458108 - Prism crashes Firefox Trunk when downloading a file [NS_ERROR_XPC_JSOBJECT_HAS_NO_FUNCTION_NAMED] [@ nsJAR::Open(nsIFile*) ]

Status: RESOLVED FIXED

(Mozilla Labs :: Prism, defect)

Description

[IU]

User-Agent:       Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b1pre) Gecko/20081001 Firefox/2.0.0.11
Build Identifier: 0.2.1

If Prism is used with the 20081001 Firefox Trunk, the browser is crashed whenever saving a file.  The following exception is generated just before the crash:

Error: 'JavaScript component does not have a method named: "onDownloadStateChange"' when calling method: [nsIDownloadProgressListener::onDownloadStateChange] = NS_ERROR_XPC_JSOBJECT_HAS_NO_FUNCTION_NAMED

Breakpad IDS:
http://crash-stats.mozilla.com/report/index/720efa61-8fd6-11dd-86e0-001cc45a2c28
http://crash-stats.mozilla.com/report/index/0f21df78-8fd6-11dd-a188-001cc45a2c28

Browser Build ID: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9.1b1pre) Gecko/20081001

Reproducible: Always

Steps to Reproduce:
1. Install the lastest Firefox 3.1 Trunk (tested on Win XP SP3)
2. Start Firefox and go to Tools --> Options... --> Main
3. Select "Always ask me where to save files."  Click OK
4. Open Error Console and make sure logging of "Errors" is enabled
5. Install Prism for Firefox 0.2.1
6. Visit http://fileforum.betanews.com/detail/Sysinternals_Suite/1194884530/1
7. Click the "Download Now" link and save the file someplace other than Desktop
8. Note the exception in Error Console
9. Firefox Crashes

Comment 1

[Martijn Wargers (dead)]

Well, this might be a Prism bug, because it somehow allows to pass a null argument to pass to the zip-reader open method, but that function still should not crash when a null argument is passed, so I filed bug 458158 for that.

Comment 2

[-fullmetaljacket-]

http://crash-stats.mozilla.com/report/index/176b39e8-9017-11dd-80cf-001a4bd43ef6

Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.1b2pre) Gecko/20081008 Minefield/3.1b2pre ID:20081008032217

Comment 3

[Matthew Gertner]

Could have sworn we knew about this. I remember discussing the nsJAR:Open() issue with someone a while back. Anyway, I agree: the problem is both Prism (shouldn't pass in NULL) and Firefox (should check the parameter, not just crash).

Comment 4

[David Baron :dbaron: (⌚️UTC-4, no longer working on Mozilla)]

Some reports with better stacks (i.e., non-Windows)
http://crash-stats.mozilla.com/report/index/5f3955a9-f02b-4f27-9061-4ea820081118
http://crash-stats.mozilla.com/report/index/14f9f753-6108-4352-8d4d-64f420081118

Comment 5

[(not currently active) Ted Mielczarek]

I hit this with 100% reproducibility on a trunk Windows build with refractor installed.

Comment 6

[Mark Finkle (:mfinkle) (use needinfo?)]

Attachment: patch

This patch adds some checks to refractor-overlay.js to keep a null file from being passed to WebAppInstall.install(file)

The patch also adds onDownloadStateChange, which keeps the error from appearing in the console. This part seemed unrelated to the crash.

Comment 7

[André Mondri]

Confirm the crashes even if you try to save a Images as ..... with 3.1 Beta 2

Comment 8

[Matthew Gertner]

Comment on attachment 353948 [details] [diff] [review]
patch

I can't for the life of me remember who I discussed this with, but you might want to check with someone on the Firefox side as well since I don't think it should crash even if you pass in a NULL pointer.

Comment 9

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to comment #8)
> (From update of attachment 353948 [details] [diff] [review])
> I can't for the life of me remember who I discussed this with, but you might
> want to check with someone on the Firefox side as well since I don't think it
> should crash even if you pass in a NULL pointer.

bug 458158

Comment 10

[Mark Finkle (:mfinkle) (use needinfo?)]

Modified: mozilla-code\mozilla-prism\mozilla\prism\extension\chrome\content\refractor-overlay.js  
Sending content: mozilla-code\mozilla-prism\mozilla\prism\extension\chrome\content\refractor-overlay.js  
Completed: At revision: 21373

Comment 11

[IU]

So does this mean the next Prism build will not longer crash Firefox, in spite of the bug 458158 being moribund?

Comment 12

[Mark Finkle (:mfinkle) (use needinfo?)]

(In reply to comment #11)
> So does this mean the next Prism build will not longer crash Firefox, in spite
> of the bug 458158 being moribund?

Correct