Mozilla can just suck up memory. Especially when you come to pages that take a lot of memory to render. Some pages, like recursive frames can suck up megs and megs of memory. Mozilla needs to keep track of how much memory different parts can use, and shut off the valve if the mem starts to become a large percentage of the computer's memory. Possibly, people could even set how much mem is the max it can use.

I got positive feedback from Jesse and he asked me to add him.

See also bug 30942, Browser should remain responsive during most infinite loops.

See also bug 8065 where Mozilla should shut off a web page after it has sucked up too much mem.

Changed it to per window because browser windows aren't the only thing we want to limit memory usage on.

> Changed it to per window because browser windows aren't the only thing we want > to limit memory usage on. Not sure about that one. Mailnews can (currently) suck up large amounts of mem. You'd have to allow 70+ MB for the Mailnews 3pane window, but 70MB are way too much for a browser window. I suggest to limit mem useage for foreign content, e.g. per <browser> section.

What is the 3pane window? If it uses 70+MB, then that is way too much and that needs to be fixed in a seperate bug. Limiting mem usage might mean that certain things will need to be adjusted or won't work such as that 3pane window. Maybe there can be a list of exceptions for the memory limiting code until they are fixed. If the window is, say 70MB because it has 70MB of messages loaded, then it should only load a few at a time from files.

Any ideas what a good algorithm would be? An application needs to have more memory per window if there is more free system memory, but there still needs to be a bottom line that must be fulfilled even if the free memory is low. If that bottom line can't be fulfilled, then Mozilla should stop allocating mem and put any part of the app that requires more mem in a sort of "standby" mode (bug 70824).

-> not mine.

testcase if you dare ;-) -> http://mozilla.netdemonz.com/bugs/testcase2.html

That testcase bloated my mozilla mem usage to 94 mb. I believe that if I had made that page larger (say 100MB) Mozilla would have kept trying to open it till it used all the hard disk virtual mem and would bring the system to a standstill and cause a reboot to be the only discourse.

Sending to XPApps (it sounded like it belonged there)

probably a dup of some other perf bug, reassigning to cathleen

I think the odds of this ever being fixed are pretty low.

I've seen firefox suck up over 1.8GB of my RAM. I've had it open about 4 hours and the usage is at 494MB with a peak of 523MB.

We should understand if we really want this feature. Otherwise mark it as WONTFIX.

We recently removed some limits of this nature from the JS engine because they're an enormous pain. And also see comment 13. I will close this.