Why Isn't the Memory Leak Being Fixed?

Discussion of general topics about Mozilla Firefox
Locked
schapel
Posts: 3483
Joined: November 4th, 2002, 10:47 pm
Location: Ann Arbor, Michigan
Contact:

Post by schapel »

HenryG wrote:The first one (131456).


I was able to figure out how to reliably reproduce that bug, and actually made a separate bug report on it. However, I never was able to get Mozilla to use more than about 10 MB more RAM than it should because of that bug. I doubt it's repsonsible for the excessive memory usage (200-300 MB or more) and systemwide slowdowns that users are complaining about. But maybe I'm wrong and this has been the problem all along! [-o<
Daemon42
Posts: 7
Joined: February 22nd, 2005, 11:14 pm

Re: 139MB of real memory and 750MB of Virtual memory in used

Post by Daemon42 »

schapel wrote:Just because memory use did not return to the level it was originally does not mean there was a memory leak or any other sort of bug.

Actually that *is*, in fact, the very definition of a memory leak
despite your belief to the contrary, but I'm pretty sure that
ridiculous line of argument has already run its course in this
thread so I'm not going to rehash it. Oddly though none
of the other processes on my machine, which BTW has been
up for 36 days, are using anywhere close to 100MB of real
and 300-750MB of VM.

Of course, if the memory use keeps climbing without limit until the whole computer slows down, now <i>that's</i> a problem. No one has posted a set of instructions for how to reproduce that behavior.


Nah, if I continue to use Firefox heavily for more than about
3-4 days it doesn't crash the machine, it just stops handling
popup windows properly (it ignores the popup blocker
exceptions and blocks everything) and I eventually have to
kill it and restart to fix that bug so it never gets a chance to
grow so huge that it eats the machine.

BTW, because it is a *leak* the memory is allocated, used,
never freed and lost track of by the app, it's never referenced
again, so it'll just keep getting swapped out to disk quietly.
Won't kill the machine until it runs out of swap space.

ian
ronin2040
Posts: 43
Joined: February 24th, 2005, 8:08 pm

Post by ronin2040 »

Im at a loss here....is this entire convo boiling down to like 2 people arguing against everyone else that no, there isnt a problem?

because from the instant i started using (and loving) firefox, i noted that it had the unfortunate quality that memory usage inevitibly continues to grow.

For those who think that its simply because of the current browser's needs, i can tell you right now that that is not the case.
As an experiment, I loaded up Firefox and opened a bunch of empty tabs while watching the task manager. Expectedly, memory usage grew. I then closed all tabs and loaded up an about:blank page, and memory usage simply grew a few more Kb.

One would expect that there would be atleast a semi-significant drop in memory usage--if the browser needs X memory for 10 tabs, one would assume that it needs LESS than X memory for 1 tab. If for some reason it needs more, then that is a seperate issue altogether.

Regardless, I would hope that this issue will get taken care of sooner rather than later. Getting people enthusiastic about a new, "sleek" browser is a tad hard when they realize that its a bit of a memory hog....

I kind of understand where you're coming from Schapel, but then again, I kind of dont. I dont think for an instant that you're some newbie; you obviously have a decent understanding of the situation. But to deny there is a problem because youre not experiencing it, or because it SHOULDNT be a problem, is naieve.

and yes, the memory usage DOES slow down both browsing and desktop usage...
Lost User 66751
Posts: 0
Joined: December 31st, 1969, 5:00 pm

Post by Lost User 66751 »

goatroper wrote:2. <a href="https://bugzilla.mozilla.org/votes.cgi?action=show_user&bug_id=131456">Vote</a> for bug <a href="https://bugzilla.mozilla.org/show_bug.cgi?id=131456">#131456</a>.

