Common bits shared by various functions (currently setting
non-blocking) should not be copy+pasted around.
Change-Id: I95056940ddc26b65f63eedaeaab6882edaef6317
In some situations we want to set the SO_PRIORITY socket option
to determine the in-kernel priority of packets generated by this
socket.
Change-Id: I89abffcd125e6d073338a5c6437b9433220e1823
Related: SYS#5427
DSCP is a 6-bit value (0..63) stored in the upper 6 bits of what was
formerly known as TOS bits. We must
* make sure the user can only specify 0..63
* shift the value by two bits when using the IP_TOS socket option
We achieve the latter by using the recently-added osmo_sock_set_dscp()
helper.
Change-Id: I64fee56b04d0ecd128bf661699d5071817ea96ec
Closes: OS#5136
At least on Linux, sockets have a IP_TOS socket option that can be
configured to set the TOS. However, TOS (of RFC791) was replaced
by the DSCP (of RFC2474) in 1998.
As the DCSP bits are only the upper 6 bits of the TOS bits, let's
introduce a helper to get, mask and set the DSCP values in the TOS
bits.
Related: OS#5136, SYS#5427
Change-Id: Ia4ba389a5b7e3e9d5f17a742a900d6fd68c08e40
When doing a "show ns", let's also dump the state of the frame
relay network, with all its links and DLCs (if any).
Change-Id: I798af3e97dc014b6e0fcde86560a1809852f7510
Related: OS#4877
Defined in 48.018 10.5.2.82.
This will be used by Channel Mode Modify for VAMOS.
Related: SYS#4895 SYS#5315
Change-Id: I9bad6e7121af43dfa9706635e58279ce672a4e14
Also add functions to convert between VAMOS and non-VAMOS speech modes.
Related: SYS#4895 SYS#5315
Change-Id: Ie0ea592da5610ae70290106d004e549cf3212a89
e7dfeac8dc introduced a regression in the block/unblock check
as it was using the priv->initiate_block instead of priv->om_blocked.
The initiate_block tracks who is responsible to unblock the NSVC.
Fixes: e7dfeac8dc ("gprs_ns2_vty: print a response to vty `nsvc <nsvci> (block|unblock|reset)")
Change-Id: I516faea223e30b120a297faed10636daa554be8a
The library specific sub-systems are kind of special, because their
position in the 'osmo_log_info' may vary depending on the number of
application specific sub-systems. This is why their associated
constant values (like DLGLOBAL) are negative, and this is what
the LOGP() macro expects as the first argument.
Before this change, invoking 'logp' command with any library
specific logging sub-system would result in getting messages
printed with the fall-back DLGLOBAL sub-systems.
Change-Id: If86563e169fe1243adfa7b09c9d65d9f88c8a99e
This will be used by osmo-bts-omldummy to parse features strings from
the cmdline.
Note that osmo_bts_feature_name() already exists to return the longer
descriptive value_strings from osmo_bts_features_descs (_descs!).
Luckily that misses the plural 'features' in the name, so that I can
still add a properly named osmo_bts_features_name() function that only
returns the name, matching the common pattern used in osmocom code.
Related: SYS#4895
Change-Id: I699cd27512887d64d824be680303e70fff3677c1
Fix counting of values missed because of FIFO overflow in
osmo_stat_item_get_next(), by assigning a new item value id effectively
as item->value[n + 1].id = item->value[n].id + 1, instead of increasing
a global_value_id that is shared between all items and groups. With
global_value_id, the count of values missed was wrong for one item, as
soon as a new value was added to another item.
This partially reverts b27b352e ("stats: Use a global index for stat
item values") from 2015, right after stats was added to libosmocore. It
was supposed to make multiple readers (reporters) possible, which could
read independently from stat_item (and later added comments explain it
like that). But this remained unused, stats has implemented multiple
reporters by reading all stat_items once and sending the same data to
all enabled reporters. The patch caused last_value_index in struct
osmo_stat_item to always remain at -1.
Replace this unused last_value_index with stats_next_id, so stats can
store the item-specific next_id in the struct again. It appears that
stats is the only direct user of osmo_stat_item, but if there are
others, they can bring their own item-specific next_id: functions in
stat_item.c still accept a next_id argument.
Related: OS#5088
Change-Id: Ie65dcdf52c8fc3d916e20d7f0455f6223be6b64f
A SNS configuration can be done over a NSVC, however this initial NSVC doesn't
need to be part of the configuration.
Those NSVC need to be removed when the configuration is done.
This wrong behaviour can be seen in the vty `show ns` on
NSEI 00001: UDP, ALIVE
FSM Instance Name: 'GPRS-NS2-SNS-BSS(NSE00001-SNS)[0x55c72c09b420]', ID: 'NSE00001-SNS'
Log-Level: 'DEBUG', State: 'CONFIGURED'
Maximum number of remote NS-VCs: 8192, IPv4 Endpoints: 8192, IPv6 Endpoints: 8192
Local IPv4 Endpoints:
10.0.0.1:23000, Signalling Weight: 1, Data Weight: 1
Remote IPv4 Endpoints:
10.0.2.2:23000, Signalling Weight: 1, Data Weight: 0
10.0.2.2:23001, Signalling Weight: 0, Data Weight: 1
3 NS-VC:
NSVCI none: UNBLOCKED DYNAMIC data_weight=1 sig_weight=0 udp)[10.0.0.1]:23000<>[10.0.2.2]:23000
NSVCI none: UNBLOCKED DYNAMIC data_weight=0 sig_weight=1 udp)[10.0.0.1]:23000<>[10.0.2.2]:23001
NSVCI none: UNCONFIGURED DYNAMIC data_weight=1 sig_weight=1 udp)[10.0.0.1]:23000<>[10.0.2.2]:8888
The UNCONFIGURED NSVC should not be present in when SNS is in CONFIGURED.
