Pepper: Remove trusted plugin loadable module.

Pepper: Remove trusted plugin loadable module.

This change links the trusted plugin into the renderer, making it easier for
the trusted plugin to use libraries in Chromium. This removes the
ppGoogleNaClPlugin loadable module file on various platforms.

This is part of a larger effort to remove the "trusted plugin" used to
bootstrap NaCl plugins.

It introduces an "internal_module" interface for setting the value returned by
pp::Module::Get(). This is so that both the trusted plugin and the remoting
plugin can be linked into the renderer. However, I believe this is safe because
the trusted plugin runs in-process and the remoting plugin runs out-of-process.

BUG=394497
R=dmichael@chromium.org, mseaborn@chromium.org, phajdan.jr@chromium.org, thestig@chromium.org, wez@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=287071

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=287580

[dmichael (off chromium), 2014-07-18 18:08:37]

I think the approach overall looks good, if it will help you to do this. I buy
that it's probably worth doing now because (a) it lets you use Chromium
primitives during the transition and (b) it lets you sort out some of the issues
with removing the actual plugin sooner rather than later.

Come to think of it, I think it will also get you on the same heap, so you'll
probably be able to start doing more sophisticated types across PPB_NaCl_Private
if it helps.

https://codereview.chromium.org/397243004/diff/40001/chrome/common/chrome_pat...
File chrome/common/chrome_paths.cc (right):

https://codereview.chromium.org/397243004/diff/40001/chrome/common/chrome_pat...
chrome/common/chrome_paths.cc:65: // information is stored relative to file
path.
Hmm, chromoting seems to use a special string:
internal-remoting-viewer

It might be good to follow up to find out if your suspicion is right. If so, I
think I agree with leaving it alone until the plugin is all-the-way gone. Then
we're left with the somewhat strange question of what to do (if anything) for
users who had the plugin disabled. Probably we accept that some users will be
affected, but that they still have enough control since they can separately
disable PNaCl.

https://codereview.chromium.org/397243004/diff/40001/ppapi/cpp/internal_modul...
File ppapi/cpp/internal_module.cc (right):

https://codereview.chromium.org/397243004/diff/40001/ppapi/cpp/internal_modul...
ppapi/cpp/internal_module.cc:10: static Module* g_module_singleton;
Would it be overkill to explicitly initialize to NULL?

https://codereview.chromium.org/397243004/diff/40001/ppapi/cpp/internal_modul...
ppapi/cpp/internal_module.cc:21: }  // namespace pp
It seems like you could put this in ppapi/cpp/private?

[teravest, 2014-07-28 18:40:08]

dmichael for ppapi/
mseaborn for components/nacl.gyp (and sanity)
sky for chrome/
wez for remoting/

https://codereview.chromium.org/397243004/diff/40001/chrome/common/chrome_pat...
File chrome/common/chrome_paths.cc (right):

https://codereview.chromium.org/397243004/diff/40001/chrome/common/chrome_pat...
chrome/common/chrome_paths.cc:65: // information is stored relative to file
path.
On 2014/07/18 18:08:37, dmichael wrote:
> Hmm, chromoting seems to use a special string:
> internal-remoting-viewer
> 
> It might be good to follow up to find out if your suspicion is right. If so, I
> think I agree with leaving it alone until the plugin is all-the-way gone. Then
> we're left with the somewhat strange question of what to do (if anything) for
> users who had the plugin disabled. Probably we accept that some users will be
> affected, but that they still have enough control since they can separately
> disable PNaCl.

Done.

https://codereview.chromium.org/397243004/diff/40001/ppapi/cpp/internal_modul...
File ppapi/cpp/internal_module.cc (right):

https://codereview.chromium.org/397243004/diff/40001/ppapi/cpp/internal_modul...
ppapi/cpp/internal_module.cc:21: }  // namespace pp
On 2014/07/18 18:08:37, dmichael wrote:
> It seems like you could put this in ppapi/cpp/private?

Done.

[Wez, 2014-07-28 19:49:34]

remoting/ LGTM

[Mark Seaborn, 2014-07-28 20:28:37]

It's great that this works!

I have a suspicion that you might run into troubles that are not covered by the
trybots.

e.g. Removing a file generally isn't tested well by the trybots' incremental
build.  Ninja/make don't remove files that no longer have Gyp targets, so if the
build still depends on a file that has been removed, the trybots will not catch
that.  So maybe try a clobber build to test that?

I wonder if the perf size checks will fail if you remove a file that they test?

I suspect the Chrome OS packaging scripts refer to the plugin, but I don't know
where they live now.  (It used to be this repo, but it hasn't been changed since
August 2013:
http://git.chromium.org/gitweb/?p=chromiumos/overlays/chromiumos-overlay.git)

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_co...
File chrome/common/chrome_content_client.cc (right):

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_co...
chrome/common/chrome_content_client.cc:179: // Specifically, Portable Native
Client is only enabled by the command line
You don't mean this, surely?  PNaCl is enabled by default now.

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_pa...
File chrome/common/chrome_paths.cc (right):

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_pa...
chrome/common/chrome_paths.cc:63: const base::FilePath::CharType
kInternalNaClPluginFileName[] =
Is it difficult to remove the use of kInternalNaClPluginFileName rather than
making this a dummy value?

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_pa...
chrome/common/chrome_paths.cc:310: cur =
cur.Append(kInternalNaClPluginFileName);
The use is here...  If it's difficult to remove FILE_NACL_PLUGIN, maybe add a
TODO comment saying what it's still used for?

https://codereview.chromium.org/397243004/diff/180001/chrome/sync_integration...
File chrome/sync_integration_tests.isolate (right):

https://codereview.chromium.org/397243004/diff/180001/chrome/sync_integration...
chrome/sync_integration_tests.isolate:22: '<(PRODUCT_DIR)/nacl_irt_x86_64.nexe',
What about the 32-bit version?  Were these dependencies missing before, and this
problem is revealed by other parts of your changeset?

https://codereview.chromium.org/397243004/diff/180001/ppapi/cpp/private/inter...
File ppapi/cpp/private/internal_module.h (right):

https://codereview.chromium.org/397243004/diff/180001/ppapi/cpp/private/inter...
ppapi/cpp/private/internal_module.h:5: #ifndef PPAPI_CPP_INTERNAL_MODULE_H_
nit: missing "PRIVATE"

https://codereview.chromium.org/397243004/diff/180001/ppapi/native_client/src...
File ppapi/native_client/src/trusted/plugin/plugin.gyp (left):

https://codereview.chromium.org/397243004/diff/180001/ppapi/native_client/src...
ppapi/native_client/src/trusted/plugin/plugin.gyp:51: 'target_name':
'ppGoogleNaClPluginChrome',
How about renaming this?

 * camelCase isn't the usual Gyp style for targets.

 * There might still be packaging scripts that refer to
ppGoogleNaClPluginChrome.{dll,so,dylib}, e.g. in the Chrome OS packaging scripts
or in Linux distros' packaging scripts.  They might accidentally pick up the new
ppGoogleNaClPluginChrome and either go wrong or just waste space.  (Maybe that's
not very likely if this target becomes ppGoogleNaClPluginChrome.a, but it might
stay as ppGoogleNaClPluginChrome.so in a modules/components build.)

[teravest, 2014-07-28 22:04:22]

I'll do a clobber build for the trybots-- thanks for suggesting that.

Thankfully, the perf sizes check seems to play nice for nonexistent files:
https://code.google.com/p/chromium/codesearch#chromium/tools/build/scripts/sl...

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_co...
File chrome/common/chrome_content_client.cc (right):

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_co...
chrome/common/chrome_content_client.cc:179: // Specifically, Portable Native
Client is only enabled by the command line
On 2014/07/28 20:28:36, Mark Seaborn wrote:
> You don't mean this, surely?  PNaCl is enabled by default now.

Sorry, looks like I had a bad rebase and this got removed and put back. Removed.

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_pa...
File chrome/common/chrome_paths.cc (right):

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_pa...
chrome/common/chrome_paths.cc:63: const base::FilePath::CharType
kInternalNaClPluginFileName[] =
On 2014/07/28 20:28:36, Mark Seaborn wrote:
> Is it difficult to remove the use of kInternalNaClPluginFileName rather than
> making this a dummy value?

Yes, there's some more work involved that I'd rather do in a separate change.
TODOs added.

https://codereview.chromium.org/397243004/diff/180001/chrome/common/chrome_pa...
chrome/common/chrome_paths.cc:310: cur =
cur.Append(kInternalNaClPluginFileName);
On 2014/07/28 20:28:36, Mark Seaborn wrote:
> The use is here...  If it's difficult to remove FILE_NACL_PLUGIN, maybe add a
> TODO comment saying what it's still used for?

I've added a TODO here.

https://codereview.chromium.org/397243004/diff/180001/chrome/sync_integration...
File chrome/sync_integration_tests.isolate (right):

https://codereview.chromium.org/397243004/diff/180001/chrome/sync_integration...
chrome/sync_integration_tests.isolate:22: '<(PRODUCT_DIR)/nacl_irt_x86_64.nexe',
On 2014/07/28 20:28:36, Mark Seaborn wrote:
> What about the 32-bit version?  Were these dependencies missing before, and
this
> problem is revealed by other parts of your changeset?

This looks to be the result of a bad rebase. Dependencies removed.

https://codereview.chromium.org/397243004/diff/180001/ppapi/cpp/private/inter...
File ppapi/cpp/private/internal_module.h (right):

https://codereview.chromium.org/397243004/diff/180001/ppapi/cpp/private/inter...
ppapi/cpp/private/internal_module.h:5: #ifndef PPAPI_CPP_INTERNAL_MODULE_H_
On 2014/07/28 20:28:37, Mark Seaborn wrote:
> nit: missing "PRIVATE"

Done.

https://codereview.chromium.org/397243004/diff/180001/ppapi/native_client/src...
File ppapi/native_client/src/trusted/plugin/plugin.gyp (left):

https://codereview.chromium.org/397243004/diff/180001/ppapi/native_client/src...
ppapi/native_client/src/trusted/plugin/plugin.gyp:51: 'target_name':
'ppGoogleNaClPluginChrome',
On 2014/07/28 20:28:37, Mark Seaborn wrote:
> How about renaming this?
> 
>  * camelCase isn't the usual Gyp style for targets.
> 
>  * There might still be packaging scripts that refer to
> ppGoogleNaClPluginChrome.{dll,so,dylib}, e.g. in the Chrome OS packaging
scripts
> or in Linux distros' packaging scripts.  They might accidentally pick up the
new
> ppGoogleNaClPluginChrome and either go wrong or just waste space.  (Maybe
that's
> not very likely if this target becomes ppGoogleNaClPluginChrome.a, but it
might
> stay as ppGoogleNaClPluginChrome.so in a modules/components build.)

Sounds reasonable. Renamed to nacl_trusted_plugin

[teravest, 2014-07-29 12:59:17]

Hmm, I'm still getting telemetry failures from a fatal message in
pepper_plugin_registry.cc.

Hold off looking at this again for now, I'll spend some time on this today and
give an update when I've got it figured out.

[teravest, 2014-07-29 17:46:05]

I was able to figure out the test failures, caused by deleting too much code
related to FILE_NACL_PLUGIN.

Mark,
For now I'd like to keep FILE_NACL_PLUGIN around; it matches our behavior for
remoting, which is also an "internal plugin" linked to the renderer. I added
some TODOs stating what that code is for, and I'll remove it when this "internal
plugin" for NaCl support is gone.

[Mark Seaborn, 2014-07-29 18:13:18]

LGTM

https://codereview.chromium.org/397243004/diff/330001/ppapi/native_client/src...
File ppapi/native_client/src/trusted/plugin/ppapi_entrypoints.h (right):

https://codereview.chromium.org/397243004/diff/330001/ppapi/native_client/src...
ppapi/native_client/src/trusted/plugin/ppapi_entrypoints.h:5: #ifndef
NATIVE_CLIENT_SRC_TRUSTED_PLUGIN_PPAPI_ENTRYPOINTS_H_
Nit: Add PPAPI_ prefix

[teravest, 2014-07-29 20:03:45]

https://codereview.chromium.org/397243004/diff/330001/ppapi/native_client/src...
File ppapi/native_client/src/trusted/plugin/ppapi_entrypoints.h (right):

https://codereview.chromium.org/397243004/diff/330001/ppapi/native_client/src...
ppapi/native_client/src/trusted/plugin/ppapi_entrypoints.h:5: #ifndef
NATIVE_CLIENT_SRC_TRUSTED_PLUGIN_PPAPI_ENTRYPOINTS_H_
On 2014/07/29 18:13:18, Mark Seaborn wrote:
> Nit: Add PPAPI_ prefix

Done.

[dmichael (off chromium), 2014-07-29 23:08:40]

lgtm

[teravest, 2014-07-30 19:56:06]

+phajdan.jr@chromium.org for chrome/installer/linux/common/installer.include

[Paweł Hajdan Jr., 2014-07-31 10:02:04]

chrome/installer/linux LGTM

[teravest, 2014-07-31 14:13:57]

-sky
+jochen for chrome/ since sky is now OOO.

[teravest, 2014-07-31 16:01:23]

-jochen +thestig
to get a reviewer in a US timezone.

[Lei Zhang, 2014-07-31 19:33:20]

Can you update the corresponding .gn files with the same changes as your .gyp[i]
files? Even if the changes in the .gn files are commented out with TODOs, it's
good to keep them consistent as much as possible.

Otherwise LGTM.

[teravest, 2014-08-01 15:44:02]

The CQ bit was checked by teravest@chromium.org

[commit-bot: I haz the power, 2014-08-01 15:45:05]

CQ is trying da patch. Follow status at
 https://chromium-status.appspot.com/cq/teravest@chromium.org/397243004/370001

[teravest, 2014-08-01 19:38:31]

Committed patchset #8 manually as 287071 (presubmit successful).

[xiyuan, 2014-08-01 20:48:31]

Could this CL causing the chromeos build failure here.

http://build.chromium.org/p/chromiumos.chromium/builders/X86%20%28chromium%29...

chromeos-chrome-38.0.2111.0_alpha-r1: Copying Chrome tests into
/build/x86-generic/tmp/portage/chromeos-base/chromeos-chrome-38.0.2111.0_alpha-r1/work/test_src
chromeos-chrome-38.0.2111.0_alpha-r1:  * Installing test targets:
libffmpegsumo.so media_unittests sandbox_linux_unittests
ppapi_example_video_decode video_decode_accelerator_unittest
video_encode_accelerator_unittest libppapi_tests.so chrome_sandbox
chromeos-chrome-38.0.2111.0_alpha-r1: i686-pc-linux-gnu-strip:
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_example_video_decode':
No such file
chromeos-chrome-38.0.2111.0_alpha-r1: i686-pc-linux-gnu-strip:
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/libppapi_tests.so':
No such file
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/lib32': No
such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/lib64': No
such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/mock_nacl_gdb':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_newlib.nmf':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_glibc.nmf':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_newlib_x32.nexe':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_newlib_x64.nexe':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_newlib_arm.nexe':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_glibc_x32.nexe':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_glibc_x64.nexe':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/ppapi_nacl_tests_glibc_arm.nexe':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/test_case.html':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/test_case.html.mock-http-headers':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/test_page.css':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/test_url_loader_data':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: rsync: link_stat
"/home/chrome-bot/chrome_root/src/chrome/test/functional" failed: No such file
or directory (2)
chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were not
transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/chrome/test/functional': No such file
or directory
chromeos-chrome-38.0.2111.0_alpha-r1: rsync: link_stat
"/home/chrome-bot/chrome_root/src/content/common/gpu/testdata" failed: No such
file or directory (2)
chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were not
transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/content/common/gpu/testdata': No such
file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: rsync: change_dir
"/home/chrome-bot/chrome_root/src/third_party/accessibility-developer-tools/gen"
failed: No such file or directory (2)
chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were not
transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/third_party/accessibility-developer-tools/gen/axs_testing.js':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: E_MACHINE not defined!
chromeos-chrome-38.0.2111.0_alpha-r1: Copying Telemetry Framework into
/build/x86-generic/tmp/portage/chromeos-base/chromeos-chrome-38.0.2111.0_alpha-r1/work/telemetry_src
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/tools/cros/': No such file or
directory
chromeos-chrome-38.0.2111.0_alpha-r1: rsync: change_dir
"/home/chrome-bot/chrome_root/src/third_party/android_tools/sdk" failed: No such
file or directory (2)
chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were not
transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/third_party/android_tools/sdk/platform-tools':
No such file or directory
chromeos-chrome-38.0.2111.0_alpha-r1: rsync: change_dir
"/home/chrome-bot/chrome_root/src/third_party/android_tools/sdk" failed: No such
file or directory (2)
chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were not
transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
'/var/cache/chromeos-chrome/chrome-src/src/third_party/android_tools/sdk/platform-tools':
No such file or directory
...
chromeos-chrome-38.0.2111.0_alpha-r1:
/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/libppGoogleNaClPluginChrome.so
does not exist and is required.
chromeos-chrome-38.0.2111.0_alpha-r1: You can bypass this error with --sloppy.

