1449950 - Website of Lenovo Webshop does not load to the end

Status: RESOLVED FIXED

(Web Compatibility :: Site Reports, defect, P1)

Description

[Zitronella]

Attachment: mozregression-lenovo-win10-64bit.png

This URL https://www3.lenovo.com/de/de/desktops-and-all-in-ones/thinkcentre/m-series-tiny/ThinkCentre-M910q/p/10MVCTO1WWDEDE1/customize 
does not load to the end since Firefox 58 on Windows (64bit)
I started mozregression and the pushlog is https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=4ab539ae8b5b2927e219b219d5735f2557e82425&tochange=b1204c61ba568e8579b5cf9e697b031890790271

It says, Ticket https://bugzilla.mozilla.org/show_bug.cgi?id=1407607 is involved

It seems, that only windows OS is affected. On Linux (OS MX 17.1 with Firefox 59.0.2) and with Macintosh; Intel Mac OS X 10.11 this site works also with Firefox 58 and above.

Comment 1

[Jan de Mooij [:jandem]]

Thanks for the report.

Do you see anything in the web console (Tools -> Web Developer -> Web Console) when this happens? I wonder if it's an overrecursion on Windows.

Comment 2

[Zitronella]

Object { country: "DE", currencycode: "EUR", language: "de", cc: "DE", lc: "de", asmindicator: "off", bu: "PUBLIC_CONSUMER", viewport: "width=device-width,initial-scale=1.0", robots: "INDEX,FOLLOW", area: "PUBLIC_CONSUMER", … }
AdobeAnalyticsEvent.js:346:1
testing customize:3061:1
popupcalled customize:3837:20
Process data: 388.5ms singlePage.js:26417:5
checkCompatibleForGroup singlePage.js:26420:5
Check compatible rules: 882.8ms singlePage.js:26427:5
PT :2.73 satelliteLib-cd0127785d50cbe3d3047e8ff57496baf66459fb.js:12:9582
unreachable code after return statement[Weitere Informationen] js:70:1092
hbs loaded fs.js:36:288
datacommon 
Object { config: {}, allWidgetsLoaded: true, eventStack: [], store: Storage, cid: "view1", el: widget, options: {…}, "$el": {…}, isTelesalesAgent: false, bootstrapDialog: BootstrapDialog(), … }
fs.js:36:288
data summary 
Object { promotions: (1) […] }
fs.js:36:288
PT :2.73 fs.js:36:288

Comment 3

[Ryan VanderMeulen [:RyanVM]]

I can reproduce this as well with the URL given. All I see in the browser console is an "unreachable code after return statement" warning in this minimized monstrosity: https://www.googletagmanager.com/gtag/js?id=

> unreachable code after return statement js:69:1092

Which corresponds to this line:
a)},oc=function(){this.prefix="gtm.";this.values={}};oc.prototype.set=function(a,b){this.values[this.prefix+a]=b};oc.prototype.get=function(a){return this.values[this.prefix+a]};oc.prototype.contains=function(a){return void 0!==this.get(a)};var pc=function(){var a=ac.sequence||0;ac.sequence=a+1;return a},qc=function(a,b,c){return a&&a.hasOwnProperty(b)?a[b]:c},rc=function(a){var b=!1;return function(){if(!b)try{a()}catch(c){}b=!0}};var sc=function(){var a=function(a){return{toString:function(){return a}}};return{G:a("function"),Yb:a("instance_name"),Zb:a("live_only"),$b:a("malware_disabled"),ac:a("once_per_event"),tb:a("once_per_load"),ub:a("setup_tags"),bc:a("tag_id"),vb:a("teardown_tags")}}();var tc=new oc,uc={},xc={set:function(a,b){ra(vc(a,b),uc)},get:function(a){return wc(a,2)},reset:function(){tc=new oc;uc={}}},wc=function(a,b){return 2!=b?tc.get(a):yc(a)},yc=function(a,b,c){var d=a.split(".");var e=function(a,b){for(var c=0;void 0!==a&&c<d.length;c++){if(null===a)return!1;a=a[d[c]]}return void 0!==a||1<c?a:b.length?e(zc(b.pop()),b):Ac(d)};return e(uc.eventModel,[b,c]);return Ac(d)},Ac=function(a){for(var b=uc,c=0;c<a.length;c++){if(null===


If there's anything else you'd like me to try, let me know.

Comment 4

[Jan de Mooij [:jandem]]

Hm, I can also repro this on Windows. It even happens with Baseline/Ion disabled and in a debug build; that makes me wonder how bug 1407607 is related. Will try a few other things.

Comment 5

[Jan de Mooij [:jandem]]

For me it also repros with an October 11 Nightly, before bug 1407607 landed.

Ryan, maybe someone else can bisect so we can see if that gives us a different range?

