| Commit message (Collapse) | Author | Age | Files | Lines |
|
|
|
|
| |
Fix missing set of api upon flow_alloc.
Various locking fixes.
|
|
|
|
| |
Padding was miscalculated, causing lockups in the fast path.
|
|
|
|
| |
The check whether the du map is empty should be first.
|
| |
|
| |
|
|
|
|
|
|
|
|
|
| |
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.
|
|
|
|
|
|
|
| |
This adds a flow_select() call that will sleep until an SDU can be
read on a flow. It returns the file descriptor for which an SDU is
ready. It takes as optional argument a timespec struct to specify a
timeout.
|
|
|
|
|
|
|
|
|
|
| |
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.
|
|
|
|
|
|
| |
Correct implementation of allocation for the Protocol Control
Information. Also removes stale code checking for a corner case when
the du map still wrapped multi-block SDU's.
|
|
|
|
|
| |
All includes of <errno.h> within ouroboros are replaced with the
ouroboros errno.
|
|
|
|
|
|
|
|
|
|
|
|
| |
The pid of the IRMd is stored in a lockfile in shared memory. This
makes checking if the IRMd is running independent of the
configuration, as previously the IRMd pid was stored at the end of the
shm_du_map, which could not be read by an IRMd that would be compiled
with different configuration options.
Also corrects some unnecessary includes
Fixes #21.
|
|
|
|
|
| |
Added check if the owner of the du map is actually running so a new
IRMd instance can remove a stale DU map.
|
|
|
|
|
| |
clean_sdus now has a boolean flag to let it know the process has
exited.
|
|
|
|
|
|
|
|
| |
When an application closes the shm_du_map, it will clean all remaining
sdu's for that application. Adds a function to clean the shm_du_map on
close.
Fixes #20.
|
|
|
|
|
|
|
|
| |
The stack used pid 0 (the scheduler) to indicate an invalid process
instance, probably as a leftover from the deprecated application
process instance id. Using -1 is a better solution.
Fixes #16.
|
|
|
|
|
|
|
|
| |
This commit introduces a first version of the revised CDAP
specification. CACEP (for authentication purposes) has been separated
from CDAP. Application developers may use CDAP if they find it
useful. Within Ouroboros CDAP will be used to perform operations on
the RIB of an IPCP.
|
|
|
|
|
|
|
| |
shm_du_map now fully supports multi-block SDU's when the
SHM_DU_MAP_MULTI_BLOCK compilation flag is set (enabled by default).
Fixes #11.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
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.
|
|
|
|
|
|
|
|
| |
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.
|
|
|
|
|
|
|
|
|
| |
This sets the correct install directories for all the binaries,
library and header files. It also sets the right permissions on the
sockets and shared memory so that regular users can also use the
ouroboros library. Root privileges are required to run the irmd.
Fixes #7
|
|
|
|
|
|
|
|
|
|
|
|
| |
Added the pid of the irmd to the shm_du_map. The IRMd will check for
an existing shm_du_map. If there is an existing file, it will exit if
the owner IRMd is running or remove it if the owner IRMd is not
running.
Also simplifies calculation of the shm_du_map pointers and corrects
exiting calls for the IRMd.
Fixes #8.
|
| |
|
|
|
|
|
| |
this allows disabling support for SDU's that are larger than a block
in the DU MAP by defining the variable SHM_MAP_SINGLE_BLOCK
|
|
|
|
|
|
|
| |
There was a check missing upon garbage collection. If the whole DU map
has been used once, all DUs will have their garbage bit set to
one. This meant that the garbage collector would keep moving the tail
indefinitely.
|
|
|
|
|
| |
When moving the tail pointer in the DU map, blocks was never reset to
zero. This caused it to sometimes clean up too many DUs.
|
|
|
|
|
|
| |
Adds a missing modulo calculation that would result in index -1 being
returned instead of zero when the ringbuffer skipped back to the
beginning.
|
|
|
|
|
| |
cleanup of file descriptors for the shared memory.
also adds a missing NULL check and some invalidation in dev.c
|
|
|
|
|
|
|
| |
The test had a synchronisation issue. Due to rebuild of the shm_du_map
for random access, there was also a hack in the shm_du_map
specifically to run this test. Because of this hack, the decision was
made to drop this test.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
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.
|
| |
|
|
|
|
| |
logs should be included in source, not in header
|
|
|
|
|
|
|
|
| |
Updated tests. Added code to drop a corner case (packet where the tail
PCI would cross the edge of the ring buffer) because solving this very
rare case is not worth the performance hit on the ringbuffer the extra
code would incur. This means the ringbuffer might drop a very small
percentage of packets.
|
|
|
|
|
|
|
|
| |
Got rid of divisions and modulo operations.
Change in the meaning of a #define.
SHM_DU_MAP_SIZE: is now an exponent of 2. Default is 10, allowing
for 1024 blocks in the map.
|
|
|
|
|
|
| |
Elements must be created/destroyed in order, but IPCPs can access PCI
while the PDU is in the ring buffer.
Test updated, no more waits are needed.
|
|
The shared memory map will allow passing SDU's between IPCPs
efficiently. It is a single block of managed memory. SDU's are stored
in a shm_du_buff datastructure, which is a list of blocks inside the
shared memory region. The mechanism for passing SDU's is not yet
implemented and could be either within the region or using another
local IPC mechanism.
The following things are configured at compile time:
SHM_FILE_NAME : the name. Default is "ouroboros_du_map".
SHM_DU_BUFF_BLOCK_SIZE: size of a block inside the shared memory
region. Default is the pagesize of the system.
SHM_DU_MAP_SIZE : the approximate size in bytes of the shared
memory map. The actual filesize may differ by
a small margin. Default is 4MB.
|