Closed Bug 425183 Opened 16 years ago Closed 16 years ago

Spinning wait cursor happens very frequently

Categories

(Firefox :: Shell Integration, defect)

PowerPC
macOS
defect
Not set
critical

Tracking

()

RESOLVED DUPLICATE of bug 235853

People

(Reporter: kelly.a.campbell, Unassigned)

References

()

Details

(Keywords: hang, regression)

Attachments

(1 file)

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.4; en-US; rv:1.9b5pre) Gecko/2008032504 Minefield/3.0b5pre
Build Identifier: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.4; en-US; rv:1.9b5pre) Gecko/2008032504 Minefield/3.0b5pre

I've noticed a major increase in the number of spinning wait cursors on MacOS in the last couple of daily builds. It generally happens when waiting on network operations, so it's very noticeable on slower connections, or on ajax apps like gmail with gtalk.

I've begun running Spin Control from the mac developer tools to try to gather more info about this. In one instance today the cnn url above took nearly a minute to load, with 10 separate spin control "detected hangs". The first hang was 14.6 seconds and I'm attaching a text report of that profile to this bug.

Reproducible: Sometimes

Steps to Reproduce:
1. Use minefield on ajax websites.
2.
3.
Actual Results:  
Spinning cursor (http://en.wikipedia.org/wiki/Spinning_wait_cursor) is frequent.

Expected Results:  
Spinning wait cursor should almost never appear.
Are you using a proxy with PAC?  If so, this is a dup of bug 235853.

Otherwise, I'm afraid you'll have to compile Firefox yourself (with function-name symbols) and re-profile.  XRE_GetFileFromPath does not call itself, so your stack traces probably suffer from a lack of symbols for other functions.
It happens with or without a proxy. Whatever is causing this was a change in the past couple of days. That proxy bug looks like it's been around a long time. I can try to bisect some builds to narrow this issue down.

I think the text file makes it hard to see the useful stacks. Within the Spin Control UI I can see that most threads are stuck waiting on semaphores but I can't tell which lock they're waiting on. NS_GetComponentRegistrar_P seems to be a common predecessor to many of the semaphores, as well as XRE_GetFileFromPath and Flash_EnforceLocalSecurity.
Keywords: hang, regression
Also, just to be clear, Firefox on MacOS has always exhibited more than it's fair share of spinning cursors. But I think this is a regression because the nightly builds (which I've been using for 2-3 months now) exhibit this behavior muc more frequently to the point that it makes Firefox near impossible to use.
On further investigation, I think this is actually a dup of the proxy bug. I've experienced much fewer wait cursors since turning proxies off completely and disabling a proxy-related extension.
Status: UNCONFIRMED → RESOLVED
Closed: 16 years ago
Resolution: --- → DUPLICATE
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: