1462019 - Developer tools are unable to attach to applications with embedded GeckoView when targetSDKversion is 26

Status: RESOLVED FIXED

(GeckoView :: General, defect, P1)

Description

[Randall Barker [:rbarker]]

After setting: mRuntime.getSettings().setRemoteDebuggingEnabled(true); the developer tools are unable to attach to an embedded GeckoView. It fails with the error message:

05-16 08:38:41.016   793   793 E adbd    : failed to connect to socket 'localfilesystem:/data/data/org.mozilla.vrbrowser/firefox-debugger-socket': Permission denied

Comment 1

[:Harald Kirschner :digitarald]

ni? :jchen, who had some ideas for fixing this ("may be able to use abstract unix sockets").

Comment 2

[Randall Barker [:rbarker]]

As :jchen found, it is related to target SDK and not Android version. I have to target API 26 for VR apps and debugging is broken on Android N as well.

Comment 3

[(inactive) Jim Chen [:jchen] [:darchons]]

(In reply to :Harald Kirschner :digitarald from comment #1)
> ni? :jchen, who had some ideas for fixing this ("may be able to use abstract
> unix sockets").

Linux supports an abstract namespace for unix domain sockets that is not tied to the filesystem. The idea is maybe that will get around the Android N+ restriction on access to the app-private data directory.

Comment 4

[David Bolter [:davidb] (NeedInfo me for attention)]

Doesn't technically block but we need this asap.

Comment 5

[Makoto Kato [:m_kato]]

(In reply to Jim Chen [:jchen] [:darchons] from comment #3)
> (In reply to :Harald Kirschner :digitarald from comment #1)
> > ni? :jchen, who had some ideas for fixing this ("may be able to use abstract
> > unix sockets").
> 
> Linux supports an abstract namespace for unix domain sockets that is not
> tied to the filesystem. The idea is maybe that will get around the Android
> N+ restriction on access to the app-private data directory.

I see.  devtoos/shared/security/socket.js uses domain socket, so it might be better to change to TCP socket.  I need research chrome's way.

Comment 6

[Brian Birtles (:birtles)]

Makoto, can you please report back here once you've investigated the issue and our options for addressing this. Thanks.

Comment 7

[James Willcox (:snorp) ([e-mail]) (he/him)]

In Fennec Nightly, I get a the following error on startup when USB debugging is enabled:

05-31 09:23:32.909 28527 28546 D GeckoBrowser: Unable to start USB debugger server: Error: Can't create listener, remote debugging disabled

Apparently it fails to even create the socket, which seems like a different problem.

Comment 8

[Makoto Kato [:m_kato]]

Chrome still has domain socket code for debugger protocol, and they has additional adb forward code (for network debugging?).  It may be better to use adb forward for some devices.  (add adb forward command to adbhelper?).

Also, even if Android O, my Nexus 5X (latest 8.1) can connect to domain socket. (It means that USB debug works well on Android 8.1).  So it depends on device security.

Comment 9

[(inactive) Jim Chen [:jchen] [:darchons]]

It depends on the target SDK version, so for testing you have to change the target SDK to something higher, for example 26.

Comment 10

[:Harald Kirschner :digitarald]

Should we open a new bug for comment 7, or do we expect the same root cause for both?

Comment 11

[:Harald Kirschner :digitarald]

Updating because of impact on productivity (and to prioritize the diagnosis work) for visibility.

Comment 12

[Panos Astithas (he/him) [:past] (please ni?)]

I believe we resolved comment 7 in Slack last week. The problem was that the pref that controls remote debugging was out of sync with the setting that is controlled by the UI. Perhaps that merits another bug, but I'm not sure if we can get good STR.

Comment 13

[:Harald Kirschner :digitarald]

The first STR/error is different though; or did that also get resolved?

Comment 14

[Panos Astithas (he/him) [:past] (please ni?)]

Right, the original STR point to a problem with the newer SDK versions, or at least this is our current theory.

Comment 15

[Makoto Kato [:m_kato]]

(In reply to Jim Chen [:jchen] [:darchons] from comment #9)
> It depends on the target SDK version, so for testing you have to change the
> target SDK to something higher, for example 26.

Since bug 1352015 is landed, I think Nightly already use SDK 26.  So I am confused...

(In reply to :Harald Kirschner :digitarald from comment #10)
> Should we open a new bug for comment 7, or do we expect the same root cause
> for both?