Comment 6

[Ryan VanderMeulen [:RyanVM]]

I got the same regression range as #c0.

Comment 7

[Jan de Mooij [:jandem]]

I'll double check but I'm pretty sure I had a 10/11 Nightly and it also didn't finish loading. I can send you a link to see if that one works for you...

Comment 8

[Ryan VanderMeulen [:RyanVM]]

WFM when I launch the 2017-10-11 nightly via mozregression, i.e. |mozregression -a https://www3.lenovo.com/de/de/desktops-and-all-in-ones/thinkcentre/m-series-tiny/ThinkCentre-M910q/p
/10MVCTO1WWDEDE1/customize --repo central --launch 2017-10-11|

INFO: Downloading build from: https://archive.mozilla.org/pub/firefox/nightly/2017/10/2017-10-11-22-01-13-mozilla-central/firefox-58.0a1.en-US.win64.zip
===== Downloaded 100% =====
INFO: Running mozilla-central build for 2017-10-11
INFO: Launching c:\Users\Ryan\AppData\Local\Temp\tmpfl_vzi\firefox\firefox.exe
INFO: Application command: c:\Users\Ryan\AppData\Local\Temp\tmpfl_vzi\firefox\firefox.exe https://www3.lenovo.com/de/de/desktops-and-all-in-ones/thinkcentre/m-series-tiny/ThinkCentre-M910q/p/10MVCTO1WWDEDE1/customize -profile c:\users\ryan\appdata\local\temp\tmp6fvybl.mozrunner
INFO: application_buildid: 20171011220113
INFO: application_changeset: 3d918ff5d63442d7b88e1b7e9cb03b832bc28fdf
INFO: application_name: Firefox
INFO: application_repository: https://hg.mozilla.org/mozilla-central
INFO: application_version: 58.0a1

Comment 9

[Jan de Mooij [:jandem]]

Ryan, does it work for you if you disable the baselinejit pref in about:config?

I'll try again later today or tomorrow because this is pretty mysterious.

Comment 10

[Ryan VanderMeulen [:RyanVM]]

Launching an affected build with javascript.options.baselinejit set to false doesn't appear to help from what I can see (and I verified in about:config that it was showing as false).

Comment 11

[Jan de Mooij [:jandem]]

Hm this doesn't work consistently for me with an older Nightly, sometimes it loads okay but then a reload will trigger the hang, etc.

I even get this to fail with Nightly on OS X sometimes. I wonder if it could be a timer or something.

Comment 12

[Jan de Mooij [:jandem]]

I spent some more time on this yesterday. The website has a weird mix of arbitrary setInterval and AJAX requests; I'm pretty sure it's a problem on their end, also considering it repros with JITs disabled. I'll see if I can come up with a reduced testcase though.

Comment 13

[Jan de Mooij [:jandem]]

Pretty sure this is overrecursing. This website is written in a horrible way. Now waiting for Windows Try builds that set a linker flag to double our stack size to 4 MB, to see if that helps.

Comment 14

[Ted Campbell [:tcampbell]]

When I looked at this before (and I guess forgot to comment), it seemed to be the site keeps recursing until certain elements load. Changes to our perf or recursion limits will cause the hang. Website is really to blame.

Comment 15

[Jan de Mooij [:jandem]]

Bumping our thread stack size on Win64 from 2 MB to 4 MB seems to fix this (but the website also still hangs sometimes). We could do that but it's harder on Win32 (much smaller virtual address space) and really the website is broken.

Comment 16

[Jan de Mooij [:jandem]]

We used to have the same problem on Linux, bug 1412773.

Comment 17

[Steven DeTar [:sdetar]]

Jan, any ideas on what the next steps are with bug?

Comment 18

[Jan de Mooij [:jandem]]

This website broken - there are multiple places where it can easily hit overrecursion. I don't think there's anything we should do on our end.

Comment 19

[Ryan VanderMeulen [:RyanVM]]

Adam, I don't suppose you'd have a way of getting in touch with Lenovo's web development team? :)

Comment 20

[Adam Stevenson [:adamopenweb]]

Ryan, sorry for the delay. Will get on this, don't think we have a contact but will find one.

Comment 21

[Adam Stevenson [:adamopenweb]]

Found a contact and reached out by email.

Comment 22

[Adam Stevenson [:adamopenweb]]

Our contact has passed this along to the appropriate team to troubleshoot and fix.

Comment 23

[Liz Henry (:lizzard) (relman/hg->git project)]

Seems to load OK now, in 63 nightly. Zitronella, do you still see the issue ?

Comment 24

[Zitronella]

no i can not see the issue any more, also not on Firefox 61.0.2 release.

Comment 25

[Mike Taylor [:miketaylr]]

Thanks!