Bug 32357 – SVG test case crashes WebKit (invalid font URL)

• Status: RESOLVED
• Resolution: FIXED
• Component: SVG

[sideshowbarker]

Running  r51881. Tried to open http://dev.w3.org/SVG/profiles/1.2T/test/svgHarness/animate-elem-227-t.svg but crashes.

[Alexey Proskuryakov]

SVGFontFaceUriElement::loadFont() tries to call setSVGFont on a null m_cachedFont. And m_cachedFont is null because we're trying to resolve "../images/SVGFreeSans.svg#ascii" in an about:blank document, and of course fail.

It's definitely a bug loadFont() that it crashes when font URL is invalid. It may be a bug elsewhere that the document base URL is about:blank.

[Justin Schuh]

Created attachment 44810 [details]
Proposed patch

Checks for NULL m_cachedFont before calling setSVGFont().

[Justin Schuh]

Created attachment 44811 [details]
Minimized test case

[Justin Schuh]

I ran into the same thing last week <http://crbug.com/29890>. Here's a short patch and a minimized test case.

[Alexey Proskuryakov]

Would you be willing to submit a patch for review, as described in <http://webkit.org/coding/contributing.html>?

[Justin Schuh]

(In reply to comment #5)
> Would you be willing to submit a patch for review, as described in
> <http://webkit.org/coding/contributing.html>?

Yep. I had to set up a proper WebKit build environment, but I'll be submitting a patch today.

[Justin Schuh]

Created attachment 45072 [details]
Patch with layout tests and changelog

This patch just checks for a NULL m_cachedFont before continuing. It follows the submission guidelines and should be ready for review.

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 45072 [details] without any errors.

Comment on attachment 45072 [details]
Patch with layout tests and changelog

> +    This test is to ensure that we do not crash when loading a SVG image without an invalid font-face-uri

Did you mean “*with* an invalid”? I think this kind of test can be done entirely in SVG.

[Justin Schuh]

(In reply to comment #9)
> Did you mean “*with* an invalid”? 

Yes I did. I'll fix that before resubmitting.

> I think this kind of test can be done entirely in SVG.

I don't know, but probably. This is my first crack at submitting a patch. So, I copied format and style from the text-font-invalid.html test, which looked similar to this case.

[Justin Schuh]

Created attachment 45104 [details]
Patch with layout tests

I fixed the typo and condensed SVG and HTML into a single file. There's still an expected output file, because that appears to be how other crash tests were done.

[WebKit Review Bot]

style-queue ran check-webkit-style on attachment 45104 [details] without any errors.

[WebKit Commit Bot]

Comment on attachment 45104 [details]
Patch with layout tests

Clearing flags on attachment: 45104

Committed r52300: <http://trac.webkit.org/changeset/52300>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.

*** Bug 32712 has been marked as a duplicate of this bug. ***