Related: SYS#5416
Change-Id: I4045ac6c033ae084743b17a16eef4fcff76589b9
The SNS fsms also track the NSE however since the NSVC
starts now in ALIVE for SNS the SNS must check when synchronize
the alive state when entering the ST_CONFIGURED.
Related: SYS#5416
Change-Id: Ib6a1cc1fd84959e69c07b72ef780642205d2cd18
The start_procedure() can't be called after ns2_nse_notify_unblocked()
because ns2_nse_notify_unblocked() might free the nsvc.
Otherwise the fsm will do use-after-free on the NSVC memory.
Related: SYS#5416
Change-Id: If97dfd123eefd71fc6c3fe886a243a21784aeeb4
Let osmo_stat_item_get_next, osmo_stat_item_discard,
osmo_stat_item_discard_all consistently refer to their next_id arg as
such (and not idx or next_idx). It refers to an ID (item->values[i].id),
not an index (item->values[i]), and it is always the next one, never the
current one.
Do the same change for _index/_idx variables in stats.c, which are used
as arguments to these functions. Replace rd_ with next_id_ in
stats_test.c, too.
Related: OS#5088
Change-Id: I5dd566b08dff7174d1790f49abd2d6ac020e120e
Prior to this patch, we would unconditionally allocate new memory
for the local SNS IP endpoints. This results in a memory leak
on every SNS-SIZE procedure.
Let's move to talloc_realloc() which recycles any previously allocated
memory.
Change-Id: I12cb670e087c6d6190f3f5bf8483ea62008ae06f
We may very well have any number of binds configured, but which
are not part of the current NSE. When creating the "full mesh" of
NS-VCs after SNS-CONFIG, we must only iterate over those binds that
are part of the NSE (or 'ipa-sns-default bind' in cae of SGSN role),
but not over all the other binds that may exist in the system.
Closes: OS#5092
Change-Id: Ida361fa02ad1d86844d54c8f0664c996ed28e30a
dump_nse() should only print persistant NSE when persistant_only
argument is set. A NSE can be persistant or dynamic configured.
Depending on the NSE all NS-VC must be the same as the NSE.
Change-Id: Ibd2c6962eda39a850ab61cf347063934378d2fdc
The binds also print a list of associated NSVC when
dumping the bind.
However the binds using their own representation of
printing the NSVC which is different to `show ns entities`.
Use the same function to print NS-VC.
Before:
NSVCI 00000: udp)[127.0.0.1]:23000<>[127.0.0.1]:22000
After:
NSVCI none: UNCONFIGURED DYNAMIC data_weight=1 sig_weight=1 udp)[127.0.0.1]:23000<>[127.0.0.1]:22000
Change-Id: If31ec6c1c07dc134ab1ddeb915bc89747c7be048
Introduce 2 new logging sub systems for signal and unit data.
Unify log messages so all log messages look similiar.
Log also Rx PDUs. Ensure dropped Tx packets (BLOCK/RESET on SNS)
contain *Tx*.
Change-Id: I34b8fde2955ecc010d1dcd9512e1bba9211e2c0d
This fixes some bug introduced in Change-Id I1638f04ba45fef3ba0b237948dff6022267141fb
Related: OS#3373
Change-Id: Ic0873e63f1f046b674c7898480ff070a18a7abc7
the previous commit fixed the state transition, but also caused
neither Tns-alive nor Tns-test to run
Change-Id: I07ea3fd5cec3d4acf4051930a1a3c746d6fd597c
In the IP-SNS SGSN role, we need to inform the BSS of our local
IP endpoints. For statically configured NSEs, those are explicitly
stated on a per-NSE level.
For dynamically created IP-SNS NSEs, we are adding a new VTY
command, using which the administrator can configure which binds
should be advertised as IP endpoints to such BSS.
Change-Id: Id01c29b07e9203c9305f2129361a4f5aaefa2c52
Related: OS#3373
As per 3GPP TS 48.016 section 6.2.4.1, we need to perform some
consistency checks during the SNS-SIZE procedure. Let's implement them.
Change-Id: I1638f04ba45fef3ba0b237948dff6022267141fb
Related: OS#3373
It is the SGSN's job to ensure sufficient NS-VC capacity. As the SGSN
doesn't tell the BSS, we should not make assumptions of only 4.
Change-Id: I41f493643cf51d7853959ab9c7bbc0ffae4e1f4b
The SNS-SIZE sent from BSS to SGSN contains the actual number of local
IP endpoints on the BSS side, and not the maximum number of remote IP
endpoints supported.
Change-Id: I62a8bca4a3f7c47bcb9f292b045fa867d8877a09
In BSS role, we can clear local + remote endpoints when sending
SNS-CONFIG, as we are first.
In SGSN role, we must not clear remote endpoints when sending
SNS-CONFIG, as we are last, and the BSS has just previously told
us its IP endpoints in the BSS-originated SNS-CONFIG.
Change-Id: I58549707ac5a3a0aae5f9348ed76f16c09ad3e46
Related: OS#3373
gprs_ns2_create_nse() doesn't allow the caller to specify if the
BSS or the SGSN role of IP-SNS shall be implemented. Add
gprs_ns2_create_nse2() to fix that.
Change-Id: I6db8c36f7c69b592d7d0fbcf323804f7e9912be2
Related: OS#3373
Clearly separate those states that only occur on the BSS side from
those of the SGSN side (which already have a prefix)
Change-Id: Iebc97637063b2676abff4a7dce50272d9b4e336b
Related: OS#3373
Some functions are only to be called when operating in BSS role,
let's enforce that by sprinkling some OSMO_ASSERT() around.
Related: OS#3373
Change-Id: I20d1bd8df62e0bb84696e291756e2d96acf49246
This adds the core functionality of implementing the SGSN-side of IP-SNS
inside a new FSM. The SGSN-side FSM shares states, events, structures
and code with the BSS-side whenever possible.
Note that this commit does not yet use the SGSN-side FSM, that is part
of subsequent commits.
