Closed Bug 251380 Opened 20 years ago Closed 17 years ago

When saving a picture, HUGE memory leak! Also slows machine down!

Categories

(Firefox :: General, defect)

Product:
Firefox
Component:
General
Platform:
x86
All
Type:
defect
Priority:
Not set
Severity:
critical

Tracking

()

Status:
RESOLVED WORKSFORME

People

(Reporter: xpw, Unassigned)

References

Details

(Keywords: memory-leak)

Attachments

(2 files)

shows how firefox eats 500 MB memory
113.47 KB, image/jpeg
Details
screenshot of ProcessExplorer showing what was bogging down the CPU
34.59 KB, image/png
Details 
User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; de-DE; rv:1.7) Gecko/20040707 Firefox/0.9.2
Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 5.1; de-DE; rv:1.7) Gecko/20040707 Firefox/0.9.2

If I save some pictures (5-8 jpg pictures of a resolution of about 800x1000 are
already enough), firefox slows dramatically down and eats up *a* *lot* of
memory. This was present already in V0.9.1
After 10 saved Pictures memory usage increases from ~25 MB to more thann 200 MB.
(I have 640 MB RAM and a 1 GB swap file)

I don't know if it has sth. to do with AntiVir, but according to the Task
manager it's firefox who's eating up all the memory.


Reproducible: Always
Steps to Reproduce:
1. surf to some site with large (800x1000 or greater) pictures on it
2. Open some (6..10) pictures in new tabs by pressing the mid button on the JPG
link. (Or only 1 pic in a new tab, save it, close tab, open the next, and so on
--- doesn't matter)
3. right click the picture an choose "Save picture as"
4. repeat 5 to 10 times from step 1

Watch the memory usage in task manager: it grows RAPIDLY. *MUCH* faster than the
file size of the images. Seems, the allocated RAM for the already DECOMPRESSED
JPEP-Picture isn't freed. 
(IMHO it's strange enough that so much RAM is allocated when *SAVING* a picture)
Actual Results:  
Firefox slows down RAPIDLY  (you have to wait more than a minute, to use firefox
again when the fith or sixth or later image is saved)
And firefox eats an never frees muuuuuuuuch memory.

Expected Results:  
free the memory or better: don't allocate it at all (what's the sake of
allocating memory, when saving a picture from memory, that already *IS* in the
memory?)
I'm running Windows XP Pro and:
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7) Gecko/20040707 Firefox/0.9.2

I got 512MB RAM and I just tried this bug out. I opened up 15 jpegs size
1920*1080, while having them open memory usage went up from ~60MB (which seems
to be my "normal" usage) up to ~130MB, but saving them and closing down the tabs
gradually lowered the usage back down to ~60MB.

The images collected file size is about 2,35MB so I can agree memory usage seems
a bit high when having a lot of images open, but I didn't experience any slow
downs due to it.
Severity: blocker → critical
I can also confirm this behavior:

I'm running Firebird 0.9.2 on WinXP, Pentium 4 3.20 GHz with 2.0 GB of RAM. 
What is interesting is that you can use just one tab and still get the same
problem.    Closing the tab does not recover the memory.  Only shutting down the
application.
You should add mlk to keywords and block Aviary 1.0
I experience severe memory leak as well.  Firefox become unusable (crash) after
browsing internet for 30min.  Memory image ~100MB.  I do nothing special to
recreate.  I simply browse web normally.  I notice that when downloading files
as well causes memory to go up.

WinXP SP2, Dell D600 laptop
needs mlk keyword
mlk keyword needed?
For me it's not just with large images but the volume of images. From time to
time I visit google image search to download celebrity pics. Often I'll want to
pull down 40-100 images in a session. It gets progressively slower with each
image saved. Tonight I got to about 30 before FireFox 1.0PR got so slow that I
considered it hung (waited about 5 mins after clicking Save Image As). 

Following a bug search I tried clearing the download manager but that hasn't
helped. I used the Clear All button, I didn't delete the file completely.

I agree with a comment elsewhere that the download manager should probably be
bypassed when doing an image save.

It would probably also be a good idea if FF saved directly from the cache rather
than checking the website for a fresh one. After all, FF should have already
refreshed the page as per the cache settings so it shouldn't need to
double-guess those settings.

I'm seeing this behaviour on mozilla 1.8a5. When I load lots of images small
size or a few big ones the memory usage jumps up very quickly. However when I
close the browser window the memory doesn't get freeded. This only happens when
I close mozilla fully. I can reproduce this every time I open mozilla and load a
big set of images
Confirming with Mozilla/5.0 (Windows; U; WinNT4.0; en-US; rv:1.7.5)
Gecko/20041110 Firefox/1.0.

I have NT SP6 box with 256MB mem. Opened 6 tabs with average news-site type of
pages. Tried "Save As" (Page complete) on the 6th tab and it took 20 secs for
the "download complete" pop up to appear. Did additional 3-4  saves, and with
each one  the time the system hung until released with "download complete" grew
longer - on the 4th save took more than 7 minutes. Memory grew from 50MB to
183MB during the saves.
Mozilla 1.4 and FF 0.8 do not show this behavior, and they save those pages quickly.
Needs mlk key word and Firestone1.1 as the Target Milestone.
Confirming, linux, firefox 1.0. This happens after log time browsing. Now saving
image consumes more and more memory. Firefox is getting really sluggish 

before image save:
  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
18341 pribya    15   0  212m 129m  34m S  0.0 52.0   9:32.60 firefox-bin
after image saved.
18341 pribya    15   0  227m 144m  34m S  1.0 57.8   9:42.25 firefox-bin

As you may notice also to size of firefox itself is not usuall on system with
256MB of RAM.
After firefox restart:
  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
26191 pribya    16   0  112m  26m  30m S  0.0 10.4   0:01.83 firefox-bin

The TIME+ is a pure runtime of application, not a realtime - it was started
approx. 5 days ago.
Keywords: mlk
OS: Windows XP → All
I have the same problem, and it is very reproduceable.  If I open 10 to 12 tabs
and right click / save on an image of 30 to 60 Kb each on each page, the memory
use by firefox goes up and up from 4 Mb until at 250 to 260 Mb, firefox locks
up taking 99% cpu.  It appears that the same happens when saving a lot of entire
web pages.  

The problem seems to have improved after going to tools / options / downloads and
clearing the "show download manager..."

Other symptoms that something is not right with the download manager include
the fact that on some saves the download manager window would appear, but in 
most cases it would not.  Likewise, the download complete flag would pop up
from the right side of the toolbar, but usually not.

I'm using the released ver 1.0
-------
OS:XP PRO SP2 ...
HW: AMD X P1.7GHz, 128MB, NVIDIA MX400 64MB ...
FIREFOX v1.0 ... 
STATUS: CONFIRMED REPRODUCIBLE - 
-------
After downloading and "saving-as" to disk approximately TEN (10) small .JPG
images (TOTAL FILESIZE ~2.5MB) from the site
[http://jeh.5u.com/christmas/victoria.htm], the PAGEFILE memory in the XP TASK
MANAGER "Performance" tab starting increasing, eventually peeking at ~300MB. I
had no more then FOUR tabs open (eventually, I had only TWO open when I gave up
the process).

Another commenter lists the pixel size of the images (and opens them in separate
tabs); size of the images doesn't appear to be as relevant as the number of
images (most of the images on the page I included are far smaller then 800x1000).

After exiting FIREFOX v1.0, PAGEFILE memory usage was reduced to ~140MB (system
startup PAGEFILE memory usage, with zero apps running, is ~120MB). 

Oddly, you can VIEW the example page I've listed without problem (in a single
tab, at least); it's when you start "saving-as" the small (size range of 14KB to
214KB) .JPG images that the memory leak (?) apparently occurs. 

