| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
| |
The framerelay code is seldomly used and the socket clean ups introduced
a regression. Create a testcase that will work as a user not having the
right capabilities to create raw sockets.
We have to make sure that this test is working even when not ran as root.
The easiest way to do this is to provide our own socket implementation.
This is done with dlopen/dlsym to convert the raw socket request to an UDP
one.
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
getaddrinfo returns EAI_SERVICE (-8) if that combination is used.
More information available in here:
http://sourceware.org/bugzilla/show_bug.cgi?id=15015
Reported by Holger Hans Peter Freyther.
While at it, this patch also removes hints.ai_flags = 0 as memset
to zero already happened just a bit before that.
|
| |
|
| |
|
|
|
|
|
|
|
|
| |
When building out-of-srcdir, paths such as "src/gsm" will not find any
source files. Since the Doxyfiles are preprocessed, we can prepend
@srcdir@ to fix that.
Signed-off-by: Alex Badea <vamposdecampos@gmail.com>
|
|
|
|
|
|
|
|
|
| |
When building out-of-srcdir, "../../config.h" fails to reach config.h
because the compiler is invoked in $builddir/tests/, not
$builddir/tests/timer/. Use "../config.h" instead; this also works
for in-srcdir builds.
Signed-off-by: Alex Badea <vamposdecampos@gmail.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This fixes the following complaint by autoconf 2.69-1, automake 1.13.1-1.
: configure.ac:80: error: 'AM_CONFIG_HEADER': this macro is obsolete.
: You should use the 'AC_CONFIG_HEADERS' macro instead.
: /usr/share/aclocal-1.13/obsolete-err.m4:12: AM_CONFIG_HEADER is expan
: configure.ac:80: the top level
Automake 1:1.11.3-1ubuntu2, autoconf 2.68-1ubuntu2 don't even emit a warning
without, and work just fine with this patch.
Signed-off-by: Christian Vogel <vogelchr@vogel.cx>
|
| |
|
|
|
|
|
|
|
|
| |
This is essentially http://patchwork.diac24.net/patch/271/ forward
ported to libosmovty
Original-by: Paul Jakma <paul@quagga.net>
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
|
|
|
|
|
|
|
|
|
|
|
| |
The second loop in osmo_revbytebits_buf() in src/bits.c grabs
4 bytes each iteration, which can easily go past the supplied
input in some cases.
Compiled with -fstack-protector , I get a "stack smashing detected"
in the bits test.
From: Nils O. Selåsdal <noselasd@fiane.dyndns.org>
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
|
|
|
|
|
|
|
| |
This file is created in ./configure so we shouldn't remove it with make.
Otherwise ./configure && make clean && make check fails with:
make[3]: *** No rule to make target `atconfig', needed by `check-local'.
Stop.
|
|
|
|
|
|
|
|
| |
lapd_test.c: In function ‘ms_to_bts_l1_cb’:
lapd_test.c:192:6: warning: variable ‘rc’ set but not used [-Wunused-but-set-variable]
lapd_test.c:199:1: warning: no return statement in function returning non-void [-Wreturn-type]
lapd_test.c: In function ‘main’:
lapd_test.c:355:2: warning: implicit declaration of function ‘osmo_init_logging’ [-Wimplicit-function-declaration]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This was found while implementing handover on a sysmobts. When we
receive a channel release request for a channel that was never really
activated (set_lapdm_context() was not called) we segfault in
lapd_recv_dlsap().
We now return early with -EINVAL in rslms_rx_rll() if we receive a
message that assumes set_lapdm_context() was already called.
These are:
* RSL_MT_UNIT_DATA_REQ
* RSL_MT_DATA_REQ
* RSL_MT_SUSP_REQ
* RSL_MT_REL_REQ
A test case was added to trigger the issue.
|
|
|
|
|
| |
This is required for CPUs < armv6, to access 16 and 32 values at right
memory locations.
|
|
|
|
| |
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
|
|
|
| |
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
|
|
|
| |
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
|
|
|
| |
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
| |
|
| |
|
| |
|
|
|
|
|
|
| |
Extracted from a patch by Max Suraev Max.Suraev@fairwaves.ru>
Signed-off-by: Sylvain Munaut <tnt@246tNt.com>
|
|
|
|
| |
Log message that are either too big or have the C/R bit set as error.
|
| |
|
|
|
|
|
|
| |
GCC 4.7.2 was already smart enough to see that the table is const
so there is no change in the generated assembly code. For some reason
the dispatch is still going through one relocation.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
When OpenBSC is handling more than one message at a time it is difficult
to see which log message belongs to which SMR instance. Introduce a
uint64_t id that can be set to the row_id/message_id and prefix all
log messages with SMR(ID).
This change is ABI and API incompatible with previous versions of
libosmogsm.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
When OpenBSC is handling more than one message at a time it is difficult
to see which log message belongs to which SMC. Introduce a uint64_t id
that can be set to the row_id/message_id and prefix all log messages
with SMC(ID).
This change is ABI and API incompatible with previous versions of
libosmogsm.
Example:
SMC(100) instance created
SMC(100) message MNSMS-EST-REQ received in state IDLE
|
| |
|
|
|
|
|
|
|
|
| |
msgb_pull returns a pointer to the new begin of the
buffer, unlike msgb_get(), where those functions
were originally taken from.
Signed-off-by: Steve Markgraf <steve@steve-m.de>
|
| |
|
|
|
|
| |
The code embeds the timer so one needs to include the header file.
|
|
|
|
|
|
| |
When the connection may not released print the name of the current
state to ease with debugging and verification that this is not a
valid state transition.
|
|
|
|
|
| |
This code got copied from OpenBSC and we used the network specific
and private numbering plan. Remove it as it can be easily set.
|
|
|
|
|
| |
<001c> gsm0411_smr.c:175 TX: MNSMS-REL-REQ adds little value so
I am reducing it to a debug statement.
|
| |
|
|
|
|
|
|
|
|
| |
The bits test will fail on Debian Stable for x86. Somehow the load
of a static const array is ending up accessing random memory. This
only happens when not using any -OX options and this is caused by
us. Do not override the default debian CFLAGS.. and the tests will
pass on debian stable.
|
|
|
|
|
| |
This avoids an unaligned access on platforms that don't support them,
such as ARMv4/ARMv5.
|
|
|
|
|
| |
Without the "extern" keyword the variables in this header file will be
seen as empty definitions when compiled on OSX.
|
|
|
|
|
| |
Add a check to not use --version-script linker flag if compiled on OSX
since it doesn't exist there
|
|
|
|
|
|
|
|
|
|
| |
Use FreeBSD struct iphdr definition for OSX also. From the commentary in
the source file:
On BSD the IPv4 struct is called struct ip and instead of iXX
the members are called ip_XX. One could change this code to use
struct ip but that would require to define _BSD_SOURCE and that
might have other complications. Instead make sure struct iphdr
is present on FreeBSD.
|
|
|
|
|
| |
When you are in the config-log VTY node, you expect "end" to work
like in any other sub-node of config.
|
|
|
|
|
|
|
| |
Prior to this fix, a persistent file or syslog log configuration
didn't work across an application re-start, as the
"logging filter all 1" was never saved and thus no messages were
logged.
|
| |
|
| |
|
|
|
|
|
| |
Freeing memory of registered timers and file descriptors cause corrupt
lists.
|
| |
|
|
|
|
|
| |
Add a testcase for the logging facility. The test is coming from
the OpenBSC code.
|
|
|
|
|
| |
log_parse_category_mask(), skip log category name right away if
name is NULL to prevent passing a NULL ptr to strlen.
|
|
|
|
| |
extension character before doing anything else
|