Closes: OS#3373
Change-Id: I780413ffacd9cb71f64cc38057aae781eeb59b2b
This reverts commit d290439b4a, which
caused "stats values skipped" messages to appear even if they were not
skipped. Revert for now, replace with a proper version in the future.
Related: SYS#4877
Change-Id: Ib43bd53188a4d31d771feb921ea14abe1a3ec877
Let the user know when the stats were not consumed fast enough for the
given FIFO length.
Related: SYS#4877
Change-Id: If0e8ab55103007693101538fb6ea310075217774
Setting the NSE dialect possibly involves the creating (or destruction)
of a IP-SNS FSM. Encapsulate that, rather than having every caller
re-implement that.
Change-Id: I24fdc26fbcfda039bd58ea166f4d5c2fd1801da1
We should never report multiple values for a metric. It is confusing for
the log reporter and wrong for statsd. Statsd will record only one value,
but will it be the first, last, ...?
This can happen if an osmo_stat_item changes more than once within the
same reporting interval.
With this patch only one aggregate value is sent to the log reporters.
The value reported is the maximum during this interval. Other
aggregations could be possible (min, last), but reporting a (useful)
average is not because the values don't include a timestamp and most
osmo_stat_items change at unregular intervals.
Change-Id: I366ab1c66f4ae6363111ea4e41b66b7d5bcade9c
Related: SYS#4877
Let's move computing of the local IP endpoints to a separate function,
so it can not only be used when entering the SNS_SIZE state. Preparation
for SGSN-side IP-SNS.
Change-Id: I5d7ce419135a8ef538cf9abcb76a49049ed7d5f9
Related: OS#3373
We don't really handle inbound SNS-ACK yet (as we don't originate
SNS ADD/DEL/CHGWEIGHT procedures yet). However, the message
receive code should already dispatch those events to the FSM,
just like we do for the (equally unimplemented) handling of inbound
SNS-SIZE which is already dispatched to the FSM.
Change-Id: If6d5f96d85e6f05534fe49376f4473d912535ad4
We already have similar log statements for all NS-{RESET,BLOCK,UNBLOCK,...}
and it's confusing that even a debug-level NS log file doesn't show
transmitting SNS messages at all.
Change-Id: I0d31c3911d3acd6ac7296c370e920d53412d8289
When freeing the NS-VC there are cases when the NSE is still alive.
This means that gprs_ns2_free_nsvc() calls ns2_prim_status_ind(),
which in turn calls ns2_count_transfer_cap(). The latter must deal
with such a situation rather than OSMO_ASSERT()
Change-Id: I4f32dee705c9886717339fe7cce5b4cd8d2c18f7
To support SGSN oriented RESET introduce a role flag to
track what's running the gprs_bssgp (local side).
Related: OS#3879
Change-Id: Ibcbaffa94cbdc4296a8a7c372304ac11d50d9559
The warning period encoding was wrong, resulting in way too short
warning periods being encoded than intended/specified by the caller.
Change-Id: Idf3cae48a6ab45550d7bbd937bb49a0e1a4e8aed
Even though the value is only between 0..120s, they didn't encode
it 1:1 in the uint8_t, but 3GPP chose to use the same encoding
as for the warning period (which has a much larger range).
Let's fix this in our implementation.
Before this patch, osmo-cbc wanted to send 30s keep-alive repetition
period, but a spec-compliant receiver actually decoded this as 80s.
Change-Id: I04baa6b6b99b092fa0512b3b6138a363c7f3a13d
We currently only have probes for the logging sub-system.
This patch adds two tracepoints for tracing the performance
impact of statistics reporting: stat_start and stat_done.
They can be used to trace the amount of time a libosmocore-using
application spends in reporting/exporting statistics. This includes
both the CPU time for encoding the statistics, as well as the system
calls for sending them.
Change-Id: I7208c45f6d051505dd2435305c67b4d26c0b1dd2
Related: OS#4311
Related: SYS#4877
If the SNS fsm isn't freed early, the SNS code will re-create a NSVC
when calling free_nsvc().
Fixes libasan heap-use-after-free.
Change-Id: If350df1d8d6dcea5715dd23b8bd1d684098cdb1f
A NS Status can contain the original NS message which might result
in a NS PDU which exceeds the MTU of the NS-VC.
Truncate the original message to the maximum possible.
Based on truncate BSSGP status message.
Related: OS#4889
Change-Id: I35d8f8bf0eae890f4db56423da0b23b638d24311
When the MTU of the frame relay device changes, update the bind
and notify all NSEs.
Related: OS#4889
Change-Id: I946f7655c9526ffd98dabdce219c6a419b71e00c
An event which originates by a received PDU is prefixed by RX.
An event which originates by code gets a REQ prefix.
Fixes: OS#5014
Change-Id: Ia8a6378cdca19b086e89058b1cc055f45c0bba7b
Implement the load sharing based on modulo of the LSP. As long the gprs_ns2 doesn't
support the resource distribution function (48.016 § 4.4a) this simple
approach is good enought.
Fixes: OS#4836
Change-Id: I8c2fe5d647694886ac600470fca6ea5d5d210a85
Introduce a `ip-sns-bind BINDID` vty command within a `nse` vty object.
The ip-sns-bind defines the binds which will be used by the dynamic
configuration with IP-SNS.
This is only the first part which only uses the binds when doing a
new SNS configuration.
The outgoing add procedure will be supported in a later patch
when the SNS fsm supports outgoing procedures.
This is a behaviour change of the API and must be synchronized with
the osmo-pcu. Otherwise SNS won't work with osmo-pcu.
Related: SYS#5354
Change-Id: I9ab8092bf286e7d90e92f5702a5404425e959c84
This allows differentiating threads withing an application, while still
keeping same numbering for single-threaded application (since first
thread ID is always the same as the process group ID).
Related: OS#5027
Change-Id: I33da02524fc064e133b2b762af7060139c4cfd81
Recent commit filling this field forgot to convert it to network byte
order.