The included page URL is one example; I've earlier noted similar behaviour on
several occasions, and with the v.9 FIREFOX. 

Exiting FIREFOX will regain most (all?) your pagefile memory. 

From other commenters, it seems not to matter what the size of physical RAM is
(range of 128MB to 512MB listed); for whatever reason, the issue seems to be in
the XP PAGEFILE cache. HTH
FOLLOWUP TO PREVIOUS EVENING BUG REPORT (same machine):

Okay. Another commenter noted that by turning OFF the FIREFOX v1.0 DOWNLOAD
MANAGER, the "memory leak" seemed lessened/manageable. So I did this (last
night, prior to knocking off), and this morning went back to the site
(previously posted), started WINDOWS TASK MANAGER, started FIREFOX v1.0, and
started downloading images while noting the PAGEFILE SIZE in TASK MANAGER.
PAGEFILE size was 123MB prior to starting FIREFOX, and 168MB after starting FIREFOX.

I can confirm that after "saving-to-disk" 26 small .JPG images (832KB total),
there was NO EVIDENCE of memory leakage (so some code in the "Download Manager"
may be the problem source??), but also noted an additional datapoint that may be
of use in pinpointing the issue.

Clicking one of the images inadvertently opened a new instance [window] of
FIREFOX. I noted a PAGEFILE size increase to 193MB. I "saved-to-disk" several
small images in this new instance [window] of FIREFOX; there was no increase in
the 193MB PAGEFILE size. I then CLOSED the "new" instance [window] of FIREFOX,
and the PAGEFILE size (of 193MB) **stayed the same**, i.e., the
post-closing-the-"new" window PAGEFILE size remained at 193MB.

Now the apparent anomaly. I purposely opened ANOTHER "new" instance of FIREFOX
on the site (which is heavy with images), and the PAGEFILE size **_decreased_
from 193MB to 172MB**. Hmm. I, at least, thought this rather odd behaviour for a
"memory leak" (but what do I know). "Saving-to-disk" several (large) images
didn't increase this 172MB appreciably.

SUMMARY
-----
1. PAGEFILE size no apps: 123MB
2. PAGEFILE size FIREFOX start: 163MB
3. PAGEFILE size after multiple image save-to-disk: ~163MB
   NOTE: no evidence of memory leak with _NO_ DOWNLOAD MANAGER 
4. PAGEFILE size after OPENING "new" FIREFOX window: 193MB
5. PAGEFILE size after multiple image saves in "new" window: 193MB
6. PAGEFILE size after CLOSING "new" window: 193MB
7. PAGEFILE size after OPENING another "new" window: 172MB
   NOTE: so PAGEFILE size *decreased*; in other words, the initial
         30MB PAGEFILE increase (from 163MB --> 193MB) wasn't released
         until after closing the first additional instance, and opening
         another new instance

Prior to a bug-fix, I'd recommend that anyone experiencing this memory leakage
problem:
   A. turn off the DOWNLOAD MANAGER (via TOOLS-->OPTIONS,DOWNLOADS menu), and
   B. keep in mind that opening a new window WILL increase PAGEFILE size, and
      this "new" window's memory resources will NOT be decreased until you 
      close the "new" window and open a THIRD window.

...if I note anything else I'd think would help pinpoint the problem, I'll followup.
(In reply to comment #1)
> I'm running Windows XP Pro and:
> Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7) Gecko/20040707
Firefox/0.9.2
> 
> I got 512MB RAM and I just tried this bug out. I opened up 15 jpegs size
> 1920*1080, while having them open memory usage went up from ~60MB (which seems
> to be my "normal" usage) up to ~130MB, but saving them and closing down the tabs
> gradually lowered the usage back down to ~60MB.
> 
> The images collected file size is about 2,35MB so I can agree memory usage seems
> a bit high when having a lot of images open, but I didn't experience any slow
> downs due to it.

