This issue actually occured on many other pages as well. Though i have no evidence, i evidently found this issue could be detonated by continually refreshing.

I mean continually refreshing page.

Moving this bug report about memory usage (which is not related to the memory allocator) to Untriaged.

I have tested this issue using Ubuntu 64bit 18.04 lts and also on Windows 10 64bit

• Firefox Latest Nightly release 69.0a1 (2019-06-03)
  1)Leaving only one Firefox windw open with only 1 tab, youtube video running https://www.youtube.com/watch?v=bKOTKHtbM54
  2) Hit refresh several times (more than 20 times)
  The memory consumption for has increased to 1.7gb used and keeps increasing each time i hit refresh.

• Firefox Beta 68.0b6: No issue when doing a refresh several times (memory consumption does not reach levels of nightly)

• Firefox release 67.0 : No issue when doing a refresh several times (memory consumption does not reach levels of nightly)

Hillwood

For firefox version 60.4.0 i used ESR release on Ubuntu UBUNTU 18.04 64 bits, but i was not able to replicate that amount of memory consumption shown in your image, as soon as it gets about 700mb consumption it always drops.

Please let me know what linux distro and version are you using, and the link of the website are you using to refresh the page constantly, i would like to replicate the original problem using the same site you used.

thanks,
Pablo

Mike, can you help me give this bug a correct component? Thanks.

(In reply to Pablo from comment #8)

Hillwood

For firefox version 60.4.0 i used ESR release on Ubuntu UBUNTU 18.04 64 bits, but i was not able to replicate that amount of memory consumption shown in your image, as soon as it gets about 700mb consumption it always drops.

Please let me know what linux distro and version are you using, and the link of the website are you using to refresh the page constantly, i would like to replicate the original problem using the same site you used.

thanks,
Pablo

openSUSE Leap 15.0.

But i can not reappear this problem on firefox 60.6.3 with openSUSE Leap 15.1. Maybe it could be fixed.

Considering the fact that the component was reset to untriaged, I will now set the bug's component to (Core) DOM: Core&HTML. If incorrect, please set it to a more appropriate component rather than resetting it to Untriaged. Thank you.

(In reply to Pablo from comment #7)

I have tested this issue using Ubuntu 64bit 18.04 lts and also on Windows 10 64bit

• Firefox Latest Nightly release 69.0a1 (2019-06-03)
  1)Leaving only one Firefox windw open with only 1 tab, youtube video running https://www.youtube.com/watch?v=bKOTKHtbM54
  2) Hit refresh several times (more than 20 times)
  The memory consumption for has increased to 1.7gb used and keeps increasing each time i hit refresh.

• Firefox Beta 68.0b6: No issue when doing a refresh several times (memory consumption does not reach levels of nightly)

• Firefox release 67.0 : No issue when doing a refresh several times (memory consumption does not reach levels of nightly)

Sounds there's a new a regression from 69 that's concerning.
Andrew, would you please chime in how we can move this forward? Thanks!

(In reply to Hsin-Yi Tsai [:hsinyi] from comment #12)

Andrew, would you please chime in how we can move this forward? Thanks!

There's clear steps to reproduce, so I think running mozregression to figure out when the behavior changed would be the next step.

I'm a little surprised that this used to be okay. We do have some kind of mechanism to force a GC when memory gets low. Maybe that broke, or maybe something else changed to stop it from freeing the page.

(In reply to Pablo from comment #7)

I have tested this issue using Ubuntu 64bit 18.04 lts and also on Windows 10 64bit

• Firefox Latest Nightly release 69.0a1 (2019-06-03)
  1)Leaving only one Firefox windw open with only 1 tab, youtube video running https://www.youtube.com/watch?v=bKOTKHtbM54
  2) Hit refresh several times (more than 20 times)
  The memory consumption for has increased to 1.7gb used and keeps increasing each time i hit refresh.

• Firefox Beta 68.0b6: No issue when doing a refresh several times (memory consumption does not reach levels of nightly)

• Firefox release 67.0 : No issue when doing a refresh several times (memory consumption does not reach levels of nightly)

Hey Pablo, any chance you would help run mozregression to get where this regressed from? Thank you.

Hi , these are the results from mozregression:

build_date: 2019-06-01 01:42:01.782000
build_url: https://queue.taskcluster.net/v1/task/UwPs5PQaSEGC6DVU8htxxg/runs/0/artifacts/public%2Fbuild%2Ftarget.zip
changeset: 72fc109fe0f203a360003b1f4f7a775ab161cfe4
pushlog_url: https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=a55afb0346770d95e0d978538ce79023597df785&tochange=72fc109fe0f203a360003b1f4f7a775ab161cfe4

Retested again Using windows 10 with the latest versions and i was not able to reproduce the problem.
Firefox nightly 70.0a1
Firefox beta 69.0b4
firefox release 68.0

I will test again with ubuntu 18

(In reply to Pablo from comment #15)

Hi , these are the results from mozregression:

build_date: 2019-06-01 01:42:01.782000
build_url: https://queue.taskcluster.net/v1/task/UwPs5PQaSEGC6DVU8htxxg/runs/0/artifacts/public%2Fbuild%2Ftarget.zip
changeset: 72fc109fe0f203a360003b1f4f7a775ab161cfe4
pushlog_url: https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=a55afb0346770d95e0d978538ce79023597df785&tochange=72fc109fe0f203a360003b1f4f7a775ab161cfe4

(In reply to Pablo from comment #16)

Retested again Using windows 10 with the latest versions and i was not able to reproduce the problem.
Firefox nightly 70.0a1
Firefox beta 69.0b4
firefox release 68.0

I will test again with ubuntu 18

Thanks Pablo.
Per the pushlog, bug 1167452 sounds suspicious. A stack of patches started to land after May 30, then some pieces got backed out around June 11. That may explain the result from comment #16.

Thanks for doing the mozregression. I'm going to move this to the GC component.

It sounds like we can mark this fixed, then?

You could confirm the fix using mozregression, but the dates of everything in comment 17 line up so well, plus you can't reproduce this on trunk, so I don't think that's necessary.

Steve, here's another regression from your patches to be aware of before you reland the rest of bug 1167452.