Add ability to load pnacl resources from DIR_PNACL_COMPONENT.

Add ability to load pnacl resources from DIR_PNACL_COMPONENT.
Currently hidden behind a flag --enable-pnacl.

This currently reuses the manifest-system of resolving URLs.
Much of that can be simplified later on.

BUG= http://code.google.com/p/nativeclient/issues/detail?id=2365
TEST= 

(1) out/Debug/chrome --user-data-dir=/tmp/temp_profile/ --enable-pnacl --component-updater-debug=fast-update

(2) wait a while for it to download pnacl

(3) surf to a pnacl webpage

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=151303

[jvoung (off chromium), 2012-06-27 19:51:27]

This depends on:
http://codereview.chromium.org/10662006/

(so ignore the changes in the patch except for the ones under
src/trusted/plugin/)

This is sort of the least-surgery-required approach, but I'm sure a bunch of the
indirection through the manifest could be cut... if we can get the
reverse-service interface to not use a manifest... Or, I guess we can keep the
manifest stuff and do it this way too.

[jvoung (off chromium), 2012-08-06 20:53:12]

Okay, should work again -- PTAL

[jvoung (off chromium), 2012-08-06 20:57:09]

On 2012/08/06 20:53:12, jvoung (chromium) wrote:
> Okay, should work again -- PTAL

err... hold on -- I think there is one more patch for windows support.

[jvoung (off chromium), 2012-08-06 23:38:50]

On 2012/08/06 20:57:09, jvoung (chromium) wrote:
> On 2012/08/06 20:53:12, jvoung (chromium) wrote:
> > Okay, should work again -- PTAL
> 
> err... hold on -- I think there is one more patch for windows support.

ok whew -- windows tryjob is now green. PTAL

[jvoung (off chromium), 2012-08-08 21:06:07]

On 2012/08/06 23:38:50, jvoung (chromium) wrote:
> On 2012/08/06 20:57:09, jvoung (chromium) wrote:
> > On 2012/08/06 20:53:12, jvoung (chromium) wrote:
> > > Okay, should work again -- PTAL
> > 
> > err... hold on -- I think there is one more patch for windows support.
> 
> ok whew -- windows tryjob is now green. PTAL

+robertm -- if you have any feedback too

[robertm, 2012-08-08 21:51:11]

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/pnacl_coordinator.cc (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_coordinator.cc:39: explicit
PnaclManifest(const pp::URLUtil_Dev* url_util,
now that the constructor has a second arg the "explicit" is not really necessary
anymore.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/pnacl_resources.cc (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_resources.cc:141: full_url);
why not follow the model above and just break here.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/pnacl_resources.h (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_resources.h:45: // these resources.
 URLs for resources are resolved by the manifest
some typo

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_resources.h:65: // Get file descs
by name. Only valid after all_loaded_callback_ has been run.
would CHECK( resource_wrappers_.size() != 0)
be too brutal?

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/service_runtime.cc (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:248: void
PluginReverseInterface::OpenManifestEntry_MainThreadContinuation(
maybe say what has happened so far when this is call
and what this function does.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:296:
NaClXCondVarBroadcast(&cv_);
add comment for what is being woken up here

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:319:
NaClXCondVarBroadcast(&cv_);
dito

[jvoung - send to chromium..., 2012-08-08 22:47:04]

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/pnacl_coordinator.cc (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_coordinator.cc:39: explicit
PnaclManifest(const pp::URLUtil_Dev* url_util,
On 2012/08/08 21:51:11, robertm wrote:
> now that the constructor has a second arg the "explicit" is not really
necessary
> anymore. 

Done.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/pnacl_resources.cc (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_resources.cc:141: full_url);
On 2012/08/08 21:51:11, robertm wrote:
> why not follow the model above and just break here.

Oops, yeah it could.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/pnacl_resources.h (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_resources.h:45: // these resources.
 URLs for resources are resolved by the manifest
On 2012/08/08 21:51:11, robertm wrote:
> some typo

Done.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/pnacl_resources.h:65: // Get file descs
by name. Only valid after all_loaded_callback_ has been run.
On 2012/08/08 21:51:11, robertm wrote:
> would CHECK( resource_wrappers_.size() != 0)
> be too brutal?

