Closed Bug 424226 Opened 16 years ago Closed 16 years ago

Immediate Fx crash when starting venkman [@ nsPluginHostImpl::TrySetUpPluginInstance]

Categories

(Core Graveyard :: Plug-ins, defect, P2)

x86
Windows XP
defect

Tracking

(Not tracked)

RESOLVED FIXED

People

(Reporter: alta88, Assigned: timeless)

Details

(Keywords: crash, regression)

Crash Data

Attachments

(1 file)

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9b5pre) Gecko/2008031004 Minefield/3.0b5pre
Build Identifier: 


new profile, venkman 0.9.87.3.  crash reporter does not start.

AppName: firefox.exe	 AppVer: 1.9.0.2991	 ModName: npjp2.dll
ModVer: 6.0.100.11	 Offset: 00002398

Reproducible: Always
Uhm, this is not a reproducible bug report, given that it does not contain steps to reproduce.

0) how did you start Firefox? What pages were loaded? Did you do anything in particular before starting Venkman?
1) how did you start Venkman? commandline flag, toolbar button, tools menu, something else?
2) did you try to reproduce this with different pages?
3) Does this problem occur on stable builds (Firefox 2) or on the latest nightly (as the one you have is 10 days old)?
4) Do you have a regression range?
5) Did you try a debug build, or did you try using the symbol server with your nightly build if building yourself is too much trouble?
6) Is this problem windows-only? Did you try Linux or Mac?

PS: the dll file in question is some JPEG2000 or "Algo Vision Luratech" component, according to Google. Have you tried disabling the plugin? It could be that it does not work with trunk.
(In reply to comment #1)
> Uhm, this is not a reproducible bug report, given that it does not contain
> steps to reproduce.
> 
> 0) how did you start Firefox? What pages were loaded? Did you do anything in
> particular before starting Venkman?

firefox.exe -Profilemanager, then selected totally clean profile.  home page is set to blank, so no page loaded.

> 1) how did you start Venkman? commandline flag, toolbar button, tools menu,
> something else?

Tools menu.

> 2) did you try to reproduce this with different pages?

happens on blank page.

> 3) Does this problem occur on stable builds (Firefox 2) or on the latest
> nightly (as the one you have is 10 days old)?

same profile works fine on Fx2.  i'm avoiding the latest trunk as a very recent checkin has broken userchrome.js and am waiting for a fix there, but will try it if necessary.

> 4) Do you have a regression range?

best i can estimate is that venkman was working last week of february on trunk, as i released an Fx extension march 1 and would have been using it before that.

> 5) Did you try a debug build, or did you try using the symbol server with your
> nightly build if building yourself is too much trouble?

i'm not familiar with debug builds, nor do i build my own.

> 6) Is this problem windows-only? Did you try Linux or Mac?
> 

only tried winxp sp1.

> PS: the dll file in question is some JPEG2000 or "Algo Vision Luratech"
> component, according to Google. Have you tried disabling the plugin? It could
> be that it does not work with trunk.
> 

that dll does not sound familiar at all, but i'm sure you know wildly overlaid storage can cause false trails.  i've disabled each and every plugin, restarted with 1 blank homepage, and gotten the same crash, the only difference being that crash reporter did start.  so there are now 3 reports that occurred within 15 minutes of this post.

i did have the latest jre, 1.6.0.10 which installs a Java Quick Starter extension into Fx.  but both the plugin and extension are disabled.
Thanks for your answers.
<snip>

> > PS: the dll file in question is some JPEG2000 or "Algo Vision Luratech"
> > component, according to Google. Have you tried disabling the plugin? It could
> > be that it does not work with trunk.
> > 
> 
> that dll does not sound familiar at all, but i'm sure you know wildly overlaid
> storage can cause false trails.  i've disabled each and every plugin, restarted
> with 1 blank homepage, and gotten the same crash, the only difference being
> that crash reporter did start.  so there are now 3 reports that occurred within
> 15 minutes of this post.
> 
> i did have the latest jre, 1.6.0.10 which installs a Java Quick Starter
> extension into Fx.  but both the plugin and extension are disabled.

Crashreporter does not allow me to search for just 15 minutes, and I would doubt that yours were the only crashes in that timeframe. As it is, there are about 60 crashes (32 unique ones) in the hour before your post. Could you just paste crash IDs (check about:crashes for them) ? Without them I don't think I'll be able to find yours back. Thanks!

16f2973d-f6da-11dc-887a-001a4bd43e5c	2008.Mar.20	06:02 PM
f5b04440-f6d9-11dc-9bd0-001a4bd43ef6	2008.Mar.20	06:01 PM
d102bbba-f6d9-11dc-8e20-001a4bd43e5c	2008.Mar.20	06:00 PM

fyi, the crash reporter not starting is a known bug with the flash plugin.
Awesome, no stack trace for any of them... :-(

I can try to repro this later, busy now. Thanks for the help, I'll comment with my own findings...
Uh, actually, I take that back. I'm not used to the new crash reports yet, obviously. So both of those crash reports blame plugins. This might make it hard to reproduce. I wonder why it's still crashing there if all the plugins are disabled... Timeless, got opinions?
Severity: major → critical
Keywords: crash
Nice, thanks! Requesting blocking, moving to correct component (though still no idea why this is triggered by Venkman starting, pretty sure it's not our fault)
Assignee: timeless → nobody
Status: ASSIGNED → NEW
Component: JavaScript Debugger → Plug-ins
Flags: blocking1.9?
Product: Other Applications → Core
QA Contact: venkman → plugins
Version: unspecified → Trunk
Assignee: nobody → timeless
Status: NEW → ASSIGNED
Summary: Immediate Fx crash when starting venkman → Immediate Fx crash when starting venkman [@ nsPluginHostImpl::TrySetUpPluginInstance]
Not sure of the details here (don't know venkman, etc.).  Can you please explain why you think this should block if it's not our fault per Comment #8 ?
Damon: Sorry for not giving a rationale immediately.

I believe this should block, because it seems to me, if I read timeless' comments and patch correctly, that this is an instance of the plugin API causing crashes if plugins are disabled. Crashes are bad. Also, per comment #2 this is a regression compared to Fx 2 (and the code does seem to have changed a lot quite recently, from a quick look at bonsai). Crashes that are regressions *used* to be blockers, I think. I'm not sure if that's still the case this close to release, but the patch seems trivial and low-risk enough, anyway.
Keywords: regression
Ah yes, if it is indeed our plugin API that's crashing and it's a regression, then yeah, we fix.  

Just to clarify, Timless' patch indeed fixes this issue?

Thanks for the info, btw.

Flags: blocking1.9? → blocking1.9+
Priority: -- → P2
it's likely i'm not clear on the patch or what's causing the crash, but it occurs with all plugins _enabled_, as well as with all plugins _disabled_.  with the flash plugin enabled, crash reporter does not start, per bug 422308.
Attachment #310980 - Flags: superreview+
Attachment #310980 - Flags: review?(jst)
Attachment #310980 - Flags: review+
Attachment #310980 - Flags: approval1.9b5?
Comment on attachment 310980 [details] [diff] [review]
yeah, it's disabled which means we don't find it :)

a=shaver
Attachment #310980 - Flags: approval1.9b5? → approval1.9b5+
Comment on attachment 310980 [details] [diff] [review]
yeah, it's disabled which means we don't find it :)

mozilla/modules/plugin/base/src/nsPluginHostImpl.cpp 	1.606

if there are other crashes, please file more bugs the original crash in comment 0 doesn't actually match the crash i picked from comment 4.
Status: ASSIGNED → RESOLVED
Closed: 16 years ago
Resolution: --- → FIXED
Crash Signature: [@ nsPluginHostImpl::TrySetUpPluginInstance]
Product: Core → Core Graveyard
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Creator:
Created:
Updated:
Size: