Closed Bug 321661 Opened 20 years ago Closed 20 years ago

Cache memory not releasing images

Categories

(Core :: Networking: Cache, defect)

1.8 Branch
x86
Windows 2000
defect
Not set
normal

Tracking

()

RESOLVED INVALID

People

(Reporter: mstrumyla, Assigned: bryner)

Details

(Keywords: regression)

Steps to reproduce: 1. in about:config, set browser.cache.memory.capacity to 4096 2. Options/Privacy/Cache/Clear Cache Now 3. open about:cache?device=memory 4. open 2nd tab and type www.identityblog.com 5. refresh 1st tab. memory cache shows 7306 kb of data stored. 6. go to 2nd tab, type in images.google.com, do a search on "cats" 7. open 9 different Result Pages in new tabs. so you have 11 tabs total. 8. refresh 1st tab. memory cache shows 15052 kb Actual Results: _none_ of the images from identityblog.com were removed from memory cache. Expected Results: should be no images from identityblog.com hanging in memory cache. this is a regression from firefox1.0. firefox1.0 after step 8 shows 7549 kb and there's no images from identityblog.com.
Reporter, are you running Adblock or another extension ?
ok, this appears to be a side effect of the blazingly fast back feature. if you set browser.sessionhistory.max_total_viewers to 0, the images are released from cache. this could be a dup of Bug 306862.
(In reply to comment #1) > Reporter, are you running Adblock or another extension ? I don't. I tested this with a new profile.
Assignee: nobody → darin
Component: Tabbed Browser → Networking: Cache
Product: Firefox → Core
QA Contact: tabbed.browser → networking.cache
Version: 1.5 Branch → 1.8 Branch
So is the memory released when the document viewer is evicted (that is, after you load 5 more pages)? Note that bug 306862 is about dropping the cached viewers when we actually run out of memory, which I bet you're not.
Over to bryner based on comment #2
Assignee: darin → bryner
The identityblog.com images are removed when the page is evicted from bfcache or the tab is closed, so they aren't leaked. This is definitely the expected behavor of fastback -- restoring the previous page requires these decoded images to be kept around in memory. I'm planning to do some work in the next month or two to tweak the memory consumption characteristics to better account for memory-heavy pages, but in the meantime I think this bug can be closed out.
Status: NEW → RESOLVED
Closed: 20 years ago
Resolution: --- → INVALID
(In reply to comment #4) > So is the memory released when the document viewer is evicted (that is, after > you load 5 more pages)? I'm on vacation right now. I will be able to test this after Jan 7.
You need to log in before you can comment on or make changes to this bug.