JPEGs are compressed files so the file size is always smaller than the amount of
memory used when the image is displayed. I have JPEG files from my digital
camera that have a disk size of 730KB but use up 5MB or more when displayed.
I'm running SUSE 9.2 and have the  same kind of memory leak expressed by the
other users.  Saving many small pictures has the same result..eventually
the  download manager quits responding.

Check if this is still happening after this is fixed:
https://bugzilla.mozilla.org/show_bug.cgi?id=131456
Sounds like a duplicate of Bug 273244.

-------
OS:XP PRO SP2 ...
HW: AMD X P1.7GHz, 384MB, NVIDIA MX400 64MB ...
FIREFOX v1.0.1 ... 
STATUS: REPRODUCIBLE 
-------

This post is also a followup to my earlier posts describing the "increasing
cache" behaviour. NOTE: The HW I'm using has changed a minor amount (with a
physical memory upgrade of an additional 256MB), and Firefox has been [auto]
updated to v1.01.

Further, a google search on the terms "Firefox memory leak" - which eventually
brought me to the webpage "Secrets of Firefox"
<http://windowssecrets.com/041202/> - suggested that changing
browser.cache.memory.capacity might help with the problem. The current setting
is 65536. 

And the suggested browser.cache.memory.capacity "fix" *did* alleviate the
problem (meaning the problem with physical cache growth was slowed during a
session). And memory was released quicker to the "static state" of the memory
cache upon exit (i.e., I no longer had to exit Firefox, AND re-start Firefox
*and* "re-"exit Firefox to effect a "full" release of the cache, as I reported
in an earlier, pre- "browser.cache.memory.capacity" setting post).

I would like to add two more data points. 

1. When I minimize Firefox during an ongoing session for a several minute period
(to perform other activities in the foreground), there's a significant delay and
*much* drive activity before Firefox will maximize to the foreground again.
(Sorry, I haven't checked the minimize times nor the cache amount; I'll followup.)

2. I'm now using the Session Saver extension. Exiting Firefox for a few minutes,
and restarting the SAME session results in DRAMATICALLY lower "current cache"
usage. (For example, last night the PF setting as reported in Task Manager
reached approximately 285MB; restarting Firefox this morning with the "saved
session" was 165MB.) That datum seems ...odd ...to me. After all, it's the same
session, with the same number of open tabs & web pages being viewed. Why
wouldn't the memory cache be the same?

As per comments #18 & #19, I checked comments in Bugs #273244 & Bug #131456
(especially in the latter ...although I didn't follow all the other BR threads),
and they both *seem* related from the descriptions to the problem I and others
posted here (although the emphasis in this BR has been on viewing and
downloading graphic images). 

Hmm. Another datapoint: lately my online sessions - including last night - have
been less about downloading, viewing & saving multiple JPEGs of various sizes,
then about opening multiple tabs to pages with minimal graphical content, and
then downloading various Linux distros - of hundreds of megabytes - in .ISO
format (yeah, for the obvious reason: MS security is getting to be
counter-productive, particularly online work).

And, as others have suggested, this "JPEG" problem *may* be (along with the
other BR's) one of multiple problems ...with the single symptom of an untoward
increase in PF usage (cache size), and [cache] memory not being released when a
tab is closed.

The following is just for thought: the "behaviour" of opening & closing tabs, so
you can quickly refer to a page, in Firefox ...is done in IE by opening a new
session of IE. When you exit the various open sessions of IE, your memory is
restored (well, I'd assume: I really should verify this ...it's been awhile
since I've used IE, but I don't recall *this* as being an issue). Firefox has
made the SAME *behaviour* more convenient by allowing "sessions" to be tabbed
(I'd assume this is akin to Opera ...but it's been an even longer time since I
used Opera, and I do NOT recall being as impressed with the use of the tab
feature then in Opera as I am now in Firefox). 

I said all that to say this - and I am NOT a programmer (and so I don't know
what's going on in the code) - why does opening & then closing a Firefox tab
*not* release the entire [cache] memory used by the tab? I mean, if the cache is
165MB, and then I open a new tab and the cache goes to 174MB, and I subsequently
close that new tab ...it would seem that the PF usage [cache] should drop to the
original 165MB, wouldn't it? (And when you consider the effect of exiting and
reoping using Session Saver, it's even more dramatic.) I dunno, but something
about this seems counter-intuitive.

Ah, but what do I know ...you guys must have already been through all this.
This is related to bug #259672.
As per comment 21, Bug #259672 seems related. The focus of that report is on
X-server. Symptoms appear to be the same.

Bug #258672 links to Bug #257016. The focus of that report is on X-server.
Symptoms appear to be the same, but comment 2 describes "...X-server memory as
growing 'three times as fast' ".

Bug #257016 links to Bug #213391. This report appears to be an *exact* duplicate
of the problem described here (ie, bug #251380). A real programmer (I'm not)
might want to note comment 4 (in #258672): "I tested on older builds, this is a
regression. The leak has been introduced between 1.3alpha and 1.3beta."
I find the same massive slow-down when downloading multiple files (in this case,
using "save link as..." to download online story chapters). The more files I
download during a Firefox session, the slower it gets. If I close Firefox and
restart it, the problem clears up for a while. Noticed this in 0.9 and still see
it in 1.0 (installing 1.0.1 when I close this session).
Running Win98se on AthlonXP 2400, 512MB, 32MB Matrox GFX, 9+GB free on C:
Flags: blocking-aviary1.1?
System: WinXP SP2
Browser: Firefox 1.0.1
System cfg: Athlon 1700+, 768 MB ram, 768 MB swap

Huge memory link, this session i used a lot 'save image as' and as you can see
from the screenshot it has eaten a lot fo memory.
Confirming this issue for Firefox 1.01, 1.02. Running on XP SP1 and XP SP2. 
Apparently it is very easy to reproduce just by going to a web site with large 
images (like http://gadgetphotogallery.i4u.com/squito/index.php?photo_id=15) 
and there is no need to actually save the images. Just view them as you would 
normally do, with your Task Manager window open and you will see how with 
opening of every single JPG file, the memory usage rises. Also use of tabs may 
help this, but I never use tabs and I still have my Firefox slow down and 
crash when the memory usage grows to 240-250MB (my RAM is 1.5GB)
*** Bug 292730 has been marked as a duplicate of this bug. ***
Keywords: crash, hang
Keywords: crash, hang
Summary: Whend saving a picture, memory is eaten up / HUGE memory leak! Also slows machine down! → When saving a picture, HUGE memory leak! Also slows machine down!
*** Bug 285490 has been marked as a duplicate of this bug. ***
This sounds like the same thing I am experiencing with Firefox 1.0.4.

When browsing sites with large images, and opening images in new tabs and
saving them and closing tabs, Firefox's memory usage climbs to hundreds of megs
and the user interface becomes totally unresponsive for several seconds at a
time each time clicking on a link or closing a tab or saving an image etc. (my
experience was with easynews.com saving images from usenet, so i can't offer
any helpful links to test this)

I'm running Windows XP SP2, P4 2.4GHz, 2GB RAM, GeForce 4 ti4600 128MB.

I have several extensions installed, but regardless of that, this is still a
bug. I will attempt a clean install on a new profile later to see if the same
problem occurs.

