Closed Bug 1820025 Opened 3 years ago Closed 2 years ago

Memory leak on https://nikonservice-dresden.de (js-realm - elements/normal)

Categories

(Web Compatibility :: Site Reports, defect)

defect

Tracking

(Not tracked)

RESOLVED WONTFIX

People

(Reporter: whimboo, Unassigned)

References

()

Details

(Keywords: memory-leak, webcompat:site-wait)

Attachments

(2 files, 1 obsolete file)

Mozilla/5.0 (Macintosh; Intel Mac OS X 10.15; rv:109.0) Gecko/20100101 Firefox/111.0 ID:20230216185941

Having the following page open leaks quite a lot of memory over time. Having the tab open for about a day the memory usage is 1GB.

https://nikonservice-dresden.de/pages/nikon-service-support-kontakt-offnungszeiten-autorisierung-nikon-service-point-fachhandel

A verbose log from about:memory can be found as attachment. The main leak seems to happen here:

│  │  │  │  │  ├──653,699,264 B (67.95%) -- class(Object)/objects
│  │  │  │  │  │  ├──450,574,912 B (46.84%) -- malloc-heap
│  │  │  │  │  │  │  ├──450,378,688 B (46.82%) ── elements/normal
│  │  │  │  │  │  │  └──────196,224 B (00.02%) ── slots
│  │  │  │  │  │  └──203,124,352 B (21.12%) ── gc-heap

So, does something hint it is an issue in Firefox?

If you can reproduce the leak, could you create GC/CC logs before the memory usage goes too much up.
(But it is rather common for web pages themselves just keeping more and more objects alive.)

Severity: -- → S3
Component: DOM: Core & HTML → General
Priority: -- → P3

@ryn.and0 please file your issue as a separate bug. It's not related.

Attachment #9321144 - Attachment is obsolete: true

After running Chrome for about 5 hours I can see that the memory usage of the of the web content process has been increased from ~80MB to 1.5GB as well. So it is most likely a website issue. I'm going to provide the requested GC/CC logs soon.

A memory report is available at: https://www.dropbox.com/s/ze6my5fqx5md0ai/gc-logs.zip?dl=0

Olli, I hope it has the necessary information that you requested. If not let me know and I an re-create.

Flags: needinfo?(smaug)

The Bugbug bot thinks this bug should belong to the 'Core::JavaScript: GC' component, and is moving the bug to that component. Please correct in case you think the bot is wrong.

Component: General → JavaScript: GC

The component has been changed since the backlog priority was decided, so we're resetting it.
For more information, please visit auto_nag documentation.

Priority: P3 → --
Component: JavaScript: GC → Desktop
Product: Core → Web Compatibility

The last uploaded file only contained cc logs. I was requested by Olli to also upload the gc equivalents:
https://www.dropbox.com/s/xxwsfeqrd6gj53z/gc-logs.zip?dl=0

There are lots of objects around, many of them have 8 properties.
I think the ownership goes (one possible path of it): setTimeout() -> js function -> js array -> js function (as an element in the array) -> (something here, possibly a property called 'c' ) -> (property called 'prevObject') -> object with 8 properties
Browser can't release any of that, since at least the function callback in setTimeout is keeping it all alive.

Flags: needinfo?(smaug)

Thank you for the analysis Olli! Lets hope that it will already help to figure out what the problem is.

FYI I've sent an email to the company by adding this bug as reference.

I've just tested and the memory leak is still around.

Verified this issue and still reproduces on Firefox 123 and 125.

Performance profile: https://share.firefox.dev/4bYZKua

The memory report is attached.

Tested with:

Browser / Version: Firefox Release 123.0 (64-bit)/ Firefox Nightly 125.0a1 (2024-02-29) (64-bit) /Firefox Beta 124.0b5(64-bit)/ Chrome Version 122.0.6261.70 (Official Build)
Operating System: Mac Ventura 13.1

Attached file memory-report.json.gz

We can probably close this as this is a site issue and happens in other browsers as well.

Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → WONTFIX
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: