Proxy PPB_ArrayBuffer_Dev, make them work over Messaging

Proxy PPB_ArrayBuffer_Dev, make them work over Messaging

BUG=103435
TEST=


Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=116594

[dmichael (off chromium), 2011-12-27 22:20:47]

sehr primary, brettw for ppapi/tests/test_post_message.cc

I decided to go the simplest way possible for now, but it turns out that SRPC is
still failing for big buffers. That's what I was checking with the test
update...  if I pass a buffer 1000000 bytes long, SRPC fails with this output:
[SRPC:NACL:29414,4278125760:18:16:50.147000] NaClSrpcMessageChannelSend: first
send failed, 65552 != -14.
[SRPC:NACL:29414,4278125760:18:16:50.147000]
NaClSrpcInvokeV(channel=0xfefc0bb0): rpc request send failed
[29414,3577083648:11:16:50.158506] munmap: rounded length to 0x100000

I'm tempted to work towards passing with shared memory to get around the SRPC
problem... unless you're confident you'll get that fixed in 18.

http://codereview.chromium.org/9022021/diff/5016/ppapi/native_client/src/shar...
File ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc (left):

http://codereview.chromium.org/9022021/diff/5016/ppapi/native_client/src/shar...
ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc:96: size_t sum =
value1 + value2;
size_t is 32-bit in NaCl and the 32-bit trusted builds, so this check seemed
wrong to me. Let me know if I'm misunderstanding something.

http://codereview.chromium.org/9022021/diff/5016/ppapi/native_client/src/shar...
ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc:121: if
(std::numeric_limits<uint32_t>::max() == string_length ||
And this part doesn't seem necessary if AddWouldOverflow works...

[brettw, 2011-12-28 20:05:14]

lgtm

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
File ppapi/native_client/src/shared/ppapi_proxy/array_buffer_proxy_var.h
(right):

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
ppapi/native_client/src/shared/ppapi_proxy/array_buffer_proxy_var.h:40:
ArrayBufferProxyVar();  // Not implemented, do not use.
Are you sure you need this to prevent the empty constructor from getting
instantiated?

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
File ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc (right):

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc:239:
RoundedStringBytes(buffer_length) - buffer_length);
This should be indented to after the ( above.

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
File ppapi/native_client/src/shared/ppapi_proxy/plugin_globals.cc (right):

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
ppapi/native_client/src/shared/ppapi_proxy/plugin_globals.cc:90:
GetBrowserInterfaceSafe(PPB_VAR_ARRAY_BUFFER_DEV_INTERFACE));
Only indent 4 spaces.

[sehr (please use chromium), 2012-01-05 19:49:26]

One nit.  LGTM.

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
File ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc (right):

http://codereview.chromium.org/9022021/diff/8002/ppapi/native_client/src/shar...
ppapi/native_client/src/shared/ppapi_proxy/object_serialize.cc:99: uint64_t sum
= uint64_t(value1) + value2;
nit, we've generally used static_cast<uint64_t> for this, but you pick.

[commit-bot: I haz the power, 2012-01-05 22:55:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/dmichael@chromium.org/9022021/18001

[commit-bot: I haz the power, 2012-01-06 00:09:56]

Change committed as 116594