If this is intended, we need to fix chromeos-chrome-9999.ebuild as well.

[teravest, 2014-08-01 21:05:19]

Yes. This change removes that file. I looked through the chromeos overlay
repository but didn't find a mention of that file name anywhere.

On Friday, August 1, 2014, <xiyuan@chromium.org> wrote:

> Could this CL causing the chromeos build failure here.
>
> http://build.chromium.org/p/chromiumos.chromium/builders/
> X86%20%28chromium%29/builds/810/steps/BuildPackages/logs/stdio
>
> chromeos-chrome-38.0.2111.0_alpha-r1: Copying Chrome tests into
> /build/x86-generic/tmp/portage/chromeos-base/chromeos-chrome-38.0.2111.0_
> alpha-r1/work/test_src
> chromeos-chrome-38.0.2111.0_alpha-r1:  * Installing test targets:
> libffmpegsumo.so media_unittests sandbox_linux_unittests
> ppapi_example_video_decode video_decode_accelerator_unittest
> video_encode_accelerator_unittest libppapi_tests.so chrome_sandbox
> chromeos-chrome-38.0.2111.0_alpha-r1: i686-pc-linux-gnu-strip:
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_example_video_decode':
> No such file
> chromeos-chrome-38.0.2111.0_alpha-r1: i686-pc-linux-gnu-strip:
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/libppapi_tests.so':
> No such file
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/lib32':
> No
> such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/lib64':
> No
> such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/mock_nacl_gdb':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_newlib.nmf':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_glibc.nmf':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_newlib_x32.nexe':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_newlib_x64.nexe':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_newlib_arm.nexe':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_glibc_x32.nexe':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_glibc_x64.nexe':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/ppapi_nacl_tests_glibc_arm.nexe':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/test_case.html':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/test_case.html.mock-http-headers':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/test_page.css':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/out_x86-
> generic/Release/test_url_loader_data':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync: link_stat
> "/home/chrome-bot/chrome_root/src/chrome/test/functional" failed: No such
> file
> or directory (2)
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were
> not
> transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/chrome/test/functional': No
> such file
> or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync: link_stat
> "/home/chrome-bot/chrome_root/src/content/common/gpu/testdata" failed: No
> such
> file or directory (2)
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were
> not
> transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/content/common/gpu/testdata':
> No such
> file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync: change_dir
> "/home/chrome-bot/chrome_root/src/third_party/accessibility-
> developer-tools/gen"
> failed: No such file or directory (2)
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were
> not
> transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/third_party/
> accessibility-developer-tools/gen/axs_testing.js':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: E_MACHINE not defined!
> chromeos-chrome-38.0.2111.0_alpha-r1: Copying Telemetry Framework into
> /build/x86-generic/tmp/portage/chromeos-base/chromeos-chrome-38.0.2111.0_
> alpha-r1/work/telemetry_src
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/tools/cros/': No such file or
> directory
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync: change_dir
> "/home/chrome-bot/chrome_root/src/third_party/android_tools/sdk" failed:
> No such
> file or directory (2)
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were
> not
> transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/third_party/
> android_tools/sdk/platform-tools':
> No such file or directory
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync: change_dir
> "/home/chrome-bot/chrome_root/src/third_party/android_tools/sdk" failed:
> No such
> file or directory (2)
> chromeos-chrome-38.0.2111.0_alpha-r1: rsync error: some files/attrs were
> not
> transferred (see previous errors) (code 23) at main.c(1052) [sender=3.0.9]
> chromeos-chrome-38.0.2111.0_alpha-r1: cp: cannot stat
> '/var/cache/chromeos-chrome/chrome-src/src/third_party/
> android_tools/sdk/platform-tools':
> No such file or directory
> ...
> chromeos-chrome-38.0.2111.0_alpha-r1:
> /var/cache/chromeos-chrome/chrome-src/src/out_x86-generic/Release/
> libppGoogleNaClPluginChrome.so
> does not exist and is required.
> chromeos-chrome-38.0.2111.0_alpha-r1: You can bypass this error with
> --sloppy.
>
> If this is intended, we need to fix chromeos-chrome-9999.ebuild as well.
>
> https://codereview.chromium.org/397243004/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[xiyuan, 2014-08-01 21:12:27]

On 2014/08/01 21:05:19, teravest wrote:
> Yes. This change removes that file. I looked through the chromeos overlay
> repository but didn't find a mention of that file name anywhere.
> 

Think libppGoogleNaClPluginChrome.so is defined in chromite/lib/chrome_util.py
and used by deploy_chrome.py to install chrome binaries.

Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
ppapi_nacl_tests_newlib.nmf. Those are defined in chromeos-chrome-9999.ebuild.

[teravest, 2014-08-01 21:24:15]

Reverted at r287090.

Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?

On Fri, Aug 1, 2014 at 3:12 PM,  <xiyuan@chromium.org> wrote:
> On 2014/08/01 21:05:19, teravest wrote:
>>
>> Yes. This change removes that file. I looked through the chromeos overlay
>> repository but didn't find a mention of that file name anywhere.
>
>
>
> Think libppGoogleNaClPluginChrome.so is defined in
> chromite/lib/chrome_util.py
> and used by deploy_chrome.py to install chrome binaries.
>
> Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
> ppapi_nacl_tests_newlib.nmf. Those are defined in
> chromeos-chrome-9999.ebuild.
>
> https://codereview.chromium.org/397243004/

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Lei Zhang, 2014-08-01 21:28:18]

When I needed to change .ebuild files, I get a CrOS checkout by
following http://www.chromium.org/chromium-os/developer-guide

I don't know if there is an easier way to make a small change like
that. I already have a CrOS checkout. If you want me to upload a CL,
let me know.

On Fri, Aug 1, 2014 at 2:23 PM, Justin TerAvest <teravest@chromium.org> wrote:
> Reverted at r287090.
>
> Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?
>
> On Fri, Aug 1, 2014 at 3:12 PM,  <xiyuan@chromium.org> wrote:
>> On 2014/08/01 21:05:19, teravest wrote:
>>>
>>> Yes. This change removes that file. I looked through the chromeos overlay
>>> repository but didn't find a mention of that file name anywhere.
>>
>>
>>
>> Think libppGoogleNaClPluginChrome.so is defined in
>> chromite/lib/chrome_util.py
>> and used by deploy_chrome.py to install chrome binaries.
>>
>> Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
>> ppapi_nacl_tests_newlib.nmf. Those are defined in
>> chromeos-chrome-9999.ebuild.
>>
>> https://codereview.chromium.org/397243004/

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[xiyuan, 2014-08-01 21:28:27]

Thank you for looking into this.

chromeos-chrome-9999.ebuild is
 in chromium/chromiumos/overlays/chromiumos-overlay:

https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/ma...


On Fri, Aug 1, 2014 at 2:23 PM, Justin TerAvest <teravest@chromium.org>
wrote:

> Reverted at r287090.
>
> Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?
>
> On Fri, Aug 1, 2014 at 3:12 PM,  <xiyuan@chromium.org> wrote:
> > On 2014/08/01 21:05:19, teravest wrote:
> >>
> >> Yes. This change removes that file. I looked through the chromeos
> overlay
> >> repository but didn't find a mention of that file name anywhere.
> >
> >
> >
> > Think libppGoogleNaClPluginChrome.so is defined in
> > chromite/lib/chrome_util.py
> > and used by deploy_chrome.py to install chrome binaries.
> >
> > Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
> > ppapi_nacl_tests_newlib.nmf. Those are defined in
> > chromeos-chrome-9999.ebuild.
> >
> > https://codereview.chromium.org/397243004/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[teravest, 2014-08-01 21:45:20]

Looking at that ebuild file, it looks like copying those test files is
intentional, I suppose for running tests on ChromeOS. I'm not sure why
those files are not getting built in that configuration, so I'll take
a look at that.

On Fri, Aug 1, 2014 at 3:27 PM, Xiyuan Xia <xiyuan@chromium.org> wrote:
> Thank you for looking into this.
>
> chromeos-chrome-9999.ebuild is in
> chromium/chromiumos/overlays/chromiumos-overlay:
>
>
https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/ma...
>
>
> On Fri, Aug 1, 2014 at 2:23 PM, Justin TerAvest <teravest@chromium.org>
> wrote:
>>
>> Reverted at r287090.
>>
>> Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?
>>
>> On Fri, Aug 1, 2014 at 3:12 PM,  <xiyuan@chromium.org> wrote:
>> > On 2014/08/01 21:05:19, teravest wrote:
>> >>
>> >> Yes. This change removes that file. I looked through the chromeos
>> >> overlay
>> >> repository but didn't find a mention of that file name anywhere.
>> >
>> >
>> >
>> > Think libppGoogleNaClPluginChrome.so is defined in
>> > chromite/lib/chrome_util.py
>> > and used by deploy_chrome.py to install chrome binaries.
>> >
>> > Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
>> > ppapi_nacl_tests_newlib.nmf. Those are defined in
>> > chromeos-chrome-9999.ebuild.
>> >
>> > https://codereview.chromium.org/397243004/
>
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[teravest, 2014-08-04 20:02:23]

I suspect that chromeos-chrome-9999.ebuild is missing some dependencies for the
files it requires.

Xiyuan,
If i follow the instructions at
http://www.chromium.org/chromium-os/developer-guide and build an image, will it
perform the build steps listed in chromeos-chrome-9999.ebuild?