No, comment #7 is prefs issue like comment #12

Comment 16

[(inactive) Jim Chen [:jchen] [:darchons]]

The target SDK version [1] is currently set to 23 [2][3] (that's different from the compile SDK version of 26)

[1] https://searchfox.org/mozilla-central/rev/292d295d6b084b43b70de26a42e68513bb7b36a3/mobile/android/base/AndroidManifest.xml.in#17
[2] https://searchfox.org/mozilla-central/rev/292d295d6b084b43b70de26a42e68513bb7b36a3/build/autoconf/android.m4#128
[3] https://searchfox.org/mozilla-central/rev/292d295d6b084b43b70de26a42e68513bb7b36a3/old-configure.in#2034

Comment 17

[Makoto Kato [:m_kato]]

(In reply to Jim Chen [:jchen] [:darchons] from comment #16)
> The target SDK version [1] is currently set to 23 [2][3] (that's different
> from the compile SDK version of 26)

Thanks

Comment 18

[Hiroyuki Ikezoe (:hiro)]

OK, I can connect geckoview_example which was locally built with SDK 26.

I did change setRemoteDebuggingEnabled(true) in loadSettings [1].  So the problem was solved?

[1] https://hg.mozilla.org/mozilla-central/file/d8f180ab7492/mobile/android/geckoview_example/src/main/java/org/mozilla/geckoview_example/GeckoViewActivity.java#l142

Comment 19

[Makoto Kato [:m_kato]]

chrome uses abstract socket (@chrome_devtools_remote).  So I think that we should try abstract socket whether this can be fixed.

Comment 20

[Makoto Kato [:m_kato]]

Agh, NSPR and nsIServerSocket don't support abstract socket such as bug 979535 ...

Comment 21

[Makoto Kato [:m_kato]]

When I create WIP that is abstract socket version, it can connect to Fennec that is target sdk version is 26.  But we need fix NSPR, adbhelper and etc...

Comment 22

[Petru-Mugurel Lingurar [:petru]]

We are working on finishing the work needed to support Oreo (metas bug 1450450 and bug 1349309) with the plan to start targeting sdk 26 before the code freeze for 62 which is next Thursday, June 14.
Is this bug (and it's dependencies) a blocker for that?

Comment 23

[Makoto Kato [:m_kato]]

(In reply to Petru-Mugurel Lingurar[:petru] from comment #22)
> We are working on finishing the work needed to support Oreo (metas bug
> 1450450 and bug 1349309) with the plan to start targeting sdk 26 before the
> code freeze for 62 which is next Thursday, June 14.
> Is this bug (and it's dependencies) a blocker for that?

I think that it may be blocker on Android O+ device since remote debugger via USB connection doesn't work on Android O+ device after setting target SDK is 26+.  But this support requires a lot of changes for various components such as NSPR, necko and etc

But Fennec team should answer for this.

Comment 24

[James Willcox (:snorp) ([e-mail]) (he/him)]

I'm pretty sure it was already broken in other cases, SDK 26 just makes it worse. This bug shouldn't block the SDK 26 change IMHO.

Comment 25

[Jan Henning [:JanH]]

(In reply to James Willcox (:snorp) ([e-mail]) from comment #24)
> I'm pretty sure it was already broken in other cases, SDK 26 just makes it
> worse. This bug shouldn't block the SDK 26 change IMHO.

If I only had an Oreo phone I'd object strongly, but since I'm still stuck on Marshmallow for better or for worse this doesn't affect me personally. Unless things really are already completely broken on Android O, this might affect other people, though...

(In reply to Panos Astithas [:past] (please ni?) from comment #12)
> I believe we resolved comment 7 in Slack last week. The problem was that the
> pref that controls remote debugging was out of sync with the setting that is
> controlled by the UI. Perhaps that merits another bug, but I'm not sure if
> we can get good STR.

Bug 1468070.

Comment 26

[Panos Astithas (he/him) [:past] (please ni?)]

NSPR and ADB helper changes have landed. I assume the next step here is to change the default value for devtools.debugger.unix-domain-socket and ship the ADB helper update?

Comment 27

[Makoto Kato [:m_kato]]

(In reply to Panos Astithas [:past] (please ni?) from comment #26)
> NSPR and ADB helper changes have landed. I assume the next step here is to
> change the default value for devtools.debugger.unix-domain-socket and ship
> the ADB helper update?

At first, I will add new method to nsIServerSocket and write test (inc. add new method to nsISocketTransportService), then call new method in devtools/shared/security/socket.js.

Comment 28

[Makoto Kato [:m_kato]]

https://treeherder.mozilla.org/#/jobs?repo=try&revision=b5ecb68cff4957d7833846e702c0f0cde7743027

Comment 29

[Makoto Kato [:m_kato]]

Attachment: Part 1. Support abstract socket address on nsIServerSocket and nsISocketTransportService

Comment 30

[Makoto Kato [:m_kato]]

Attachment: Part 1. Support abstract socket address on nsIServerSocket and nsISocketTransportService

Comment 31

[Makoto Kato [:m_kato]]

Comment on attachment 8989676 [details] [diff] [review]
Part 1. Support abstract socket address on nsIServerSocket and nsISocketTransportService

Actually, nsIServerSocket doesn't support abstract socket address (that support on Linux only) of Unix domain socket.  Android O doesn't allow file path based Unix domain socket from other process due to permission, so we need it to support remote debugging via USB connection.  (Chrome uses this socket address type via USB).

- nsIServerSocket.initWithAbstractAddress creats the Unix domain socket that uses abstract socket address.
- nsISocketTransportService.createUnixDomainAbstractAddressTransport opens the Unix domain socket that uses abstract socket address

Although nsISocketTransportService.createUnixDomainAbstractAddressTransport doesn't require Android's remote connection since this method is client side, it is for unit test.

And, NSPR side is bug 1466771 and it is already landed in NSPR upstream.

Comment 32

[Makoto Kato [:m_kato]]

Attachment: Part 2. devtools should use abstract socket address for USB connection (WIP)

Comment 33

[Chris Peterson [:cpeterson]]

I am marking status-firefox62=wontfix because I assume we will not want to uplift NSPR changes to Beta 62. Plus this is a developer debugging tool and probably less important for Beta.

Comment 34

[Sebastian Kaspari (:sebastian; :pocmo)]

> I am marking status-firefox62=wontfix because I assume we will not want to uplift NSPR changes to Beta 62.

In this case we want to hide the setting from Focus for the release that's using GV 62.0. I'll file an issue.

Comment 35

[Honza Bambas (:mayhemer)]

Comment on attachment 8989676 [details] [diff] [review]
Part 1. Support abstract socket address on nsIServerSocket and nsISocketTransportService

Review of attachment 8989676 [details] [diff] [review]:
-----------------------------------------------------------------

sorry for the delay, I'm pretty busy these days.

::: netwerk/base/nsServerSocket.cpp
@@ +322,5 @@
>  NS_IMETHODIMP
> +nsServerSocket::InitWithAbstractAddress(const nsACString& aName,
> +                                        int32_t aBacklog)
> +{
> +#if defined(XP_LINUX)

#ifdef XP_...

and there is somewhat a mishmash between XP_LINUX and XP_UNIX here.  The existing code for domain sockets is XP_UNIX.

please explain well why XP_LINUX is the thing here (android?)

@@ +326,5 @@
> +#if defined(XP_LINUX)
> +  // Create an abstract socket address PRNetAddr referring to the name
> +  PRNetAddr addr;
> +  if (aName.Length() > sizeof(addr.local.path) - 2)
> +    return NS_ERROR_FILE_NAME_TOO_LONG;

always brace the body:

if () {
}

::: netwerk/base/nsSocketTransport2.cpp
@@ +938,1 @@
>          return NS_ERROR_FILE_NAME_TOO_LONG;

when you are here please:

if () {
}

@@ +943,5 @@
> +        mHost.Assign(name + 1, length - 1);
> +#else
> +        return NS_ERROR_SOCKET_ADDRESS_NOT_SUPPORTED;
> +#endif
> +    } else {

please try to write this a way that there is no else after return in the non-linux case

::: netwerk/base/nsSocketTransportService2.cpp
@@ +843,5 @@
> +nsSocketTransportService::CreateUnixDomainAbstractAddressTransport(
> +                              const nsACString& aName,
> +                              nsISocketTransport **result)
> +{
> +   // Abstract socket address is supported on Linux only

indent

::: netwerk/test/unit/test_unix_domain.js
@@ +556,5 @@
>  }
> +
> +function test_abstract_address_socket()
> +{
> +  const socketname = "abstractsocket";

isn't the limit only 5 bytes?

Comment 36

[Chris Peterson [:cpeterson]]

This bug is important for developers, but I'm tagging it [geckoview:klar:p2] because this is not a Focus+GV blocker.