Related: OS#5027
Fixes: bb149ecda2
Change-Id: I50857f35cb28138fa6f28100afeaa00f492f303a
This API wraps conventional gettid() linux-specific API, which even in
Linux itself is sometimes not properly supported/announced.
This API also allows future porting to other platforms if needed, and so
far falls back to getpid() if no gettid(9 can be found.
Code ported from osmo-trx.git, see commit 7a07de1efd4eb7cc11c33d3ad25cb2df70aa1ef1.
Related: OS#5027
Change-Id: Id7534beeb22fcd50813dab76dd68818e2ff87ec2
It was recently discovered that PID field in gsmtap log messages was
always set to 0. Before this patch, the field was never being set.
The approach of this patch is to record the PID of process one, in
order to avoid calling getpid() syscall on each
log line to be sent. The counterpart of this optimization is that
eventual fork() calls would still keep the old incorrect value, but I
think nobody can safely assume that fork() is possible once all this
kind of infrastructure has already been configured (fork() should only
be done really at the start of the program before any osmocom foo is
initialized, or to immediatelly call exec()).
Related: OS#5027
Change-Id: I7db00d1810f0860166bffa0bda8566caa82e06a9
The BSSGP layer needs to know the MTU of the NS UNIDATA payload.
The MTU can be 0 if the NSE doesn't contain any NSVC.
Every status indication will contain the mtu value.
The MTU in the status indication contains the maximum transfer
unit of a BSSGP message. From NS side the maximum SDU.
Related: OS#4889
Change-Id: I5016b295db6185ec131d83089cf6c806e34ef1b6
CGI-PS type doesn't exist in GSM 08.08 Cell Id lists. That type of cell
id is osmocom-specific and used internally. In here CGI-PS is
automatically converted to CGI (since the later is an extension of this
one).
The encode/decode_cell_id_u are left intact (comment added) since those
can still be used (and are used by RIM code) to encode/decode TS 48.018
Cell Identifiers.
Related: SYS#4909
Change-Id: Id74f4577c397c1ba696f00395858311bd82cb2c8
It's a static internal function, so it makes sense to have it at start
of its related section.
It will be used by other functions in follow up patches.
Change-Id: I60f61f8f7bb6543feb068bdcee76d3b752565c95
If the BSS (or SGSN) has sent a BVC-RESET PDU for a BVCI to the SGSN (or
BSS) and is awaiting a BVC-RESET- ACK PDU in response, but instead
receives a BVC-RESET PDU indicating the same BVCI, then this shall be
interpreted as a BVC-RESET ACK PDU and the T2 timer shall be stopped.
Related: OS#4974
Change-Id: I4d15733f9f205cb563b66ef9e41dc8df50151900
The log line sneaked in when fixing the alive ms
Fixes: ab0e8646c4 ("gprs_ns2_vc_fsm: use CLOCK_MONOTONIC for alive elapsed timer")
Change-Id: Iffe367b240f47c39232bbc26991c19752a1c75ad
bssgp_bvc_get_features_* are fsm "methods" and the name should indicate
that just lika all other function names in bssgp_bvc_fsm.h
Change-Id: I30fbbe36cdabf9635eaf4dfb1e93c8ce0f667b39
Add functions to get/set the maximum supported BSSGP PDU size by the NS
layer.
IPv4 and IPv6 should not matter since we can just enable IP
fragmentation and send NS PDUs up to 2**16 + bytes. Frame relay does not
support fragmentation and this is the reason we need to be aware of the
maximum PDU size. Luckily with 1600 bytes the MTU in frame relay can hold a
regular IP packet including NS/BSSGP overhead.
On the NS layer this corresponds to the size of an NS SDU in NS-UNITDATA
(3GPP TS 48.016 Ch. 9.2.10)
Change-Id: I9bb82ead27366b7370c9ff968e03ca2113ec11f0
Related: OS#4889
It seems like we still don't have NS2 VTY tests running in libosmocore
so this only got caught once osmo-sgsn/osmo-gbproxy builds failed.
Change-Id: Id3cd407b05457a4703ee38c4b1b1b65800bbd30e
Related: OS#4887
The alive elapsed timeout was only set once on the start of the
test procedure but not every time an ALIVE PDU was sent.
Fixes: OS#4997
Change-Id: I029696dfff21919f97ac4c33cdd82162b5ab1555
gettimeofday can jump and the comment says it should not be used for elapsed timer.
Related: OS#4997
Change-Id: I41989d8f9f82f4d1f7b97f11577653699365c8ae
Allow to assign a signalling and data weight to UDP binds.
Those weights will be used when doing dynamic configuration over
IP-SNS.
This is only the first part which only uses the assigned weights
when doing a new SNS configuration.
The outgoing change weight procedure will be supported in a later patch
when the SNS fsm supports outgoing procedures.
Related: SYS#5354
Change-Id: I5133e4229377d44772a9af28628a2bc420fea34b
<0026> gprs_ns2_fr.c:515 BIND(hdlcnet1) Can not create AF_PACKET socket. Are you root or have CAP_NET_RAW?
=================================================================
==3872359==ERROR: AddressSanitizer: heap-use-after-free on address 0x6130000030c0 at pc 0x7fef120aa92e bp 0x7ffebf6b5c20 sp 0x7ffebf6b5c18
READ of size 8 at 0x6130000030c0 thread T0
#0 0x7fef120aa92d in osmo_fr_link_free (/usr/local/lib/libosmogb.so.11+0x16992d)
#1 0x7fef1205105a in free_bind (/usr/local/lib/libosmogb.so.11+0x11005a)
Change-Id: I23c0f1697edd5734085fa18b0a2f253c0f206c53
The followign happens if osmo-gbproxy is started without CAP_NET_RAW:
<0026> gprs_ns2_fr.c:515 BIND(hdlcnet1) Can not create AF_PACKET socket. Are you root or have CAP_NET_RAW?
gprs_ns2_fr.c:176:2: runtime error: member access within null pointer of type 'struct msgb' AddressSanitizer:DEADLYSIGNAL
the second line is free_bind() iterating overr the backlog while
destroying the not-yet-fully-initialized bind.
