Native SSID extraction moved to platform code on Android

On Android, move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.

Committed: https://crrev.com/35f59b4376232b4685c2a2e922f00a3bce2fbfde
Cr-Commit-Position: refs/heads/master@{#374743}

[Ryan Hamilton, 2016-01-26 20:15:50]

lgtm

https://codereview.chromium.org/1633733005/diff/1/net/android/network_library.h
File net/android/network_library.h (right):

https://codereview.chromium.org/1633733005/diff/1/net/android/network_library...
net/android/network_library.h:85: // Returns the current SSID if device is
connected to Wi-Fi network.
nit: "connected to a Wi-Fi network" (add "a")

[Ryan Sleevi, 2016-01-26 20:22:26]

rsleevi@chromium.org changed reviewers:
+ rsleevi@chromium.org

[Ryan Sleevi, 2016-01-26 20:22:26]

Please do *not* commit this yet. I have an important design  request here

Further, I'd like to suggest a reword of the commit description:

On Android, Move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
File net/base/network_quality_estimator.cc (right):

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
net/base/network_quality_estimator.cc:782: network_id.id =
android::GetCurrentWifiConnectionSsid();
DESIGN: Given that network_interfaces is already designed in a
platform-abstracted way to avoid these #ifdefs, I would instead recommend the
following approach:

In network_interfaces_linux.cc, guard GetWifiSSID by a

#if !defined(OS_ANDROID)
std::string GetWifiSSID() {
  ...
}
#endif

Then, implement a network_interfaces_android.cc which implements the GetWifiSSID
function, and either calls android::GetCurrentWifiConnectionSsid() or simply
move that implementation logic (the JNI call) into the _android file.

Add the _android file to the .gyp and .gn files, conditionally as appropriate.

This ensures the API semantics of net::GetWifiSSID() remain the same, and also
avoids the inconsistent naming being introduced here (SSID vs Ssid)

[Ryan Hamilton, 2016-01-26 20:24:48]

not lgtm

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
File net/base/network_quality_estimator.cc (right):

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
net/base/network_quality_estimator.cc:782: network_id.id =
android::GetCurrentWifiConnectionSsid();
On 2016/01/26 20:22:26, Ryan Sleevi wrote:
> DESIGN: Given that network_interfaces is already designed in a
> platform-abstracted way to avoid these #ifdefs, I would instead recommend the
> following approach:
> 
> In network_interfaces_linux.cc, guard GetWifiSSID by a
> 
> #if !defined(OS_ANDROID)
> std::string GetWifiSSID() {
>   ...
> }
> #endif
> 
> Then, implement a network_interfaces_android.cc which implements the
GetWifiSSID
> function, and either calls android::GetCurrentWifiConnectionSsid() or simply
> move that implementation logic (the JNI call) into the _android file.
> 
> Add the _android file to the .gyp and .gn files, conditionally as appropriate.
> 
> This ensures the API semantics of net::GetWifiSSID() remain the same, and also
> avoids the inconsistent naming being introduced here (SSID vs Ssid)

Oo! Good idea.

[ripp, 2016-01-27 12:07:24]

Description was changed from

==========
Native SSID extraction moved to platform code on Android

Some devices go to reboot, right after the start of the browser.
We managed to found out that this happens when native function
GetWifiSSID is called from the
NetworkQualityEstimator::GetCurrentNetworkID() method - this leads
to kernel panic and the device goes to reboot. Apparently this is due
to some error in the kernel. We have replaced the native call to the
platform one and the problem has gone. Affected devices are:
HTC One S
Lenovo A820
asus zenfone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.
==========

to

==========
On Android, move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.
==========

[ripp, 2016-01-27 12:09:06]

https://codereview.chromium.org/1633733005/diff/1/net/android/network_library.h
File net/android/network_library.h (right):

https://codereview.chromium.org/1633733005/diff/1/net/android/network_library...
net/android/network_library.h:85: // Returns the current SSID if device is
connected to Wi-Fi network.
On 2016/01/26 20:15:49, Ryan Hamilton wrote:
> nit: "connected to a Wi-Fi network" (add "a")

Done.

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
File net/base/network_quality_estimator.cc (right):

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
net/base/network_quality_estimator.cc:782: network_id.id =
android::GetCurrentWifiConnectionSsid();
On 2016/01/26 20:22:26, Ryan Sleevi wrote:
> DESIGN: Given that network_interfaces is already designed in a
> platform-abstracted way to avoid these #ifdefs, I would instead recommend the
> following approach:
> 
> In network_interfaces_linux.cc, guard GetWifiSSID by a
> 
> #if !defined(OS_ANDROID)
> std::string GetWifiSSID() {
>   ...
> }
> #endif
> 
> Then, implement a network_interfaces_android.cc which implements the
GetWifiSSID
> function, and either calls android::GetCurrentWifiConnectionSsid() or simply
> move that implementation logic (the JNI call) into the _android file.
> 
> Add the _android file to the .gyp and .gn files, conditionally as appropriate.
> 
> This ensures the API semantics of net::GetWifiSSID() remain the same, and also
> avoids the inconsistent naming being introduced here (SSID vs Ssid)

Done.

https://codereview.chromium.org/1633733005/diff/1/net/base/network_quality_es...
net/base/network_quality_estimator.cc:782: network_id.id =
android::GetCurrentWifiConnectionSsid();
On 2016/01/26 20:24:48, Ryan Hamilton wrote:
> On 2016/01/26 20:22:26, Ryan Sleevi wrote:
> > DESIGN: Given that network_interfaces is already designed in a
> > platform-abstracted way to avoid these #ifdefs, I would instead recommend
the
> > following approach:
> > 
> > In network_interfaces_linux.cc, guard GetWifiSSID by a
> > 
> > #if !defined(OS_ANDROID)
> > std::string GetWifiSSID() {
> >   ...
> > }
> > #endif
> > 
> > Then, implement a network_interfaces_android.cc which implements the
> GetWifiSSID
> > function, and either calls android::GetCurrentWifiConnectionSsid() or simply
> > move that implementation logic (the JNI call) into the _android file.
> > 
> > Add the _android file to the .gyp and .gn files, conditionally as
appropriate.
> > 
> > This ensures the API semantics of net::GetWifiSSID() remain the same, and
also
> > avoids the inconsistent naming being introduced here (SSID vs Ssid)
> 
> Oo! Good idea.

Done.

[Ryan Sleevi, 2016-01-27 18:12:45]

https://codereview.chromium.org/1633733005/diff/20001/net/android/network_lib...
File net/android/network_library.h (right):

https://codereview.chromium.org/1633733005/diff/20001/net/android/network_lib...
net/android/network_library.h:86: NET_EXPORT std::string
GetCurrentWifiConnectionSsid();
Please see the previous remarks about matching style (Ssid vs SSID)

Also, could you clarify why this is named differently than the //net method?
That is, why not just

android::GetWifiSSID() ?

[ripp, 2016-01-28 09:12:59]

https://codereview.chromium.org/1633733005/diff/20001/net/android/network_lib...
File net/android/network_library.h (right):

https://codereview.chromium.org/1633733005/diff/20001/net/android/network_lib...
net/android/network_library.h:86: NET_EXPORT std::string
GetCurrentWifiConnectionSsid();
On 2016/01/27 18:12:45, Ryan Sleevi wrote:
> Please see the previous remarks about matching style (Ssid vs SSID)
> 
> Also, could you clarify why this is named differently than the //net method?
> That is, why not just
> 
> android::GetWifiSSID() ?

There was no particular reason. My mistake.
changed to GetWifiSSID.

[bengr, 2016-01-30 01:12:30]

https://codereview.chromium.org/1633733005/diff/40001/net/android/java/src/or...
File net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java (right):

https://codereview.chromium.org/1633733005/diff/40001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:238: return
wifiInfo == null ? "" : wifiInfo.getSSID();
The Android API documentation says WifiInfo#getSSID() could return "<unknown
ssid>" if there is no network currently connected but isn't clear if this is
ever returned in practice because WifiManager#getConnectionInfo() only returns
non-null if there currently is an active WiFi connection. It would be good to
check (either in java or c++) that you do not return "<unknown ssid>".

https://codereview.chromium.org/1633733005/diff/40001/net/base/network_interf...
File net/base/network_interfaces_android.cc (right):

https://codereview.chromium.org/1633733005/diff/40001/net/base/network_interf...
net/base/network_interfaces_android.cc:1: // Copyright (c) 2016 The Chromium
Authors. All rights reserved.
Current convention is to remove the "(c)".

[ripp, 2016-02-01 09:33:47]

https://codereview.chromium.org/1633733005/diff/40001/net/android/java/src/or...
File net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java (right):

https://codereview.chromium.org/1633733005/diff/40001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:238: return
wifiInfo == null ? "" : wifiInfo.getSSID();
On 2016/01/30 01:12:30, bengr wrote:
> The Android API documentation says WifiInfo#getSSID() could return "<unknown
> ssid>" if there is no network currently connected but isn't clear if this is
> ever returned in practice because WifiManager#getConnectionInfo() only returns
> non-null if there currently is an active WiFi connection. It would be good to
> check (either in java or c++) that you do not return "<unknown ssid>".

Added check against <unknown ssid> and added removing of surround double
quotation marks from returned ssid. As documentation says - if ssid is UTF8
string it is surronded by quotes - I've missed this in previous patch

https://codereview.chromium.org/1633733005/diff/40001/net/base/network_interf...
File net/base/network_interfaces_android.cc (right):

https://codereview.chromium.org/1633733005/diff/40001/net/base/network_interf...
net/base/network_interfaces_android.cc:1: // Copyright (c) 2016 The Chromium
Authors. All rights reserved.
On 2016/01/30 01:12:30, bengr wrote:
> Current convention is to remove the "(c)".

Done.

[ripp, 2016-02-05 09:17:05]

Dear colleagues, are there any more comments or I can commit?

[bengr, 2016-02-05 17:42:51]

sleevi@: Would you like to take another look? I'll take a pass now as well.

[Ryan Hamilton, 2016-02-05 17:44:02]

this lgtm now, but please refrain from landing until sleevi signs off.

[Ryan Sleevi, 2016-02-05 17:46:25]

lgtm

[bengr, 2016-02-05 18:02:49]

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
File net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java (right):

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:235: * Returns
the current SSID if device is connected to a Wi-Fi network.
if -> if the

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:242: if
(wifiInfo != null) {
You can return early here:

if (wifiInfo == null) {
  return "";
}

if (ssid == null || "<unknown ssid>".equals(ssid)) {
  return "";
}

return ssid.replaceAll(...);

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:245: if (ssid
!= null && !"<unknown ssid>".equals(ssid)) {
Just to verify, when <unknown ssid> is returned, it does not have double quotes,
right?

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:246: // Remove
surround double quotation marks and return result.
Remove surround -> Remove the surrounding
return -> return the

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:247: return
ssid.replaceAll("^\"|\"$", "");
Is the string guaranteed to be quoted? If so, what you have should work. It also
removes a double quote if it appears at the beginning or end of the string and
not both. 

Suggest either changing the comment to "Remove a double quotation mark if it
appears at the beginning or end of the string." or checking explicitly that both
the first and last character are double quotation marks and removing them both
only if both are present.

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:251: return "";
Can a valid SSID be the empty string? If so, you'd probably want to return
something different when there isn't an SSID. Consider also the case of a string
that is '""'.

[ripp, 2016-02-09 08:02:05]

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
File net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java (right):

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:235: * Returns
the current SSID if device is connected to a Wi-Fi network.
On 2016/02/05 18:02:49, bengr wrote:
> if -> if the

Done.

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:242: if
(wifiInfo != null) {
On 2016/02/05 18:02:49, bengr wrote:
> You can return early here:
> 
> if (wifiInfo == null) {
>   return "";
> }
> 
> if (ssid == null || "<unknown ssid>".equals(ssid)) {
>   return "";
> }
> 
> return ssid.replaceAll(...);

Done.

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:245: if (ssid
!= null && !"<unknown ssid>".equals(ssid)) {
On 2016/02/05 18:02:49, bengr wrote:
> Just to verify, when <unknown ssid> is returned, it does not have double
quotes,
> right?

That's right.

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:246: // Remove
surround double quotation marks and return result.
On 2016/02/05 18:02:49, bengr wrote:
> Remove surround -> Remove the surrounding
> return -> return the

Comment replaced with self explanatory method removeSurroundingQuotes

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:247: return
ssid.replaceAll("^\"|\"$", "");
On 2016/02/05 18:02:49, bengr wrote:
> Is the string guaranteed to be quoted? If so, what you have should work. It
also
> removes a double quote if it appears at the beginning or end of the string and
> not both. 
> 
> Suggest either changing the comment to "Remove a double quotation mark if it
> appears at the beginning or end of the string." or checking explicitly that
both
> the first and last character are double quotation marks and removing them both
> only if both are present.

Added explicit checking for quotation marks at the beginning and at the end of
line

https://codereview.chromium.org/1633733005/diff/60001/net/android/java/src/or...
net/android/java/src/org/chromium/net/AndroidNetworkLibrary.java:251: return "";
On 2016/02/05 18:02:49, bengr wrote:
> Can a valid SSID be the empty string? If so, you'd probably want to return
> something different when there isn't an SSID. Consider also the case of a
string
> that is '""'.

In fact, AndroidNetworkLibrary::getWifiSSID method is the implementation of
net::GetWifiSSID function that already has such an interface.

[bengr, 2016-02-09 15:33:14]

lgtm

[ripp, 2016-02-09 16:03:26]

The CQ bit was checked by ripp@yandex-team.ru

[ripp, 2016-02-09 16:03:26]

The patchset sent to the CQ was uploaded after l-g-t-m from rch@chromium.org,
rsleevi@chromium.org
Link to the patchset: https://codereview.chromium.org/1633733005/#ps100001
(title: "Comment fixed")

[commit-bot: I haz the power, 2016-02-09 16:04:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1633733005/100001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1633733005/100001

[commit-bot: I haz the power, 2016-02-09 16:12:31]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-02-09 16:12:32]

Try jobs failed on following builders:
  android_chromium_gn_compile_dbg on tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_chro...)
  android_chromium_gn_compile_rel on tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/android_chro...)

[ripp, 2016-02-10 14:48:03]

The CQ bit was checked by ripp@yandex-team.ru

[ripp, 2016-02-10 14:48:04]

The patchset sent to the CQ was uploaded after l-g-t-m from bengr@chromium.org,
rch@chromium.org, rsleevi@chromium.org
Link to the patchset: https://codereview.chromium.org/1633733005/#ps120001
(title: "Fixed gn gen")

[commit-bot: I haz the power, 2016-02-10 14:48:21]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1633733005/120001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1633733005/120001

[commit-bot: I haz the power, 2016-02-10 17:44:58]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-02-10 17:44:59]

Try jobs failed on following builders:
  linux_android_rel_ng on tryserver.chromium.android (JOB_FAILED,
https://build.chromium.org/p/tryserver.chromium.android/builders/linux_androi...)

[ripp, 2016-02-10 18:48:24]

The CQ bit was checked by ripp@yandex-team.ru

[ripp, 2016-02-10 18:48:24]

The patchset sent to the CQ was uploaded after l-g-t-m from bengr@chromium.org,
rch@chromium.org, rsleevi@chromium.org
Link to the patchset: https://codereview.chromium.org/1633733005/#ps140001
(title: "Added permission for net_unittests")

[commit-bot: I haz the power, 2016-02-10 18:51:29]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1633733005/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1633733005/140001

[commit-bot: I haz the power, 2016-02-10 22:10:40]

Description was changed from

==========
On Android, move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.
==========

to

==========
On Android, move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.
==========

[commit-bot: I haz the power, 2016-02-10 22:10:40]

Committed patchset #8 (id:140001)

[gone, 2016-02-11 01:53:16]

On 2016/02/10 22:10:40, commit-bot: I haz the power wrote:
> Committed patchset #8 (id:140001)

I've synced past this CL and I'm immediately crashing on startup because this CL
accesses a function that requires a permission:

  02-10 20:47:13.685 21394 21536 W System.err: java.lang.SecurityException:
WifiService: Neither user 10196 nor current process has
android.permission.ACCESS_WIFI_STATE.
  02-10 20:47:13.686 21394 21536 W System.err: 	at
android.os.Parcel.readException(Parcel.java:1546)
  02-10 20:47:13.686 21394 21536 W System.err: 	at
android.os.Parcel.readException(Parcel.java:1499)
  02-10 20:47:13.686 21394 21536 W System.err: 	at
android.net.wifi.IWifiManager$Stub$Proxy.getConnectionInfo(IWifiManager.java:991)
  02-10 20:47:13.686 21394 21536 W System.err: 	at
android.net.wifi.WifiManager.getConnectionInfo(WifiManager.java:1176)
  02-10 20:47:13.686 21394 21536 W System.err: 	at
org.chromium.net.AndroidNetworkLibrary.getWifiSSID(AndroidNetworkLibrary.java:240)

Other folks on the team have hit this, as well.  Don't know how it got past the
bots, unless the code path doesn't get triggered in tests.

[mdjones, 2016-02-11 02:00:25]

A revert of this CL (patchset #8 id:140001) has been created in
https://codereview.chromium.org/1688103002/ by mdjones@chromium.org.

The reason for reverting is: Causes instant crash on Android KK, L, and M..

[commit-bot: I haz the power, 2016-02-16 22:31:33]

Description was changed from

==========
On Android, move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.
==========

to

==========
On Android, move Wifi SSID extraction to a Java implementation

On some Android devices, the device will restart right
after the browser is launched. This is due to a kernel bug on
these devices when using the Linux GetWifiSSID implementation,
which causes a kernel panic and device restart.

To avoid this, a Java-based implementation is used for Android
devices, which avoids this. Examples of affected devices include:
HTC One S
Lenovo A820
Asus ZenFone 2
MediaPad 10 FHD

R=bengr@chromium.org,rch@chromium.org
BUG=555067
TEST=Take the problem device, install browser and start it, wait for
10 seconds. Make 10 experiments, cleaning data before each run.
Device should not reboot.

Committed: https://crrev.com/35f59b4376232b4685c2a2e922f00a3bce2fbfde
Cr-Commit-Position: refs/heads/master@{#374743}
==========

[commit-bot: I haz the power, 2016-02-16 22:31:34]

Patchset 8 (id:??) landed as
https://crrev.com/35f59b4376232b4685c2a2e922f00a3bce2fbfde
Cr-Commit-Position: refs/heads/master@{#374743}