Comment 37

[Makoto Kato [:m_kato]]

Attachment: Part 2. devtools should use abstract socket address for USB connection

Comment 38

[Makoto Kato [:m_kato]]

Comment on attachment 8992534 [details] [diff] [review]
Part 2. devtools should use abstract socket address for USB connection

I would like to change USB connection way to use abstract socket address since we supports Android 8+.  Part 1's patch adds abstract socket address to server socket, so devtools should use it when socket name isn't abstract path.

I still keep old way if using abstract file path on devtools.debugger.unix-domain-socket prefs.  But GeckoView's app only supports new way that uses abstract socket address.

Also, abstract socket name for USB connection is "<package name>/firefox-debugger-socket".  If OK, I will send PR/merge this change to adbhelper.

Comment 39

[J. Ryan Stinnett [:jryans] (Use needinfo, replies may be slow)]

Comment on attachment 8992534 [details] [diff] [review]
Part 2. devtools should use abstract socket address for USB connection

Review of attachment 8992534 [details] [diff] [review]:
-----------------------------------------------------------------

Thanks, it looks good to me!

Comment 40

[Makoto Kato [:m_kato]]

(In reply to (away till 13.8.) Honza Bambas (:mayhemer) from comment #35)
> Comment on attachment 8989676 [details] [diff] [review]
> Part 1. Support abstract socket address on nsIServerSocket and
> nsISocketTransportService
> 
> Review of attachment 8989676 [details] [diff] [review]:
> -----------------------------------------------------------------
> 
> sorry for the delay, I'm pretty busy these days.
> 
> ::: netwerk/base/nsServerSocket.cpp
> @@ +322,5 @@
> >  NS_IMETHODIMP
> > +nsServerSocket::InitWithAbstractAddress(const nsACString& aName,
> > +                                        int32_t aBacklog)
> > +{
> > +#if defined(XP_LINUX)
> 
> #ifdef XP_...
> 
> and there is somewhat a mishmash between XP_LINUX and XP_UNIX here.  The
> existing code for domain sockets is XP_UNIX.

UNIX domain socket supprots non-Windows.  (Windows 10 the latest supports it as OS level, but NSPR doesn't)

> please explain well why XP_LINUX is the thing here (android?)

Yes.  this supports on Linux and Android.  Abstract socket address only supports on Linux kernel.

> ::: netwerk/test/unit/test_unix_domain.js
> @@ +556,5 @@
> >  }
> > +
> > +function test_abstract_address_socket()
> > +{
> > +  const socketname = "abstractsocket";
> 
> isn't the limit only 5 bytes?

This limitation is https://searchfox.org/mozilla-central/source/nsprpub/pr/include/prio.h#178

Comment 41

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/6fb8d7cb6c29
Part 1. Support abstract socket address on nsIServerSocket and nsISocketTransportService. r=mayhemer
https://hg.mozilla.org/integration/mozilla-inbound/rev/0b0036d13e81
Part 2. Use abstract socket address for USB remote debugging. r=jryans

Comment 42

[Stefan Hindli [:stefan_hindli]]

Backed out 2 changesets (Bug 1462019) for org.mozilla.geckoview.test.AccessibilityTest.testMoveByCharacter failures on a CLOSED TREE

Failure: https://treeherder.mozilla.org/#/jobs?repo=mozilla-inbound&filter-resultStatus=testfailed&filter-resultStatus=busted&filter-resultStatus=exception&filter-classifiedState=unclassified&selectedJob=190209634
Failure log: https://treeherder.mozilla.org/logviewer.html#?job_id=190209634&repo=mozilla-inbound&lineNumber=1543+

[task 2018-07-26T09:50:44.997Z] 09:50:44  WARNING -  TEST-UNEXPECTED-FAIL | org.mozilla.geckoview.test.AccessibilityTest.testCheckbox | status -2
[task 2018-07-26T09:50:44.998Z] 09:50:44     INFO -  TEST-INFO took 22061ms
[task 2018-07-26T09:50:45.088Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: id=AndroidJUnitRunner
[task 2018-07-26T09:50:45.088Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: current=12
[task 2018-07-26T09:50:45.089Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: class=org.mozilla.geckoview.test.AccessibilityTest
[task 2018-07-26T09:50:45.089Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: stream=
[task 2018-07-26T09:50:45.090Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: numtests=163
[task 2018-07-26T09:50:45.090Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: test=testMoveByCharacter
[task 2018-07-26T09:50:45.091Z] 09:50:45     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS_CODE: 1
[task 2018-07-26T09:50:45.091Z] 09:50:45     INFO -  TEST-START | org.mozilla.geckoview.test.AccessibilityTest.testMoveByCharacter
[task 2018-07-26T09:50:47.494Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: id=AndroidJUnitRunner
[task 2018-07-26T09:50:47.495Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: current=12
[task 2018-07-26T09:50:47.495Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: class=org.mozilla.geckoview.test.AccessibilityTest
[task 2018-07-26T09:50:47.496Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: stream=
[task 2018-07-26T09:50:47.497Z] 09:50:47     INFO -  org.mozilla.geckoview.test | Error in testMoveByCharacter(org.mozilla.geckoview.test.AccessibilityTest):
[task 2018-07-26T09:50:47.497Z] 09:50:47     INFO -  org.mozilla.geckoview.test | java.lang.RuntimeException: java.io.IOException: No such file or directory
[task 2018-07-26T09:50:47.498Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rdp.RDPConnection.<init>(RDPConnection.java:247)
[task 2018-07-26T09:50:47.498Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule.waitForOpenSession(GeckoSessionTestRule.java:1339)
[task 2018-07-26T09:50:47.499Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule.openSession(GeckoSessionTestRule.java:1326)
[task 2018-07-26T09:50:47.499Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule.prepareStatement(GeckoSessionTestRule.java:1298)
[task 2018-07-26T09:50:47.500Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule$4.evaluate(GeckoSessionTestRule.java:1485)
[task 2018-07-26T09:50:47.500Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.support.test.internal.statement.UiThreadStatement$1.run(UiThreadStatement.java:44)
[task 2018-07-26T09:50:47.500Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.app.Instrumentation$SyncRunnable.run(Instrumentation.java:1719)
[task 2018-07-26T09:50:47.501Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.os.Handler.handleCallback(Handler.java:730)
[task 2018-07-26T09:50:47.501Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.os.Handler.dispatchMessage(Handler.java:92)
[task 2018-07-26T09:50:47.502Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.os.Looper.loop(Looper.java:137)
[task 2018-07-26T09:50:47.502Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.app.ActivityThread.main(ActivityThread.java:5103)
[task 2018-07-26T09:50:47.503Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at java.lang.reflect.Method.invokeNative(Native Method)
[task 2018-07-26T09:50:47.503Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at java.lang.reflect.Method.invoke(Method.java:525)
[task 2018-07-26T09:50:47.503Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:737)
[task 2018-07-26T09:50:47.503Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:553)
[task 2018-07-26T09:50:47.504Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at dalvik.system.NativeStart.main(Native Method)
[task 2018-07-26T09:50:47.504Z] 09:50:47     INFO -  org.mozilla.geckoview.test | Caused by: java.io.IOException: No such file or directory
[task 2018-07-26T09:50:47.505Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.net.LocalSocketImpl.connectLocal(Native Method)
[task 2018-07-26T09:50:47.505Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.net.LocalSocketImpl.connect(LocalSocketImpl.java:258)
[task 2018-07-26T09:50:47.505Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.net.LocalSocket.connect(LocalSocket.java:108)
[task 2018-07-26T09:50:47.506Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rdp.RDPConnection.<init>(RDPConnection.java:240)
[task 2018-07-26T09:50:47.506Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	... 15 more
[task 2018-07-26T09:50:47.507Z] 09:50:47     INFO -  org.mozilla.geckoview.test |
[task 2018-07-26T09:50:47.507Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: numtests=163
[task 2018-07-26T09:50:47.508Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: stack=java.lang.RuntimeException: java.io.IOException: No such file or directory
[task 2018-07-26T09:50:47.508Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rdp.RDPConnection.<init>(RDPConnection.java:247)
[task 2018-07-26T09:50:47.509Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule.waitForOpenSession(GeckoSessionTestRule.java:1339)
[task 2018-07-26T09:50:47.509Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule.openSession(GeckoSessionTestRule.java:1326)
[task 2018-07-26T09:50:47.510Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule.prepareStatement(GeckoSessionTestRule.java:1298)
[task 2018-07-26T09:50:47.510Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rule.GeckoSessionTestRule$4.evaluate(GeckoSessionTestRule.java:1485)
[task 2018-07-26T09:50:47.511Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.support.test.internal.statement.UiThreadStatement$1.run(UiThreadStatement.java:44)
[task 2018-07-26T09:50:47.511Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.app.Instrumentation$SyncRunnable.run(Instrumentation.java:1719)
[task 2018-07-26T09:50:47.511Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.os.Handler.handleCallback(Handler.java:730)
[task 2018-07-26T09:50:47.512Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.os.Handler.dispatchMessage(Handler.java:92)
[task 2018-07-26T09:50:47.513Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.os.Looper.loop(Looper.java:137)
[task 2018-07-26T09:50:47.514Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.app.ActivityThread.main(ActivityThread.java:5103)
[task 2018-07-26T09:50:47.515Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at java.lang.reflect.Method.invokeNative(Native Method)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at java.lang.reflect.Method.invoke(Method.java:525)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at com.android.internal.os.ZygoteInit$MethodAndArgsCaller.run(ZygoteInit.java:737)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at com.android.internal.os.ZygoteInit.main(ZygoteInit.java:553)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at dalvik.system.NativeStart.main(Native Method)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | Caused by: java.io.IOException: No such file or directory
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.net.LocalSocketImpl.connectLocal(Native Method)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.net.LocalSocketImpl.connect(LocalSocketImpl.java:258)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at android.net.LocalSocket.connect(LocalSocket.java:108)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	at org.mozilla.geckoview.test.rdp.RDPConnection.<init>(RDPConnection.java:240)
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | 	... 15 more
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test |
[task 2018-07-26T09:50:47.516Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS: test=testMoveByCharacter
[task 2018-07-26T09:50:47.612Z] 09:50:47     INFO -  org.mozilla.geckoview.test | INSTRUMENTATION_STATUS_CODE: -2
[task 2018-07-26T09:50:47.613Z] 09:50:47  WARNING -  TEST-UNEXPECTED-FAIL | org.mozilla.geckoview.test.AccessibilityTest.testMoveByCharacter | status -2

Comment 43

[Pulsebot]

Backout by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/0f171ed1fd00
Backed out 2 changesets for  org.mozilla.geckoview.test.AccessibilityTest.testMoveByCharacter failures on a CLOSED TREE

Comment 44

[Makoto Kato [:m_kato]]

Attachment: Part 3. GeckoView test should use abstract socket address

Comment 45

[Makoto Kato [:m_kato]]

Comment on attachment 8995410 [details] [diff] [review]
Part 3. GeckoView test should use abstract socket address

Since I change remote debugger socket to abstract socket address, GeckoView's test  should use it.

Comment 46

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/mozilla-inbound/rev/f2273043a332
Part 1. Support abstract socket address on nsIServerSocket and nsISocketTransportService. r=mayhemer
https://hg.mozilla.org/integration/mozilla-inbound/rev/ef06cf98451c
Part 2. Use abstract socket address for USB remote debugging. r=jryans
https://hg.mozilla.org/integration/mozilla-inbound/rev/45da2e0ebe71
Part 3. GeckoView test should use abstract socket address. r=snorp

Comment 47

[Andreea Pavel [:apavel]]

https://hg.mozilla.org/mozilla-central/rev/f2273043a332
https://hg.mozilla.org/mozilla-central/rev/ef06cf98451c
https://hg.mozilla.org/mozilla-central/rev/45da2e0ebe71

Comment 48

[Markus Stange [:mstange]]

Should this be RESOLVED FIXED now?

Comment 49

[Hiroyuki Ikezoe (:hiro)]

I believe so.

Comment 50

[Brian Birtles (:birtles)]

Bear in mind that we still need to release a new version of ADB Helper before this will actually work.

Comment 51

[:Harald Kirschner :digitarald]

(In reply to Brian Birtles (:birtles) from comment #50)
> Bear in mind that we still need to release a new version of ADB Helper
> before this will actually work.

Filed as bug 1480259 to track progress.

Comment 52

[James Willcox (:snorp) ([e-mail]) (he/him)]

:m_kato, can you request uplift for 62 please?

Comment 53

[Makoto Kato [:m_kato]]

(In reply to James Willcox (:snorp) ([e-mail]) from comment #52)
> :m_kato, can you request uplift for 62 please?

This requires NSPR 4.20 beta+, but NSPR 4.20 will be in Gecko 63 according to bug 1477680.  If we have to land this in 62, NSPR have to update to 4.20, but it isn't released yet.

Comment 55

[Chris Peterson [:cpeterson]]

status-geckoview62=wontfix because we won't be able to uplift the NSPR fixes to the GECKOVIEW_62_RELBRANCH release branch.