Let's make sure the backlog llist_head is always initialized properly.
Change-Id: I4d2fa50955c5897cd469fee68d4ddc65a9f5688f
Move it above the place where the bit is set, since the bit represents
whether Extension Information is available, not whether R99 is
available.
Change-Id: Ice592acc50a24efd7fe4cf1a91f1d48fd74f38d8
In prepration to introduce more commands e.g. ip-sns-bind rename the ip-sns-remote
Related: SYS#5354
Change-Id: Ida979f3b9daa5f7280a629441e4006a7635653b0
The SNS must know when all NS-VC have failed. Further more
there might be a corner case when the SNS configuration succeeds but
no NS-VC comes up afterwards.
Related: OS#5355
Change-Id: Ie72da9adeefe0c2850d49a9208b2d0a4556f9101
When writing to the AF_PACKET socket, we have to distinguish the
pseudo-errors like -ENOBUFS (where we do want to add to the backlog)
from real errors like -ENETDOWN, -EMSGSIZE, ... where we don't want
to add the failed packet to the backlog.
Change-Id: Ibbb6805da0f118466c4c91e458e62b63b84cb794
This patch improves the behavior of the newly-added backlog in
situations where the interface goes up/down.
* don't add new packets to the backlog while if_running == false
* flush the backlog on both ifup and ifdown events
Change-Id: Ib35d099526544fe2cff64566fd56147a906adab9
There's not really any point in storing multiple LMI messages,
and then transmitting them in inverse order, as the existing code
does. Instead, we shall store only the last (failed) LMI message
and try to transmit that at highest priority, before any NS messages
in the actual queue.
Change-Id: I5407a76a34d7e687966fe1a915febf3a87256593
Reading a log line like this:
<0026> gprs_ns2_vc_fsm.c:808 GPRS-NS2-VC(FR-hdlcnet1-DLCI16-NSE02001-NSVC00001)[0x6120000024a0]{UNBLOCKED}: Received Event RESET
is very ambiguous. Does it mean we received a NS-RESET message? Does it
mean the FSM was instructed to send a NS-RESET? ...
Let's make sure the human-readable names give a very clear indication
of what exactly is happening.
Change-Id: I8b7615b3eca04212831163ff0ea4aea35069cd0e
Fix the encoding of the asynchronous Q.933 STATUS message
we send at DLC creation time.
Change-Id: Id1460ffa6266be85f44579c447421eed12481b02
Closes: OS#5010
this is a bit of a hack. Q.933 explicitly forbids either side from ever
sending a sequence number of '0'. Values start from '1' and are modulo 256,
but '0' is always skipped. So if the peer is sending us a "last received
sequence number of '0' it means it has not yet received any packets from us,
which in turn can only mean that it has just been restarted. Let's treat
this as "service affecting condition" and notify upper layers. This helps
particularly in recovering from rapidly re-starting peers, where the Q.933
nor NS have time to actually detect the connection was lost.
Change-Id: I960a7b17f2550cb49a7b9d72ed87cd271bb64122
Related: OS#4974
Comparing struct gprs_ra_id using memcmp can be error prone, so lets add
a compare function to compare two struct gprs_ra_id values reliably.
Change-Id: I4d7558c04d9d01761516526086be5104bb2eeada
Related: SYS#5103
This adds a --enable-systemtap configure option, which will then
add static tracepoints to the generated libosmocore binary.
At this point, only two tracepoints are supported: log_start
and log_done. They can be used to trace the amount of time
a libosmocore-using application spends in potentiall blocking calls to
log to stderr or to files.
Related: OS#4311
Change-Id: I7e1ab664241deb524c9582cbd1bec31af46c747e
All bind function should work in the same way.
Also fixing a null pointer assignment if no **result is giving.
Change-Id: Idd0c2190d2af39804c18c4786a997079db9a4330
ST_ALIVE isn't quite clear. This state means this NS-VC is still dead.
ST_ALIVE is used for NS-VC with vc_mode ALIVE (e.g. UDP NS-VC). It's
the first state when the remote hasn't yet answered or when
the NS-VC lost ALIVEs.
Related: OS#4958
Change-Id: I0b29e8ce6444546d90bde68b1f957f1013238784
There's no point in sending NS-RESET (or any other) messages to the
underlying FR layer if the FR DLC has not been marked as
available/active yet.
Change-Id: Id4e7565ba166ca1d12f8800c643d9f2bc4d66873
Closes: OS#4999
There are use cases in which the NS-VC FSM should go back to
unconfigured _without_ being re-started immediately. Let's permit
the caller to do that.
Change-Id: I31fe695a83d38ea1f10f5444e840633d7fa68442
If we receive messages for a DLC which has not yet reported as being
available by Q.933 LMI, drop the incoming message. Otherwise we would
dispatch it to the user, and the user wants to respond - but then
we reject the transmission due to the inactive DLC.
Change-Id: Ia4a045fdf165b526f429f4617e0fdc76036480bd
Related: OS#4999
Using 'uint8_t' for the length argument is definitely a bad idea.
Because of this, packing more than 255 septets would not work as
expected. Deprecate the old function and use 'size_t' instead.
Change-Id: Ib1aac538afeb0a5c76a1df472d555139a496e12e
If we are the 'user' side of FR and a link has just recovered,
we should ensure the next STATUS is for "full status". This way
we learn about the present DLCs as quickly as possible, saving up
to 10 seconds of further delay in link recovery.
Related: OS#4999
Change-Id: I6f905a18a7d130a3c02b4a3e7a2a2dc24afc0ea1
Log the NS-STATUS primitives from within the library, so we don't have
to keep related code in each and every application.
Change-Id: I368883acfc8ea76529befcd429bf8f2445a60a94
ENOSPC is used with non-volatile (disk) storage, while ENOMEM is
customarily used for RAM allocation failures.
