1980023 - Memory leak in main process over extended (2+ days) sessions

Status: UNCONFIRMED

(Core :: Performance: Memory, defect)

Description

[Eddie Carswell]

User Agent: Mozilla/5.0 (X11; Linux x86_64; rv:143.0) Gecko/20100101 Firefox/143.0

Steps to reproduce:

Open a Firefox session with several tabs (mine just restores previous session).

Wait for several (2+) days and observe memory usage of main process.

Note: my session usually has hundreds of tabs, but only a few are loaded at once.

Actual results:

Huge memory allocation for heap-unclassified (~80% or more) in the main process. This often reaches several GB. If I leave the browser open for a week or more, I might see it reach over 10GB in heap-unclassified.

This has nothing to do with private browsing and is not a duplicate of bug 1973462.

Expected results:

The main process shouldn't be leaking memory over long running sessions.

Comment 1

[Eddie Carswell]

Attachment: 2025-06-12_about-support.json

Comment 2

[Eddie Carswell]

Attachment: 2025-06-12_memory-report.json.gz

Comment 3

[Eddie Carswell]

Attachment: 2025-07-29_memory-report.json.gz

Comment 4

[Eddie Carswell]

Attachment: 2025-07-29_about-support.json

Comment 5

[Eddie Carswell]

This has been a problem for several months now, as can be seen by the attached reports.

Running the latest Nightly (at time of browser launch), OS is Fedora 42 with KDE on Wayland.

Comment 6

[Eddie Carswell]

Attachment: 2025-07-29_memory-report_fresh-launch-1min.json.gz

Comment 7

[Eddie Carswell]

Attachment: 2025-07-29_memory-report_fresh-launch-5mins.json.gz

Comment 8

[Eddie Carswell]

I've attached two more after restarting the browser (and auto updating).

About one minute in, it balloons to 4GB in heap-unclassified, but then about 5 minutes in, it drops to ~300MB.

Over time that number will continue to rise until I restart the browser.

Comment 9

[:Gijs (he/him)]

:mccr8, can you advise on how to get more actionable information from the increase in heap-unclassified here?

(In reply to Eddie Carswell from comment #0)

Wait for several (2+) days and observe memory usage of main process.

Just to clarify, I'm assuming you're actually using the browser over these 2 days? Or is loading a session with, say, 100 instances of (literally!) https://example.com/ and doing nothing for 2 days enough to reproduce?

Comment 10

[Eddie Carswell]

(In reply to :Gijs (he/him) from comment #9)

Just to clarify, I'm assuming you're actually using the browser over these 2 days? Or is loading a session with, say, 100 instances of (literally!) https://example.com/ and doing nothing for 2 days enough to reproduce?

I've observed the memory leak in both cases of regular daily usage, and several days without use. While I have hundreds of tabs, only a few are actually loaded, the rest start discarded (including my pinned tabs) until I need to use them.

I'm not sure how easy/difficult it'll be to reproduce this issue but if there's other things I can do to gather further details let me know.

Comment 11

[Andrew McCreight [:mccr8]]

I don't think there's any more reliable approach to figuring out heap-unclassified than using DMD with a Nightly build.

Comment 12

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

The severity field is not set for this bug.
:jlink, could you have a look please?

For more information, please visit BugBot documentation.