Another note, the last time I was experiencing the problem, I opened
sysinternals' Process Explorer and viewed Firefox, it showed that all the CPU
was being consumed by jpeg_free_small during the unresponsive times after
clicking links or opening/closing new tabs (screenshot attached). I don't know
what that means, but maybe this will help track down the bug?
Try to disable FF internal download manager, I am susspicious that this is DM issue.
*** Bug 299167 has been marked as a duplicate of this bug. ***
see the comments on 
https://bugzilla.mozilla.org/show_bug.cgi?id=299167

Firefox jumps to 100% CPU usage when saving many small pictures
worked with Windows XP and Linux
minusing for now, this is most likely either a dupe of/tightly related bug
131456 or Bug 259672.
Flags: blocking-aviary1.1? → blocking-aviary1.1-
(In reply to comment #33)
> minusing for now, this is most likely either a dupe of/tightly related bug
> 131456 or Bug 259672.

This is most probably not tab related, but download manager and definitely is
not unix only as 259672 is. However 259672 could be part of this problem.



FF 1.0.4 still has this behavior.

Testcase: a local page and local file for download:
<html>
<a href="Titan IV.mpg">Titan IV.mpg</a>
</html>

after several "Save As..." (sorted pmap output):

08989000    30304K rwxp [heap]
08048000     9172K r-xp /.share/usr/app/firefox-1.0.4/firefox-bin
401b3000     2436K r-xp /.share/usr/app/gtk+-2.4.3/usr/lib/libgtk-x11-2.0.so.0.400.3
41205000     2312K rwxp [ anon ]
bf401000     2044K rwxp [ anon ]
bf201000     2044K rwxp [ anon ]
bf001000     2044K rwxp [ anon ]
bee01000     2044K rwxp [ anon ]
408b6000     1080K r-xp /app/glibc-2.3.2/lib/libc-2.3.2.so
40b63000     1024K rwxp /dev/zero
40f08000      992K ---p [ anon ]
4059c000      752K r-xp /.share/usr/X11R6-4.4.0/lib/libX11.so.6.2
40080000      620K r-xp /.share/usr/app/firefox-1.0.4/libxpcom.so
407fe000      592K r-xp /.share/usr/app/gcc-3.3.3/lib/libstdc++.so.5.0.5
40526000      464K r-xp /.share/usr/app/glib-2.4.2/lib/libglib-2.0.so.0.400.2
40013000      420K r-xp /.share/usr/app/firefox-1.0.4/libmozjs.so
40720000      392K r-xp /.share/usr/app/firefox-1.0.4/libsoftokn3.so
40e44000      384K rwxs /SYSV00000000
40c8c000      384K rwxs /SYSV00000000
406bd000      376K r-xp /.share/usr/app/firefox-1.0.4/libnss3.so
40420000      372K r-xp /.share/usr/app/gtk+-2.4.3/usr/lib/libgdk-x11-2.0.so.0.400.3
0893d000      304K rwxp /.share/usr/app/firefox-1.0.4/firefox-bin
40786000      284K r-xp /.share/usr/X11R6-4.4.0/lib/libXt.so.6.0
409fe000      280K r-xp /.share/usr/X11R6-4.4.0/lib/libfreetype.so.6.3.3
4016d000      268K rwxp [ anon ]
40d2f000      256K rwxp [ anon ]
40cec000      256K rwxp [ anon ]

Doing "Save As..." once again - 4 megs leak:

