| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This PR enhances the shared memory providing recovery if a process
crashes. It adds a SHM_DU_TIMEOUT_MICROS variable, setting an
expiration time for SDU's when shared memory is full. If an
application doesn't read a blocking SDU within this time, the shared
memory will be cleansed of all SDU's for this application and the
application's rbuff will be cleared.
Some refactoring of the API's. Fixed wrong pthread checks in IRMd.
Fixes #13
Fixes #14
|
|
|
|
| |
Changes the variable name as well to API instead of PID.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This changes the IRM API after discussions with Dimitri. The register
operation is now split into a bind and register operation. The same
for unregister; unbind and unregister. PIDs are now used as the
application instance name. A name for a PID is only provided for
scriptability in bash. It is therefore also no longer passed down to
the IPCP. Every operation on an IPCP through the IRM API has to use
the PID. Quering of the PIDs by name is possible. The IRM tool has
been updated to use this new API as well. A subcommand 'ipcp' has been
added for operations that take effect on IPCPs only.
Fixes #12
|
|
|
|
|
| |
shim UDP would deadlock upon a dealloc call. fd_clr cannot be called
under wrlock.
|
|
|
|
|
| |
getsockname() can give bad behaviour when haphazardly looping fd's,
added check that the fd has a flow. Replaced LOG_DBGF with LOG_DBG.
|
|
|
|
|
|
|
|
| |
Update to POSIX 200112L to allow use of robust mutexes in the
shm_du_map. Removed the implementation of the rw_lock in favor of
pthread_rwlock_t.
Placeholder for the shm_du_map_sanitize function.
|
|
|
|
| |
Unlock missing when compiled without DNS support.
|
|
|
|
|
|
|
| |
Adds a thread that cleans up stale pending flows. Compile time option
IRMD_FLOW_TIMEOUT allows setting the timeout period. Default timeout is
5 seconds. The thread wakes up 20 times per period to check the
flows.
|
|
|
|
|
| |
The ipcpds didn't unlock the flows lock when the dst_name is
unbeknownst to the IRMd.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
The registration function has been moved to the irm tool, applications
now need to be registered by an administrator. Currently only supports
one instance per registered name, and an AP can be registered under
only one name.
The irmd can now start a registered server application on demand.
For the full functionality of the tool, execute "irm register".
AP name removed from flow allocation. Flow allocation does not send
the source ap name as it is quite useless. The accept() call now only
returns the AE name.
|
| |
|
|
|
|
|
| |
When the ring buffer is empty, a read call will sleep. A write call on
an empty ring buffer will wake up sleeping readers.
|
|
|
|
|
|
|
| |
This stops IPC processes from hanging due to bad locks. It first
informs all threads that the IPCP is shutting down, then cancels all
threads. Cancellation is still required because threads may be in
accept() or receivefrom().
|
|
|
|
|
| |
The shim UDP did a non-blocking write call, causing loss of
SDU's. This has been changed to a blocking write call.
|
|
|
|
|
|
|
|
| |
In linux, the select call changes the timeout variable to the time not
spent in sleep, causing the timeout to decrease and that loop to grind
the cpu. The timeval is now reset to the correct value every loop.
Also removed a LOG_ERR message that was hampering performance.
|
|
|
|
| |
Conf was not checked on bootstrap.
|
| |
|
| |
|
|
|
|
|
|
| |
The shim UDP now supports deallocating a flow end-to-end. Contains
some stability fixes for flow allocation and some missing close()
calls in lib/sockets.
|
| |
|
|
|
|
|
| |
There could be a theoretical race condition that the server sends a
message before the FD is set on the client side, which is resolved.
|
|
|
|
|
|
|
|
|
|
|
| |
There were two race conditions. The first was where client received
the response message before the client UDP port was connected to the
server UDP port and the first message of the client was sent to the
LISTEN port instead of the server application port. This was solved by
reordering the operations so the response is sent after the port is
correctly connected. The second race condition is that the FD is not
added to the set in time at the server side before the first SDU
arrives. This was solved by adding a variable proteceted by a mutex.
|
|
|
|
|
| |
some debug code for locking and a sleep check are removed from the
shim-udp.
|
|
|
|
|
|
|
|
|
| |
It uses UDP port 0x0D1F on all hosts to send and receive flow
allocation messages. It supports communication between server and
client AP over a single shim IPCP.
Implementation of full flow deallocation is pending. Both the client
and the server still have to call flow_dealloc();
|
|
|
|
|
|
| |
Adds a missing modulo calculation that would result in index -1 being
returned instead of zero when the ringbuffer skipped back to the
beginning.
|
| |
|
|
|
|
|
|
|
|
|
|
| |
Simplified locking to take only two locks: the first lock guards the
state of the ipcp. This lock must be held for writing on bootstrap and
closing, and held for reading during all other operations. The second
lock guards operations on flows, and must be held for writing during
allocation and deallocation, and held for reading when sending sdu's.
After adding a fd to FD_SET, the shim will wait for 1 ms to ensure
that the FD is added to the select call.
|
|
|
|
|
| |
This takes a lock before notifying the IRMd of a new flow request, to
avoid race conditions when trying to retrieve the fd by port-id.
|
| |
|
|
|
|
|
| |
The fd_to_port_id function has no internal locking and must now be
called under lock.
|
|\
| |
| |
| | |
lib, ipcpd, irmd: Add QoS cube definition
|
| |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
This adds the QoS cube definition, which is an enum to select which
QoS is needed in the IPCP. An application has to use the qos_spec in
qos.h to define what it needs. The IRMd will map this unto a qos cube
definition.
Some headers are now also no longer installed on the system, since
they are only to be used within the irmd and ipcps.
|
| | |
|
|/
|
|
|
|
|
| |
Added necessary locks for the shim-udp. This PR also improves thread
management, the main thread now starts a mainloop thread, which spawns
sdu handler threads when it the IPCP is enrolled. If the IPCP exits
the enrolled state, the sdu loop is cancelled.
|
|
|
|
|
| |
The shim_udp now correctly destroys its ringbuffer on exit. Now
handles sdu's in a single thread using select() with a 10 us timeout.
|
|
|
|
|
| |
A value that was unitialized was being treated as an error in the
release build with gcc.
|
|
|
|
|
| |
This adds nslookup support to the shim UDP, so that a custom DNS
server can be queried.
|
|
|
|
|
|
| |
This allows someone to disable the use of DNS in the shim UDP. It also
adds a config file specific for the shim UDP (which holds the nsupdate
location). Certain defines were also moved to the global config file.
|
|
|
|
| |
caused the shim to run out of file descriptors fast...
|
|
|
|
|
| |
returning -1 as uint32_t leads to bugs.
also changed types in GPB to sint to use zigzag encoding.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
This commit has a first implementation of flow allocation (the "slow
path") and read/write (the "fast path") for ouroboros. It provides
basic but unstable communications over the shared memory.
It required a lot of changes all over the stack, and fixes a number of
previously undetected issues.
This PR still need heavy revision regarding data model, locking and
cleanup.
lib/dev: modifications to the API. It now uses an ap_init() call to
set the AP name and sets the Instance ID to the pid of the process. It
also binds the AP to the shared memory and creates tables for mappings
in the fast path. A call to ap_fini() releases the resources.
lib/shm_ap_rbuff: added ring buffer for data exchange between
processes in the fast path. It passes an index in the shm_du_map.
lib/shm_du_map: rewrote API to work with calls from dev.c. Garbage
collector added. Tests updated to new API.
ipcpd/ipcp-data: removed everything related to flows, as these are
universal for all ap's and kept in ap_data (dev.c), or similar structs
for shim ipcps.
shim-udp: added flow allocator and read/write functions and shm
elements.
irmd: revised data model and structures necessary for flow allocation.
tools: echo updated to new dev.h API.
messaging system was updated to comply with new flow allocation
messages. All exchanges use pid and port_id to bootstrap the fast
path.
|
|
|
|
|
| |
This adds support for adding the application name to a BIND DNS server
through the nsupdate utility.
|
|
|
|
|
| |
flow allocation now propagates on the client side up to the IPCP.
added UNKNOWN_AP and UNKNOWN_AE definitions to dev.h
|
|
|
|
|
| |
This abstracts away the IRMd messages by calling the functions from
ipcp.h in the ouroboros library. It also fixes some formatting issues.
|
|
|
|
|
|
| |
This commits adds the gethostbyname call to resolve the dst_name for a
flow allocation using DNS. The DNS server should be specified using
the method mandated by the system (e.g. added to /etc/resolv.conf).
|
|
|
|
|
|
|
|
| |
All instance-id's in ouroboros will be set by the system to the pid of
the process associated with this application process instance. This
means that the user has no way to choose the instance id's. Function
calls that assumed manually defined instance id's have been replaced
throughout the system.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
Unregistering ap's now works. An AP now registers/unregisters its AP-I
by sending its AP name and its pid to the IRMd. The IPCPs register
whatevercast names. An AP name is currently mapped on a whatevercast
name represented by the same string literal. The IRMd allows
registration of only one AP-I per AP. A Name Space Management system
is needed in the processing system so we can resolve this completely.
Changing the stack to register whatevercast names required some changes
all over the ipcpd implemented and in the library.
|
| |
|
|
|
|
|
|
| |
Initial code for application registration. Specifying "*" will (for
now) register with the first IPCP available in the system. Modified
the echo server not to barf messages on failed accept()
|