Add an interface for nacl to create delete-on-close temp files,

Add an interface for nacl to create delete-on-close temp files,
to be used by pnacl for scratch files during compilation and linking.

This is in-lieu of the current use of pepper temp files, which are not
allowed in incognito mode, and are more cumbersome to clean up on surfaway.

Use this interface for the .o file in pnacl coordinator.  The .nexe still uses
pepper temp files, since we haven't moved that over for caching yet.
We will clean that up later.

Also do some crude quota enforcement for these files, using the
existing reverse service interface.

BUG= http://code.google.com/p/nativeclient/issues/detail?id=2683
TEST= none -- pnacl compilation still works

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=148964

[jvoung (off chromium), 2012-07-24 22:18:27]

Hi,

Brett could you review the parts not under native_client/.
David and Derek, could you review the native_client/ parts.

Thanks!

http://codereview.chromium.org/10815080/diff/1/chrome/test/nacl_test_injectio...
File chrome/test/nacl_test_injection/buildbot_nacl_integration.py (right):

http://codereview.chromium.org/10815080/diff/1/chrome/test/nacl_test_injectio...
chrome/test/nacl_test_injection/buildbot_nacl_integration.py:83:
'--integration_bot=1',
temporary for testing.

[brettw, 2012-07-25 23:33:39]

Is it possible to have IDL for this? We've tried to do IDL for everything else.
You can wrap everything in a "#inline c" block to quote it.

[jvoung (off chromium), 2012-07-25 23:58:57]

On 2012/07/25 23:33:39, brettw wrote:
> Is it possible to have IDL for this? We've tried to do IDL for everything
else.
> You can wrap everything in a "#inline c" block to quote it.

I will try converting the pp_file_handle.h to IDL, and I could try moving
ppapi/api/private/finish_writing_these/ppb_flash_file.idl out of
"finish_writing_these".

ppb_nacl_private_impl.h/idl  might be a bit trickier.  I remember trying to move
it outside of "finish_writing_these" before, but NaCl had used "bool" instead of
PP_Bool and it got a bit messy (maybe required cross-repo changes... I don't
remember exactly):
http://code.google.com/p/chromium/issues/detail?id=89968#c13

[brettw, 2012-07-26 17:40:04]

If the flash file stuff is difficult to make IDL, getting IDL for the new file
is the most important thing.

[jvoung (off chromium), 2012-07-26 20:46:30]

On 2012/07/26 17:40:04, brettw wrote:
> If the flash file stuff is difficult to make IDL, getting IDL for the new file
> is the most important thing.

It think the flash file IDL is pretty close to the .h, but I wasn't sure about
the versioning (.h has 3.0 and 2.0, but the idl only had 0.1).

Did the pp_file_handle in a separate cl:
http://codereview.chromium.org/10824048/

[brettw, 2012-07-27 22:38:17]

I don't have a very good understanding of the nacl plugin code, so I barely even
skimmed it. I assume some of the other reviewers will look at that. The Chrome
stuff LGTM.

http://codereview.chromium.org/10815080/diff/4002/chrome/browser/nacl_host/pn...
File chrome/browser/nacl_host/pnacl_file_host.cc (right):

http://codereview.chromium.org/10815080/diff/4002/chrome/browser/nacl_host/pn...
chrome/browser/nacl_host/pnacl_file_host.cc:105: if
(!file_util::Delete(file_path, false)) {
Unless I'm mistaken, it looks like this is already done in CreatePlatformFile
when you specify DELETE_ON_CLOSE.

http://codereview.chromium.org/10815080/diff/4002/chrome/renderer/pepper/ppb_...
File chrome/renderer/pepper/ppb_nacl_private_impl.cc (right):

http://codereview.chromium.org/10815080/diff/4002/chrome/renderer/pepper/ppb_...
chrome/renderer/pepper/ppb_nacl_private_impl.cc:36: 
Got an extra blank in here.

http://codereview.chromium.org/10815080/diff/4002/ppapi/native_client/src/tru...
File ppapi/native_client/src/trusted/plugin/pnacl_streaming_translate_thread.cc
(right):

http://codereview.chromium.org/10815080/diff/4002/ppapi/native_client/src/tru...
ppapi/native_client/src/trusted/plugin/pnacl_streaming_translate_thread.cc:96:
PluginReverseInterface* llc_reverse =
Check indenting.

[sehr (please use chromium), 2012-07-27 23:38:27]

Modulo brettw's comments, LGTM.

[jvoung - send to chromium..., 2012-07-28 01:30:05]

http://codereview.chromium.org/10815080/diff/4002/chrome/browser/nacl_host/pn...
File chrome/browser/nacl_host/pnacl_file_host.cc (right):

http://codereview.chromium.org/10815080/diff/4002/chrome/browser/nacl_host/pn...
chrome/browser/nacl_host/pnacl_file_host.cc:105: if
(!file_util::Delete(file_path, false)) {
On 2012/07/27 22:38:17, brettw wrote:
> Unless I'm mistaken, it looks like this is already done in CreatePlatformFile
> when you specify DELETE_ON_CLOSE.

Oops make sense -- removed.

http://codereview.chromium.org/10815080/diff/4002/chrome/renderer/pepper/ppb_...
File chrome/renderer/pepper/ppb_nacl_private_impl.cc (right):

http://codereview.chromium.org/10815080/diff/4002/chrome/renderer/pepper/ppb_...
chrome/renderer/pepper/ppb_nacl_private_impl.cc:36: 
On 2012/07/27 22:38:17, brettw wrote:
> Got an extra blank in here.

Done.

http://codereview.chromium.org/10815080/diff/4002/ppapi/native_client/src/tru...
File ppapi/native_client/src/trusted/plugin/pnacl_streaming_translate_thread.cc
(right):

http://codereview.chromium.org/10815080/diff/4002/ppapi/native_client/src/tru...
ppapi/native_client/src/trusted/plugin/pnacl_streaming_translate_thread.cc:96:
PluginReverseInterface* llc_reverse =
On 2012/07/27 22:38:17, brettw wrote:
> Check indenting.

Done.