diff options
Diffstat (limited to 'en/devices/architecture/hidl/threading.html')
| -rw-r--r-- | en/devices/architecture/hidl/threading.html | 24 |
1 files changed, 18 insertions, 6 deletions
diff --git a/en/devices/architecture/hidl/threading.html b/en/devices/architecture/hidl/threading.html index 5b8fa038..26c2cc12 100644 --- a/en/devices/architecture/hidl/threading.html +++ b/en/devices/architecture/hidl/threading.html @@ -48,11 +48,11 @@ available, it blocks until one is available.</p> <p>If the server has only one thread, then calls into the server are completed in order. A server with more than one thread may complete calls out of order -even if the client has only one thread. As <code>oneway</code> calls do not -block the client, multiple <code>oneway</code> calls may be processed -simultaneously or out of order by a server with more than one thread, and -<code>oneway</code> calls may be processed concurrently with a subsequent -blocking call.</p> +even if the client has only one thread. However, for a given interface object, +<code>oneway</code> calls are guaranteed to be ordered (see +<a href="#model">Server threading model</a>). For a multi-threaded server that +hosts multiple interfaces, <code>oneway</code> calls to different interfaces +may be processed concurrently with each other or other blocking calls.</p> <p>Multiple nested calls will be sent on the same hwbinder thread. For instance, if a process (A) makes a synchronous call from a hwbinder thread into process (B), @@ -160,7 +160,19 @@ server returns a <code>Return<void></code> object.</p> <h3 id=oneway>Oneway calls</h3> <p>When a function is marked <code>oneway</code>, the client returns immediately -and does not wait for the server to complete its function call invocation.</p> +and does not wait for the server to complete its function call invocation. At the +surface (and in aggregate), this means the function call takes half the +time because it is executing half the code, but when writing implementations that +are performance sensitive, this has some scheduling implications. Normally, +using a oneway call causes the callee to continue to be scheduled whereas +using a normal synchronous call causes the scheduler to immediately transfer +from the callee to the caller process. This is a performance optimization in +binder. For services where the oneway call must be executed in the target process +with a high priority, the scheduling policy of the receiving service can be +changed. In C++, using <code>libhidltransport</code>'s method +<code>setMinSchedulerPolicy</code> with the scheduler priorities and policies +defined in <code>sched.h</code> ensures that all calls into the service run at +least at the set scheduling policy and priority.</p> </body> </html> |