According to https://bugzilla.mozilla.org/show_bug.cgi?id=131456 the Status is RESOLVED, Resolution : FIXED
schapel
Posts: 3483
Joined: November 4th, 2002, 10:47 pm
Location: Ann Arbor, Michigan
Contact:

Post by schapel »

ronin2040 wrote:Im at a loss here....is this entire convo boiling down to like 2 people arguing against everyone else that no, there isnt a problem?


No, no one is saying there isn't a problem. ](*,)
schapel
Posts: 3483
Joined: November 4th, 2002, 10:47 pm
Location: Ann Arbor, Michigan
Contact:

Re: 139MB of real memory and 750MB of Virtual memory in used

Post by schapel »

Daemon42 wrote:Actually that *is*, in fact, the very definition of a memory leak
despite your belief to the contrary, but I'm pretty sure that
ridiculous line of argument has already run its course in this
thread so I'm not going to rehash it.


Can you give a reference to your definition of "memory leak"?
Masaki
Posts: 11
Joined: May 9th, 2003, 8:24 am

Post by Masaki »

anjanesh wrote:
goatroper wrote:2. <a href="https://bugzilla.mozilla.org/votes.cgi?action=show_user&bug_id=131456">Vote</a> for bug <a href="https://bugzilla.mozilla.org/show_bug.cgi?id=131456">#131456</a>.

According to https://bugzilla.mozilla.org/show_bug.cgi?id=131456 the Status is RESOLVED, Resolution : FIXED
Any chance of someone backporting the changes to firefox 1.0.x?

The trunk looks a bit too rough to be used atm
User avatar
scratch
Posts: 4942
Joined: November 6th, 2002, 1:27 am
Location: Massachusetts

Post by scratch »

no chance whatsoever.
markmurdoch
Posts: 51
Joined: November 26th, 2004, 6:04 am
Location: Gold Coast Austalia

Post by markmurdoch »

phulla i am running windows 2000 professional processer 533 and 256 ram,.And have the same problem, but i use freshUI 2004 And just keep resetting the memory which is my only choice until i have the time to work something else out.
markmurdoch
Posts: 51
Joined: November 26th, 2004, 6:04 am
Location: Gold Coast Austalia

Post by markmurdoch »

Iam running intel celeron 533 and 256RAM with windows 2000 professional.And i have a huge problem with ram leakage,i just use freshUI 2004 TO FIX the problem until i have got the time and effort to fix it .And that seems to work fine for the time been.
User avatar
Branstrom
Posts: 97
Joined: March 16th, 2004, 9:45 am
Location: Sweden
Contact:

Post by Branstrom »

So it's going to be fixed in Firefox 1.1? No more memory leakage?

Yay. I've been waiting for this one a long time.
User avatar
scratch
Posts: 4942
Joined: November 6th, 2002, 1:27 am
Location: Massachusetts

Post by scratch »

one specific memory leak will be fixed in 1.1.
Daemon42
Posts: 7
Joined: February 22nd, 2005, 11:14 pm

Post by Daemon42 »

schapel wrote:No, no one is saying there isn't a problem. ](*,)


No you've just chosen to redefine what "a problem" is so
as to exclude what most people consider to be a memory
leak. Ever increasing VM size.

My definition of "memory leak" is generally in line with these
http://www.google.com/search?q=define%3 ... S:official
http://www.webopedia.com/TERM/M/memory_leak.html
http://searchwin2000.techtarget.com/sDe ... 33,00.html
They all vary some, but they also all agree on one common thing..
"an application allocations memory and then never returns
it to the system"
Just because you haven't seen an application or OS crash yet
doesn't mean that it's not leaking. Just because the OS can free
up that memory when the application is finally shut down, does
not mean that the application wasn't leaking up to that point.
While it's running, that is memory that can't be used by another
app. If you run out of swap space, the OS is every bit as
out of memory as if it runs out of physical memory. It just
takes much longer to run out of VM.

Now in an almost unbelievable twist of irony, while I was
composing a reply to your message, my machine which as
previous mentioned had been up for 34 days, chose to crash
with a BSOD a few minutes ago, forcing me to hard reset, reboot,
start over.. etc.. See http://thud.us/images/bsod.jpg
I kid you not.

Anyway.. what I was going to say before I was so rudely
interrupted by an OS crash which *probably* wasn't directly
caused by Firefox but may have been a contributing favor due
to excessive memory allocation.. was this..
I happen to be a software engineer so I actually have
some real experience with this topic. My team develops and
supports a large (half a million lines of code) set of DLLs
which all run under a single multithreaded server app that serves
up requests for healthcare transactions. It services 30,000 users,
processing 200-400 thousand realtime transactions per day. It
runs continously for typically about two weeks between
releases (which force us to restart). It allocates and frees
blocks of memory anywhere from a single byte to 30MB at
a time millions of times a day and at the end of a week,
typically the real memory usage is around 45MB, and VM
is about 80MB and it leaks, yes leaks, about 2 or 3MB per day
which is better than it used to be, but not as good as we want.
By comparison.. My Firefox 1.0.1 process which had been running
for about 26 hours before the machine died, used fairly heavily
by one user, me, showed as having allocated 184MB of real memory
and 830MB of VM before the machine crashed. If I closed all but
one window, VM usage would drop to about 700MB and then
stay there indefinitely. That is a big memory leak.

And to answer someone's earlier questions.. Observed memory
leakage was the same regardless of whether plugins were enabled or
not. Memory usage was reduced slightly with the latest nightly
build, but it was buggy in so many other ways (lost such essential
features as "Find") that I had to go back to 1.0.1. .

To everyone else, sorry for the long post. That's why I didn't
want to get into this again, but a leak is a leak and this one is a
doozy. BTW, I wouldn't care, if I didn't really like Firefox a lot
so don't ever get the idea that I'm just bashing on the product.

ian
schapel
Posts: 3483
Joined: November 4th, 2002, 10:47 pm
Location: Ann Arbor, Michigan
Contact:

Post by schapel »

Daemon42 wrote:
schapel wrote:No, no one is saying there isn't a problem. ](*,)


No you've just chosen to redefine what "a problem" is so
as to exclude what most people consider to be a memory
leak. Ever increasing VM size.


Ok, if you want to file a bug report in Bugzilla about the "problem" you experience, go ahead.

Daemon42 wrote:My definition of "memory leak" is generally in line with these
http://www.google.com/search?q=define%3 ... S:official
http://www.webopedia.com/TERM/M/memory_leak.html
http://searchwin2000.techtarget.com/sDe ... 33,00.html
They all vary some, but they also all agree on one common thing..
"an application allocations memory and then never returns
it to the system"


No, they do not all agree on that one common thing. Only one of those definitions mentions returning memory to the system. A memory leak occurs when a program fails to release memory <i>internally</i>. Generally, that memory is <i>not</i> immediately returned to the operating system, but is reused by the program. If the memory were not freed and made available for reuse, the program would keep requesting more and more memory until the computer's memory was exhausted.

You should know this as a software developer. I'm also one, and I wrote a C++ utility for finding memory leaks.


Anyway, all this talk is distracting us from the problem at hand. You and others <i>are</i> having a serious issue with memory usage. In order to get it fixed, we need to file a bug report in Bugzilla. That will require a set of steps that will reproduce the extreme memory usage. As soon as someone posts such a set of instructions, I will help file the bug report.
User avatar
scratch
Posts: 4942
Joined: November 6th, 2002, 1:27 am
Location: Massachusetts

Post by scratch »

schapel wrote:Ok, if you want to file a bug report in Bugzilla about the "problem" you experience, go ahead.


there are already numerous bugzilla bugs on it.

Daemon42 wrote:If the memory were not freed and made available for reuse, the program would keep requesting more and more memory until the computer's memory was exhausted.

You should know this as a software developer. I'm also one, and I wrote a C++ utility for finding memory leaks.


this is exactly what happens. did you not see the part where he said firefox was using *1 GB* of memory?
Locked