44961 - memory usage, freeing/leaks. general.

Status: VERIFIED INVALID

(Core :: Networking: Cache, defect, P3)

Description

[janne hayrynen]

there are several bugs about mozilla's memory consumption, this is an attempt to
explain the biggest issue here and narrow this down some.

pre: memory cache is set to 0 kbytes.
start up mozilla, it takes the usual ~20mbytes.

query bugzilla for ALL bugs, after loading the single huge document the process
size has grown to about 290mbytes.
keep on browsing to whatever other pages, the process size doesn't go down,
pressing the "clear memory cache" wont affect it (well, it was set to 0 anyway).
the process size keeps growing until mozilla finally crashes to whatever reason.
opening a new browser window and closing the old one doesn't affect it either.

i'm using such a large page as an example here to demonstrate the extremity of
this issue. i'm sorry i don't have the knowhow to investigate this further, but
i think the fact that mozilla seems to leave the ~270mbytes of memory reserved
after visiting and leaving _one_ large page says there's be something REALLY
wrong here... can this be due to *minor* leaks??

Comment 1

[Asa Dotzler [:asa]]

(janne hayrynen) could you please include the build ID with any reports you
submit.  Thanks

Comment 2

[janne hayrynen]

just tested this to still happen with build 2000070909.

Comment 3

[Joseph Elwell]

my comm build 2000071211 eats up 300+ megs, and I freezes up too much to try
this bug fully with the URL listed. The file in question, when I viewed it, was
17Megs. As a note, I tried viewing the file locally, and it still eats up hoards
of memory - so it doesn't appear to be network related.

If I search on bugzilla for all Unconfirmed, new, assigned, or reopened bugs;
then Mozilla goes from 30megs to 82 megs - and never down, even after I click
clear memory and clear disk cache buttons in the cache preferences.

Comment 4

[Asa Dotzler [:asa]]

reassigning

Comment 5

[David Baron :dbaron: (⌚️UTC-5, no longer working on Mozilla)]

This doesn't necessarily have anything to do with the cache...

Comment 6

[neeti]

Yes, this is not necessarily a cache bug. There is a bug filed about the "clear 
memory cache", bug 30670. This is a very general bug about memory leaks in 
mozilla. 

Reassigning to warren for now to get it on his radar.

Neeti

Comment 7

[Warren Harris]

Reassigning to kandrot.

Comment 8

[Phil Peterson]

Not holding PR3 for this. Marking nsbeta3-. Please nominate for rtm if there's
something here we need to fix before shipping seamonkey, with explanation about
which specific leak we need to fix, how big it is, etc.

Comment 9

[James Green]

*** Bug 65742 has been marked as a duplicate of this bug. ***

Comment 10

[gordon]

This seems unrelated to the necko cache.

Comment 11

[Edward Kandrot]

Linux does not report memory usage correctly.  The size you see is the size of
the process mulitplied by the number of threads the process is using.  Clearing
our caches will have little to no effect on the memory reported, since the OS
counts memory that it has cached as belonging to us.  The memory is not really
in use by us, and will be used by another process if needed, but is cached by
the OS to make apps faster.  I believe the crash was due to other problems,
since it was back when the release was less stable.

Comment 12

[benc]

-> cache
qa to me.
VERIFIED: Invalid.

If there is set of steps that creates a test case for Mozilla 0.9 or later, 
please put it here and re-open.