On 17 Jan (17:26:13), Jim Dumont wrote:
> Hi there,
> 
> Was there a follow-up to the series of questions raised by Daniel below and 
> referenced here:
> 
> http://lists.lttng.org/pipermail/lttng-dev/2013-November/021844.html
> 
> I didn't see one in the mailing list.
> 
> BR,
> 
> /Jim
> 
> 
> > 1) Live support. It's now possible to stream traces (kernel and UST) 
> > while they being extracted meaning a relayd can now receive traces 
> > live and a viewer can attach to it to process the trace while running. 
> > The "--live" has been added to lttng create command that sets up a 
> > live tracing session.
> >
> > See doc/live-reading-howto.txt and doc/live-reading-protocol.txt for 
> > more detail on that feature.
> 
>    Does this mean you can't do a "live" trace to local storage?

You can use local storage for the live feature BUT you always need a
relayd thus there will be some streaming happening on the loopback
interface.

However, you *can't* start a live viewer by directly pointing to a trace
on disk. It may works but we do NOT recommend it since you'll probably
end up in a bad state or have synchronization issue.

> 
> > 2) Java Util Logging support. With a special LTTngAgent located in 
> > lttng-ust project (liblttng-ust-jul) to trace JUL events and gather 
> > UST traces with them.
> >
> > The -j/--jul domain has been added to lttng command line so you can 
> > list the possible JUL Logger object and enable/disable them. It is the 
> > same as using the UST or Kernel domain.
> >
> > As of now, filters and loglevel are ignored with the JUL domain.
> 
>    Saying "It is the same as using the UST or Kernel domain" leaves a lot of 
> questions unanswered, since the kernel and user-space behave *very* 
> differently on a lot of points.
> 
> 1) Does --jul require root privileges?

No. JUL actually acts exactly like UST since it uses it thus a Java
application connect to the session daemon that it can talk to.

> 2) I suppose the various commands have --kernel pre-empting --userspace 
> pre-empting --ul?

Yup. This is actually not mention anywhere but looking at "lttng"
command line code, the domain's order is: kernel, ust, jul. Probably we
should actually output an error if the user uses multiple domain...

> 3) add-context: does --jul have the same possible context types as 
> --userspace (i.e. ip, pthread_id, procname, vpid, vtid)?

Nope.

> 4) enable-channel:
> 4.1) Are the --jul --subbuf-size, --num-subbuf and --read-timer defaults the 
> same as with --userspace?

Nope. JUL uses it's internal channel so basically enable channel does
NOT support the jul domain thus not having -j in the help/man.

> 4.2) Presumably --output can only be mmap?

Yes. Same as UST but again you can't use enable-channel with JUl domain.

> 4.3) Are the --jul buffering schemes the same as with --userspace (i.e. 
> --buffers-pid and --buffers-uid)?

Always per UID. You can't change it.

> 4.4) If yes, won't they all degenerate to the virtual machine's user and 
> process IDs?

Not sure I understand this one?

> 5) enable-event:
> 5.1) Can you assign --jul events to multiple channels like you can in 
> --userspace?

No.

> 5.2) Is the --jul event type --tracepoint or something else entirely?

It is tracepoint yes. For the JUL domain, ALL UST tracepoints are
enabled (there is only one for now with the lttng-ust library we use
(liblttng-ust-jul)).

> 5.3) Can the trailing wild card (and exclusion option) be used in the event 
> name when enabling/disabling --jul events?

Wild card are NOT supported for the JUL domain. Thus for instance "lttng
enable-event -j my_event_subsystem42_*" will not work. Only the "*" will
work enabling every Loggers.

> 6) list: Does bug #654 also apply to --jul?

Yes most probably. The session name is ignored.


Now after answering all this, I feel like either creating a lttng-jul
man page or adding a special section in "lttng.1" for the JUL domain
explaining all this. Thoughts?

Cheers!
David

> 
> Daniel U. Thibault
> Protection des syst?mes et contremesures (PSC) | Systems Protection & 
> Countermeasures (SPC) Cyber s?curit? pour les missions essentielles (CME) | 
> Mission Critical Cyber Security (MCCS) R & D pour la d?fense Canada - 
> Valcartier (RDDC Valcartier) | Defence R&D Canada - Valcartier (DRDC 
> Valcartier)
> 2459 route de la Bravoure
> Qu?bec QC  G3J 1X5
> CANADA
> Vox : (418) 844-4000 x4245
> Fax : (418) 844-4538
> NAC : 918V QSDJ <http://www.travelgis.com/map.asp?addr=918V%20QSDJ>
> Gouvernement du Canada | Government of Canada 
> <http://www.valcartier.drdc-rddc.gc.ca/>
> 
> 
> 
> ------------------------------
> 
> Message: 2
> Date: Fri, 15 Nov 2013 21:10:00 +0000 (UTC)
> From: Mathieu Desnoyers <[email protected]>
> To: lttng-dev <[email protected]>
> Subject: [lttng-dev] [RELEASE] LTTng-UST 2.4.0-rc1
> Message-ID:
>       <[email protected]>
> Content-Type: text/plain; charset=utf-8
> 
> LTTng-UST, the Linux Trace Toolkit Next Generation Userspace Tracer, is port 
> of the low-overhead tracing capabilities of the LTTng kernel tracer to 
> user-space. The library "liblttng-ust" enables tracing of applications and 
> libraries.
> 
> Main changes: Java Util Logging support, live streaming support, event 
> "excluders" (courtesy of Mentor), base address tracing for symbol lookup 
> (courtesy of Mentor).
> 
> Changelog:
> 2013-11-15 lttng-ust 2.4.0-rc1
>         * doc/examples: error out when a subdir make fails
>         * JUL: fix out of tree build
>         * Java/jul: fix make dist
>         * JUL: install documentation and jar file
>         * Add missing copyrights to test scripts
>         * Escape minus signs in lttng-ust-cyg-profile manpage
>         * Fix spelling mistake in lttng-ust manpage
>         * Fix tests/benchmark
>         * Implement base-address-state tracing
>         * Base-address tracing for dlopen and dlclose
>         * Add a check against excluders
>         * Add handler for LTTNG_UST_EXCLUSION in UST ABI
>         * Store exclusions to enablers
>         * Add excluders to enabler structure
>         * Send exclusion data through ustcomm
>         * Define exclusion structure for UST ABI
>         * Define a new command LTTNG_UST_EXCLUSION
>         * Fix: package of LTTngUst.h was renamed without renaming target in 
> makefile
>         * Fix: check for java .class in package subfolders
>         * Fix: java library to use user define JDK at compile time
>         * Add liblttng-ust-jul for JUL support
>         * Fix: application SIGBUS when starting in parallel with sessiond
>         * Fix: Minor corrections to the lttng-ust man page
>         * Documentation cleanup: Overhaul of the lttng-gen-tp man page
>         * Warning cleanup: add missing padding initializer
>         * Fix: socket connect hang on heavy loads
>         * Cleanup: fix typo
>         * Fix: ust-comm recvmsg should handle partial receive
>         * gcc warning fix: -Wextra
>         * Add usage reference count for tracepoints
>         * tracepoint.c: Move add_callsite/remove_callsite further down in file
>         * Store the callsites into the library callsite list
>         * Fix linking warning
>         * Per-stream ioctl to get the current timestamp
>         * LTTng ringbuffer ABI calls for index generation
>         * Specialize lttng_ust_lib_ring_buffer_client_cb
>         * Add tracing instrumentation for pthread mutex lock functions
> 
> Project website: http://lttng.org
> Download link: http://lttng.org/download (please refer to the README file for 
> installation instructions)
> 
> --
> Mathieu Desnoyers
> EfficiOS Inc.
> http://www.efficios.com
> 
> 
> 
> ------------------------------
> 
> Message: 3
> Date: Fri, 15 Nov 2013 21:12:06 +0000 (UTC)
> From: Mathieu Desnoyers <[email protected]>
> To: lttng-dev <[email protected]>
> Subject: [lttng-dev] [RELEASE] LTTng modules 2.4.0-rc1
> Message-ID:
>       <[email protected]>
> Content-Type: text/plain; charset=utf-8
> 
> The LTTng modules provide Linux kernel tracing capability to the LTTng
> 2.x tracer toolset.
> 
> Major changes: live streaming support, index generation by consumerd support, 
> 3.12 Linux kernel support.
> 
> Changelog:
> 2013-11-15 LTTng modules 2.4.0-rc1
>         * Update ext4 instrumentation to Linux 3.12
>         * Update RCU instrumentation to Linux 3.12
>         * Update btrfs instrumentation to 3.12 kernel
>         * Update vmscan instrumentation to Linux 3.12
>         * update: kmem instrumentation 3.12 kernel support
>         * lttng-statedump-impl: Use generic hard irqs for Linux >= 3.12
>         * lttng-statedump-impl: Fix lttng_list_interrupts for empty 
> implementation
>         * Import fix from LTSI: 3.4+ RT kernels use CONFIG_PREEMPT_RT_FULL
>         * Add new tracepoint in KVM
>         * Blacklist kernels 3.10.13 and 3.11.2
>         * LTTng trace-clock: shrink kernel blacklist
>         * Metadata flush writes data from the cache
>         * Per-stream ioctl to get the current timestamp
>         * LTTng ringbuffer ABI calls for index generation
>         * Prepare LTTng override of ring buffer fops
>         * Blacklist Linux kernels 3.10+
> 
> Project website: http://lttng.org
> Download link: http://lttng.org/download
> (please refer to the README files for installation instructions and
> lttng-tools doc/quickstart.txt for usage information)
> 
> -- 
> Mathieu Desnoyers
> EfficiOS Inc.
> http://www.efficios.com
> 
> 
> 
> ------------------------------
> 
> Message: 4
> Date: Sat, 16 Nov 2013 08:42:15 -0500
> From: David Goulet <[email protected]>
> To: ??? <[email protected]>
> Cc: lttng-dev <[email protected]>
> Subject: Re: [lttng-dev] ??? how does ust tracer map share lttng
>       buffer
> Message-ID: <20131116134215.GA7300@thessa>
> Content-Type: text/plain; charset="utf-8"
> 
> On 16 Nov (07:18:45), ??? wrote:
> > >Indeed, from a sessiond point of view, this will typically use the same FD 
> > >number, but it's really associated with a different posix SHM each time.
> > 
> > 
> >   Yes,what confused me the most is that sessiond sends the same FD number 
> > to UST every time,but UST receives the different FD number each time and 
> > can map it to the right per-cpu share buffer sucessfully.
> 
> Yes, in per UID buffers, for every application using UST we send the FD
> corresponding to the already allocated buffer (being per UID).
> 
> Please note that sending a FD through a Unix socket dups it so basically
> UST receives that, access the shared memory buffer and can then operate
> correctly.
> 
> Cheers!
> David
> 
> > 
> > 
> > Thanks
> > MuXiu.Cao
> > 
> > 
> > 
> > 
> > 
> > From: "???" <[email protected]>
> > To: "mathieu.desnoyers" <[email protected]>
> > Sent: Friday, November 15, 2013 11:37:11 AM
> > Subject: how does ust tracer map share lttng buffer
> > 
> > 
> > Hi,Mathieu
> > 
> > 
> >    After lttng sub-command enable-event is issued at first time, 
> > lttng-consumerd will be informed by lttng-sessiond to open and mmap share 
> > memory regions .Then lttng-sessiond get all shmfds from lttng-consumerd.But 
> > how does ust tracer map to these share memory regions?
> > 
> > 
> >   AFAIK,lttng-sessiond will call  do_consumer_create_channel()  ,then 
> > send_channel_uid_to_ust(). What's confused me is in 
> > send_channel_uid_to_ust(),lttng-sessiond dup() one shmfd , then 
> > ust_consumer_send_stream_to_ust() to ust tracer, then close the shmfd and 
> > loop all shmfds. So all duped fds are the same fd-number (say A)and send to 
> > ust tracer through 
> > ust_consumer_send_stream_to_ust()->ustctl_send_stream_to_ust()->ustctl_send_stream()->ustcomm_send_fds_unix_sock()(unix
> >  socket is $HOME/.lttng/lttng-ust-sock-5).Why does it send the same fds to 
> > ust tracer?
> >   At ust tracer side,it calls 
> > handle_message()->ustcomm_recv_stream_from_sessiond() and mmap every 
> > received fd to share memory region.However,every received fd is different 
> > from A.So I must miss something anywhere.How does ust tracer map share 
> > lttng buffer?
> > 
> > 
> > In lttng-sessiond, we send file descriptors for each posix SHM associated 
> > to every per-cpu buffer to the application over unix socket. So it's more 
> > or less : receive FD from consumer, send FD over unix socket to UST, close 
> > FD (we loop on this). Indeed, from a sessiond point of view, this will 
> > typically use the same FD number, but it's really associated with a 
> > different posix SHM each time. David can probably answer your question in 
> > more detail.
> > 
> > 
> > 
> > Thanks,
> > 
> > 
> > 
> > Mathieu
> > 
> > 
> > 
> > 
> > 
> > 
> > 
> > PS:code is from lttng-2.3.0 .
> > 
> > 
> > Thanks
> > MuXiu.Cao
> > 
> > 
> > 
> > 
> > 
> > 
> > 
> > -- 
> > 
> > Mathieu Desnoyers
> > EfficiOS Inc.
> > http://www.efficios.com
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: signature.asc
> Type: application/pgp-signature
> Size: 620 bytes
> Desc: Digital signature
> URL: 
> <http://lists.lttng.org/pipermail/lttng-dev/attachments/20131116/afbf97a4/attachment-0001.pgp>
> 
> ------------------------------
> 
> Message: 5
> Date: Sun, 17 Nov 2013 11:36:19 +0200
> From: Amit Margalit <[email protected]>
> To: Mathieu Desnoyers <[email protected]>
> Cc: [email protected]
> Subject: Re: [lttng-dev] [PATCH v3 2/2] Implement base-address-state
>       tracing
> Message-ID:
>       <of5ea7c09a.2f75fe97-onc2257c26.00348fd3-c2257c26.0034c...@il.ibm.com>
> Content-Type: text/plain; charset="us-ascii"
> 
> I guess I missed something in the discussion - is this going to be active 
> all the time, or can I disable it somehow?
> 
> Maybe a command-line switch to turn this off?
> 
> For me this is overhead that I don't need and don't want.
> 
> Thanks,
> 
> Amit Margalit
> IBM XIV - Storage Reinvented
> XIV-NAS Development Team
> Tel. 03-689-7774
> Fax. 03-689-7230
> 
> 
> 
> From:   Mathieu Desnoyers <[email protected]>
> To:     Paul Woegerer <[email protected]>
> Cc:     [email protected]
> Date:   11/14/2013 09:30 PM
> Subject:        Re: [lttng-dev] [PATCH v3 2/2] Implement 
> base-address-state tracing
> 
> 
> 
> ----- Original Message -----
> > From: "Paul Woegerer" <[email protected]>
> > To: [email protected], "mathieu desnoyers" 
> <[email protected]>
> > Sent: Thursday, November 14, 2013 1:39:49 PM
> > Subject: [PATCH v3 2/2] Implement base-address-state tracing
> > 
> > Dump the base-address state (executable and shared objects) into session
> > on session-enable (per-session events).
> 
> Merged, thanks !
> 
> Mathieu
> 
> > 
> > Signed-off-by: Paul Woegerer <[email protected]>
> > ---
> >  include/lttng/ust-tracepoint-event.h     | 14 +++++++
> >  liblttng-ust-baddr/Makefile.am           |  4 +-
> >  liblttng-ust-baddr/lttng-ust-baddr.c     | 72
> >  ++++++++++++++++++++++++++++++++
> >  liblttng-ust-baddr/ust_baddr_statedump.c | 21 ++++++++++
> >  liblttng-ust-baddr/ust_baddr_statedump.h | 60 
> ++++++++++++++++++++++++++
> >  liblttng-ust/lttng-events.c              |  2 +
> >  liblttng-ust/lttng-tracer-core.h         |  3 ++
> >  liblttng-ust/lttng-ust-comm.c            | 60 
> ++++++++++++++++++++++++++
> >  8 files changed, 235 insertions(+), 1 deletion(-)
> >  create mode 100644 liblttng-ust-baddr/ust_baddr_statedump.c
> >  create mode 100644 liblttng-ust-baddr/ust_baddr_statedump.h
> > 
> > diff --git a/include/lttng/ust-tracepoint-event.h
> > b/include/lttng/ust-tracepoint-event.h
> > index bb3a05d..be58030 100644
> > --- a/include/lttng/ust-tracepoint-event.h
> > +++ b/include/lttng/ust-tracepoint-event.h
> > @@ -480,6 +480,18 @@ size_t
> > __event_get_align__##_provider##___##_name(_TP_ARGS_PROTO(_args))      \
> >  #define TP_FIELDS(...) __VA_ARGS__
> > 
> >  /*
> > + * For state dump, check that "session" argument (mandatory) matches 
> the
> > + * session this event belongs to. Ensures that we write state dump data 
> only
> > + * into the started session, not into all sessions.
> > + */
> > +#undef _TP_SESSION_CHECK
> > +#ifdef TP_SESSION_CHECK
> > +#define _TP_SESSION_CHECK(session, csession)   (session == csession)
> > +#else /* TP_SESSION_CHECK */
> > +#define _TP_SESSION_CHECK(session, csession)   1
> > +#endif /* TP_SESSION_CHECK */
> > +
> > +/*
> >   * Using twice size for filter stack data to hold size and pointer for
> >   * each field (worse case). For integers, max size required is 64-bit.
> >   * Same for double-precision floats. Those fit within
> > @@ -506,6 +518,8 @@ void
> > __event_probe__##_provider##___##_name(_TP_ARGS_DATA_PROTO(_args))   \
> >                                                                      \
> >                if (0)                \
> >                                (void) __dynamic_len_idx; /* don't warn 
> if unused */    \
> > +              if (!_TP_SESSION_CHECK(session, __chan->session))   \
> > +                              return;       \
> >                if 
> (caa_unlikely(!CMM_ACCESS_ONCE(__chan->session->active)))     \
> >                                return;       \
> >                if (caa_unlikely(!CMM_ACCESS_ONCE(__chan->enabled)))    \
> > diff --git a/liblttng-ust-baddr/Makefile.am 
> b/liblttng-ust-baddr/Makefile.am
> > index afa9489..0d3cf28 100644
> > --- a/liblttng-ust-baddr/Makefile.am
> > +++ b/liblttng-ust-baddr/Makefile.am
> > @@ -5,7 +5,9 @@ lib_LTLIBRARIES = liblttng-ust-baddr.la
> >  liblttng_ust_baddr_la_SOURCES = \
> >                lttng-ust-baddr.c \
> >                ust_baddr.c \
> > -              ust_baddr.h
> > +              ust_baddr.h \
> > +              ust_baddr_statedump.c \
> > +              ust_baddr_statedump.h
> >  liblttng_ust_baddr_la_LIBADD = \
> >                -L$(top_builddir)/liblttng-ust/.libs \
> >                -llttng-ust
> > diff --git a/liblttng-ust-baddr/lttng-ust-baddr.c
> > b/liblttng-ust-baddr/lttng-ust-baddr.c
> > index f24a171..a856965 100644
> > --- a/liblttng-ust-baddr/lttng-ust-baddr.c
> > +++ b/liblttng-ust-baddr/lttng-ust-baddr.c
> > @@ -34,6 +34,7 @@
> > 
> >  #define TRACEPOINT_DEFINE
> >  #include "ust_baddr.h"
> > +#include "ust_baddr_statedump.h"
> > 
> >  int
> >  lttng_ust_push_baddr(void *so_base, const char *so_name)
> > @@ -62,3 +63,74 @@ lttng_ust_pop_baddr(void *so_base)
> >                tracepoint(ust_baddr, pop, so_base);
> >                return 0;
> >  }
> > +
> > +static int
> > +extract_soinfo_events(struct dl_phdr_info *info, size_t size, void 
> *data)
> > +{
> > +              int j;
> > +              int num_loadable_segment = 0;
> > +
> > +              for (j = 0; j < info->dlpi_phnum; j++) {
> > +                              char resolved_path[PATH_MAX];
> > +                              struct stat sostat;
> > +                              void *base_addr_ptr;
> > +
> > +                              if (info->dlpi_phdr[j].p_type != PT_LOAD)
> > +                                              continue;
> > +
> > +                              /* Calculate virtual memory address of 
> the loadable segment */
> > +                              base_addr_ptr = (void *) info->dlpi_addr
> > +                                              + 
> info->dlpi_phdr[j].p_vaddr;
> > +
> > +                              num_loadable_segment += 1;
> > +                              if ((info->dlpi_name == NULL || 
> info->dlpi_name[0] == 0)
> > +                                                              && 
> num_loadable_segment == 1) {
> > +                                              /*
> > +                                               * If the iterated 
> element is the executable itself we
> > +                                               * have to use Dl_info to 
> determine its full path
> > +                                               */
> > +                                              Dl_info dl_info = { 0 };
> > +                                              if 
> (!dladdr(base_addr_ptr, &dl_info))
> > +                                                              return 0;
> > +                                              if 
> (!realpath(dl_info.dli_fname, resolved_path))
> > +                                                              return 0;
> > +                              } else {
> > +                                              /*
> > +                                               * For regular 
> dl_phdr_info entries we have to check if
> > +                                               * the path to the shared 
> object really exists
> > +                                               */
> > +                                              if 
> (!realpath(info->dlpi_name, resolved_path)) {
> > +                                                              /* Found 
> vDSO, put the 'path' into brackets */
> > + snprintf(resolved_path, PATH_MAX - 1, "[%s]",
> > +              info->dlpi_name);
> > +                                              }
> > +                              }
> > +
> > +                              if (stat(resolved_path, &sostat)) {
> > +                                              sostat.st_size = 0;
> > +                                              sostat.st_mtime = -1;
> > +                              }
> > +
> > +                              tracepoint(ust_baddr_statedump, soinfo,
> > +                                                              (struct 
> lttng_session *) data, base_addr_ptr,
> > + resolved_path, sostat.st_size, sostat.st_mtime);
> > +
> > +                              /*
> > +                               * We are only interested in the base 
> address (lowest virtual
> > +                               * address associated with the memory 
> image), skip the rest
> > +                               */
> > +                              break;
> > +              }
> > +              return 0;
> > +}
> > +
> > +int
> > +lttng_ust_baddr_statedump(struct lttng_session *session)
> > +{
> > +              /*
> > +               * Iterate through the list of currently loaded shared 
> objects and
> > +               * generate events for loadable segments using 
> extract_soinfo_events
> > +               */
> > +              dl_iterate_phdr(extract_soinfo_events, session);
> > +              return 0;
> > +}
> > diff --git a/liblttng-ust-baddr/ust_baddr_statedump.c
> > b/liblttng-ust-baddr/ust_baddr_statedump.c
> > new file mode 100644
> > index 0000000..75f74ca
> > --- /dev/null
> > +++ b/liblttng-ust-baddr/ust_baddr_statedump.c
> > @@ -0,0 +1,21 @@
> > +/*
> > + * Copyright (C) 2013  Paul Woegerer <[email protected]>
> > + *
> > + * This library is free software; you can redistribute it and/or
> > + * modify it under the terms of the GNU Lesser General Public
> > + * License as published by the Free Software Foundation; either
> > + * version 2.1 of the License, or (at your option) any later version.
> > + *
> > + * This library is distributed in the hope that it will be useful,
> > + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> > + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> > + * Lesser General Public License for more details.
> > + *
> > + * You should have received a copy of the GNU Lesser General Public
> > + * License along with this library; if not, write to the Free Software
> > + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 
> 02110-1301
> > USA
> > + */
> > +
> > +#define TRACEPOINT_CREATE_PROBES
> > +#define TP_SESSION_CHECK
> > +#include "ust_baddr_statedump.h"
> > diff --git a/liblttng-ust-baddr/ust_baddr_statedump.h
> > b/liblttng-ust-baddr/ust_baddr_statedump.h
> > new file mode 100644
> > index 0000000..77a9af4
> > --- /dev/null
> > +++ b/liblttng-ust-baddr/ust_baddr_statedump.h
> > @@ -0,0 +1,60 @@
> > +#undef TRACEPOINT_PROVIDER
> > +#define TRACEPOINT_PROVIDER ust_baddr_statedump
> > +
> > +#if !defined(_TRACEPOINT_UST_BADDR_STATEDUMP_H) ||
> > defined(TRACEPOINT_HEADER_MULTI_READ)
> > +#define _TRACEPOINT_UST_BADDR_STATEDUMP_H
> > +
> > +#ifdef __cplusplus
> > +extern "C" {
> > +#endif
> > +
> > +/*
> > + * Copyright (C) 2013  Paul Woegerer <[email protected]>
> > + *
> > + * Permission is hereby granted, free of charge, to any person 
> obtaining a
> > copy
> > + * of this software and associated documentation files (the 
> "Software"), to
> > deal
> > + * in the Software without restriction, including without limitation 
> the
> > rights
> > + * to use, copy, modify, merge, publish, distribute, sublicense, and/or 
> sell
> > + * copies of the Software, and to permit persons to whom the Software 
> is
> > + * furnished to do so, subject to the following conditions:
> > + *
> > + * The above copyright notice and this permission notice shall be 
> included
> > in
> > + * all copies or substantial portions of the Software.
> > + *
> > + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, 
> EXPRESS
> > OR
> > + * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF 
> MERCHANTABILITY,
> > + * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT 
> SHALL
> > THE
> > + * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR 
> OTHER
> > + * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, 
> ARISING
> > FROM,
> > + * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER 
> DEALINGS IN
> > THE
> > + * SOFTWARE.
> > + */
> > +
> > +#include <stdint.h>
> > +#include <unistd.h>
> > +#include <lttng/ust-events.h>
> > +
> > +#define LTTNG_UST_BADDR_STATEDUMP_PROVIDER
> > +#include <lttng/tracepoint.h>
> > +
> > +TRACEPOINT_EVENT(ust_baddr_statedump, soinfo,
> > +              TP_ARGS(struct lttng_session *, session, void *, baddr, 
> const char*,
> > sopath, int64_t, size, int64_t, mtime),
> > +              TP_FIELDS(
> > +                              ctf_integer_hex(void *, baddr, baddr)
> > +                              ctf_string(sopath, sopath)
> > +                              ctf_integer(int64_t, size, size)
> > +                              ctf_integer(int64_t, mtime, mtime)
> > +              )
> > +)
> > +
> > +#endif /* _TRACEPOINT_UST_BADDR_STATEDUMP_H */
> > +
> > +#undef TRACEPOINT_INCLUDE
> > +#define TRACEPOINT_INCLUDE "./ust_baddr_statedump.h"
> > +
> > +/* This part must be outside ifdef protection */
> > +#include <lttng/tracepoint-event.h>
> > +
> > +#ifdef __cplusplus
> > +}
> > +#endif
> > diff --git a/liblttng-ust/lttng-events.c b/liblttng-ust/lttng-events.c
> > index 26601a6..21e3639 100644
> > --- a/liblttng-ust/lttng-events.c
> > +++ b/liblttng-ust/lttng-events.c
> > @@ -293,6 +293,8 @@ int lttng_session_enable(struct lttng_session 
> *session)
> >                /* Set atomically the state to "active" */
> >                CMM_ACCESS_ONCE(session->active) = 1;
> >                CMM_ACCESS_ONCE(session->been_active) = 1;
> > +
> > +              lttng_ust_sockinfo_session_enabled(session->owner, 
> session);
> >  end:
> >                return ret;
> >  }
> > diff --git a/liblttng-ust/lttng-tracer-core.h
> > b/liblttng-ust/lttng-tracer-core.h
> > index f643a7e..e7f549e 100644
> > --- a/liblttng-ust/lttng-tracer-core.h
> > +++ b/liblttng-ust/lttng-tracer-core.h
> > @@ -45,4 +45,7 @@ const char *lttng_ust_obj_get_name(int id);
> > 
> >  int lttng_get_notify_socket(void *owner);
> > 
> > +void lttng_ust_sockinfo_session_enabled(void *owner,
> > +                              struct lttng_session *session_enabled);
> > +
> >  #endif /* _LTTNG_TRACER_CORE_H */
> > diff --git a/liblttng-ust/lttng-ust-comm.c 
> b/liblttng-ust/lttng-ust-comm.c
> > index a6e4ba3..6ef3006 100644
> > --- a/liblttng-ust/lttng-ust-comm.c
> > +++ b/liblttng-ust/lttng-ust-comm.c
> > @@ -34,6 +34,7 @@
> >  #include <time.h>
> >  #include <assert.h>
> >  #include <signal.h>
> > +#include <dlfcn.h>
> >  #include <urcu/uatomic.h>
> >  #include <urcu/futex.h>
> >  #include <urcu/compiler.h>
> > @@ -106,6 +107,7 @@ struct sock_info {
> > 
> >                char wait_shm_path[PATH_MAX];
> >                char *wait_shm_mmap;
> > +              struct lttng_session *session_enabled;
> >  };
> > 
> >  /* Socket from app (connect) to session daemon (listen) for 
> communication */
> > @@ -122,6 +124,8 @@ struct sock_info global_apps = {
> >                .notify_socket = -1,
> > 
> >                .wait_shm_path = "/" LTTNG_UST_WAIT_FILENAME,
> > +
> > +              .session_enabled = NULL,
> >  };
> > 
> >  /* TODO: allow global_apps_sock_path override */
> > @@ -135,6 +139,8 @@ struct sock_info local_apps = {
> > 
> >                .socket = -1,
> >                .notify_socket = -1,
> > +
> > +              .session_enabled = NULL,
> >  };
> > 
> >  static int wait_poll_fallback;
> > @@ -176,6 +182,7 @@ static const char *cmd_name_mapping[] = {
> > 
> >  static const char *str_timeout;
> >  static int got_timeout_env;
> > +static void *ust_baddr_handle;
> > 
> >  extern void lttng_ring_buffer_client_overwrite_init(void);
> >  extern void lttng_ring_buffer_client_overwrite_rt_init(void);
> > @@ -235,6 +242,39 @@ void print_cmd(int cmd, int handle)
> >  }
> > 
> >  static
> > +void *lttng_ust_baddr_handle(void)
> > +{
> > +              if (!ust_baddr_handle) {
> > +                              ust_baddr_handle = dlopen(
> > + "liblttng-ust-baddr.so.0", RTLD_NOW | RTLD_GLOBAL);
> > +                              if (ust_baddr_handle == NULL)
> > +                                              ERR("%s", dlerror());
> > +              }
> > +              return ust_baddr_handle;
> > +}
> > +
> > +static
> > +int lttng_ust_baddr_statedump(struct lttng_session *session)
> > +{
> > +              static
> > +              int (*lttng_ust_baddr_init_fn)(struct lttng_session *);
> > +
> > +              if (!lttng_ust_baddr_init_fn) {
> > +                              void *baddr_handle = 
> lttng_ust_baddr_handle();
> > +                              if (baddr_handle) {
> > +                                              lttng_ust_baddr_init_fn = 
> dlsym(baddr_handle,
> > + "lttng_ust_baddr_statedump");
> > +                                              if 
> (lttng_ust_baddr_init_fn == NULL)
> > +                                                              ERR("%s", 
> dlerror());
> > +                              }
> > +                              if (!lttng_ust_baddr_init_fn)
> > +                                              return -1;
> > +              }
> > +
> > +              return lttng_ust_baddr_init_fn(session);
> > +}
> > +
> > +static
> >  int setup_local_apps(void)
> >  {
> >                const char *home_dir;
> > @@ -1143,6 +1183,13 @@ restart:
> >                                                ret = 
> handle_message(sock_info, sock, &lum);
> >                                                if (ret) {
> > ERR("Error handling message for %s socket", sock_info->name);
> > +                                              } else {
> > +                                                              struct 
> lttng_session *session =
> > +  sock_info->session_enabled;
> > +                                                              if 
> (session) {
> > +  sock_info->session_enabled = NULL;
> > +  lttng_ust_baddr_statedump(session);
> > +                                                              }
> >                                                }
> >                                                continue;
> >                                default:
> > @@ -1379,6 +1426,12 @@ void __attribute__((destructor)) 
> lttng_ust_exit(void)
> >                 * cleanup the threads if there are stalled in a syscall.
> >                 */
> >                lttng_ust_cleanup(1);
> > +
> > +              if (ust_baddr_handle) {
> > +                              int ret = dlclose(ust_baddr_handle);
> > +                              if (ret)
> > +                                              ERR("%s", dlerror());
> > +              }
> >  }
> > 
> >  /*
> > @@ -1456,3 +1509,10 @@ void ust_after_fork_child(sigset_t 
> *restore_sigset)
> >                ust_after_fork_common(restore_sigset);
> >                lttng_ust_init();
> >  }
> > +
> > +void lttng_ust_sockinfo_session_enabled(void *owner,
> > +                              struct lttng_session *session_enabled)
> > +{
> > +              struct sock_info *sock_info = owner;
> > +              sock_info->session_enabled = session_enabled;
> > +}
> > --
> > 1.8.4.2
> > 
> > 
> 
> -- 
> Mathieu Desnoyers
> EfficiOS Inc.
> http://www.efficios.com
> 
> _______________________________________________
> lttng-dev mailing list
> [email protected]
> http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
> 
> 
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: 
> <http://lists.lttng.org/pipermail/lttng-dev/attachments/20131117/1387c63e/attachment.html>
> 
> ------------------------------
> 
> _______________________________________________
> lttng-dev mailing list
> [email protected]
> http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev
> 
> 
> End of lttng-dev Digest, Vol 67, Issue 48
> *****************************************
> 
> _______________________________________________
> lttng-dev mailing list
> [email protected]
> http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Attachment: signature.asc
Description: Digital signature

_______________________________________________
lttng-dev mailing list
[email protected]
http://lists.lttng.org/cgi-bin/mailman/listinfo/lttng-dev

Reply via email to