I have to correct the issue: The phone only crashes when you try to assign values of the "struct" after the function call to other variables (see https://jsfiddle.net/f2Lommf5/6798). In the first test case only fails but the phone does not crash.

Tested the issue using #66.0.3359.158 on Android Samsung J7 7.0 as per the steps mentioned in original comment.No crash is seen upon navigating to  https://jsfiddle.net/f2Lommf5/6798.

@Reporter: Could you please update to latest chrome version 67.0.3396.87 and check if issue persists. If yes, please help us by providing crash id from chrome://crashes.

Thanks!!

Thanks for the feedback! I asked my colleague to do a new test with your recommended setup. BTW: You might want to verify the issue with a mobile device that has a GPU from Qualcomm.

Besides: If https://jsfiddle.net/f2Lommf5/6830/ does not crash your phone but the colored output is red, there is still a problem with the device.

Thank you for providing more feedback. Adding the requester to the cc list.

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

Repeated from https://github.com/mrdoob/three.js/issues/14137#issuecomment-397094343

Verified that this issue does occur with the latest Chrome version available (67.0.3396.87). There are no crash IDs listed at chrome://crashes

I'm presuming this is because it isn't just causing chrome to crash, but rather causing the whole phone to crash (like full, hard restart), and that is occurring before the problem can be recorded.

To elaborate on Michael's comment: the Samsung J7 appears to have either a Qualcomm MSM8939 Snapdragon 615 (with Adreno 405 GPU) or Exynos 7580 Octa (with Mali-T720MP2) depending on the exact model. This issue appears to manifest on phones run by a Qualcomm Snapdragon 425 (with Adreno 308 GPU)

I've been testing on an LG Aristo and an LG Fiesta. The overarching issue has been reported on 15 different specific devices from 7 manufacturers[1], most (all?) of which have the same Snapdragon 425 (though it's unconfirmed whether this crash on this exact test case occur on all of those; I'm currently charging a Samsung J3 with this same processor to verify)

[1]: https://github.com/mrdoob/three.js/issues/14137#issue-326298141

Repeated from https://github.com/mrdoob/three.js/issues/14137#issuecomment-397167706

Just tested on a Samsung Galaxy J3 Luna Pro (SM-S327VL) which also has the same Snapdragon 425/Adreno 308, but is running Android 6.0.1 (security patch July 1, 2017). As reported in the original issue on the AFrame repo (https://github.com/aframevr/aframe/issues/3523#issuecomment-391611976), this version of Android does _not_ exhibit the errors described in this issue.

It seems no official update is available for this device, so I'm still unable to personally confirm the issues on a non-LG phone (though, as previously mentioned, numerous others have already done so: https://github.com/mrdoob/three.js/issues/14137#issue-326298141)

note for anyone trying to repro this: this issue causes the phone to crash completely and hard-restart, so make sure you don't have anything important open that you'll lose, and I'd recommend testing it in an incognito tab to avoid getting stuck in the same restart loop that I did: https://github.com/mrdoob/three.js/issues/14137#issuecomment-396348889

As per comment #!0 this issue seems to be hardware related.Hence adding TE-NeedsTriageHelp label for further inputs on same.

Thanks!

Adding some more labels. Need to potentially add this to the WebGL conformance suite and report this to Qualcomm, though I'm not optimistic about the drivers for these phones being fixed.

Could repro this issue on LG Aristo/NRD90U with M67

Steps:
1. Launch Chrome 
2. Go to https://jsfiddle.net/f2Lommf5/6798
3. Phone crashes not just browser

kbr@, I have the device with me, I can drop off at your desk.

Submitter: have created a WebGL conformance test for this in this pull request:
https://github.com/KhronosGroup/WebGL/pull/2663

It fails on the affected phone, but doesn't crash the phone. Perhaps the JSFiddles are crashing because they're repeatedly rendering with the affected shader.

If you can figure out how to modify the conformance test to crash the phone, please put up another pull request once the above one lands. Alternatively, the tests in conformance/rendering/ may be a better starting point.

Is there anything actionable here on the Chrome side? If this is not a Chrome issue, should we remove the release-block label?

I was holding off marking this ExternalDependency until new Android graphics tests landed:
https://android-review.googlesource.com/718028

but let me do so now. There is nothing we can do from the Chrome side about this.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6f64860d89fdb6d3397286e722b42f18c7df7db7

commit 6f64860d89fdb6d3397286e722b42f18c7df7db7
Author: Kenneth Russell <kbr@chromium.org>
Date: Sat Jul 21 02:54:54 2018

Roll WebGL a5c263c..21dbf06

https://chromium.googlesource.com/external/khronosgroup/webgl.git/+log/a5c263c..21dbf06

Bug:  830901 , 851870, 851873,  859400 ,  866089 
Tbr: zmo@chromium.org
Tbr: kainino@chromium.org
Cq-Include-Trybots: luci.chromium.try:win_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux_angle_rel_ng;luci.chromium.try:win_angle_rel_ng
Change-Id: Icef8ddc970ccb6653e95920c28b4d003d1b27100
Reviewed-on: https://chromium-review.googlesource.com/1144306
Commit-Queue: Kenneth Russell <kbr@chromium.org>
Reviewed-by: Kenneth Russell <kbr@chromium.org>
Cr-Commit-Position: refs/heads/master@{#577063}
[modify] https://crrev.com/6f64860d89fdb6d3397286e722b42f18c7df7db7/DEPS
[modify] https://crrev.com/6f64860d89fdb6d3397286e722b42f18c7df7db7/content/test/gpu/gpu_tests/webgl2_conformance_expectations.py
[modify] https://crrev.com/6f64860d89fdb6d3397286e722b42f18c7df7db7/content/test/gpu/gpu_tests/webgl_conformance_revision.txt

Removing Release-Block-Stable label as there is nothing the Chrome team can do to work around this driver bug.