It is too easy for calling code to use the same filter and context indexes for
different filters and structs. For example, openbsc's IMSI filter and libgb's
GPRS_BVC filter both fall on index 1 even though there are plenty more indexes
to choose from. To alleviate this, have one central definition here, sort of
like ports.h does for VTY and CTRL port numbers.
Add static asserts to make sure the indexes fit in the available array and bit
mask space.
Calling code like openbsc.git and osmo-pcu need adjustments and/or should move
to using these enum values instead of their local definitions.
Taking this opportunity to also prepare for a split of struct gsm_subscriber in
openbsc into bsc_subsciber and vlr_subscriber with appropriate separate filter
index constants for both subscriber types.
Include previous LOG_FILTER_ALL in the LOGGING_FILTER_* enum, and replace its
use by (1 << LOGGING_FILTER_ALL).
Change-Id: I5c343630020f4b108099696fd96c2111614c8067
In log_set_category_filter(), passing a negative index lead to memory
corruption. Particularly dangerous since the internal logging categories have
negative values.
Fix: apply map_subsys() to interpret negative values as internal logging
categories.
As a side effect, out-of-bounds logging categories will be mapped to DLGLOBAL
instead of being dropped.
Fix the expectations in logging_test to match the fixed bug.
While at it also guard against a NULL logging target.
Change-Id: Ib0725b22bc39498c6b3970a61eb3339cf56d19f1
To check category bounds, rather use num_cat_user, to redirect all semantically
unknown categories to DLGLOBAL.
Adjust logging_test expectations accordingly: "(d)" is now also shown.
Note: subsys is and needs to be signed, while num_cat* are unsigned. Thus for a
negative subsys, 'subsys >= num_cat_user' practically always yields true. Pay
close attention to signedness and check upper bound only for positive values.
Change-Id: I4a952b759f30d90fbfb81fedcfc56a8092ea18c1
In map_subsys(), fix the '>' condition to '>=' for array bounds checking.
Also make the bounds checking more strict: after both invocations of
subsys_lib2index(), re-check validity of the array index. If the final index is
still wrong, which should never happen, exit by assertion.
Change-Id: I7ca1a1d47724e40350f1c4dfebe90bad01c965f9
For out-of-bounds logging categories, redirect to the proper DLGLOBAL array
index instead of returning -1.
Adjust test expectation which shows that the bugs tested for are fixed.
Note: there are separate bounds checking problems, left for another patch.
Change-Id: I6ea9a59e005a22e0305454291714fdb9531c346b
In the process, also:
* Change the license from AGPLv3 to GPLv2-or-later;
* correct spelling of 'sysmocom' to lowercase;
* add '2016' to the copyright;
* rename to osmo_*;
* add API docs;
* add logging category DLOAP: define id and add to internal_cat;
* redirect all oap.c logging to DLOAP.
A unit test will follow in a subsequent patch, since it needs a minor tweak for
decoding of boolean values.
The related openbsc change-id is I2f06aaa6eb54eafa860cfed8e72e41d82ff1c4cf.
Tweaked-by: Neels Hofmeyr
Change-Id: If5099e60681a215e798b6675f21813f26769c253
All DL* categories are typically negative, but DLGSUP isn't, and it's also not
in libosmocore's internal_cat array.
See: 3b6fb0880c
This means that a program using DLGSUP has to include DLGSUP in its own logging
cat array (typically not needed for DL* categories), which means for osmo-nitb
that DLGSUP (11) replaces DMGCP (also 11), and DMGCP becomes unusable.
Fix this: make DLGSUP -11 and include in internal_cat.
In gsup_test.c, no longer add DLGSUP to the logging categories array.
External follow-ups are otherwise needed only in osmo-hlr.git and some pending
patches for openbsc (Id3938267fa062e1a997d3704cd678874306f86ee).
Change-Id: Id974c7be158e4d60421a98110f5c807aefd31119
This target wraps the to-be-logged string (With metadata) into a GSMTAP
packet and sends it to the configured destination address.
Change-Id: I9a7e72b8c9c6f6f2d76d1ea2332dcdee12394625
Some targets might not want to receive only an opaque, pre-formatted
string, but rather the unformatted arguments with metadata like
sub-system/level/file/line. We solve this by introducing a
log_target->output_raw() function pointer. If a target specifies this
function, it takes precedence over the regular log_target->output()
function.
Change-Id: I9dc9205d70dce9581458e7e9dc2d8a92991897bd
There doesn't seem to be a reason not to check the filter. Update
and extend the test. Currently the filter function will be called
once for the log check and once for the output of it.
This commit adds this predicate function which can be used to
avoid the execution of code if a certain log level is not enabled.
The function will only return 0 (false), if it is sure that a logging
call for the same facility and level will not produce any output.
This safety criterion shall ensure, that no logging output is lost
due to the use of this predicate as a guard. On the other hand, even
if the predicate returns != 0 (true), no logging output might get
generated by a similar logging command.
Note that the current implementation is not focussed on performance,
which could be improved by using a lookup table instead of iterating
through every target.
Sponsored-by: On-Waves ehf
This log level is used by the stats subsystem log reporter to report
statistics to level INFO. Note that the default level of DLSTATS is
NOTICE.
Sponsored-by: On-Waves ehf
Currently when using 'logging print extended-timestamp 1', the
subsecond part (milliseconds) of the printed timestamp is always 0.
This makes it difficult to correlate log entries with PCAP file
entries if there are many of them per second.
This patch changes _output in logging.c to use gettimeofday() instead
of time() when extended timestamps are enabled and replaces the '000'
by the milliseconds computed from tv_usec.
Sponsored-by: On-Waves ehf
We tried to fix it but it isn't that easy. The original fix was
cd6ed82d1f but we had to revert it
as "everything" is present/used in existing config files.
If we ever change the ABI we can make everything be something
that is > 0. For now use a wording that makes it obvious that
people should not use "everything".
We want to see from which category/subsystem a certain log message
is coming from and use a different timestamp format as well. Add
two new bitfields. This doesn't change the size of the structure
and on 32bit we still have 27bits left.
The extended timestamp will take preference over the current and
default timestamp format.
Fixes: SYS#602
The filter_fn has not been copied into the new structure breaking
the imsi and other filters in OpenBSC. Looking at the code we should
also introduce a callback for the reset of the context so we could
use subscr_get/subscr_put on the subscriber structure.
Commit cd6ed82d1f made "EVERYTHING"
map to LOGL_DEBUG but when writing out the configuration the following
would be written:
logging level all unknown 0x0
This happend because no string was found for the value 0. Address it
by adding a legacy check for 0 and write out the str from the index
0. Currently this is "EVERYTHING".
As Holger points out "logging level XXX everything" wasn't working, as
it sets category->loglevel to 0, which is checked in osmo_vlogp() and
will never get logged.
Introduce a print_filename attribute for each logtarget. Initialize it
with 1 to be backward compatible with earlier versions. The bit is taken
from an existint bitfield. There were at least six bits left of the byte.
Applications should keep the log area in a static const area. Mark
the pointer as const to address compiler warnings in OpenBSC, cast
the const away for the osmo_log_info as it is not declared as const.
Make sure the declaration and definition match, add const to
the functions called by logp/logp2.
Compile output:
logging.c:317: error: conflicting types for 'logp'
../include/osmocom/core/logging.h:34: note: previous declaration of 'logp' was here
logging.c:327: error: conflicting types for 'logp2'
../include/osmocom/core/logging.h:168: note: previous declaration of 'logp2' was here
make[3]: *** [logging.lo] Error 1
* All loging prefixes in libraries should be DL like DLINP
* All signals / subsystems should be called S_L_* SS_L_*
* All command nodes should be called L_*_NODE
This makes sure existinc code still compiles as expected
This adds the libosmo-abis logging subsystems to libosmocore,
it uses the new change that harald proposed based on negative
numbers for library logging subsystems.
In 8256076722 it was attempted to fix
a bug previously introduced by logging related changes. The problem
is that a va_list can be corrupted after it has been used once, so
we need to va_copy before each successive use.
And if we copy it, we also need to use the copy, and not the original ;)
We do this by using a trick: library-internal log categories use
negative subsystem numbers, which are converted into positive
array indexes at the time of logging.
library-internal log categories need to be knwo at compile-time,
while application-specified categories now are of unlimited number,
as they are dynamically allocated.
Harald reported a problem in the logging:
http://lists.osmocom.org/pipermail/openbsc/2011-May/002896.html
Reverting 81e9636454 seems to
fix the problem. However, that workaround looks ugly.
Holger gives us another clue on what was wrong:
http://lists.osmocom.org/pipermail/openbsc/2011-May/002905.html
While digging in the manpage, I found this:
"The functions vprintf(), vfprintf(), vsprintf(), vsnprintf()
are equivalent to the functions printf(), fprintf(), sprintf(),
snprintf(), respectively, except that they are called with a
va_list instead of a variable number of arguments. These functions
do not call the va_end macro. Consequently, the value of ap is
undefined after the call. The application should call va_end(ap)
itself afterwards."
This patch removes a workaround to fix some strange memory corruption
now that _output() has been completely reworked and we make use of
snprintf appropriately.
This patch reworks _output() to handle snprintf() return value
appropriately and to use one single buffer to build the logging
string, instead of four.
OpenBSC> logging level
all Global setting for all subsystems <----- this description was missing
rll A-bis Radio Link Layer (RLL)
[...]
This problem was introduced by myself in:
"vty: integration with logging framework"
04139f14b6
This patch moves all GSM-specific definitions to include/osmocom/gsm.
Moreover, the headers in include/osmocore/ have been moved to
include/osmocom/core.
This has been proposed by Harald Welte and Sylvain Munaunt.
Tested with `make distcheck'.
Signed-off-by: Pablo Neira Ayuso <pablo@gnumonks.org>
The logging categories are registered by the applications, like
osmo-nitb, during startup. Thus, the VTY have to provide the logging
commands according to the logging categories that the application
has registered.
Before this patch, the VTY contained the logging categories
hardcoded. Thus, any change in the logging categories by the
application would not be shown by the VTY.
So far, this was not a problem because all applications used the
same logging categories. However, according to what Harald told
me, this may be a problem in the future.
This patch resolve the lack of integration between the logging
framework and the VTY by generating the VTY logging commands
from the logging categories registered.
Since this patch changes one function of the libosmocore API,
it follows another patch for the openbsc application to get in
sync with the new function layout.
I have reworked and renamed the functions:
* log_vty_category_string()
* log_vty_level_string()
to provide the new ones that generate the exact output that VTY
requires.
This patch does not release the memory allocated by
talloc_zero_size() to store the VTY strings for the commands
and the description. I found no exit function that can clean
up resources that were allocated.
Before this patch, there was a bug in the code caused by a memcpy
from one data structure to another. unfortuantely the data structures
were not the same, so we have to explicitly iterate over the array
and assign the structure members manually.