Memory leak in Firefox at resale.info when using the code inspector
Categories
(DevTools :: General, defect, P3)
Tracking
(Not tracked)
People
(Reporter: 5228, Unassigned)
References
Details
(Whiteboard: dt-leak)
Attachments
(2 files)
User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:83.0) Gecko/20100101 Firefox/83.0
Steps to reproduce:
To repeat this bug you need to go to the site resale.info click on the link "Machine Search" (or directly at https://www.resale.info/machine-search.php?lang=en).
Open the Code Inspector and try searching on any selector (CSS or XPath).
Actual results:
I have 16 GB of RAM and all this memory fills up in 15-20 seconds, after which my Ubuntu stops responding.
The critical memory leak bug was tested in Ubuntu 18.04 in Firefox Developer Edition 83.0b6 and also in Firefox Browser 82.0 (64-bit).
I provide an illustration of this in the GIF in the attached files.
The same was tested in Chrome 86.0.4240.111 (Official Build) (64-bit) - no problems with critical memory leaks were detected.
P.S: For those who will test this bug, I advise you to close Firefox before it has time to fill all the free RAM, otherwise your system may freeze.
Expected results:
There should be no memory leak
Comment 2•5 years ago
|
||
Thanks for the report. It sounds like this issue is triggered by using the code inspector, so I'm going to move this over to DevTools.
Updated•5 years ago
|
Comment 3•5 years ago
|
||
The severity field is not set for this bug.
:Honza, could you have a look please?
For more information, please visit auto_nag documentation.
Comment 4•5 years ago
•
|
||
I can't see the mem leak, but I am experiencing tab crash:
https://crash-stats.mozilla.org/report/index/8e351904-632b-4598-b829-666e10201127
Looks like workers might be involved
MOZ_RELEASE_ASSERT(mWorkerThread && mWorkerThread->IsOnCurrentThread()) (not on worker thread!)
My STRs:
- Load https://www.resale.info/machine-search.php?lang=en
- Open DevTools and select the Inspector panel
- Wait (you might try to search for something in the Inspector panel, but I am not sure if it's related)
Andrew, any ideas?
Honza
Comment 5•5 years ago
|
||
This is an IPC error that I believe corresponds to using an IPC actor on the wrong thread. https://searchfox.org/mozilla-central/rev/8a4aa0c699d9ec281d1f576c9be1c6c1f289e4e7/ipc/glue/MessageChannel.h#558-559
Unfortunately, the crash report in comment 4 did not properly symbolicate/get a backtrace, so it's not clear what's going wrong in this specific situation.
(Which is not to say DOM Workers might not be involved; they could be! But we'd want a better backtrace.)
Description
•