- From: Brian Smith <bsmith@mozilla.com>
- Date: Mon, 27 Jun 2011 11:14:39 -0700 (PDT)
- To: Devdatta Akhawe <dev.akhawe@gmail.com>
- Cc: public-web-security@w3.org, Gervase Markham <gerv@mozilla.org>
Devdatta Akhawe wrote: > Gervase Markham wrote: > > Isn't there a risk here? HTTP vhosting exists, but TLS vhosting does > > not (really) yet. So the owner of the website which is at > > http://www.foo.com/ might find that accessing https://www.foo.com/ > > actually gives them the HTML content of https://www.bar.com/, which > > is > hosted on the same machine, but controlled by someone else > > entirely. > > Wouldn't this throw a certificate error (and with HSTS, die without > warning) ? I think Devdatta is right. Gervase's concern should definitely be added to the security considerations. It is possible that your domain is foo.example.org and somebody else has a wildcard certificate for *.example.org. But, if it hurts when you do that then...don't do that. We must make sure that CSP composes well with other security mechanisms. Without this change, sites using CSP will break when any of the subresource sources enable HSTS and/or otherwise start redirecting http:// -> https://. This would add risk to deploying HSTS that would slow (in some cases, even prevent) its adoption. Also, I think most CSP source directives should be of the form *.example.org, so that the linked-to site is free to move a commonly hot-linked resource to a subdomain, especially a CDN. I expect that many sites will do such reorganization as part of the effort of TLS-enabling their sites. It would be helpful to provide this advise in the spec, and to change examples in the spec that use third-party sources to use this style, where it makes sense to do so. - Brian
Received on Monday, 27 June 2011 18:15:17 UTC