diff options
| author | dmaclach@chromium.org <dmaclach@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2010-11-16 21:14:58 +0000 |
|---|---|---|
| committer | dmaclach@chromium.org <dmaclach@chromium.org@0039d316-1c4b-4281-b951-d872f2087c98> | 2010-11-16 21:14:58 +0000 |
| commit | 293fd7a9eb27dac7599a752e847c5b15f449519f (patch) | |
| tree | acc6e4a4c1d5d2ea551a64b92e8223dc94bad728 /chrome/chrome_common.gypi | |
| parent | c18215ab75f2f05e1da3f901fd385393a5d8def6 (diff) | |
| download | chromium_src-293fd7a9eb27dac7599a752e847c5b15f449519f.zip chromium_src-293fd7a9eb27dac7599a752e847c5b15f449519f.tar.gz chromium_src-293fd7a9eb27dac7599a752e847c5b15f449519f.tar.bz2 | |
Add multi_process_lock to base
Platform abstraction for a shared lock between processes. The process
that owns the lock will release it on exit even if exit is due
to a crash.
For cloud-print and remoting we want to be able to have a singleton service-process that can run independently of the browser process. This service process will communicate with the browser process via the standard IAC channels, but we want to have a way of signaling to other processes that a) there is a service-process running and b) that it is in a state where it is ready to be communicated with. The multi_process_lock class is intended to work as a simple flag that can be queried from multiple processes. If the service-process should crash, we would like the flag to be cleared automatically so that there is never confusion about the state of the service-process.
Other approaches considered for some Unix/Mac:
- Standard unix domain sockets depend on the file system and don't clean up properly in the case of a crash.
- Shared memory on unix depend on the file system and don't clean up properly in the case of a crash.
- System V semaphores on unix again depend on the file system.
- named_mach_ports on Mac OS. Bootstrap_register_name has been deprecated on 10.6, so we are doing essentially the same thing using CFMessagePort.
On Windows it is implemented as an event.
On Mac it is implemented using a CFMessagePort name.
On Linux it is implement using an abstract name port socket.
TEST=none
BUG=none
Review URL: http://codereview.chromium.org/4721001
git-svn-id: svn://svn.chromium.org/chrome/trunk/src@66323 0039d316-1c4b-4281-b951-d872f2087c98
Diffstat (limited to 'chrome/chrome_common.gypi')
| -rw-r--r-- | chrome/chrome_common.gypi | 4 |
1 files changed, 4 insertions, 0 deletions
diff --git a/chrome/chrome_common.gypi b/chrome/chrome_common.gypi index 89aeadc..f7f44fc 100644 --- a/chrome/chrome_common.gypi +++ b/chrome/chrome_common.gypi @@ -86,6 +86,10 @@ 'common/message_router.h', 'common/metrics_helpers.cc', 'common/metrics_helpers.h', + 'common/multi_process_lock.h', + 'common/multi_process_lock_linux.cc', + 'common/multi_process_lock_mac.cc', + 'common/multi_process_lock_win.cc', 'common/nacl_cmd_line.cc', 'common/nacl_cmd_line.h', 'common/nacl_messages.cc', |