On 2014/08/01 21:45:20, teravest wrote:
> Looking at that ebuild file, it looks like copying those test files is
> intentional, I suppose for running tests on ChromeOS. I'm not sure why
> those files are not getting built in that configuration, so I'll take
> a look at that.
> 
> On Fri, Aug 1, 2014 at 3:27 PM, Xiyuan Xia <mailto:xiyuan@chromium.org> wrote:
> > Thank you for looking into this.
> >
> > chromeos-chrome-9999.ebuild is in
> > chromium/chromiumos/overlays/chromiumos-overlay:
> >
> >
>
https://chromium.googlesource.com/chromiumos/overlays/chromiumos-overlay/+/ma...
> >
> >
> > On Fri, Aug 1, 2014 at 2:23 PM, Justin TerAvest
<mailto:teravest@chromium.org>
> > wrote:
> >>
> >> Reverted at r287090.
> >>
> >> Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?
> >>
> >> On Fri, Aug 1, 2014 at 3:12 PM,  <mailto:xiyuan@chromium.org> wrote:
> >> > On 2014/08/01 21:05:19, teravest wrote:
> >> >>
> >> >> Yes. This change removes that file. I looked through the chromeos
> >> >> overlay
> >> >> repository but didn't find a mention of that file name anywhere.
> >> >
> >> >
> >> >
> >> > Think libppGoogleNaClPluginChrome.so is defined in
> >> > chromite/lib/chrome_util.py
> >> > and used by deploy_chrome.py to install chrome binaries.
> >> >
> >> > Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
> >> > ppapi_nacl_tests_newlib.nmf. Those are defined in
> >> > chromeos-chrome-9999.ebuild.
> >> >
> >> > https://codereview.chromium.org/397243004/
> >
> >
> 
> To unsubscribe from this group and stop receiving emails from it, send an
email
> to mailto:chromium-reviews+unsubscribe@chromium.org.

[xiyuan, 2014-08-04 20:12:51]

On 2014/08/04 20:02:23, teravest wrote:
> I suspect that chromeos-chrome-9999.ebuild is missing some dependencies for
the
> files it requires.
> 
> Xiyuan,
> If i follow the instructions at
> http://www.chromium.org/chromium-os/developer-guide and build an image, will
it
> perform the build steps listed in chromeos-chrome-9999.ebuild?
> 

Yes. After you do "cros_workon --board=${BOARD} start chromeos-chrome",
"CHROME_ORIGIN=LOCAL_SOURCE build_packages" or "emerge-${BOARD} chromeos-chrome"
should use chromeos-chrome-9999.ebuild. Otherwise, it would use the one of the
rc ebuild (forgot which one it would use by default).

[Mark Seaborn, 2014-08-04 20:22:27]

It troubles me that this change got reverted for breaking the Chromium OS
build, because there was an explicit decision earlier this year that the
Chromium OS buildbots should not be tree closers.  The cros_* trybots got
removed as a result, which leaves developers with no simple way to avoid
breaking the Chromium OS build.  (See
https://code.google.com/p/chromium/issues/detail?id=363401)

As far as I can tell from https://chromium-status.appspot.com/status_viewer,
the buildbot failure from this change didn't actually close the tree
(unless it occurred when the tree was already closed from another change?).

I'm just raising this because it creates hassle to have changes reverted
for breaking builds that aren't readily testable...

Cheers,
Mark

On 1 August 2014 14:23, Justin TerAvest <teravest@chromium.org> wrote:

> Reverted at r287090.
>
> Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?
>
> On Fri, Aug 1, 2014 at 3:12 PM,  <xiyuan@chromium.org> wrote:
> > On 2014/08/01 21:05:19, teravest wrote:
> >>
> >> Yes. This change removes that file. I looked through the chromeos
> overlay
> >> repository but didn't find a mention of that file name anywhere.
> >
> >
> >
> > Think libppGoogleNaClPluginChrome.so is defined in
> > chromite/lib/chrome_util.py
> > and used by deploy_chrome.py to install chrome binaries.
> >
> > Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
> > ppapi_nacl_tests_newlib.nmf. Those are defined in
> > chromeos-chrome-9999.ebuild.
> >
> > https://codereview.chromium.org/397243004/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[teravest, 2014-08-04 20:49:41]

It sounds like I was too eager to revert this. After committing this, I saw a
buildbot failure email, and a CrOS-related build failure making the chromium
waterfall red, but it sounds like that was an unrelated build error.

I have a change out for review to fix chromite, and I'll spend some time looking
over the ebuild file to try to figure out what's wrong, but if that takes too
long, I'll leave that to CrOS developers to fix.



On 2014/08/04 20:22:27, Mark Seaborn wrote:
> It troubles me that this change got reverted for breaking the Chromium OS
> build, because there was an explicit decision earlier this year that the
> Chromium OS buildbots should not be tree closers.  The cros_* trybots got
> removed as a result, which leaves developers with no simple way to avoid
> breaking the Chromium OS build.  (See
> https://code.google.com/p/chromium/issues/detail?id=363401)
> 
> As far as I can tell from https://chromium-status.appspot.com/status_viewer,
> the buildbot failure from this change didn't actually close the tree
> (unless it occurred when the tree was already closed from another change?).
> 
> I'm just raising this because it creates hassle to have changes reverted
> for breaking builds that aren't readily testable...
> 
> Cheers,
> Mark
> 
> On 1 August 2014 14:23, Justin TerAvest <mailto:teravest@chromium.org> wrote:
> 
> > Reverted at r287090.
> >
> > Xiyuan, in what repository can I find chromeos-chrome-9999.ebuild?
> >
> > On Fri, Aug 1, 2014 at 3:12 PM,  <mailto:xiyuan@chromium.org> wrote:
> > > On 2014/08/01 21:05:19, teravest wrote:
> > >>
> > >> Yes. This change removes that file. I looked through the chromeos
> > overlay
> > >> repository but didn't find a mention of that file name anywhere.
> > >
> > >
> > >
> > > Think libppGoogleNaClPluginChrome.so is defined in
> > > chromite/lib/chrome_util.py
> > > and used by deploy_chrome.py to install chrome binaries.
> > >
> > > Besides this, we also miss a bunch of ppapi_nacl_tests* files, e.g.
> > > ppapi_nacl_tests_newlib.nmf. Those are defined in
> > > chromeos-chrome-9999.ebuild.
> > >
> > > https://codereview.chromium.org/397243004/
> >
> 
> To unsubscribe from this group and stop receiving emails from it, send an
email
> to mailto:chromium-reviews+unsubscribe@chromium.org.

[xiyuan, 2014-08-04 21:08:41]

On 2014/08/04 20:49:41, teravest wrote:
> It sounds like I was too eager to revert this. After committing this, I saw a
> buildbot failure email, and a CrOS-related build failure making the chromium
> waterfall red, but it sounds like that was an unrelated build error.
> 
> I have a change out for review to fix chromite, and I'll spend some time
looking
> over the ebuild file to try to figure out what's wrong, but if that takes too
> long, I'll leave that to CrOS developers to fix.
> 
> 
> 
> On 2014/08/04 20:22:27, Mark Seaborn wrote:
> > It troubles me that this change got reverted for breaking the Chromium OS
> > build, because there was an explicit decision earlier this year that the
> > Chromium OS buildbots should not be tree closers.  The cros_* trybots got
> > removed as a result, which leaves developers with no simple way to avoid
> > breaking the Chromium OS build.  (See
> > https://code.google.com/p/chromium/issues/detail?id=363401)
> > 
> > As far as I can tell from https://chromium-status.appspot.com/status_viewer,
> > the buildbot failure from this change didn't actually close the tree
> > (unless it occurred when the tree was already closed from another change?).
> > 
> > I'm just raising this because it creates hassle to have changes reverted
> > for breaking builds that aren't readily testable...
> > 
> > Cheers,
> > Mark
> > 

Justin, thank you for promptly reverting the CL and that helped to make Cros
bots happy.

I cannot say I agree with the policy to make Cros bots not tree closers. From my
perspective, this makes it harder for cros sheriffs/gardeners to do their jobs.
There are so many different components in chromium. Without help from chromium
devs, it is very difficult to keep cros bots green.

[teravest, 2014-08-05 15:38:05]

On Mon, Aug 4, 2014 at 3:08 PM,  <xiyuan@chromium.org> wrote:
> On 2014/08/04 20:49:41, teravest wrote:
>>
>> It sounds like I was too eager to revert this. After committing this, I
>> saw a
>> buildbot failure email, and a CrOS-related build failure making the
>> chromium
>> waterfall red, but it sounds like that was an unrelated build error.
>
>
>> I have a change out for review to fix chromite, and I'll spend some time
>
> looking
>>
>> over the ebuild file to try to figure out what's wrong, but if that takes
>> too
>> long, I'll leave that to CrOS developers to fix.
>
>
>
>
>> On 2014/08/04 20:22:27, Mark Seaborn wrote:
>> > It troubles me that this change got reverted for breaking the Chromium
>> > OS
>> > build, because there was an explicit decision earlier this year that the
>> > Chromium OS buildbots should not be tree closers.  The cros_* trybots
>> > got
>> > removed as a result, which leaves developers with no simple way to avoid
>> > breaking the Chromium OS build.  (See
>> > https://code.google.com/p/chromium/issues/detail?id=363401)
>> >
>> > As far as I can tell from
>> > https://chromium-status.appspot.com/status_viewer,
>> > the buildbot failure from this change didn't actually close the tree
>> > (unless it occurred when the tree was already closed from another
>> > change?).
>> >
>> > I'm just raising this because it creates hassle to have changes reverted
>> > for breaking builds that aren't readily testable...
>> >
>> > Cheers,
>> > Mark
>> >
>
>
> Justin, thank you for promptly reverting the CL and that helped to make Cros
> bots happy.
>
> I cannot say I agree with the policy to make Cros bots not tree closers.
> From my
> perspective, this makes it harder for cros sheriffs/gardeners to do their
> jobs.
> There are so many different components in chromium. Without help from
> chromium
> devs, it is very difficult to keep cros bots green.

I fought with this yesterday afternoon and this morning, and still
can't get a CrOS build to use my local chromium repository, even when
passing "USE=chrome_internal" and "--chrome_root=..." to cros_sdk, and
setting CHROME_ORIGIN=LOCAL_SOURCE.

I've tried building the specified ebuild file in two different ways:
$ ebuild
../third_party/chromiumos-overlay/chromeos-base/chromeos-chrome/chromeos-chrome-9999.ebuild
 compile

$ ./build_packages --board=x86-generic chromeos-base/chromeos-chrome

... but with both methods, I eventually hit a gles2 error as specified
in https://code.google.com/p/chromium/issues/detail?id=394423, even
though I'm using a x86-generic board type.

