Pepper API implementation for output protection.

chromeos/display/output_configurator.h

Index: chromeos/display/output_configurator.h
diff --git a/chromeos/display/output_configurator.h b/chromeos/display/output_configurator.h
index 5891785381cb22161e41415a98e62b8fd00679b1..45e94379d90bdef2c322d67d7761624caca01f12 100644
--- a/chromeos/display/output_configurator.h
+++ b/chromeos/display/output_configurator.h
@@ -25,6 +25,10 @@ typedef XID RROutput;
 typedef XID RRCrtc;
 typedef XID RRMode;
 
+namespace chrome {
+class PepperOutputProtectionHost;
+}  // namespace chrome
+
 namespace chromeos {
 
 // Used to describe the state of a multi-display configuration.
@@ -218,6 +222,18 @@ class CHROMEOS_EXPORT OutputConfigurator
     // Sends a D-Bus message to the power manager telling it that the
     // machine is or is not projecting.
     virtual void SendProjectingStateToPowerManager(bool projecting) = 0;
+
+    // Query link status and protection status. Returns true on success.
+    virtual bool QueryOutputProtectionStatus(
+        const chrome::PepperOutputProtectionHost* client,

[DaleCurtis, 2013/09/12 01:23:13]

Why not just use a void* if you don't intend to access the class.

[kcwu, 2013/09/12 18:22:08]

Done.

[dmichael (off chromium), 2013/09/12 20:10:15]

This is better, but my preference is still for an IDMap (or similar). If what
you want is a unique ID, I think your intent is clearer if you use a unique ID
rather than treat a pointer as one. Also, there's a slight chance the allocator
could reuse that pointer for a new client before the task runs, which would make
the "new" client disappear.

[kcwu, 2013/09/16 11:57:37]

Done.
I changed my code to register for an integer as id. I don't use IDMap because I
don't need the reverse mapping.

+        uint32_t* link_mask,
+        uint32_t* protection_mask) = 0;
+
+    // Set desired protection methods. Returns true when the protection request

[ddorwin, 2013/09/12 04:22:13]

s/Set/Request the/

[kcwu, 2013/09/12 18:22:08]

Done.

+    // has been made.
+    virtual bool EnableOutputProtection(
+        const chrome::PepperOutputProtectionHost* client,
+        uint32_t desired_method_mask) = 0;
   };
 
   // Helper class used by tests.
@@ -348,6 +364,25 @@ class CHROMEOS_EXPORT OutputConfigurator
   // so that time-consuming ConfigureOutputs() won't be called multiple times.
   void ScheduleConfigureOutputs();
 
+  // Query link status and protection status.
+  // |link_mask| is the type of connected output links, which is a bit-mask

[ddorwin, 2013/09/12 04:22:13]

nit: "bit mask" is 5x more common in the Chromium code and "bitmask" is even
more common, though that includes variable names.

[kcwu, 2013/09/12 18:22:08]

Done.

+  // of PP_OutputProtectionLinkType_Private values. |protection_mask| is the
+  // desired protection methods, which is a bit-mask of the
+  // PP_OutputProtectionMethod_Private values.
+  // Returns true on success.
+  virtual bool QueryOutputProtectionStatus(
+      const chrome::PepperOutputProtectionHost* client,
+      uint32_t* link_mask,
+      uint32_t* protection_mask);

[DaleCurtis, 2013/09/12 01:23:13]

OVERRIDE?

[kcwu, 2013/09/12 18:22:08]

Done.

[kcwu, 2013/09/16 11:57:37]

I found these methods are not override, so I reverted. Also removed "virtual"
keyword.

+
+  // Set desired protection methods.
+  // |protection_mask| is the desired protection methods, which is a bit-mask
+  // of the PP_OutputProtectionMethod_Private values.
+  // Returns true when the protection request has been made.
+  virtual bool EnableOutputProtection(
+      const chrome::PepperOutputProtectionHost* client,
+      uint32_t desired_method_mask);

[DaleCurtis, 2013/09/12 01:23:13]

Ditto?

[kcwu, 2013/09/12 18:22:08]

Done.

+
  private:
   // Configure outputs.
   void ConfigureOutputs();