Particularly in case both sides initiate a BLOCK or UNBLOCK procedure
at almost the same time, it can happen thet we're already in BLOCKED
state and receive a late BLOCK-ACK or in UNBLOCKED state and receive a
late UNBLOCK-ACK.
Let's just silently discard them instaed of generating NS-STATUS which
may confuse the peer.
Change-Id: I2e5b934e1cf6c6cf982d5ab1dbb32e8920b91071
The primitive normally only contains NSE + BVCI, but in a tester
we actually want to verify which NS-VC a given message has arrived on,
and hence it makes sense to add the NSVCI, too.
Change-Id: I9402bf0be47e5b93c9cfb081eb8f9fa6734c9227
This took me quite some time: Tried to use NS_PROVIDER_FR, but the
code was compiled without support for it. It just failed silently
without printing any error or ever sending any message on the FR link.
Change-Id: I96475127a2079830b3456a8e288adf4c6c908887
The Load Sharing Function distributes traffic among all unblocked NS-VC
within a NS-VCG. The sharing is done based on a LSP (Link Selector
Parameter) which is passed with the NS-UNITDATA.req primitive from BSSGP
to NS. Details are implementation specific, but NS must ensure that all
traffic of one LSP is always sent through the same NS-VC, as long as
that NS-VC is alive/unblocked.
We use the LSP as follows:
* Signaling BVC always only uses lsp 0
* PTP BVC messages unrelated to user straffic use a per-BVC static LSP,
which is the BVCI
* User traffic can set whatever LSP it wants
* NS keeps an array of NSVCs currently unblocked and uses the LSP modulo
the array size as an index into it
Change-Id: I8b960ec4d729f50cadca353bb52685311cd45eed
Related: SYS#5084
In case of a NS-VCG with multiple NS-VC, we must BVC-RESET only when the
first NS-VC becomes unblocked, i.e. as soon as we have any connection
to the peer at all. Whether we have further additional links doesn't
matter, at least not in the sense that all state should be reset.
Change-Id: I69b2e9bd919fc981f189b6671b4234c3642e2449
This is something we need to simulate more complex scenarios,
particularly in the context of frame relay.
Change-Id: If1220852785853f8a5d8de183d5053ddd6ccb958
If IP-SNS is not used we should wait for the RESET procedure to finish
before sending NS-ALIVE.
For IP-SNS start NS-ALIVE in both roles (sgsn and bss) and don't handle
NS-RESET.
Also unified the log messages a bit (received -> Rx).
Related: SYS#5002
Change-Id: Ie01fee70297255b3d9c091bc2cec75b0f915c588
We want to see useful identification for components in the log, and
hence must be giving every component a name at create() time.
Change-Id: I0fe650243953e4d85161684865acd0354b2e465f
This makes for much more readable code, and we can even do without
activating any default altsteps.
Change-Id: I4c38dd55b7c27899735f5730851d36c1410d96a8
This allows NS_Emulation to react to changes of the underlying
transport layer (e.g. Frame Relay Link/DLCI up).
Change-Id: If00e9c50dc664ce62b6c0cbde99d741e8173169b
This adds a NS_Provider_FR which interfaces between FrameRelay_Emulation
and NS_Emulation. Include support for it only if enabled at compile
time to avoid pulling in a dependency on the FrameRelay stack for every
user of NS_Emulation.
Change-Id: I42ca811d23e383e362d2527c8ff2c916a62a5b42
Harald Welte
Daniel Willmann