1678254 - Debugger not finding sources in localhost

Status: RESOLVED FIXED

(DevTools :: Debugger, defect, P2)

Description

[mattia.basaglia]

User Agent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:83.0) Gecko/20100101 Firefox/83.0

Steps to reproduce:

I was testing a web page on localhost, and opened the debugger tab in the developer tools.

Actual results:

it said "this page has no source", this happens for every page in localhost. but the same code under some other domain works fine.

It happens if I access the pages as localhost and 127.0.0.1, but if I use some other alias I have in /etc/hosts it works fine.

Expected results:

it should show the js files for that page.

Comment 1

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

Bugbug thinks this bug should belong to this component, but please revert this change in case of error.

Comment 2

[Pascal Chevrel:pascalc (PTO until April 26)]

Related to https://bugzilla.mozilla.org/show_bug.cgi?id=1677806?

Comment 3

[Valentin Gosu [:valentin] (he/him)]

Based on the user agent, this is linux, so likely not related to bug 1677806.
Could you try to use https://mozilla.github.io/mozregression/ to identify when this started happening? Thanks

Comment 4

[mattia.basaglia]

I reinstalled firefox and now sources appear sometimes but don't appear other times, bisecting might be difficult.

I've also discovered that the issue happens per tab, if I open a new tab to the same page it works fine.

Comment 5

[Greg Tatum [:gregtatum]]

I'm reproducing this frequently with the following:

• Refresh the page a few times, working on a project with WebConsole open.
• Add a debugger statement to the original sources.
• Refresh the page.
• The debugger catches and pauses.
• No sources are listed anywhere.
• The following error is in the browser console.

Error: Resource "server0.conn13.child3/source100" already exists, cannot insert {"id":"server0.conn13.child3/source100","actor":"server0.conn13.child3/source100","thread":"server0.conn13.child3/thread22","source":"sourceURL-resource://gre/modules/ExtensionContent.jsm","isBlackBoxed":false,"sourceMapBaseURL":"resource://gre/modules/ExtensionContent.jsm","sourceMapURL":null,"url":"resource://gre/modules/ExtensionContent.jsm","breakpointPositions":{},"breakableLines":null}
    insertResources resource://devtools/client/debugger/src/utils/resource/core.js:42
    update resource://devtools/client/debugger/src/reducers/source-actors.js:34
    Redux 2
    waitUntilService resource://devtools/client/debugger/src/actions/utils/middleware/wait-service.js:71
    promiseMiddleware resource://devtools/client/debugger/src/actions/utils/middleware/promise.js:46
    context resource://devtools/client/debugger/src/actions/utils/middleware/context.js:35
    thunk resource://devtools/client/debugger/src/actions/utils/middleware/thunk.js:29
    dispatch Redux
    insertSourceActors resource://devtools/client/debugger/src/actions/source-actors.js:21
    thunk resource://devtools/client/debugger/src/actions/utils/middleware/thunk.js:29
    dispatch Redux
    newGeneratedSources resource://devtools/client/debugger/src/actions/sources/newSources.js:350
    thunk resource://devtools/client/debugger/src/actions/utils/middleware/thunk.js:29
    bindActionCreator Redux
    onSourceAvailable resource://devtools/client/debugger/src/client/firefox.js:157
Promise-backend.js:964:35

This was on Mac, Dev Edition, Firefox 84.0b6.

I've reproduced this both on a built webpack project, and a non-built project with <script type="module">. These were all on localhost, but I'm not sure that it matters.

Closing the tab, and re-opening can make the problem go away. I suspect a race condition in source loading as I have trouble reproducing it reliably.

ni: ochameau since we were chatting about this on Matrix, and I wanted to bring it to your attention. This user seems to be experiencing the same issue as me.

Comment 6

[Greg Tatum [:gregtatum]]

Hmmm... re-reading the details of the original post I realize my steps are a bit different here.

Comment 7

[Alexandre Poirot [:ochameau]]

mattia.basaglia, are your also having troubles with a website using sourcemaps ?
(I'm trying to understand if that only fails with sourcemaps)

Comment 8

[Alexandre Poirot [:ochameau]]

I'm not able to reproduce this easily.

Greg, Are you having some addons installed?
Your exception mention resource://gre/modules/ExtensionContent.jsm source.
You are only using regular devtools and not the Browser toolbox? It is a bit surprising to see this file appear in the regular devtools!

Comment 9

[Alexandre Poirot [:ochameau]]

I tried simple sourcemapped/or not sourcemapped test pages locally without success,
I also tried debugging an online profile of the profiler, same, the debugger breaks and the sources appear.

If you have something I can reproduce (even intermittently) with a website available online, that could help.

Comment 10

[mattia.basaglia]

(In reply to Alexandre Poirot [:ochameau] from comment #7)

mattia.basaglia, are your also having troubles with a website using sourcemaps ?
(I'm trying to understand if that only fails with sourcemaps)

No, it's just normal files. It doesn't even show the html page in the list of sources

Comment 11

[Alexandre Poirot [:ochameau]]

Attachment: Bug 1678254 - [devtools] Stop unwatching/rewatching for resources on each page load from the source map service.

Comment 12

[Alexandre Poirot [:ochameau]]

I think this has been regressed by bug 1620280.
And bug 1644188, which landed today, not very intentionaly fixed that regression.
We might still want to proceed with comment 11's patch in order to fix backward compat usecases.

Comment 13

[Chris Peterson [:cpeterson]]

Tracking dt-fission-m3-mvp bugs for Fission MVP.

Comment 14

[Jan Honza Odvarko [:Honza] (always need-info? me)]

This is more about clean up at this point and we can move it to the reserve list.

Comment 15

[Chris Peterson [:cpeterson]]

Moving "dt-fission-m3-reserve" bugs to "dt-fission-future" because they don't block Fission MVP.

Comment 16

[Greg Tatum [:gregtatum]]

Removing needinfo since it's showing up in my bug dashboard.

Comment 17

[Alexandre Poirot [:ochameau]]

Attachment: Bug 1678254 - [devtools] Pass `commands` instead of `toolbox` to SourceMapUrlService.

Comment 18

[Alexandre Poirot [:ochameau]]

Attachment: Bug 1678254 - [devtools] Check descriptor type instead of current target.

Comment 19

[Pulsebot]

Pushed by [e-mail]:
https://hg.mozilla.org/integration/autoland/rev/9825dda94577
[devtools] Stop unwatching/rewatching for resources on each page load from the source map service. r=jdescottes
https://hg.mozilla.org/integration/autoland/rev/699bd6ddd665
[devtools] Pass `commands` instead of `toolbox` to SourceMapUrlService. r=jdescottes
https://hg.mozilla.org/integration/autoland/rev/60197e4e1341
[devtools] Check descriptor type instead of current target. r=jdescottes

Comment 20

[Iulian Moraru]

https://hg.mozilla.org/mozilla-central/rev/9825dda94577
https://hg.mozilla.org/mozilla-central/rev/699bd6ddd665
https://hg.mozilla.org/mozilla-central/rev/60197e4e1341

Comment 21

[Marco Mucci [:MarcoM]]

Hi Alex. Could you add a note in the team stand-up about the point complexity for this bug. Thanks.

Comment 22

[Jan de Mooij [:jandem]]

There was a huge JS memory usage regression a few days ago. Could it be from these patches (they are in the regression range)?

https://treeherder.mozilla.org/perfherder/graphs?timerange=1209600&series=autoland,2242037,1,4
https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=9a633c33737c933e49a5f07990e054a51e2bc43c&tochange=221d5494188c4c088861751d5bfe78331742e039

Comment 23

[Jan de Mooij [:jandem]]

Ah nevermind bug 1700955 is more likely for that, I hadn't realized these changes were for AWSY.