I assume there has to be a straightforward way to do this, but I
haven't found anything that works, after checking the documentation
and asking on IRC.

>
> https://codereview.chromium.org/397243004/

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[xiyuan, 2014-08-05 15:49:47]

On 2014/08/05 15:38:05, teravest wrote:
> I fought with this yesterday afternoon and this morning, and still
> can't get a CrOS build to use my local chromium repository, even when
> passing "USE=chrome_internal" and "--chrome_root=..." to cros_sdk, and
> setting CHROME_ORIGIN=LOCAL_SOURCE.
> 
> I've tried building the specified ebuild file in two different ways:
> $ ebuild
>
../third_party/chromiumos-overlay/chromeos-base/chromeos-chrome/chromeos-chrome-9999.ebuild
>  compile
> 
> $ ./build_packages --board=x86-generic chromeos-base/chromeos-chrome
> 
> ... but with both methods, I eventually hit a gles2 error as specified
> in https://code.google.com/p/chromium/issues/detail?id=394423, even
> though I'm using a x86-generic board type.
> 
> I assume there has to be a straightforward way to do this, but I
> haven't found anything that works, after checking the documentation
> and asking on IRC.
> 

If you passed the correct chrome_root to cros_sdk and set
CHROME_ORIGIN=LOCAL_SOURCE in the chroot's env, build_packages should work. Not
sure about ebuild, I would think it should be a wrapped ebuild-${BOARD},
something like ebuild-x86-generic.

If you have USE=chrome_internal, then you also need src-internal with chromium
check out. Maybe that's why you see the gles error.

[teravest, 2014-08-05 17:53:44]

On Tue, Aug 5, 2014 at 9:49 AM,  <xiyuan@chromium.org> wrote:
> On 2014/08/05 15:38:05, teravest wrote:
>>
>> I fought with this yesterday afternoon and this morning, and still
>> can't get a CrOS build to use my local chromium repository, even when
>> passing "USE=chrome_internal" and "--chrome_root=..." to cros_sdk, and
>> setting CHROME_ORIGIN=LOCAL_SOURCE.
>
>
>> I've tried building the specified ebuild file in two different ways:
>> $ ebuild
>
>
>
../third_party/chromiumos-overlay/chromeos-base/chromeos-chrome/chromeos-chrome-9999.ebuild
>>
>>   compile
>
>
>> $ ./build_packages --board=x86-generic chromeos-base/chromeos-chrome
>
>
>> ... but with both methods, I eventually hit a gles2 error as specified
>> in https://code.google.com/p/chromium/issues/detail?id=394423, even
>> though I'm using a x86-generic board type.
>
>
>> I assume there has to be a straightforward way to do this, but I
>> haven't found anything that works, after checking the documentation
>> and asking on IRC.
>
>
>
> If you passed the correct chrome_root to cros_sdk and set
> CHROME_ORIGIN=LOCAL_SOURCE in the chroot's env, build_packages should work.
> Not
> sure about ebuild, I would think it should be a wrapped ebuild-${BOARD},
> something like ebuild-x86-generic.
>
> If you have USE=chrome_internal, then you also need src-internal with
> chromium
> check out. Maybe that's why you see the gles error.

Okay, I finally got it to build without USE=chrome_internal locally.
The build succeeds, with the chrome patch applied.

Completed chromeos-base/chromeos-chrome-9999 (in 105m29.9s)
...
Merge complete
Done
Builds complete
INFO    : Elapsed time (build_packages): 109m57s
Done

However, if I watch the logs during the package building process, I
see the same errors that were reported by the buildbot. Are those
required for installing the package, or just debug messages that
always get printed, because something is broken in there?

I see other files that are missing as well, and I'm wondering if the
missing files besides libppGoogleNaCl... always fail to copy. I looked
through buildbot logs to see if this was the case, but we don't print
any details from this build step on success, so there's no way to
know.

Now that the chromite fix is in, I'm inclined to commit this again and
see what happens.

>
> https://codereview.chromium.org/397243004/

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[xiyuan, 2014-08-05 18:01:33]

On 2014/08/05 17:53:44, teravest wrote:
> Okay, I finally got it to build without USE=chrome_internal locally.
> The build succeeds, with the chrome patch applied.
> 
> Completed chromeos-base/chromeos-chrome-9999 (in 105m29.9s)
> ...
> Merge complete
> Done
> Builds complete
> INFO    : Elapsed time (build_packages): 109m57s
> Done
> 
> However, if I watch the logs during the package building process, I
> see the same errors that were reported by the buildbot. Are those
> required for installing the package, or just debug messages that
> always get printed, because something is broken in there?
> 
> I see other files that are missing as well, and I'm wondering if the
> missing files besides libppGoogleNaCl... always fail to copy. I looked
> through buildbot logs to see if this was the case, but we don't print
> any details from this build step on success, so there's no way to
> know.
> 
> Now that the chromite fix is in, I'm inclined to commit this again and
> see what happens.

We should be good if we see "Completed chromeos-base/chromeos-chrome-9999"
message from build_packages. I have not watched the build log closely so you are
probably right that those errors have always been there. Thank you so much for
digging through and trying it out.

[teravest, 2014-08-05 19:53:30]

Committed patchset #9 manually as 287580 (presubmit successful).