Closed Bug 536199 Opened 16 years ago Closed 16 years ago

Crash Yahoo! frontpage with 12/21 1.9.2 nightly build

Categories

(Core Graveyard :: Plug-ins, defect)

1.9.2 Branch
ARM
Maemo
defect
Not set
blocker

Tracking

(status1.9.2 final-fixed, fennec1.0+)

RESOLVED FIXED
Tracking Status
status1.9.2 --- final-fixed
fennec 1.0+ ---

People

(Reporter: aakashd, Assigned: mfinkle)

Details

(Keywords: mobile)

Attachments

(1 file)

Build Id: Mozilla/5.0 (X11; U; Linux armv7l; Nokia N900; en-US; rv:1.9.2b5pre) Gecko/20091221 Firefox/3.6b5pre Fennec/1.0b6pre Steps to Reproduce: 1. Enable Plugins 2. Go to www.yahoo.com 3. Wait for page load and for flash title and ad to start Actual Results: Fennec will crash Expected Results: Fennec shouldn't crash
Assignee: mozbugz → nobody
tracking-fennec: ? → 1.0+
this won't crash if you are on m.www.yahoo.com (auto redirect), but does on www.yahoo.com
This is a crash in libflashplayer.so
I can't get it to not redirect me to the mobile page.
Attached patch patchSplinter Review
The crash is happening in NativeImageDraw when we set the window: http://mxr.mozilla.org/mozilla-central/source/layout/generic/nsObjectFrame.cpp#5102 But only when NativeImageDraw is called from InvalidateRect: http://mxr.mozilla.org/mozilla-central/source/layout/generic/nsObjectFrame.cpp#2733 Doug mentioned that we only really need to call SetWindow if the size of the plugin changes. Therefore, this patch does just that. The size only changes when SetAbsoluteScreenPosition is called, which then calls NativeImageDraw. The patch skips the SetWindow call if "invalidRect" is null and "sizeChanged" Theremight be an upside to skipping SetWindow whenever possible. Page loading and panning both seem faster with the patch. A test build with this patch can be found here: http://people.mozilla.com/~mfinkle/fennec/fennec-1.0b6pre.en-US.linux-gnueabi-arm-flashfix.tar
Attachment #418795 - Flags: review?(mozbugz)
Comment on attachment 418795 [details] [diff] [review] patch can we move the | if (!invalidRect && sizeChanged) { | block into SetupXShm? We probably only need to do this when size changes.
(In reply to comment #5) > (From update of attachment 418795 [details] [diff] [review]) > can we move the | if (!invalidRect && sizeChanged) { > | block into SetupXShm? We probably only need to do this when size changes. I don't follow. The SetupXShm() call has a sizeChanged check already, right?
Comment on attachment 418795 [details] [diff] [review] patch discussed on IRC. This is probably a fine optimization, but we do not know if this actually fixes the crash.
Attachment #418795 - Flags: review?(mozbugz) → review+
Assignee: nobody → mark.finkle
Status: NEW → RESOLVED
Closed: 16 years ago
Resolution: --- → FIXED
Target Milestone: --- → RC
Component: General → Plug-ins
Keywords: mobile
Product: Fennec → Core
Target Milestone: RC → ---
Flags: in-testsuite?
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: