1524009 - Google Maps hangs browser tab

Status: NEW

(Core :: Graphics, defect, P3)

Description

[joseph.neal]

User Agent: Mozilla/5.0 (Windows NT 10.0; Win64; x64; rv:60.0) Gecko/20100101 Firefox/60.0

Steps to reproduce:

On a fresh install of Firefox ESR 60.50.0 (64-bit) open google maps with the traffic overlay (map view, not satellite). Scroll around map and zoom in/out and eventually the browser tab appears to hang.

This may take as little as 10 seconds, but may also take a couple minutes. It does not depend on the area/map being viewed.

I also tested back revs of the browser (ESR 60.0+) with the same results and enabled/disabled "Use hardware acceleration when available" with no change. I have not tested the mainstream channel of Firefox.

This is on a network with no proxy and has been reproduced on several machines (White Box i7-4400, Surfacebook Pro, etc...)

My System Specs:
Win10 Ent 1803
Dell OptiPlex 7040
Core i7-6700
16GB RAM
SanDiskX400M 128GB

Actual results:

Set up Procmon to watch firefox and ran into these at the time of the hang:

9:40:59.1570495 AM firefox.exe 4720 QueryInformationVolume C:\Users\XXXXXXXX\AppData\Roaming\Mozilla\Firefox\Profiles\oap2aezk.default\sessionstore-backups\recovery.jsonlz4 BUFFER OVERFLOW VolumeCreationTime: 12/27/2016 1:34:18 PM, VolumeSerialNumber: E0FC-F8C9, SupportsObjects: True, VolumeLabel: Winʆ
9:40:59.1570841 AM firefox.exe 4720 QueryAllInformationFile C:\Users\XXXXXXXX\AppData\Roaming\Mozilla\Firefox\Profiles\oap2aezk.default\sessionstore-backups\recovery.jsonlz4 BUFFER OVERFLOW CreationTime: 1/30/2019 9:34:07 AM, LastAccessTime: 1/30/2019 9:40:59 AM, LastWriteTime: 1/30/2019 9:40:59 AM, ChangeTime: 1/30/2019 9:40:59 AM, FileAttributes: A, AllocationSize: 4,096, EndOfFile: 3,156, NumberOfLinks: 1, DeletePending: False, Directory: False, IndexNumber: 0x2800000002f317, EaSize: 0, Access: Read Attributes, Synchronize, Position: 0, Mode: Synchronous IO Non-Alert, AlignmentRequirement: Word

Expected results:

Expected that the browser should not hang while using google maps.

Comment 1

[Liviu Seplecan]

retested this issue on different versions of firefox ESR: 60.0 and 60.5.0 but wasn't able to reproduce it. I've tried even with the option "Use hardware acceleration when available" unchecked. Could you provide me additional info's, description, screenshots or a short video?

Comment 2

[joseph.neal]

Attachment: hangv2.mp4

Working on getting a video of the actual hang. This is when it has already encountered the hang.

Comment 3

[joseph.neal]

Attachment: hangv3.mp4

Here's a video of the actual hang. Browser will allow limited functionality, but won't close when you click on the X.

Some times it takes a minute or two, sometimes it is 20 seconds or less.

Comment 4

[joseph.neal]

Attachment: hangv3.mp4

Uploaded the wrong one.

This is the one with the hang in it.

Comment 5

[Liviu Seplecan]

Thanks for the video. I was able to reproduce this issue on firefox ESR version 60.0 at fresh install. I add a new component to this issue so that the dev's team will handle it.

Comment 6

[joseph.neal]

Sounds good.

Let me know if any more input is required.

Comment 7

[Andrew Osmond [:aosmond] (he/him)]

Update the title, buffer overflow evokes a security flaw, whereas I believe this just means we requested data, but our buffer wasn't big enough so the call failed with a more appropriate buffer size. It may or may not be related.

Liviu, since you are able to reproduce, are you able to confirm whether or not this is a regression? Also, since you marked it as not failing on the newer builds, it would be good to find out what fixed it in case the fix is possible to uplift.

Comment 8

[Liviu Seplecan]

Hi, I've checked another time this issue but now I wasn't able to reproduce it on ESR versions: 60.0, 60.1 -> 60.5. I couldn't obtain a range regression of this bug because now I cannot reproduce it and on all additional versions (firefox release 65.0, beta 66.0b6 and nightly 67.0a1) the issue won't occur.

Comment 9

[Andrew Osmond [:aosmond] (he/him)]

See comment 7. Would it be possible for you to run mozregression to try and find a fix? Or confirm the issue is present on latest release (65) / nightly (67) builds? Thanks.

Comment 10

[Liviu Seplecan]

yes, sure

Comment 11

[Liviu Seplecan]

The issue cannot be confirmed on latest release 65.0 or 67.0a1 - nightly. After a mozregression, this is the result:

Bug 1507049 - Rename MOZ_CRASH_UNSAFE_OOL MOZ_CRASH_UNSAFE. r=froydnj

Differential Revision: https://phabricator.services.mozilla.com/D18515

Comment 12

[joseph.neal]

Good Morning,

I have tested for this and replicated on ESR versions:

60.1.0
60.2.2
60.3.0
60.4.0
60.5.0

In our environment we use the ESR version exclusively, but it doesn't sound like it is an issue in the mainstream branch.

Comment 13

[Andrew Osmond [:aosmond] (he/him)]

(In reply to Liviu Seplecan from comment #11)

The issue cannot be confirmed on latest release 65.0 or 67.0a1 - nightly. After a mozregression, this is the result:

Bug 1507049 - Rename MOZ_CRASH_UNSAFE_OOL MOZ_CRASH_UNSAFE. r=froydnj

Differential Revision: https://phabricator.services.mozilla.com/D18515

That regression range doesn't make sense. The change was never uplifted to release and only just landed on central in the last few days. Would you mind trying again?

The next EA

Comment 14

[Andrew Osmond [:aosmond] (he/him)]

*The next ESR is not until 68, so it will be a while.

Comment 15

[Liviu Seplecan]

Hi, I tried again with the latest ESR version 60.5.1 on both cases(enabled/disabled "Use hardware acceleration when available")-but without success- I cannot reproduce this issue.

Comment 16

[Liviu Seplecan]

Hi, I tried again with the latest version of ESR - 60.5.1 in both cases (enabled/disabled "Use hardware acceleration when available") but without success. I cannot reproduce this issue.

Comment 17

[joseph.neal]

I just installed the update and have been able to reproduce this hang twice.

The first hang took about a minute, the second took about 10 seconds.

Comment 18

[Liviu Seplecan]

Hi, please download Firefox Nightly from here: https://nightly.mozilla.org/ and retest the problem.
them
Can you please capture a performance profile? You can get more info on how to install and use the Cleopatra add-on (that helps you get the performance profile) by going to:
https://developer.mozilla.org/en-US/docs/Mozilla/Performance/Profiling_with_the_Built-in_Profiler
https://perf-html.io/

Comment 19

[joseph.neal]

I have given this a lot of testing under the nightly build (67.0a1) and so far have not been able to reproduce the hang.

However, this is not the ESR version.

I am able to reproduce the hang on the ESR builds:
60.1.0
60.2.2
60.3.0
60.4.0
60.5.0
60.5.1

I can try and get a performance profile, but it probably doesn't make sense to do it on the nightly build. Did you want me to capture the performance profile on the ESR build?

Comment 20

[joseph.neal]

Attachment: hang-impacted-profiler.png

This is the picture of what happened when I tried to run a performance profile.

The hang took down the profiler too.

Comment 21

[joseph.neal]

Ok, tried to run a performance profile on the ESR version (see attached).

Essentially, the browser is left in a state that is only partially usable. The profiler couldn't operate properly after the hang.

Comment 22

[Gregory Pappas [:gregp]]

Reporter, are you still experiencing this issue?

Comment 23

[BugBot [:suhaib / :marco/ :calixte]]

Redirect a needinfo that is pending on an inactive user to the triage owner.
:bhood, since the bug has recent activity, could you have a look please?

For more information, please visit BugBot documentation.