| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
The queued packets were not correctly read. The rcv_cr->seqno now
indicates the next packet the receiver application expects. A lot more
stable now, but still some further issues to be fixed.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
|\ |
|
| |
| |
| |
| |
| |
| |
| |
| | |
Multiblock writes that require wrapping will now correctly pad
unuseable space at the end of the rdrbuff.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This fixes rudimentary automated repeat-request ARQ to correctly
configure the both connection records and use the receiver seqno. The
rto variable is moved out of the connection record.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
|\| |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The QoS specs were defined in the source file instead of in the header
file, which resulted in uninitialized structs being used, which gave
rise to weird behavior in the library.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|\| |
|
| |
| |
| |
| | |
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The FRCT kept only a left window edge in the receiver connection
window, however, it needs to keep track of the left window edge
(highest ACK'd sequence number) and the highest delivered sequence
number, so it can delay ACKs that cannot be piggybacked. TCP
recommends at most 500 ms for delayed ACKs (probably good to keep it
near half of RTO).
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| | |
A lot of files were unnecessarily including qoscube.h.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This splits off the CRC from FRCT so it can be set
independently. Ouroboros now allows raw flows with error checking.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| | |
Patch for previous commit, which was broken.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
The dt_const header is a remnant from an earlier version of the DT
PCI. It can be safely removed.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| | |
Renames port_id to flow_id according to updated nomenclature.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| | |
The FLOWSFLAGS command was recently obsoleted with the removal of
online reconfiguration of FRCT.
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This shortens packet_scheduler to psched, which results in more
readable code.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This will change SDU (Service Data Unit) to packet everywhere. SDU is
OSI terminology, whereas packet is Ouroboros terminology.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This replaces Service Access Point with Endpoint Identifier in the
raptor IPCP, since it better fits the Ouroboros terminology.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
The flow allocator now passes the full qos specification to the
endpoint, instead of just a cube. This is a more flexible
architecture, as it makes QoS cubes internal to the layers.
Adds endianness transforms for the flow allocator protocol in the
normal IPCP.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Since the Ethernet IPCP now has multiple reader threads it was
possible that both exit the select call, which caused one of the two
threads to block on the recv call. This makes the socket non-blocking
so that the recv call simply fails.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This will change the build to exclusively select one raw socket API in
case multiple are present in the sytem, which will simplify the code.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
Since Linux kernel 3.14 there is the option to bypass the kernel
Qdisc. This will speed up the Ethernet IPCP.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
There were some compilation issues introduced by adding the interface
monitor to the Ethernet IPCP. Furthermore it was not possible to
select between raw sockets or netmap if both were available.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
An unlock was called twice instead of a lock/unlock sequence, causing
a data race.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This adds multiple reader and writer threads, configurabe via cmake
with IPCP_ETH_RD_THR and IPCP_ETH_WR_THR. Improves ethernet IPCP
throughput, which looks to be limited by the raw socket calls.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This merges the dt_pci code into the dt source. It simplifies the code
base, and allows inline compilation. Removes unnecessary includes.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This will check the return values of init functions so that the code
is more robust. It also removes a duplicate init in the timerwheel,
checks for buffer overflows in the RIB and checks string lengths.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| | |
Some return value checks were missing in the UDP IPCP.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
A previous commit changes the signature of the function that
calculates the routing table. The test was not changed accordingly,
making make check fail.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
There was a double close of a fd in the Ethernet IPCP. It also passes
the correct max length to a memcpy (a strlen of the source string was
used previously).
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
There were missing unlocks in certain error conditions and some NULL
pointers were passed to strcmp which is undefined behavior.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This fixes some memleaks and potential buffer overflows in the irm
tool.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This removes configuration from the FRCT protocol to send it during
flow allocation.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| | |
This removes the _DEFAULT_SOURCE definition in the endian header as it
should not be there. This avoids double and conflicting definitions.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Previously there was a separate function per routing algorithm
selection, when in fact the algorithms all take as input a graph and
output a routing table, making it possible to place them in a single
function.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|/
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This adds rudimentary support for sending and processing
acknowledgments and doing retransmission.
It replaces the generic timerwheel with a specific one for
retransmission. This is currently a fixed wheel allowing
retransmissions to be scheduled up to about 32 seconds into the
future. It currently has an 8ms resolution. This could be made
configurable in the future. Failures of the flow (i.e. rtx not
working) are indicated by the rxmwheel_move() function returning a fd.
This is currently not yet handled (maybe just setting the state of the
flow to FLOWDOWN is a better solution).
The shm_rdrbuff tracks the number of users of a du_buff. One user is
the full stack, each retransmission will increment the refs counter
(which effectively acts as a semaphore). The refs counter is
decremented when a packet is acked. The du_buff is only allowed to be
removed if there is only one user left (the "stack").
When a packet is retransmitted, it is copied in the rdrbuff. This is
to ensure integrity of the packet when multiple layers do
retransmission and it is passed down the stack again.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
|
|
|
|
|
|
|
|
| |
The management flows were considering every event to be a FLOW_PKT
event, when in fact it could also be a flow down or up event,
resulting in indefinite reads.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
|
| |
Commit a3903da659e introduced sequence number for the LSAs to make it
more robust, but the check was not done properly, and the seqno was
always changed regardless of whether it was an old LSA or not.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
This enhances the irm connect and irm disconnect command to allow
creating connections between IPCPs based on wildcard matching for the
component name. In case no component was specified it sets up
connections between all possible components.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
|
|
|
|
| |
This will add a sequence number to the Link State Advertisements so
that upon receiving a LSA we can check if we need to forward the LSA
or not. If we already have the LSA with the received seqno in the LSDB
we do not forward it. This allows for setting up a management network
that has loops since the spanning tree upon which the LSAs are sent is
pruned on every hop.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
| |
This adds the infrastructure to actively react to flow up, down and
deallocated events.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
| |
This adds a unidirectional test to operf, which is handy for testing
unidirectional streams.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
|
| |
This will add a flag so the PFF is only recalculated if the LSDB has
changed. It also removes the instant recalculation of the LSDB if a
new neighbor is added, since this might cause instabilities.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|
|
|
|
|
|
|
|
|
| |
The DT component was flagging a connection as down and passing the fd
that was down. Of course the other components expect a connection
instead of just a fd. Now the connection manager will listen to flow
up and down events, and flag the connection up or down if needed.
Signed-off-by: Sander Vrijders <[email protected]>
Signed-off-by: Dimitri Staessens <[email protected]>
|
|\ |
|
| |
| |
| |
| |
| |
| |
| | |
This will mark flows down when they are finalized.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
|\| |
|
| |
| |
| |
| |
| |
| |
| |
| | |
The RIB output length was fixed for 3 qos cubes. Now it will scale
correctly if QoS cubes are added or removed.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|
|\| |
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This replaces the time utility functions with macros. This avoids
using library functions in the tools and also slightly speeds up the
implementation.
Signed-off-by: Dimitri Staessens <[email protected]>
Signed-off-by: Sander Vrijders <[email protected]>
|