Probably good to know.  Could also/instead check that it can find a mapping for
the URL instead?

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
File ppapi/native_client/src/trusted/plugin/service_runtime.cc (right):

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:248: void
PluginReverseInterface::OpenManifestEntry_MainThreadContinuation(
On 2012/08/08 21:51:11, robertm wrote:
> maybe say what has happened so far when this is call
> and what this function does.

Done.

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:296:
NaClXCondVarBroadcast(&cv_);
On 2012/08/08 21:51:11, robertm wrote:
> add comment for what is being woken up here

There's a bunch of these Broadcasts all over the file... so I'll add a comment
to the function.  Basically each Foo() waits on Foo_MainThreadContinuation() or
some Foo_followup_MainThreadContinuation(), if Foo needs multiple steps on the
main thread.

Added a comment to the first Foo_MainThreadContinuation() function
(OpenManifestEntry_MainThreadContinuation()).

http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
ppapi/native_client/src/trusted/plugin/service_runtime.cc:319:
NaClXCondVarBroadcast(&cv_);
On 2012/08/08 21:51:11, robertm wrote:
> dito

Done.

[sehr (please use chromium), 2012-08-10 22:05:48]

On 2012/08/08 22:47:04, jvoung wrote:
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> File ppapi/native_client/src/trusted/plugin/pnacl_coordinator.cc (right):
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/pnacl_coordinator.cc:39: explicit
> PnaclManifest(const pp::URLUtil_Dev* url_util,
> On 2012/08/08 21:51:11, robertm wrote:
> > now that the constructor has a second arg the "explicit" is not really
> necessary
> > anymore. 
> 
> Done.
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> File ppapi/native_client/src/trusted/plugin/pnacl_resources.cc (right):
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/pnacl_resources.cc:141: full_url);
> On 2012/08/08 21:51:11, robertm wrote:
> > why not follow the model above and just break here.
> 
> Oops, yeah it could.
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> File ppapi/native_client/src/trusted/plugin/pnacl_resources.h (right):
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/pnacl_resources.h:45: // these
resources.
>  URLs for resources are resolved by the manifest
> On 2012/08/08 21:51:11, robertm wrote:
> > some typo
> 
> Done.
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/pnacl_resources.h:65: // Get file descs
> by name. Only valid after all_loaded_callback_ has been run.
> On 2012/08/08 21:51:11, robertm wrote:
> > would CHECK( resource_wrappers_.size() != 0)
> > be too brutal?
> 
> Probably good to know.  Could also/instead check that it can find a mapping
for
> the URL instead?
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> File ppapi/native_client/src/trusted/plugin/service_runtime.cc (right):
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/service_runtime.cc:248: void
> PluginReverseInterface::OpenManifestEntry_MainThreadContinuation(
> On 2012/08/08 21:51:11, robertm wrote:
> > maybe say what has happened so far when this is call
> > and what this function does.
> 
> Done.
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/service_runtime.cc:296:
> NaClXCondVarBroadcast(&cv_);
> On 2012/08/08 21:51:11, robertm wrote:
> > add comment for what is being woken up here
> 
> There's a bunch of these Broadcasts all over the file... so I'll add a comment
> to the function.  Basically each Foo() waits on Foo_MainThreadContinuation()
or
> some Foo_followup_MainThreadContinuation(), if Foo needs multiple steps on the
> main thread.
> 
> Added a comment to the first Foo_MainThreadContinuation() function
> (OpenManifestEntry_MainThreadContinuation()).
> 
>
http://codereview.chromium.org/10689012/diff/33003/ppapi/native_client/src/tr...
> ppapi/native_client/src/trusted/plugin/service_runtime.cc:319:
> NaClXCondVarBroadcast(&cv_);
> On 2012/08/08 21:51:11, robertm wrote:
> > dito
> 
> Done.

LGTM.