Change-Id: Ia4c16d8278dc30c7cc69169b18428cda5272738d
Avoid code duplication between three different drivers by sharing
the "core" of the bind initialization in a new, shared ns2_bind_alloc().
Change-Id: I535fc68e94fcd695de827dd922706adc1c5a2cb7
The counters had all been lost in the ns -> ns2 configuration,
let's try to implement some more of them: LOST_ALIVE, BLOCKED,
LOST_RESET.
Change-Id: I75d8637bd142722fdf9a162a2369aa9cdc980a57
When the FR code decides to drop a packet (ENOBUFS from the AF_PACKET
socket for non-signaling packet), let's report that back via the
frame_relay code into the generic NS2 code.
This way the generic NS2 code always knows if a packet was actually
successfully transmitted, or if it was dropped for some reason.
Change-Id: I4bb517fd04af69dbe6da628b132d57994ab3e5a4
Don't call the function pointer of the underlying transport
directly, as we many still want to do some common processing,
such as statistics counting (see follow-up patches).
Change-Id: I8d14c7b8d3aacc2bed033072d7934dbd6aab41df
It's always a bad idea to have an array of descriptions that's indexed
by an enum, without using named initializers. It's too easy to get
inconsistencies.
Change-Id: Id0ebd2a202a465ca0298f4245f1fb5c495235fc8
AF_PACKET sockets cannot be written-to using select(), as they
will always return "writable" but then still fail with ENOBUFS.
This also means that we cannot use osmo_wqueue() as it assumes
that a writable socket can actually be written to.
As there's no way to figure out when exactly we can again perform
a successful write, we have no other option but to start a timer
and re-try at a later time.
We will scale that timer based on the estimated duration of transmission
for the just-failed PDU on the line rate of a 31TS E1 Line.
Furthermore, with this patch, we stop queueing anything but signaling
traffic (NS-BVCI=0) or Q.933 LMI. User data (NS-BVCI != 0) will
instead be dropped in order to avoid buffer-bloat.
Change-Id: I4f79a246236c94175ceffa5f3f556c8931c6bc62
Closes: OS#4995
gprs_bssgp and gprs_bssgp_util.c also contains code related to send and
receive RIM PDUs via BSSGP and also code to encode and decode RAN
INFORMATION PDUs. Lets move this to gprs_bssgp_rim.c
Change-Id: Icda279452962b06e552cb1361d2a27b7dc8a6b04
Related: SYS#5103
This feature is used by BSC to gain knowledge on whether a given BTS
supports GPRS Cell Change Notification (CCN) related procedures on PDCH,
and as a result enable or not by default the CCN_ACTIVE bit in SI13 to
announce the support it is allowed to use the feature.
Related: SYS#4909
Change-Id: I61991266b95d0c13d51b47906cc07846e9cf1390
Older commit adding the 2 bits for Rel-4 extension forgot to increase
the length field (see TS 44.060 Table 12.24.1)
Fixes: 946bb95af1
Change-Id: I20efb4403cdf6c5bc717502a7075630044142f17
The vty configuration was wrong when any other node was
in between the vty configuration (e.g. gb or ss7).
The vty code doesn't have a relation between parent and child node.
It wasn't detected because this only happened in the sgsn.
The pcu doesn't have any binds defined.
Change-Id: I8a3d67d41baca36b4d1a951a574cd41e556355c5
The gprs_ns2_vty_init_reduced() call initialized the ns2 vty without
configuration nodes (except timeout). The PCU can be only configured
by the BTS/BSC. It should not configured by vty.
Change-Id: I4437da74bd48ba64d3f16ea67afc26c45b6d3ac9
The call was only introduced as workaround for the first implementation
of vty. There is no need for this anymore. The configuration can
just add "accept-ipaccess" to the bind to allow creation of dynamic
ipaccess NSE.
Change-Id: Ie924ead6da17657f3da334068c8ada82c8845495
Drop the vty(1) code and replace it with vty2. The vty(1) was only
used as intermediate to not develop a vty while developing a new
code base behind. Users of gprs_ns2_ has to use the new vty code.
API change which must be synchronized with osmo-pcu,
osmo-gbproxy, osmo-sgsn.
Change-Id: Ic2059e75d8ede8e5c29c4fef6be608ed79c8a97c
This reverts commit b306094448.
It was merged too quickly and patches for projects using related
features are not yet prepared.
Change-Id: I8a2aaf74a47de8f4f0adb37d16426d199788e3fe
Drop the vty(1) code and replace it with vty2. The vty(1) was only
used as intermediate to not develop a vty while developing a new
code base behind. Users of gprs_ns2_ has to use the new vty code.
API change which must be synchronized with osmo-pcu,
osmo-gbproxy, osmo-sgsn.
Change-Id: I8c3f2afecc74b78f7f914f7dce166cbcb63444eb
All public enum should have the prefix GPRS_NS2_.
API change which must be synchronized with osmo-pcu,
osmo-gbproxy, osmo-sgsn.
Change-Id: I548ff12f7277cbb7e1a630a3dc02b738ce89be72
Whenever a RIM PDU is received, log to which RIM routing information
(address) it is going to and where it is comming from.
Change-Id: Ia08d3b162a4f6257cccaa7f0764fa7ea498355ef
Related: SYS#5103
The bssgp_rim_routing_info_discr_strs string list contains whitespaces,
when the whitespaces are replaced with a "-" the log output looks better
Change-Id: I26facd3dc160603da89dcd787cccf78b19a20f02
Related: SYS#5103
RIM routing formation structs can contain different variants of address
identifiers, so it is difficult for an API user to pick the _name()
function to generate a human readable string. Lets add
bssgp_rim_ri_name() and bssgp_rim_ri_name_buf() to make printing a
routing identifier easier.
Change-Id: Idca6bdccffe663aea71a0183ca3ea5bb5b59e702
Related: SYS#5103
All functions which are exposed by gprs_ns2_internal.h should not contain
the public prefix gprs_. Internal function should only contain ns2_ prefix.
Change-Id: Icecc5a918902cd10efac72bbac20780d39aab272
The NSVCI must match the PDUs. The only exception is a RESET with
dialect ipaccess. However those will be handled later.
Change-Id: I0ebdd4f4135b03a7d44a93aff3846c8ec19f333b
The NSE must match the PDUs. The only exception is a RESET with
dialect ipaccess. However those will be handled later.
Change-Id: I00bc04f3f46f5ae8ddd8b4b7d5479fb8e6463e1e
In preparation of checking the NSEI/NSVCI in ns2_vc_rx. It
makes the function less error prone as the *tp is always defined.
Change-Id: I3cd6ed20cdc1390075669e1d0101154c51f3997f
Back in 2018 in I393907b3c9e0cc1145e102328adad0a83ee13a9f Neels
introduced "last" as an optional flag to log the file/line number at the
end of the line, rather than at the end of the header. It seems nobody
has been usingi this feature, or at least never tried to save a related
config file, as there was no code whatsoever that would ever save this
optional "last" flag.
Change-Id: I7b6245256aecc425722242aaabc154ac58ba27a0
The NSVCI must match the PDUs. The only exception is a RESET with
dialect ipaccess. However those will be handled later.
Change-Id: I8518bdb03e9e33b04b77a2a346d04c5d29544a6c
The NSE must match the PDUs. The only exception is a RESET with
dialect ipaccess. However those will be handled later.
Change-Id: Ic8fc49d850490feb0ff69171a6b0881ae8696c0b
Let's avoid open-coding the printing of log context and rather rely on
log macros to prefix each log line with the relevant context. This
helps log readability, log post processing whether by grep or more
sophisticated tools.
Change-Id: I946c0e77686d91efc5afb62031e1ac1033a9a586
When the length field is written the function writes the lower and the
higher half of the length at the same position, so the higher half is
overwritten with the lower half, this is wrong.
Change-Id: I3cce0b2796793554a421fd3ce8e749c52d22eaea
Related: CID#216670
Related: SYS#5103
The current version of libosmocore only supports NACC related RIM
application containers. If the parser detects a different application
container it returns with EINVAL. Unfortunately this means that the
caller can not distinguish if there is a lack of support or a real
parsing error.
Change-Id: Ib5ada3554c04259764352888cf95bc4460cb2a54
Related: SYS#5103
If multiple objects are printed in the VTY, only the first line of each
object should be on the first character of the line, all others should
be indented. With this patch the "snow ns entities" output becomes
much more readable:
OsmoGbProxy> show ns entities
NSEI 00102: UDP, DEAD
FSM Instance Name: 'GPRS-NS2-SNS-BSS(NSE00102-SNS)[0x6120000018a0]', ID: 'NSE00102-SNS'
Log-Level: 'DEBUG', State: 'SIZE'
Timer: 1
Maximum number of remote NS-VCs: 8, IPv4 Endpoints: 4, IPv6 Endpoints: 0
NSVCI none: UNCONFIGURED DYNAMIC data_weight=1 sig_weight=1 udp)[127.0.0.1]:23000<>[127.0.0.11]:8888
NSEI 00101: UDP, DEAD
FSM Instance Name: 'GPRS-NS2-SNS-BSS(NSE00101-SNS)[0x6120000015a0]', ID: 'NSE00101-SNS'
Log-Level: 'DEBUG', State: 'SIZE'
Timer: 1
Maximum number of remote NS-VCs: 8, IPv4 Endpoints: 4, IPv6 Endpoints: 0
NSVCI none: UNCONFIGURED DYNAMIC data_weight=1 sig_weight=1 udp)[127.0.0.1]:23000<>[127.0.0.10]:7777
Change-Id: Id1b4c80a6caef410076a68b4301adaa01ba7e57a
Similiar to the BSSGP layer prefix events with RX if it's an received PDU
or REQ if it's a request from the code.
Change-Id: I341fa28fb671d439c050d985c88ece1521430a99
A BLOCK message can be received when waiting for a UNBLOCK message
in state BLOCK
Related: SYS#5208
Change-Id: Ie7b34b3ef04aa28304143191222324e1a3786cb2
The vty should be able to block or unblock a specific NSVC.
Further more this case is special for the UNITDATA as those
can be still received until the other side response to the BLOCK PDU.
Related: OS#4939
Change-Id: Ic0ce3c5fabc8644cc1ee71a8f6dd783fadf7b84d
Usually talloc_free() and other free functions in osmocom allows
to be called with NULL which is then ignored.
Change-Id: If7b0c6916a29d4611d0a40c388414076eb83e6b5
Add a OSMO_ASSERT to all bind calls which doesn't
check if the bind is from the expected type.
The only exception is rx and tx functions (hot path).
Change-Id: Ia4f8932263c60618c7f0dfc32d50ba5a8d57602b
At the moment libosmogb offers no convinient way to send RIM PDUs. Also
parsing an incoming RIM messages into destination, source routing
info and RIM container is not available.
Change-Id: I18134fd9938040d2facb6beee3732628b167ce8c
Related: SYS#5103
The function bssgp_parse_rim_ri() and bssgp_create_rim_ri() are located
in gprs_bssgp.c, since there is now a gprs_bssgp_rim.c module it makes
more sense to put them there. Also adjust the code a bit so that its
more intuitive to read.
Change-Id: Icd667f41d5735de56cd9fb257670337c679dd258
Related: SYS#5103
BSSGP RIM uses a number of nested containers to signal RIM application
specific payload information in a generic way. Lets add the container
structurs required for NACC.
Depends: libosmocore If48f412c32e8e5a3e604a78d12b74787a4786374
Change-Id: Ibbc7fd67658e3040c12abb5706fe9d1f31894352
Related: SYS#5103
Right now we end up in situations where only a NS-VC for data (BVCI != 0)
becomes unblocked, but the BSSGP and/or user application code is
notified that the NSE has recovered.
