Commit Graph

62 Commits

Author SHA1 Message Date
Matthew Heon
35361595f3 Remove runtime lockDir and add in-memory lock manager
Remove runtime's lockDir as it is no longer needed after the lock
rework.

Add a trivial in-memory lock manager for unit testing

Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:51:09 -05:00
Matthew Heon
d4b2f11601 Convert pods to SHM locks
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:51:09 -05:00
Matthew Heon
a364b656ea Add lock manager to libpod runtime
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:51:09 -05:00
Matthew Heon
e73484c176 Move to POSIX mutexes for SHM locks
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:51:09 -05:00
Matthew Heon
f38fccb48c Disable lint on SHMLock struct
Golint wants to rename the struct. I think the name is fine. I
can disable golint. Golint will no longer complain about the
name.

Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
a21f21efa1 Refactor locks package to build on non-Linux
Move SHM specific code into a subpackage. Within the main locks
package, move the manager to be linux-only and add a non-Linux
unsupported build file.

Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
3ed81051e8 Add an SHM-backed Lock Manager implementation
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
185136cf0e Add interface for libpod multiprocess locks
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
35cc71a9e8 Improve documentation and unit tests for SHM locks
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
52d95f5072 Propogate error codes from SHM lock creation and open
Also add a few more unit tests

Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
b489feff71 Add mutex invariant to SHM semaphores.
Check value of semaphores when incrementing to ensure we never go
beyond 1, preserving mutex invariants.

Also, add cleanup code to the lock tests, ensuring that we never
leave the locks in a bad state after a test. We aren't destroying
and recreating the SHM every time, so we have to be careful not
to leak state between test runs.

Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00
Matthew Heon
27cebb780b Initial skeleton of in-memory locks
Signed-off-by: Matthew Heon <matthew.heon@gmail.com>
2019-01-04 09:45:59 -05:00