As discussed in Gerrit change I9c3bf64537ef2223e29f8082861fa32fde26bf68
remove defines that don't serve any purpose. These are defines for
reserved values and changing them later if a newer spec defined them
would break API.
Keep the comments to explain the missing values.
In some cases the phone requests a PDP context type that isn't available
no the PGW/GGSN, e.g. phone requests a combined IPv4/v6 context, but
only IPv4 is supported.
In that case the GGSN can send a Create PDP Context Response with cause
"New PDP type due to network preference" or "New PDP type due to single
address bearer only". libgtp should continue handling these cause values
like the "Request Accepted" cause. Use the new gtp_cause_successful()
function for that.
According to the spec the upf/pgw can accept a modified pdp context from
the request e.g. if an ipv4/6 context was requested, but only ipv4 is
availiable. Introduce a function that checks all cause values that are
See also: 3GPP TS 29.060 Ch 7.3.2
This option should be used for any executables which are used only
for testing, or for generating other files and are consequently never
installed. By specifying this option, we are telling Libtool that
the executable it links will only ever be executed from where it is
built in the build tree. Libtool is usually able to considerably
speed up the link process for such executables.
When using built-in static_assert() , gcc v12.2.1 fails:
In file included from gsn.c:27:
gsn.c: In function 'gtp_new':
gsn.c:444:54: error: expression in static assertion is not constant
444 | osmo_static_assert(gtp_T_defs.default_val != 0, first_default_val_not_zero);
The reason is likely that gtp_T_defs is not const, so it cannot
be assert()ed statically. With the current osmo_static_assert()
implementation, this assert does nothing. One can change the
gtp_T_defs.default_val to 0 and the code will still compile.
Related:  libosmocore.git I5ca34bc14c05e8c38c721d7df33feb1c6c41c76e
List the most recent library version where new symbols where added, so
debian properly upgrades libgtp6 when upgrading osmo-sgsn from the
version that is currently in Debian to a version from the Osmocom
This timer controls the amount of time a resp message transmitted by the
local gsn is to be stored in the resp queue. This is used in order to
detect duplicate requests received, since GTP states the exact same
response should be answered if a duplicate request is received.
Prior to this patch, this timer was hardcoded to 60 seconds.
This patch actually should be set, in general, to a value
equal than (T3-RESPONSE * N3-REQUESTS) values configured at
the peer, since that is the maximum period during which the local gsn
expects to receive req retransmissions from the peer.
Hence, this value must be user configurable to adapt it to the peers
connected to the GSN.
The 60 seconds hardcoded value is therefore changed to default to our
local (T3-RESPONSE * N3-REQUESTS), since the most common scenario for
osmo-ggsn/osmo-sgsn is to run it against a peer osmo-sgsn/osmo-ggsn,
which will have the same values by default.
This way we avoid by default caching response messages for way too long,
potentially filling the queue.
By default systemd will execute service with root directory (or home directory for user instance) which might result in
attempts to create files in unexpected place. Let's set it to 'osmocom' subdir of state directory (/var/lib for system
When using 'check_PROGRAMS', autoconf/automake generates smarter
Makefiles, so that the test programs are not being compiled during
the normal 'make all', but only during 'make check'.
QUEUE_SIZE holds the number of pending transmitted messages
which can be handled concurrently.
Current value was 1024, same as PDP contexts (PDP_MAX). However, that
seems to be a quite low amount, which can be filled under certain
conditions, for instance if recovery procedure is triggered on the GSN
which is running full (around PDP_MAX pdp contexts created).
In this scenario, the GSN would need to send around PDP_MAX concurrent
messages (DeletePDPContextReq), which means the queue would very likely
end up being full.
Hence, let's define QUEUE_SIZE based on PDP_MAX, and set it to twice the
size to make sure it won't be filled in extreme conditions.
osmo-ggsn crashes when concurrent pdp context num 1024 is created, due to
the gsn->pdpa array (of size PDP_MAX, 1024) being full.
The crash happens because return code of gtp_pdp_newpdp was not checked,
and hence a pointer "pdp" pointing to a temporary not-fully-allocated
object was being passed to gsn->cb_create_context_ind() callback.
Let's avoid crashing and instead reject the PDP context.
Coverity warns around that test code:
"Overrunning struct type in46_addr of 20 bytes by passing it to a
function which accesses it at byte offset 39."
That's basically because in64a_from_eua expects to be passed a 2 element
array to be filled. The second element, though, is only accessed in the
case where an IPv4v6 EUA is passed, so in the cases where the test
explicitly passes an IPv4 or IPv6 EUA it's not really an issue, hence
coverity throwing a false positive here.
Let's anyway rewrite the code to pass a 2 element array for completeness,
since it doesn't hurt and makes coverity happy.
Related: Coverity CID#249006
from "man strtoul":
If endptr is not NULL, strtoul() stores the address of the first invalid character in *endptr.
In particular, if *nptr is not '\0' but **endptr is '\0' on return, the entire string is valid.
This is a small standalone program (under MIT license, hence cannot make
use of libosmocore) whose only purpose is to answer GTPC (v1 and v2)
Echo Request messages with Echo Reply ones, with information provided
from the command line.
A small python script companion is provided to easily test the program.