Closed Bug 400467 Opened 17 years ago Closed 17 years ago

Java broken on Vista after Firefox 2.0.0.8 upgrade (says Java Not Found, Or Not Working)

Categories

(Firefox Build System :: General, defect)

1.8 Branch
x86
Windows Vista
defect
Not set
major

Tracking

(Not tracked)

VERIFIED FIXED
mozilla1.9beta1

People

(Reporter: msmarymac3, Assigned: robert.strong.bugs)

References

()

Details

(Keywords: regression, verified1.8.1.10, verified1.8.1.9)

Attachments

(3 files)

User-Agent: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.8.1.8) Gecko/20071008 Firefox/2.0.0.8 Build Identifier: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.8.1.8) Gecko/20071008 Firefox/2.0.0.8 I just updated my version this morning & ever since when I go to load a game that's the message that comes up I can access this site with IE version7 so I believe it is a problem with this critical update Reproducible: Always Steps to Reproduce: 1.Go to Pogo.com 2.access a game might have to create an account 3.click to load the game Actual Results: took forever to open & when it did there was the java not working, or found message. Expected Results: opened & allowed me to play the game like it did right before I installed those critical updates
Component: Download Manager → General
QA Contact: download.manager → general
Mary: Can you please use this site and report which version of Java it shows: http://www.javatester.org/version.html?
Version: unspecified → 2.0 Branch
(In reply to comment #1) > Mary: Can you please use this site and report which version of Java it shows: > http://www.javatester.org/version.html? I am running Java Version 1.60_03
I suffered this bug too (Windows Vista, Java 1.6.0 update 3 and Firefox 2.0.0.8) but I think I've found a workaround: * Right click Firefox icon and choose "Run as administrator" * Accept UAC prompt * Enter any site which use a Java applet, Java should load normally * Close Firefox * Run Firefox again, but in normal mode (without administrator privileges) Java should work correctly after that.
This works fine with Windows XP and it appears to be Vista only.
(In reply to comment #4) > I suffered this bug too (Windows Vista, Java 1.6.0 update 3 and Firefox > 2.0.0.8) but I think I've found a workaround: > * Right click Firefox icon and choose "Run as administrator" > * Accept UAC prompt > * Enter any site which use a Java applet, Java should load normally > * Close Firefox > * Run Firefox again, but in normal mode (without administrator privileges) > Java should work correctly after that. This worked & fixed it thank You !!!
Flags: blocking1.8.1.9?
Flags: blocking1.8.1.10?
rob_strong: any idea what we changed that would cause Java to need a UAC prompt? Or why that prompt is suppressed for non-admin accounts? Bug 400422 describes the specific problem of being unable to play Java-based games at pogo.com
Status: UNCONFIRMED → NEW
Ever confirmed: true
Summary: Says Java Not Found, Or Not Working → Java broken on Vista after Firefox 2.0.0.8 upgrade (says Java Not Found, Or Not Working)
(In reply to comment #5) > This works fine with Windows XP and it appears to be Vista only. > I can confirm this Bug for Vista and JRE 1.6 update 3. I need to use explicit the Option "Run as Administrator" (comment #4) to run java games at pogo. Even when i run Vista with a Admin User Account. Its a regression. Firefox 2.0.0.7 is fine (don`t need to run Firefox in "Run as Administrator" mode" with JRE 1.6.3) and 2.0.0.8.
Keywords: regression
This is almost certainly due to bug 378598: we explicitly embed a manifest to runas=invoker. This means that vista compatibility mode is no longer active, and attempts to write to the program directory or UAC-protected registry keys, which would previously have been redirected to the virtual filestore/registry, now fail. It would be good to run under filemon/regmon to see what operations are failing and notify Sun... plugins really shouldn't be failing because we stopped enabling vista compatibility mode.
Component: General → Plug-ins
Product: Firefox → Core
QA Contact: general → plugins
Version: 2.0 Branch → 1.8 Branch
Component: Plug-ins → Build Config
QA Contact: plugins → build-config
(In reply to comment #11) > what operations are failing Following has been reported to Bugzilla Japan. It was Name=CurrentVersion in next key. (version number of Gecko is held) - HKEY_LOCAL_MACHINE\SOFTWARE\Wow6432Node\mozilla.org\Mozilla (when x64) - HKEY_LOCAL_MACHINE\SOFTWARE\mozilla.org\Mozilla (when x86) (On my Win-XP SP2/32bit, CurrentVersion=1.8.1.2 was found in this key) Reporter says checked by Process Monitor. ( http://bugzilla.mozilla.gr.jp/show_bug.cgi?id=5900 , In Japanese) Note: Use of above entry is possibly a result of solution for crash of Bug 83376.
It appears that Java is trying to open that key with write access even though it doesn't appear to try to write anything. By giving my account full control of HKEY_LOCAL_MACHINE\SOFTWARE\mozilla.org\Mozilla a random Java game that wasn't working - in fact it crashed the browser - works fine.
(In reply to comment #15) > even though it doesn't appear to try to write anything. To Robert Strong: Is it true even when CurrentVersion doesn't exist in the registry key of HKEY_LOCAL_MACHINE\SOFTWARE\mozilla.org\Mozilla? (I think that workaround of comment #4 indicates; ) ( Once the entry is created/written, no further write permission is required.)
I already had HKEY_LOCAL_MACHINE\SOFTWARE\mozilla.org\Mozilla CurrentVersion = 1.9a8pre in my registry and it still exhibited this problem for me and removing the permissions I added caused the bug to reappear. That's what led me to believe it was failing due to specifying write access when opening the key. I can change the permissions on HKEY_LOCAL_MACHINE\SOFTWARE\mozilla.org\Mozilla via the installer and via helper.exe during Software Update to workaround this issue with Java. The NSIS Access Control plugin would be necessary and I have a working patch. http://nsis.sourceforge.net/AccessControl_plug-in In the patch I gave Authenticated Users Full Access which equates to Full Control and Read in the registry permissions... I believe this should be sufficient.
(In reply to comment #17) > CurrentVersion = 1.9a8pre It may be a cause of problem that workaround of comment #4 won't work in your environment. When browser sniffing, beta version is not usually recognized. I guess that similar version detection logic is used by Sun Java. I think Sun Java is not an exception of Murphy's Raw. What will happen when CurrentVersion=1.8.1.2 or CurrentVersion=1.9.1.1 is set?
WADA, running as administrator does work since then you have write access... I am proposing a fix we can deploy to users quickly without having to run as administrator. So, by granting Full Access permissions to authenticated user for that reg key the problem no longer exists. btw: I did try out release versions for CurrentVersion and it doesn't matter what the value is.
Can we make sure we get a bug report filed with Sun as well?
The logic that Java is using for updating that key value is strange. After changing it to an official release value it crashed the first time and worked the second time. So, it is entirely possible that an official release value that matches the actual release works. Also, adding the actual value to HKCU and removing the mozilla.org key from HKLM didn't help. http://crash-stats.mozilla.com/report/index/6c604e57-806a-11dc-b3fb-001a4bd43ed6?date=2007-10-22-06 I reported a bug to http://bugs.sun.com/ and it has a Review ID of 1094708 but it won't appear until they triage it
I haven't had time to verify this but it is possible we are causing this at http://lxr.mozilla.org/seamonkey/source/modules/plugin/base/src/nsPluginDirServiceProvider.cpp#258
Or line 374, but it looks like that code fails gracefully if it is unable to open/set the registry key.
Flags: blocking1.8.1.9? → blocking1.8.1.9+
It turns out that an installer modified my permissions without my knowing it which was causing the strange behavior for me and that adding the value for CurrentVersion during install will fix this. You can download the attached regedit file and import it into your registry to workaround this bug and you will be prompted by UAC to import it. Patch coming up
Assignee: nobody → robert.bugzilla
Status: NEW → ASSIGNED
Attached patch installer patchSplinter Review
All this does is add the GRE version to the value of CurrentVersion under HKEY_LOCAL_MACHINE\SOFTWARE\mozilla.org\Mozilla when the value is not the same as the version being installed. I verified that adding this value under HKCU doesn't help Java so that isn't going to help. If having this value is truly necessary for Java we should probably get them to also check under HKCU and have our code add it there when the value is incorrect under HKLM. As is, we only update it when it doesn't exist. Many thanks to Seth Spitzer for sending me his registry permissions so I could fix my own that were mucked up by some f#4$n company's installer.
Attachment #285807 - Flags: review?(sspitzer)
robert, looks good. just a few question / comments to make sure I understand: 1) the browser/installer/windows/nsis/installer.nsi is for when the user runs the installer and the browser/installer/windows/nsis/shared.nsh change is for PostUpdate, right? 2) will this change have any impact on side-by-side installs? (does this make it so the last version of firefox to update and do PostUpdate or to be installed "win", and be able to use java?) 3) do we need to delete this key on uninstall?
(In reply to comment #26) > robert, looks good. just a few question / comments to make sure I understand: > > 1) the browser/installer/windows/nsis/installer.nsi is for when the user runs > the installer and the browser/installer/windows/nsis/shared.nsh change is for > PostUpdate, right? Correct > 2) will this change have any impact on side-by-side installs? (does this make > it so the last version of firefox to update and do PostUpdate or to be > installed "win", and be able to use java?) I don't have a clue as to what the value is used for by Java since just adding 1.1 made it so Java worked. > 3) do we need to delete this key on uninstall? The uninstall process is going to have to jump through some hoops to do this properly since Java uses it and Java could be used by other Mozilla apps including XULRunner apps (e.g. Songbird, etc.) so there is no way I know of to easily tell that it is ok to remove this key on uninstall. Until there is a reliable way to tell I am not going to remove this key on uninstall.
Requesting blocking1.9 since this also affects the trunk and we have a beta release soon and a patch.
Flags: blocking1.9?
Comment on attachment 285807 [details] [diff] [review] installer patch r=sspitzer robert, thanks for answering my questions.
Attachment #285807 - Flags: review?(sspitzer) → review+
Blocking, aye. Not sure if it's an M9 blocker, but request approval when you're ready?
Flags: blocking1.9? → blocking1.9+
Comment on attachment 285807 [details] [diff] [review] installer patch It be ready matey! Arghhh
Attachment #285807 - Flags: approval1.9?
Flags: blocking1.8.1.10? → blocking1.8.1.10+
Comment on attachment 285807 [details] [diff] [review] installer patch approved for 1.8.1.9 and 1.8.1.10, a=dveditz for release-drivers For 1.8.1.9 please land on the GECKO181_20071004_RELBRANCH. 1.8.1.10 is the normal 1.8 branch
Attachment #285807 - Flags: approval1.8.1.9+
Attachment #285807 - Flags: approval1.8.1.10?
Attachment #285807 - Flags: approval1.8.1.10? → approval1.8.1.10+
Checked in to trunk Checking in mozilla/browser/installer/windows/nsis/installer.nsi; /cvsroot/mozilla/browser/installer/windows/nsis/installer.nsi,v <-- installer.nsi new revision: 1.38; previous revision: 1.37 done Checking in mozilla/browser/installer/windows/nsis/shared.nsh; /cvsroot/mozilla/browser/installer/windows/nsis/shared.nsh,v <-- shared.nsh new revision: 1.17; previous revision: 1.16 done Checked in to MOZILLA_1_8_BRANCH Checking in mozilla/browser/installer/windows/nsis/installer.nsi; /cvsroot/mozilla/browser/installer/windows/nsis/installer.nsi,v <-- installer.nsi new revision: 1.3.2.28; previous revision: 1.3.2.27 done Checking in mozilla/browser/installer/windows/nsis/shared.nsh; /cvsroot/mozilla/browser/installer/windows/nsis/shared.nsh,v <-- shared.nsh new revision: 1.3.2.11; previous revision: 1.3.2.10 done Checked in to GECKO181_20071004_RELBRANCH Checking in mozilla/browser/installer/windows/nsis/installer.nsi; /cvsroot/mozilla/browser/installer/windows/nsis/installer.nsi,v <-- installer.nsi new revision: 1.3.2.27.2.1; previous revision: 1.3.2.27 done Checking in mozilla/browser/installer/windows/nsis/shared.nsh; /cvsroot/mozilla/browser/installer/windows/nsis/shared.nsh,v <-- shared.nsh new revision: 1.3.2.10.2.1; previous revision: 1.3.2.10 done
Status: ASSIGNED → RESOLVED
Closed: 17 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla1.9 M9
To reproduce remove the following registry key and try to launch a Java game HKLM\Software\mozilla.org sample Java game http://www.flyordie.com/games/online/games.html?lang=en&game=8Ball&room=801&rs=1 and click "Play as Guest" To verify the fix remove the key as above Run the installer and try to play a Java game and After a Software Update try to play a Java game (remember to remove the key first)
Attachment #285807 - Flags: approval1.9?
Filed bug 400878 for the underlying issue
verified fixed 1.8.1.9 using Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.8.1.9pre) Gecko/2007102404 BonEcho/2.0.0.9pre and the steps for testing from rob. My tests pass and so i can play the game. Also Java is working fine. Adding verified keyword.
Thank You al for your help on V 2.0.0.8 @Fly or Die Reversi. I right clicked and ran as administrator. Problem solved and working fine.
verified fixed on trunk with Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9a9pre) Gecko/2007102605 Minefield/3.0a9pre and the steps to reproduce from comment #36 - Verified Fixed
Status: RESOLVED → VERIFIED
Flags: blocking1.8.1.11+ → blocking1.8.1.10+
verified fixed 1.8.1.10 using Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.8.1.10pre) Gecko/2007111503 BonEcho/2.0.0.10pre and pogo games - working fine - Adding verified keyword
Product: Core → Firefox Build System
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: