938587 - Make ArchiveReaderZipEvent fail in a more transparent way

Status: NEW

(Core :: DOM: Core & HTML, defect, P5)

Description

[Andrew McCreight [:mccr8]]

In bug 920978, nsJar::Open is being run by the test harness and fails, causing the test to fail:

TEST-UNEXPECTED-FAIL | chrome://mochitests/content/browser/toolkit/mozapps/extensions/test/browser/browser_dragdrop.js | uncaught exception - NS_ERROR_FAILURE: Component returned failure code: 0x80004005 (NS_ERROR_FAILURE) [nsIZipReader.open] at chrome://mochikit/content/chrome-harness.js:271

The theory is that this is somehow related to being out of memory, but "NS_ERROR_FAILURE" certainly doesn't help you figure that out one way or the other.

Chasing code along, it looks like the actual failure must happen in ArchiveReaderZipEvent::Exec.  But the only thing in there that returns NS_ERROR_FAILURE seems to be the "no central offset" part of the code.

Is my analysis right?  Does that seem like an OOM?  Is there a more useful error message we could give here so the sheriffs don't have to read the tea leaves quite so hard?

Comment 1

[Andrew McCreight [:mccr8]]

Andrea, it looks like you are most familiar with this code.

Comment 2

[Andrea Marchesini [:baku]]

(In reply to Andrew McCreight [:mccr8] from comment #1)
> Andrea, it looks like you are most familiar with this code.

Hi mccr8. Actually 'no central offset' can be generated by a malformed ZIP file. Are you suggesting to add a warning message or to change the error code? I don't think a generic "NS_ERROR_FAILURE" is fine here.

Comment 3

[Andrew McCreight [:mccr8]]

I was thinking changing the error code.  My goal is to have OOMs produce some kind of failure on TBPL that is readily identifiable as being an OOM, if that's possible.

Comment 4

[Firefox Bug Husbandry Bot]

https://bugzilla.mozilla.org/show_bug.cgi?id=1472046

Move all DOM bugs that haven’t been updated in more than 3 years and has no one currently assigned to P5.

If you have questions, please contact :mdaly.