Re: [filter-effects][css-masking] Move security model for resources to CSP

On Wed, May 29, 2013 at 11:48 AM, Dirk Schulze <dschulze@adobe.com> wrote:
> On Apr 10, 2013, at 2:18 AM, Robert O'Callahan <robert@ocallahan.org> wrote:
>> On Wed, Apr 10, 2013 at 8:51 PM, Anne van Kesteren <annevk@annevk.nl> wrote:
>> If we accept the need for a sandbox domain, same-origin loads becomes
>> an option I think. And actually, even in the face of an open redirect
>> you could fail flat the moment the target URL becomes cross-origin and
>> not fetch it. Several APIs on the platform have a request mode of
>> same-origin  (different from tainted cross-origin, which will fetch)
>> with an opt in availability for CORS.
>>
>> So we need to turn all kinds of external loads into CORS same-origin loads?
>>
>> That sounds like it would work, but be quite invasive to spec and implement.
>
> To recapitulate:
>
> This threat currently focuses on SVGs as image resources and if there are ways to let an SVG image load further resources. An initial test for <img> and CSS Images actually shows that Firefox and Chrome block any external resources of an SVG image right away - independent if the resource has the same origin or not. The bug reports on Chrome [1] and Firefox [2] and this thread actually confirm that.
>
> Maybe CSS and SVG should specify exactly that: No load of any external resources of an SVG file loaded as image. Exclusions of the restrictions can be specified later after more investigations.
>
> Is that something we can agree initially?

I'm happy with whatever roc and Anne are happy with.  ^_^

~TJ

Received on Wednesday, 29 May 2013 19:28:20 UTC