| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
| |
The ipcp-ops header was removed and merged into ipcp.h. The common
components dif_name and ipcp_type have been moved to the main ipcp
struct. After this move, ipcp_data only contained shim information, so
it was renamed to shim_data. The ipcp_init() function checks the type
and will only include the shim_data if the type is not an IPCP_NORMAL.
All ipcps have been adapted to this change in API.
|
| |
|
| |
|
|
|
|
|
|
|
| |
This allows IPCPs to bind a name, so that they can announce their name
to neighbors which can then allocate a flow to them. Registering of
the name happens by an administrator. It also moves the irmd_api to
common ground, since it is used by all IPCPs.
|
|
|
|
|
|
|
| |
These operations separe the starting and joining of the main ipcp
threads into ipcp_boot() and ipcp_shutdown() operations. This allows
the proper cleanup of user data and user threads after the IPCP is
requested to shut down.
|
|
|
|
|
|
|
|
|
|
| |
Now correctly relays the qoscube end-to-end in the stack.
A simple function specifying the cube in the spec is used for initial
testing.
The translation is now done in dev.c, but it could be moved elsewhere
when qos cabability matures and the need arises.
|
|
|
|
|
| |
This corrects the license statements on all files. Installed headers
are LGPLv2.1, the rest of the code is GPLv2.
|
| |
|
|
|
|
| |
Changes a state of the IPCP to a more correct terminology.
|
|
|
|
|
|
|
|
| |
This will split the IPCP state PENDING_ENROL into IPCP_CONFIG and
IPCP_BOOTING. IPCP_CONFIG is concerned only with configuring the IPCP
with the bare essence. When in IPCP_BOOTING, the IPCP will complete
its configuration by starting its policies, and thus making the IPCP
completely functioning.
|
|
|
|
| |
Total SDU size (including LLC header) cannot exceed 1500 bytes.
|
|
|
|
|
|
|
| |
All calls for opening rbuffs are now concentrated on the dev side.
This allows some simplifications in the np1 calls. The ipcp_fini call
will not destroy the mutex associated with the state, since the final
state needs to be checked before shutting down an IPCP.
|
| |
|
|
|
|
|
| |
Fixes a double free (conf->dif_name) and the handling of non-created
threads when the bootstrap of an IPCP fails
|
|
|
|
|
|
|
|
|
|
|
| |
The steps for flow deallocation have been further refined. An
operation ipcp_flow_fini() which wait for all SDUs to be read from a
flow has been added. The shim IPCPs and the local IPCP have been
adapted to this new API.
Deallocation messages have been removed from the shim IPCPs, since
there is insufficient state synchronisation between them to make this
work reliably.
|
|\
| |
| |
| | |
lib, irmd, ipcpd: Add name querying to IPCPs
|
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This adds the ability to query IPCPs if a name can be reached through
them, e.g. if a name is available in a DIF. This means that in the
shim-udp a DNS query is performed, in the shim-eth-llc an ARP-like
query has been added, in the local a check is done to see if the name
is registered, and in the normal currently no application is reachable
through it.
|
|/
|
|
|
| |
Fixes memory leaks associated with the dif_name (which doesn't need to
be maintained in shim DIFs) and the fqueue for the N + 1 flows.
|
|
|
|
|
| |
Remote requests for a flow that's locally deallocated should not
generate an error. Sending messages should not be under lock.
|
|
|
|
|
| |
The default flow options were not set when a flow allocation request
arrives remotely. This went unnoticed due to a previous bug.
|
|
|
|
|
| |
This has the code checked with -Wcast-qual and -Wconversion flags.
These flags were removed because SWIG generated code fails.
|
|
|
|
|
| |
This reduces the risk for some bugs, for instance due to
signed/unsigned mismatches and unused variables.
|
|
|
|
|
|
|
|
| |
The fast path will now use an incoming ring buffer per flow per
process. This necessitated the development of a new method for the
asynchronous io call, which is now based on an event queue system for
scalability (fqueue). The ipcpd's and tools have been updated to this
API.
|
|
|
|
|
| |
Both the N and N-1 entity must now call flow_dealloc for the port_id
to be released.
|
| |
|
|
|
|
| |
Now correctly handles and drops non-ouroboros traffic.
|
|
|
|
|
| |
Upon deallocation, the remote's SAP was being used to lookup the fd
instead of our own SAP. This resulted in strange behaviour.
|
|
|
|
|
|
| |
Some locks were taken at an inappropriate time, causing the
shim-eth-llc to block indefinitely upon shutdown and causing flow
allocation to halt since a response could never be given.
|
|
|
|
|
|
|
|
|
|
|
|
| |
IPCPs can now use ap_init() to initialize the memory. All flows are
accessed using flow descriptors, this greatly simplifies IPCP
development. Reverts the fast path to a single ap_rbuff per process.
Splits lib/ipcp into irmd/ipcp and lib/ipcp-dev. Adds a lib/shim-dev
holding tailored functions for shims. Moves the buffer_t to utils.h.
Fixes the shim-eth-llc length field. Removes the flow from shared.h.
Fixes #4
Fixes #5
|
| |
|
|\
| |
| |
| | |
lib: Add northbound ringbuffers
|
| |
| |
| |
| | |
Fast path is split in north and southbound paths.
|
|/
|
|
| |
Sending a dealloc to the irmd moved to its own call.
|
|
|
|
|
|
|
|
|
|
| |
The shm_du_map is renamed to shm_rdrbuff to reflect the Random
Deletion Ringbuffer used in the implementation. The close_on_exit call
is removed and SDUs are cleaned up by the application in the ap_fini()
call. This required a non-blocking peek() operation in the shm_ap_rbuff.
Some initial implementation for future support of qos cubes has been
added to the shm_rdrbuff.
|
|
|
|
| |
IPCPs will report their pid on shutdown for faster debugging.
|
|
|
|
|
|
| |
The state lock was reverted to an rwlock to avoid interference of
management functions with the fast path. IPCPs now close without
calling unsafe functions in the signal handler.
|
|
|
|
|
|
|
| |
This adds a condition variable to the IPCP state, so that upon state
changes any listeners to state changes can be notified. It also
replaces the read/write lock with a mutex in order to be able to do
so.
|
|
|
|
|
| |
Now the correct max length of an SDU is checked upon writing a frame,
as well as on receiving it.
|
|
|
|
|
|
| |
This adds the MSG_DONTWAIT flag to the send function of the
shim-eth-llc, when using PF_PACKET. Previously the send would return
only after the frame was sent, resulting in poor performance.
|
|
|
|
|
|
| |
This will notify the IRMd when the IPCP is initialized and ready to
receive messages. Previously a bootstrap could fail since the IPCP was
not listening to the socket yet.
|
|
|
|
|
|
| |
This changes the amount of time the shim-eth-llc will hold the IPCP
state lock when sending and receiving frames. Before it was holding
the lock for too long.
|
|
|
|
|
|
| |
This adds the functionality of exchanging the static DIF information
between 2 DIF members. After exchange the enrollment is stopped, and
the IPCP that initiated enrollment transitions to the enrolled state.
|
|
|
|
| |
The entry from the ringbuffer was never freed, causing memleak per SDU.
|
|
|
|
| |
ipcp-data was not correctly destroyed.
|
|
|
|
|
|
|
|
|
| |
Blocking I/O now uses condition variables in the shared memory instead
of busy waiting. Timeouts can be specified. This requires the size of
the rbuffs and du_map to be the same, to guarantee that when the
shm_du_map is not full, the ap_rbuffs can't be full either.
Added the timeout option to the flow for future use.
|
|
|
|
|
|
|
|
| |
It seems like drivers are setting the Ethernet length field wrong when
sending an LLC message. The LLC shim now writes the payload length in
the frame to circumvent the wrong information from the driver.
Also fixes deallocation.
|
|
|
|
|
|
|
|
|
|
| |
Various portability fixes for FreeBSD. POSIX requires shm file names
to start with a "/" to be portable. lseek(2) can be undefined on
POSIX shm, replaced with ftruncate(2). IRMd check on existing lockfile
more portable.
FreeBSD 11.0 is preferred as it natively supports robust mutexes.
Full working LLC implementation pending.
|
|
|
|
|
|
|
|
| |
Upon receipt of a frame, the minimum of the length reported in the
frame and the frame length reported by kernel is taken. Some device
drivers change the length in the frame. Some others add padding,
making the length reported by the kernel too high. This is a lousy
workaround to make it work on as many systems as possible.
|
| |
|
|
|
|
|
| |
All includes of <errno.h> within ouroboros are replaced with the
ouroboros errno.
|