New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 678149 link

Starred by 3 users

Issue metadata

Status: Verified
Owner:
Closed: Mar 2017
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Chrome
Pri: 1
Type: Bug-Regression



Sign in to add a comment

[Login Screen] Adding user, while connected to captive portal network crashes chrome.

Project Member Reported by aashuto...@chromium.org, Jan 4 2017

Issue description

Chrome Version: <From about:version: Google Chrome 57.0.2926.0>
Chrome OS Version: <From about:version: Platform 9147.0.0>
Chrome OS Platform: <Samus/Reks>
Network info: <Wifi>

Please specify Cr-* of the system to which this bug/feature applies (add
the label below).

Steps To Reproduce:
(1) On the login screen, Connect to captive portal network.
(2) Click Add person to add a new account on the device. 


Expected Result:
User should be able to add a account. 

Actual Result:
Captive portal authentication pop up shows up and device crashes.

How frequently does this problem reproduce? (Always, sometimes, hard to
reproduce?)
Always

What is the impact to the user, and is there a workaround? If so, what is
it?
User cannot add a new account, while connected to captive portal network.

Please provide any additional information below. Attach a screen shot or
log if possible.
CHROMEOS_RELEASE_BUILD_NUMBER=9147
CHROMEOS_RELEASE_BUILD_TYPE=Official Build
CHROMEOS_RELEASE_CHROME_MILESTONE=57
CHROMEOS_RELEASE_DESCRIPTION=9147.0.0 (Official Build) dev-channel samus 

For graphics-related bugs, please copy/paste the contents of the about:gpu
page at the end of this report.

LOGS ATTACHED.
Samus logs - log-010317-183919.tar.gz & Kernel Warnings. 
Reks logs - log-010317-190331.tar.gz

 
log-010317-183919.tar.gz
388 KB Download
kernel_warning.20170103.183527.0.kcrash
4.7 KB Download
kernel_warning.20170103.183527.0.meta
176 bytes Download
log-010317-190331.tar.gz
862 KB Download
Cc: -steve...@chromium.org kirtika@chromium.org snanda@chromium.org
Labels: -Pri-2 Pri-1
Owner: steve...@chromium.org
Status: Assigned (was: Untriaged)
stevenjb@ - assigning to you for further investigation. If you are not the right owner, please assign it back to snanda@ 



I see the following in different crash logs:

=====================
[7696:7696:0103/183850.984669:ERROR:device_event_log_impl.cc(140)] [18:38:50.984] Network: network_portal_detector_impl.cc:463 Portal detection timeout:  name=CrOS_GUESTGATE_MKII id=4aac27be-7702-4dc6-88b6-f2f5f225cbbe


=====================
[979:979:0103/185526.759323:ERROR:device_event_log_impl.cc(140)] [18:55:26.759] Network: network_portal_detector_impl.cc:463 Portal detection timeout:  name=GUESTGATE id=6cd1c253-1349-4d49-9757-ff895b700d14


=====================
[5788:5788:0103/190238.538759:ERROR:device_event_log_impl.cc(140)] [19:02:38.538] Network: network_portal_detector_impl.cc:463 Portal detection timeout:  name=GUESTGATE id=6cd1c253-1349-4d49-9757-ff895b700d14
[5788:5788:0103/190241.563858:ERROR:device_event_log_impl.cc(140)] [19:02:41.563] Network: device_event_log.cc:117 @@@ Slow method: ../../../../../../../home/chrome-bot/chrome_root/src/ash/common/system/chromeos/network/network_list_md.cc:UpdateNetworkListInternal: 84ms
[5788:5788:0103/190244.685913:ERROR:device_event_log_impl.cc(140)] [19:02:44.685] Network: device_event_log.cc:117 @@@ Slow method: ../../../../../../../home/chrome-bot/chrome_root/src/ash/common/system/chromeos/network/network_list_md.cc:UpdateNetworkListInternal: 80ms
[5788:5788:0103/190247.386888:ERROR:device_event_log_impl.cc(140)] [19:02:47.386] Network: network_state_handler.cc:920 Default network in unexpected state: GUESTGATE (/service/1)State: idle


Cc: alemate@chromium.org steve...@chromium.org xiy...@chromium.org
Owner: achuith@chromium.org
->achuith@ to triage, + others.

Looks like we need to figure out how to handle this during login.

Aashutosh: Any idea if this is an existing bug or a recent regression? Could you please try on R53 or something around that time?
This is a recent regression. I cannot produce this issue on R55 or R56. 
Cc: achuith@chromium.org
Owner: alemate@chromium.org
I suspect this may be one of Alex's changes in that case. I'll also take a look
Still an  issue on Samus R57-9292.10.0

Comment 7 by mea...@chromium.org, Feb 10 2017

Is there a chance this could be related to bug 680329? (that one has crash ids though)
Cc: dsunk...@chromium.org
still an issue Link- 9202.43.0. 
Labels: ReleaseBlock-Stable
Please find attached logs on Link 57 / 9202.43.0 / 57.0.2987.85
debug-logs_20170301-165930.tgz
186 KB Download
log-030117-165942.tar.gz
618 KB Download
alemate@ this is a stable blocker for M57. Will you be be looking at this issue?
Owner: r...@chromium.org
rkc@ can you find someone else to look at this since alemate@ is slammed?
Owner: jdufault@chromium.org
Jacob, could you take a look at this please?

Status: Started (was: Assigned)
When clicking the "password" field I get this stack trace on Chrome 59.0.3032.2.

Received signal 11 SEGV_MAPERR 000000000000
#0 0x7fd25284d91d base::debug::StackTrace::StackTrace()
#1 0x7fd25284dd10 base::debug::(anonymous namespace)::StackDumpSignalHandler()
#2 0x7fd24f708580 <unknown>
#3 0x7fd2573b8a4b password_manager::PasswordAutofillManager::OnShowNotSecureWarning()
#4 0x7fd251dbbc18 autofill::mojom::PasswordManagerDriverStubDispatch::Accept()
#5 0x7fd25290f2b9 mojo::InterfaceEndpointClient::HandleValidatedMessage()
#6 0x7fd25291440c mojo::internal::MultiplexRouter::ProcessIncomingMessage()
#7 0x7fd25291610a mojo::internal::MultiplexRouter::Accept()
#8 0x7fd25290cde7 mojo::Connector::ReadSingleMessage()
#9 0x7fd25290cf03 mojo::Connector::ReadAllAvailableMessages()
#10 0x7fd2529225f3 mojo::Watcher::OnHandleReady()
#11 0x7fd2528ecf13 base::debug::TaskAnnotator::RunTask()
#12 0x7fd25286edaf base::MessageLoop::RunTask()
#13 0x7fd25286f25e base::MessageLoop::DeferOrRunPendingTask()
#14 0x7fd252871560 base::MessageLoop::DoWork()
#15 0x7fd252871ad9 base::MessagePumpLibevent::Run()
#16 0x7fd25286eb1a base::MessageLoop::RunHandler()
#17 0x7fd2528961ed base::RunLoop::Run()
#18 0x7fd25246c783 ChromeBrowserMainParts::MainMessageLoopRun()
#19 0x7fd250e28b30 content::BrowserMainLoop::RunMainMessageLoopParts()
#20 0x7fd250e2d1a5 content::BrowserMainRunnerImpl::Run()
#21 0x7fd250e24e19 content::BrowserMain()
#22 0x7fd2523fb5a6 content::ContentMainRunnerImpl::Run()
#23 0x7fd2523fa0e9 content::ContentMain()
#24 0x7fd25087293b ChromeMain
#25 0x7fd24e355816 __libc_start_main
#26 0x7fd250872759 _start
  r8: 00000bbe48df6601  r9: fffffffc016d7bb9 r10: fffff44249b213a9 r11: 0000000000039eff
 r12: 00007ffdaca44510 r13: 00007ffdaca44550 r14: 00007ffdaca44530 r15: 00000bbe48d01c88
  di: 00007ffdaca44520  si: 00000bbe48d01ce0  bp: 00007ffdaca445c0  bx: 00007ffdaca44520
  dx: 00000bbe48d01c88  ax: 0000000000000000  cx: 0000000000000000  sp: 00007ffdaca444d0
  ip: 00007fd2573b8a4b efl: 0000000000010206 cgf: 0000000000000033 erf: 0000000000000004
 trp: 000000000000000e msk: 0000000000000000 cr2: 0000000000000000
[end of stack trace]
autofill_client_ is null here[1], which is causing the crash. Regression was likely introduced in this[2] CL.

Trying to determine if we should setup an autofill client for the authentication webview or if we check for a null autofill_client_ inside of the autofill manager.
 
1: https://cs.chromium.org/chromium/src/components/password_manager/core/browser/password_autofill_manager.cc?l=267
2: https://chromium.googlesource.com/chromium/src/+/50592c0e5e6c21d5b11f3a5f1515d9afcd5fe1a7
jdufault: That's bug 698438, estark is working on it.

Comment 18 Deleted

Thanks; I suspect these are the same issues. CL at https://codereview.chromium.org/2736733006.
Issue 698438 has been merged into this issue.
Cc: jdufault@chromium.org
Owner: est...@chromium.org
Looks like estark@ has a more complete CL at  https://codereview.chromium.org/2727813006/. Passing ownership.
Cc: aashuto...@chromium.org
estark@ please comment on whether this CL is ready to go to M57? if yes, please request for merge asap.
Cc: mea...@chromium.org
Re #23: no, I wasn't aware of this bug until a couple days ago and the fix hasn't landed yet. We might be able to land https://codereview.chromium.org/2733043005/ today as a temporary band-aid.
Project Member

Comment 25 by bugdroid1@chromium.org, Mar 7 2017

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/88c753c103d76fd3697f7fcef962011588e3975f

commit 88c753c103d76fd3697f7fcef962011588e3975f
Author: estark <estark@chromium.org>
Date: Tue Mar 07 20:56:31 2017

Disable Form-Not-Secure warning when |autofill_client_| is null

This is a temporary fix for an M57 crasher when Form-Not-Secure is
enabled.

PasswordAutofillManager doesn't generally seem to allow for the possibility that
|autofill_client_| is null, yet it's possible for it to be null, in the case of
the SimpleWebViewDialog ChromeOS captive portal login. In a follow-up, we should
figure out whether PasswordAutofillManager is intended to support a null
|autofill_client_| and also whether SimpleWebViewDialog is supposed to work
with password autofill.

BUG= 678149 
TBR=engedy@chromium.org

Review-Url: https://codereview.chromium.org/2733043005
Cr-Commit-Position: refs/heads/master@{#455213}

[modify] https://crrev.com/88c753c103d76fd3697f7fcef962011588e3975f/components/password_manager/core/browser/password_autofill_manager.cc

Labels: Merge-Request-57
Requesting a merge to M57 for the commit in comment 25. (Won't do the merge until it lands on canary tomorrow)
Btw, jdufault, looking through the bug again, how sure are we that the missing autofill client is the actual issue here? The original bug report doesn't seem to describe any interaction with the captive portal dialog, which would be necessary to trigger the crash I fixed.
Project Member

Comment 28 by sheriffbot@chromium.org, Mar 7 2017

Labels: -Merge-Request-57 Hotlist-Merge-Review Merge-Review-57
This bug requires manual review: Only 6 days from stable, we might already have a stable candidate build
Please contact the milestone owner if you have questions.
Owners: amineer@(clank), cmasso@(bling), ketakid@(cros), govind@(desktop)

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Labels: -Merge-Review-57 Merge-Approved-57
Please validate on canary before merging. Approving merge to M57 Chrome OS.
Re #27: aashutoshk@ can confirm how the crash happened, but I was testing using the same WiFi network as the initial bug report. It was serving a login page over HTTP - so I'm pretty confident it is the same issue.
After the captive portal dialog pops up, chrome immediately crashes. There is no interaction with the captive portal dialog. 
If it immediately crashes, that's probably bug 680329.
estark@ are we ready for the CL to land in M57 this morning?
meacer is trying to figure out how to test the captive portal dialog on canary.
krishnargv@/harpreet@ - can either of you help them test this? We need to get this CL landed in the next hour or so.
Which build is the fix in? I don't see it in ToT build or M58.

Aashutosh will verify the fix.
It's in 59.0.3034.0
I checked golden-eye builds webpage, the latest build available for testing is, 

2017-03-08 04:34	9349.0.0	59.0.3033.0 

Will test and update once the build is available. Let me know, if I missing something.

Note that to test this you need to:
1.) set the flag #enable-http-form-warning in chrome://flags
2.) connect to a captive portal with a login page that is served over http and has a password field
3.) start typing in the password field in the captive portal login dialog

Are you able to test with such a captive portal? meacer is working on getting one set up right now but it's non-trivial.
Yes, we do have a captive portal network meeting your requirements. 
Project Member

Comment 41 by bugdroid1@chromium.org, Mar 8 2017

Labels: -merge-approved-57 merge-merged-2987
The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/588d0414a6ef0690800b7bc1ec824ec114d8b258

commit 588d0414a6ef0690800b7bc1ec824ec114d8b258
Author: Emily Stark <estark@google.com>
Date: Wed Mar 08 21:55:17 2017

Disable Form-Not-Secure warning when |autofill_client_| is null

This is a temporary fix for an M57 crasher when Form-Not-Secure is
enabled.

PasswordAutofillManager doesn't generally seem to allow for the possibility that
|autofill_client_| is null, yet it's possible for it to be null, in the case of
the SimpleWebViewDialog ChromeOS captive portal login. In a follow-up, we should
figure out whether PasswordAutofillManager is intended to support a null
|autofill_client_| and also whether SimpleWebViewDialog is supposed to work
with password autofill.

BUG= 678149 
TBR=engedy@chromium.org

Review-Url: https://codereview.chromium.org/2733043005
Cr-Commit-Position: refs/heads/master@{#455213}
(cherry picked from commit 88c753c103d76fd3697f7fcef962011588e3975f)

Review-Url: https://codereview.chromium.org/2740673004 .
Cr-Commit-Position: refs/branch-heads/2987@{#803}
Cr-Branched-From: ad51088c0e8776e8dcd963dbe752c4035ba6dab6-refs/heads/master@{#444943}

[modify] https://crrev.com/588d0414a6ef0690800b7bc1ec824ec114d8b258/components/password_manager/core/browser/password_autofill_manager.cc

Status: Fixed (was: Started)
Labels: Merge-Request-58
Project Member

Comment 44 by sheriffbot@chromium.org, Mar 9 2017

Labels: -Merge-Request-58 Hotlist-Merge-Approved Merge-Approved-58
Your change meets the bar and is auto-approved for M58. Please go ahead and merge the CL to branch 3029 manually. Please contact milestone owner if you have questions.
Owners: amineer@(clank), cmasso@(bling), bhthompson@(cros), govind@(desktop)

For more details visit https://www.chromium.org/issue-tracking/autotriage - Your friendly Sheriffbot
Project Member

Comment 45 by bugdroid1@chromium.org, Mar 9 2017

Labels: -merge-approved-58 merge-merged-3029
The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/0bdbd12ada6095acd5f0f13b960773e758f48d5d

commit 0bdbd12ada6095acd5f0f13b960773e758f48d5d
Author: Emily Stark <estark@google.com>
Date: Thu Mar 09 22:00:54 2017

Disable Form-Not-Secure warning when |autofill_client_| is null

This is a temporary fix for an M57 crasher when Form-Not-Secure is
enabled.

PasswordAutofillManager doesn't generally seem to allow for the possibility that
|autofill_client_| is null, yet it's possible for it to be null, in the case of
the SimpleWebViewDialog ChromeOS captive portal login. In a follow-up, we should
figure out whether PasswordAutofillManager is intended to support a null
|autofill_client_| and also whether SimpleWebViewDialog is supposed to work
with password autofill.

BUG= 678149 
TBR=engedy@chromium.org

Review-Url: https://codereview.chromium.org/2733043005
Cr-Commit-Position: refs/heads/master@{#455213}
(cherry picked from commit 88c753c103d76fd3697f7fcef962011588e3975f)

Review-Url: https://codereview.chromium.org/2739013005 .
Cr-Commit-Position: refs/branch-heads/3029@{#94}
Cr-Branched-From: 939b32ee5ba05c396eef3fd992822fcca9a2e262-refs/heads/master@{#454471}

[modify] https://crrev.com/0bdbd12ada6095acd5f0f13b960773e758f48d5d/components/password_manager/core/browser/password_autofill_manager.cc

Status: Verified (was: Fixed)
Link device working as expected on R57-9202.51.0 & R58-9334.5.0 builds. 

Sign in to add a comment