08989000    34216K rwxp [heap]
08048000     9172K r-xp /.share/usr/app/firefox-1.0.4/firefox-bin
401b3000     2436K r-xp /.share/usr/app/gtk+-2.4.3/usr/lib/libgtk-x11-2.0.so.0.400.3
41205000     2312K rwxp [ anon ]
bf401000     2044K rwxp [ anon ]
bf201000     2044K rwxp [ anon ]
bf001000     2044K rwxp [ anon ]
bee01000     2044K rwxp [ anon ]
408b6000     1080K r-xp /app/glibc-2.3.2/lib/libc-2.3.2.so
40b63000     1024K rwxp /dev/zero
40f08000      992K ---p [ anon ]
4059c000      752K r-xp /.share/usr/X11R6-4.4.0/lib/libX11.so.6.2
40080000      620K r-xp /.share/usr/app/firefox-1.0.4/libxpcom.so
407fe000      592K r-xp /.share/usr/app/gcc-3.3.3/lib/libstdc++.so.5.0.5
40526000      464K r-xp /.share/usr/app/glib-2.4.2/lib/libglib-2.0.so.0.400.2
40013000      420K r-xp /.share/usr/app/firefox-1.0.4/libmozjs.so
40720000      392K r-xp /.share/usr/app/firefox-1.0.4/libsoftokn3.so
40e44000      384K rwxs /SYSV00000000
40c8c000      384K rwxs /SYSV00000000
406bd000      376K r-xp /.share/usr/app/firefox-1.0.4/libnss3.so
40420000      372K r-xp /.share/usr/app/gtk+-2.4.3/usr/lib/libgdk-x11-2.0.so.0.400.3
0893d000      304K rwxp /.share/usr/app/firefox-1.0.4/firefox-bin
40786000      284K r-xp /.share/usr/X11R6-4.4.0/lib/libXt.so.6.0
409fe000      280K r-xp /.share/usr/X11R6-4.4.0/lib/libfreetype.so.6.3.3
4016d000      268K rwxp [ anon ]
40d2f000      256K rwxp [ anon ]
40cec000      256K rwxp [ anon ]
*** Bug 294187 has been marked as a duplicate of this bug. ***
Flags: blocking-aviary2.0?
Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8) Gecko/20051111 Firefox/1.5
After visiting sites like www.deviantart.com, opening a lot of images and saving them with "Save Image As" I can see that memory usage grows and after closing tabs with images goes a little down, but not completely (I was specifically trying now to catch what is causing it and found that "Save Image As" makes memory leaks more likely, that's how I found this bug afterwards). On about:cache I see that memory cache size stays at around 27Mb, but it obviously can't explain firefox taking 135MB with only two tabs open (this bug and bugzilla search results of 200 bugs). Within a week or so firefox often takes up to 1GB of ram, because I'm using laptop and am hibernating heavily -- I don't like having to close pages I'm working with...
If someone experiencing this reliably could follow the instructions here to create a leak log, that'd be great.  http://dbaron.org/log/2006-01#e20060110a

Clearing nomination until that is complete and we can assess where the leak is happening and which product should get the bug.
Flags: blocking-firefox2?
I am seeing the same issues.  They are significantly more noticable when downloading large amounts of images.  However, I can open multiple instances of Firefox and watch the memory climb.  Memory usage stays at or near its peak until ALL instances of Firefox are closed down.

I am running Firefox 1.5.  I wasn't experiencing this feature until I upgraded.
Assignee: bross2 → nobody
I can't reproduce with Firefox 3 beta 1 on Windows XP. I did an advanced Google image search for extra large, full color, JPEGs of cats. I opened nine of the full-size images in nine different tabs, then saved each JPEG to a folder. After doing so, memory use was up by tens of megabytes. After closing the tabs and continuing to navigate to at least eight other pages (to clear the huge images out of the bfcache), memory use was back down to just 2 MB above the starting point. leak-gauge.pl reports 0 leaks.

Can someone provide steps to reproduce an obvious memory leak with saving images in recent Firefox 3 builds?
That would be a realy good news.. if it's true. Are you using download manager (or how it is called) - if there is anything like this in F3?
Yes, the Downloads window came up. I just tried to reproduce the problem again, this time searching for pictures of dogs. Memory started at 80 MB. After opening nine tabs of huge dogs pictures and downloading them, memory use was at 230 MB. After closing the tabs I opened, and then navigating to some other pages, memory use went back down to 82 MB. I did not close the Downloads window or do anything to remove the downloads from the download manager. After entering this bug report and closing the Downloads window, memory use is down to 79 MB, less than when I started. Can anyone still reproduce any memory problem with downloading files in Firefox 3?
Status: NEW → RESOLVED
Closed: 17 years ago
Resolution: --- → WORKSFORME
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: