-around glibc __nptl_setxid vs clone hang
Mathieu Desnoyers [Thu, 12 Apr 2012 01:07:13 +0000 (21:07 -0400)] 
Fix: work-around glibc __nptl_setxid vs clone hang

hash table resize threads exit end up setting a "locked" state within
libc pthread, which deadlocks with seteuid/setegid called from the
cloned process in runas.c when runas() is called exactly when a resize
thread exits.

Temporarily fix this issue by adding a mutex cross this resize
operation, which holds mutual exclusion with runas() usage.

We should investigate whether we want to properly call exec() from the
runas.c clone child before touching any non-async-signal-safe libc call.
However, given that this change is more intrusive, let's first use this
mutex-based work-around.

Before this fix, running 1000 instances of "demo-trace 300" with
sessiond running as root, and:

lttng create
lttng enable-event -u -a
lttng start

would sometimes lead to consumerd hang with the following clone child
backtrace:

setxid_mark_thread (cmdp=<optimized out>, t=0x7f52dd47c700)
    at allocatestack.c:995
995 allocatestack.c: No such file or directory.
(gdb) bt full
    at allocatestack.c:995
        ch = <optimized out>
    at allocatestack.c:1088
        t = 0x80
        signalled = <optimized out>
        result = <optimized out>
        runp = 0x7f52dd47c9c0
    at ../sysdeps/unix/sysv/linux/setegid.c:44
        __p = 0xfffffffffffffe00
        __cmd = {syscall_no = 119, id = {-1, 1000, -1}, cntr = 0}
        result = <optimized out>
        data = 0x7f52e66e1930
        writelen = <optimized out>
        writeleft = <optimized out>
        index = <optimized out>
        sendret = {i = 0, c = "\000\000\000"}
        ret = <optimized out>
        __func__ = "child_run_as"
    at ../sysdeps/unix/sysv/linux/x86_64/clone.S:112
No locals.
No symbol table info available.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix: consumer fd recv poll should be sensitive to POLLPRI too
Mathieu Desnoyers [Wed, 11 Apr 2012 22:15:24 +0000 (18:15 -0400)] 
Fix: consumer fd recv poll should be sensitive to POLLPRI too

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix consumer lttng_consumer_poll_socket poll revents check bug
Mathieu Desnoyers [Wed, 11 Apr 2012 22:06:08 +0000 (18:06 -0400)] 
Fix consumer lttng_consumer_poll_socket poll revents check bug

Was causing the sessiond to fail to receive streams under heavy load,
because this test needs to be done with a mask rather than equality.
Testing equality was failing as soon as POLLPRI (or any other flag) was
set.

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix: increase consumer open files limit
David Goulet [Wed, 11 Apr 2012 21:56:29 +0000 (17:56 -0400)] 
Fix: increase consumer open files limit

Set ulimit of consumer when root so it can scales up to the session
daemon number of open files.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoUpdate README: document dep on modprobe for kernel tracing
Mathieu Desnoyers [Wed, 11 Apr 2012 00:33:57 +0000 (20:33 -0400)] 
Update README: document dep on modprobe for kernel tracing

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix tests: add missing session2 destroy
Mathieu Desnoyers [Tue, 10 Apr 2012 21:39:40 +0000 (17:39 -0400)] 
Fix tests: add missing session2 destroy

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix: increase wait time for UST high-throughput test
David Goulet [Tue, 10 Apr 2012 17:45:46 +0000 (13:45 -0400)] 
Fix: increase wait time for UST high-throughput test

On slower system, this test failed due to the time period the test does
for applications to at least start tracing.

Add a warning on stdout telling that this is an arbitrary time and might
fail on slower system.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: wait for sessiond to stop in tests
David Goulet [Tue, 10 Apr 2012 17:42:46 +0000 (13:42 -0400)] 
Fix: wait for sessiond to stop in tests

Before returning from stop_sessiond bash function, we wait that the
sessiond daemon completely stopped. If it hungs up at that point, the
kill did not work and investigation can begin.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: tests: error handling labels are in the wrong order
Mathieu Desnoyers [Tue, 10 Apr 2012 15:38:49 +0000 (11:38 -0400)] 
Fix: tests: error handling labels are in the wrong order

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix: remove unused return variable
David Goulet [Tue, 10 Apr 2012 14:59:07 +0000 (10:59 -0400)] 
Fix: remove unused return variable

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: kernel test scripts
David Goulet [Fri, 6 Apr 2012 19:27:03 +0000 (15:27 -0400)] 
Fix: kernel test scripts

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoDon't report back error on syscalls fail for -a -k
David Goulet [Fri, 6 Apr 2012 19:24:53 +0000 (15:24 -0400)] 
Don't report back error on syscalls fail for -a -k

lttng enable-event -a -k will not report an error anymmore if enabling
syscall events has failed. Please refer to the commitdiff for a detailed
comment on why this is done like so.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: destroy context hash table being NULL
David Goulet [Thu, 5 Apr 2012 19:56:43 +0000 (15:56 -0400)] 
Fix: destroy context hash table being NULL

Passing an event unknown loglevel type to the session daemon (for UST
domain) was triggering an error code path to destroy the context hash
table of the event which is not created once the error is hit.

Fix a segfault completely killing the session daemon.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoAdd channel command in utils.sh of tests
David Goulet [Thu, 5 Apr 2012 19:49:32 +0000 (15:49 -0400)] 
Add channel command in utils.sh of tests

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoImprove UST tracing actions test
David Goulet [Thu, 5 Apr 2012 19:48:23 +0000 (15:48 -0400)] 
Improve UST tracing actions test

Test all possible lttng command for UST domain.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoAdd wildcard test to UST
David Goulet [Thu, 5 Apr 2012 19:46:52 +0000 (15:46 -0400)] 
Add wildcard test to UST

Rename test file and add two event wildcard type to the test.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: make lttng expand path for trace output opt
David Goulet [Thu, 5 Apr 2012 15:39:59 +0000 (11:39 -0400)] 
Fix: make lttng expand path for trace output opt

lttng create --output was passing the path string to the session daemon
and thus, for relative path like './mytraces', it was created in the
current directory of the session daemon.

Now lttng command line uses the realpath(3) of the --output string and
denies creation if multiple level of directory does not exist (Ex:
/tmp/foo/bar/chap, if foo/ does not exist, it is refused).

Directory creation still occurs on the session daemon side.

Reported-by: Ettore Del Negro <ettore@ettoredelnegro.me>
Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: remove tests from runall script
David Goulet [Tue, 3 Apr 2012 18:36:14 +0000 (14:36 -0400)] 
Fix: remove tests from runall script

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: return value signedness
David Goulet [Tue, 3 Apr 2012 18:29:01 +0000 (14:29 -0400)] 
Fix: return value signedness

Also remove repeated include headers.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: man page environment variable name error
David Goulet [Tue, 3 Apr 2012 16:09:16 +0000 (12:09 -0400)] 
Fix: man page environment variable name error

Reported-by: Ryan Kyser <Ryan.Kyser@jci.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: check return value on getenv USER
David Goulet [Tue, 3 Apr 2012 14:48:04 +0000 (10:48 -0400)] 
Fix: check return value on getenv USER

System with no environment variable USER, one of the easter egg caused a
segfault on certain embedded linux platforms.

Other getenv() of the project were audited to ensure that the return
value is checked.

Reported-by: Ryan Kyser <Ryan.Kyser@jci.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoRemove useless test
David Goulet [Mon, 2 Apr 2012 20:30:51 +0000 (16:30 -0400)] 
Remove useless test

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoAdd before-after test
David Goulet [Mon, 2 Apr 2012 18:44:55 +0000 (14:44 -0400)] 
Add before-after test

Test tracing before and after application execution.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoCleanup debug printf in test
David Goulet [Mon, 2 Apr 2012 16:14:53 +0000 (12:14 -0400)] 
Cleanup debug printf in test

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix tests when lttng-ust is disabled
David Goulet [Mon, 2 Apr 2012 16:10:05 +0000 (12:10 -0400)] 
Fix tests when lttng-ust is disabled

(close #200)

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoAdd low-throughput test
David Goulet [Mon, 2 Apr 2012 15:57:23 +0000 (11:57 -0400)] 
Add low-throughput test

Test during 20 minutes events at each minute, 10 minutes and 20 minutes.
It validates the event order with an incremental counter.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoAdd high-throughput test
David Goulet [Tue, 27 Mar 2012 20:04:27 +0000 (16:04 -0400)] 
Add high-throughput test

This tests high number of events being recorded with multiple concurrent
instrumented applications.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoImprove tests and change tests/ directory layout
David Goulet [Tue, 6 Mar 2012 21:59:55 +0000 (16:59 -0500)] 
Improve tests and change tests/ directory layout

Add run-report.py and test_list.py files which are an extended test framework
providing memory and CPU sampling of the session daemon during the tests.

Tests are now divided on a per package basis. The kernel/ directory contains
integration tests for lttng-modules (kernel tracer). ust/ directory contains
tests for lttng-ust and tools/ directory has the unit tests of lttng-tools
itself which does not test tracer integration.

First patch aimed at improving test coverage of the lttng 2.0 toolchain and
provide the python framework for more detailled analysis.

Use make check to run them all or use run-report.py and select which tests to
run in test_list.py.

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoUpdate version to 2.0.0 stable v2.0.0
David Goulet [Tue, 20 Mar 2012 22:32:45 +0000 (18:32 -0400)] 
Update version to 2.0.0 stable

ROCK ON!!!

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoUpdate version to v2.0.0-rc4 v2.0.0-rc4
David Goulet [Tue, 20 Mar 2012 20:42:48 +0000 (16:42 -0400)] 
Update version to v2.0.0-rc4

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: add small detail to enable-channel man page
David Goulet [Tue, 20 Mar 2012 20:19:16 +0000 (16:19 -0400)] 
Fix: add small detail to enable-channel man page

(close #192)

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: warned of arbitrary time for ust-nprocesses test
David Goulet [Tue, 20 Mar 2012 20:15:56 +0000 (16:15 -0400)] 
Fix: warned of arbitrary time for ust-nprocesses test

(close #157)

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix. add limitation to man page
David Goulet [Tue, 20 Mar 2012 20:09:51 +0000 (16:09 -0400)] 
Fix. add limitation to man page

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: minor changes to lttng.1
David Goulet [Tue, 20 Mar 2012 19:53:22 +0000 (15:53 -0400)] 
Fix: minor changes to lttng.1

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: add exit values to lttng.1 man page
David Goulet [Tue, 20 Mar 2012 19:51:14 +0000 (15:51 -0400)] 
Fix: add exit values to lttng.1 man page

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: add delay before validating apps in test nproc
David Goulet [Tue, 20 Mar 2012 19:33:10 +0000 (15:33 -0400)] 
Fix: add delay before validating apps in test nproc

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: error handling in lttng enable-event
David Goulet [Tue, 20 Mar 2012 19:32:43 +0000 (15:32 -0400)] 
Fix: error handling in lttng enable-event

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: document structure init. to 0 in lttng.h
Raphaël Beamonte [Sun, 18 Mar 2012 02:02:53 +0000 (22:02 -0400)] 
Fix: document structure init. to 0 in lttng.h

Document the fact that the structures should be initialized to zero in
lttng.h since there is no init function.

(close #135)

Signed-off-by: Raphaël Beamonte <raphael.beamonte@gmail.com>
Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: uninitialized variable
David Goulet [Tue, 20 Mar 2012 18:55:52 +0000 (14:55 -0400)] 
Fix: uninitialized variable

Signed-off-by: David Goulet <dgoulet@ev0ke.net>
12 years agoKeep track of FD used for UST applications (v2)
Mathieu Desnoyers [Tue, 20 Mar 2012 18:53:07 +0000 (14:53 -0400)] 
Keep track of FD used for UST applications (v2)

Allow to keep 25% of file descriptors reserved for commands/kernel
tracing/internal communication within the sessiond by limiting
applications to 75% of the available file descriptors. This ensures
traced applications cannot cause a sessiond denial of service.

Changelog since v1:
- Add missing fd-limit.h

Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix: lttng view, error message and exit code
Julien Desfossez [Tue, 20 Mar 2012 15:19:14 +0000 (11:19 -0400)] 
Fix: lttng view, error message and exit code

lttng view is a frontend command, when the viewer is not found in the
path, the error message should be human readable (no need for the
developper-oriented debug message).

Also the return code must indicate that something went wrong if the
viewer is not on the system.

(fix #144)

Signed-off-by: Julien Desfossez <julien.desfossez@efficios.com>
Signed-off-by: David Goulet <dgoulet@ev0ke.net>
12 years agoSmall fixes to lttng man pages (v2)
Yannick Brosseau [Tue, 20 Mar 2012 15:37:09 +0000 (11:37 -0400)] 
Small fixes to lttng man pages (v2)

Signed-off-by: Yannick Brosseau <yannick.brosseau@gmail.com>
12 years agoMerge branch 'master' of git://git.lttng.org/lttng-tools
David Goulet [Tue, 20 Mar 2012 18:35:38 +0000 (14:35 -0400)] 
Merge branch 'master' of git://git.lttng.org/lttng-tools

12 years agoFix: handle EINTR for sendmsg syscall
David Goulet [Tue, 20 Mar 2012 18:33:23 +0000 (14:33 -0400)] 
Fix: handle EINTR for sendmsg syscall

(close #170)

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: lttng UI exit value and error message
David Goulet [Tue, 20 Mar 2012 18:27:12 +0000 (14:27 -0400)] 
Fix: lttng UI exit value and error message

Change lttcomm error code starting value from 1000 to 10. This way, bash
can return the exact error code since 255 is the maximum.

Fix multiple warning, error and message output.

Some return code could still be not "standardize" but for stable it's
ok. Fixes can come later on.

(close #105)

Signed-off-by: David Goulet <dgoulet@efficios.com>
12 years agoFix: session lock use after free
Mathieu Desnoyers [Tue, 20 Mar 2012 17:39:20 +0000 (13:39 -0400)] 
Fix: session lock use after free

The session lock is broken in that it does not handle teardown correctly
(use after free). Surround each usage by the session list lock for now
to fix this issue, and don't unlock the session lock after free. Since
each session lock usage is surrounded by session list lock, no other
thread will be left waiting on this lock when the session destroy is
performed.

This effectively renders useless the per-session lock. Leave it there
for now to minimize code change before 2.0 final.

This locking scheme will be revisited for lttng 2.1.

Acked-by: David Goulet <dgoulet@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
12 years agoFix kernel_list_events memleaks
Mathieu Desnoyers [Tue, 20 Mar 2012 17:15:01 +0000 (13:15 -0400)] 
Fix kernel_list_events memleaks

- The fscanf %m allocation was leaking the "event" (needs to be freed
  explicitly).
- The error handling of realloc should free the previous pointer.

Acked-by: David Goulet <dgoulet@efficios.com>
Signed-off-by: Mathieu Desnoyers <mathieu.desnoyers@efficios.com>
This page took 0.041931 seconds and 4 git commands to generate.