566121 - HTML5 parser breaks pages that use FRAMESET if they start with NUL (ASCII 0)

Status: RESOLVED DUPLICATE of bug 563526

(Core :: DOM: HTML Parser, defect)

Description

[Screwtape]

User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.3a5pre) Gecko/20100514 Minefield/3.7a5pre
Build Identifier: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.3a5pre) Gecko/20100514 Minefield/3.7a5pre

I realise this may seem like a very obscure and unimportant corner case, but my ADSL modem's web-based-admin triggers it - where once I saw a vast array of buttons for configuring my PPPoE connection, DHCP server, etc.; with current Firefox nightlies I see only � unless I turn off the HTML5 parser.

Reproducible: Always

Steps to Reproduce:
1. Obtain a Firefox nightly with the "html5.enable" pref turned on.
2. Load the file I'm about to attach.
Actual Results:  
Firefox displays only U+FFFD REPLACEMENT CHARACTER

Expected Results:  
Firefox renders the frameset, covering the entire page with green.

* When Firefox displays only U+FFFD, View Source displays all the HTML 
  (correctly syntax-highlighted, etc.)
* When the leading NUL is removed from the attached file, Firefox displays the
  expected result.
* When "html5.enable" is set to False, Firefox displays the expected result.
* Google's Chromium displays the expected result.
* It seems that the frameset is important. Most HTML after the NUL will render
  correctly.
* However, plain text preceded by a NUL also fails to display:

    data:text/html,%00hello%20world

  ...displays only U+FFFD.

Comment 1

[Screwtape]

Attachment: NUL-prefixed page mentioned in the reproduction instructions.

Comment 2

[Ria Klaassen (not reading all bugmail)]

Confirmed on Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.3a5pre) Gecko/20100515 Minefield/3.7a5pre

Comment 3

[Masatoshi Kimura [:emk]]

Dup of bug 563526?

Comment 4

[Screwtape]

Yes, I believe it is a dupe of 563526. Sorry — I did search, but it wasn't until later that I figured out it was FRAMESET in particular.

Comment 5

[Henri Sivonen (:hsivonen)]

Resolving as duplicate. Was this bug also about a D-Link product?

Filed bug 566280 about the plain text case.

Comment 6

[Screwtape]

Yes, this bug was about the exact same model.