In the case of osmo-gbproxy, this will trigger a BVC-RESET on the
BVCI=0, but that obviously only works if the sig_weight > 0...
Closes: OS#4956
Change-Id: I933ee3969c052394d61ec6cf8c7c21d17957d9ab
gprs_sn2_fr_connect2() is the same as gprs_ns2_fr_connect() with
the lookup-and-create-on-demand of the NSE first.
Rather than copy+paste, they should simply invoke each other.
Change-Id: If835bf138f213e7f58205018e7efe3ecb772c624
The transfer cap call in ns2_prim_status_ind() asserts if
NSE is in an invalid state (nse is either alive or has NSVCS).
Change-Id: I535b3e5dd7240d19dd685652173775b250f5cc2d
The local and remote entries should be cleared on SNS Size.
Further it can be cleared when SNS failed (e.g. all NSVC become dead)
so VTY doesn't show old entries before entering SNS Size.
Related: OS#4949
Change-Id: Ie4db81acdd5f8ddf1a1f5dc7645d1144545d9c73
The vty command is used for test cases to reset NSE state.
So dynamic NSE shouldn't present.
Related: SYS#5208
Change-Id: I0a4f35c974c8c3b79c48f2f56170722c95254332
48.016 7.2: says an UNBLOCK pdu should be answered with UNBLOCK_ACK on
already unblocked NSVC
Related: SYS#5208
Change-Id: Ic92e99b2607d1e54ecb4668667065502a55a2ce0
When receiving a NS Reset over an unknown NSVC the NS code would create
a dynamic NSE. If the NSEI or NSVCI is already configured to a
persistant NSE/NSVC the packet should be ignored.
Related: SYS#5208
Change-Id: I855911e7d364f2e5b08ea05857747aa63fcf1cd3
Prevent memory corruption or segfaults by asserting that NSE and bind
link layer match. A mismatch should never happen and might cause the
bind to access invalid memory when sending because nsvc->priv doesn't
match what it expects.
Change-Id: I7ca4cd1c5dac8b5e44ffc4825b9373b2d04911ab
Related: OS#4948
The SNS code ignored the link type of the bind and just bound to every
bind it could find. This resulted in a segfault when an SNS NSVC tries
to send its UDP messages though frame relay.
Fixes: OS#4948
Change-Id: Ibb832a39876362d094cce635192f7b4f84dc2b10
The libosmocore TLV parser had a number of insufficient bounds checks
leading to reads beyond the end of the respective input buffer.
This patch
* adds proper out-of-bounds checks to all TLV types
* simplifies some of the existing checks
* introduces test cases to test all the corner cases
where either TAG, or length, or value are not fully contained
in the input buffer.
Thanks to Ilja Van Sprundel for reporting these problems.
Change-Id: I98b02c914c9e3ecf56050af846292aa6979d7508
This structure is needed in order to identify a given cell within the
BSS during RIM transactions.
The naming was made up by myself since I couldn't find any naming
reference for this kind of data (RAI + CI).
Since LAI + CI = CGI, then RAI + CI = CGI-PS
osmo_rai_name2 family of functions get a "2" suffix due to already
existing functions handling struct struct gprs_ra_id in gsm48.h
Change-Id: If48f412c32e8e5a3e604a78d12b74787a4786374
Both LAC and TAC take 2 octets and follow MCC/MNC fields on the wire.
We abuse gsm48_encode_ra() for encoding of MCC/MNC, but it can also
be abused to encode TAC in bssgp_create_rim_ri(). There is no need
to encode '0000'O and then override it with osmo_store16be().
Change-Id: I986552aa52cf38b1c5290d2e5cd3ff2d1c36a4e5
This is no event from an incoming message so rx is NULL, and we can't send a
status PDU. Also blocking the signalling BVC is not allowed (unblocking it is already
forbidden).
Change-Id: I3e384b71d57e939efc1596ac1d92380ed5eb916d
Fixes: CID#215716
Prior to this patch, it was not possible to gather SET/GET reply
information when implementing a CTRL client using libosmocontrol. This
is specially important when using the GET command, since one wants to
receive the queried value.
CTRL traps can also be handled this way by extending this patch in the
future if needed.
Change-Id: Id3c4631cd32c13e78e11b6e8194b8c16307ec4f1
This adds an inter-thread queue "it_q" to libosmocore. With it_q,
one can perform thread-safe enqueing of messages to another thread,
who will receive the related messages triggered via an eventfd
handled in the usual libosmocore select loop abstraction.
Change-Id: Ie7d0c5fec715a2a577fae014b0b8a0e9c38418ef
Writing a configuration that will be rejected by the VTY parser
is not the best solution, but still better than printing values
from previous iterations or the stack garbage. In any case,
this is unlikely to happen, just making Coverity happy.
Change-Id: I26644fe544c82c90767ec1a9709918474bd1be53
Fixes: CID#215852
In 'struct gsmtap_hdr' field 'snr_db' is defined as a signed integer,
however all functions that fill this structure accept an unsigned
integer. This is wrong, because SNR can be negative.
Let's use 'int8_t' instead of 'uint8_t'. Changing from unsigned
to signed should be relatively safe compared to the opposite.
Most of the callers I am aware of always do pass 0 anyway.
Change-Id: I9f432be5c346d563bf518111c14ff04d4a63f592
Related: SYS#5073
We used to suppress/drop any "zero length" messages, but we didn't
include the header when computing the length. However, in CBSP there
are messages (at least KEEP-ALIVE-COMPLETE) which only consist of the
header without any information elements. We cannot simply drop such
messages.
This also fixes the return value of osmo_cbsp_recv_buffered() to be
the total number of received octets (including the header).
Change-Id: Ib620128a167cb77f061ee57e8f8ad707b96b1c0d
This is a fixed-length Tag-Value IE. Our decoder already parsed
it correctly, but the encoder encoded it as TLV, which is wrong.
Change-Id: I7e1d7eab8b8e51acd9a24c38e2d3d30bbf00847a