/irc-logs / w3c / #webapps / 2008-11-14 / end
Options:
- # Session Start: Fri Nov 14 00:00:00 2008
- # Session Ident: #webapps
- # [00:03] <adrianba> he said he was going to be travelling for a few weeks - back toward the end of november
- # [00:05] * Quits: Lachy_ (Lachlan@24.234.142.17) (Ping timeout)
- # [00:11] <smaug> Hixie: so about <eventsource>
- # [00:11] <smaug> someone may inject that to a page
- # [00:11] <smaug> and then it connects to anyway
- # [00:12] <smaug> "anywhere" may be some evil server with support for AC
- # [00:12] <Hixie> hm?
- # [00:13] <smaug> I hadn't noticed that support for AC was added there at some point
- # [00:13] <smaug> (still haven't found out any good way to search changes in html 5)
- # [00:14] <smaug> Hixie: and then the evil server may dispatch events to the content
- # [00:14] <smaug> which are handled by the event listeners...
- # [00:16] <smaug> <smaug> and then it connects to anywhere
- # [00:16] <smaug> <smaug> "anywhere" may be some evil server with support for AC
- # [00:16] * smaug should learn to type, and even in English :p
- # [00:16] <Hixie> i really have no idea what you're asking :-)
- # [00:17] <Hixie> can you try again with antecedants for all your "it"s and "that"s and full sentences? :-)
- # [00:17] <smaug> bah
- # [00:18] <smaug> Hixie: so some site accepts user generated content. So probably not scripts, but elements
- # [00:18] <Hixie> ok
- # [00:19] <smaug> and user adds <eventsource src="someevilsite_with_access_control">
- # [00:19] <Hixie> and the site filters it, because <eventsource> isn't in their whitelist
- # [00:19] <smaug> are you sure sites do that
- # [00:19] <smaug> whitelisting
- # [00:19] <Hixie> they better, using a blacklist with html is the quickest way to get xss'ed
- # [00:20] <Hixie> but let's assume they do have this gaping hole and use a blacklist, and so <eventsource> goes through
- # [00:20] <Hixie> continue with the attack description
- # [00:20] <smaug> then server sends random events
- # [00:21] <smaug> although they are MessageEvents, their type can be anything
- # [00:21] <smaug> so the evil server could possible trigger any event listener on the page
- # [00:22] <Hixie> not really
- # [00:22] <Hixie> the events don't bubble
- # [00:22] <Hixie> he could trigger any capturing event listener in the chain
- # [00:22] <smaug> well, capturing event listeners
- # [00:22] <Hixie> yes
- # [00:22] <smaug> right
- # [00:23] <smaug> let's try otherway, what is the reason to have <eventsource>?
- # [00:23] <smaug> .addEventSource/.removeEventSource should be IMO enough
- # [00:23] <Dashiva> Have you ever seen a capturing listener used without it being an accident or misguided?
- # [00:23] <smaug> Dashiva: yes
- # [00:23] <smaug> all the time
- # [00:23] <Hixie> Dashiva: we've already established teh site author makes serious mistakes
- # [00:24] <Hixie> since he's not whitelisting
- # [00:24] <Hixie> smaug: so that you can declare an event source simply
- # [00:24] <Hixie> smaug: addEventSource() is a pain, and i'd rather remove that altogether frankly
- # [00:26] <smaug> why is that pain?
- # [00:27] <Hixie> far more power than necessary
- # [00:27] <Hixie> being able to do XMLHttpRequest.addEventSource() is just silly
- # [00:27] <smaug> well, with the current spec, perhaps yes
- # [00:28] <smaug> when the spec had the support for dispatching any kinds of server sent events, the situation was a bit different
- # [00:28] <smaug> server-sent-events could have been used to emulate user interaction (at least in some cases)
- # [00:29] <smaug> er, simulate
- # [00:29] <Hixie> not well, but yes
- # [00:30] <smaug> this whole <eventsource> thingie came up in our server-sent-events security review
- # [00:31] <smaug> other issue was adopting node to a new document
- # [00:31] <smaug> what should happen then
- # [00:32] <Hixie> nothing special, why?
- # [00:33] <Hixie> (you can't adopt cross-origin)
- # [00:33] <smaug> so the connection is kept alive
- # [00:33] <Hixie> sure
- # [00:33] <smaug> ok
- # [00:33] <smaug> fine to me
- # [00:33] <Hixie> good good :-)
- # [00:33] <smaug> but still, <eventsource> + evil site
- # [00:34] <smaug> what to do with that
- # [00:35] <smaug> one option could be that <eventsource> shouldn't support AC
- # [00:37] <Hixie> dude if we're assuming the site isn't using a whitelist, the site is so owned already there's really not much to discuss
- # [00:37] <smaug> eh
- # [00:39] <Hixie> e.g. what if the user slides in an <output> element with an onforminput="" attribute?
- # [00:42] <smaug> yeah, that is bad
- # [00:42] <smaug> for the site
- # [00:43] <smaug> and for the vendor which is the first implementator of that feature
- # [00:43] <smaug> (vendor X browser users hacked site FooBar)
- # [00:45] <Hixie> opera's already implemented <output>
- # [00:45] <Hixie> has for years
- # [00:45] <Hixie> blacklists are a fast track to getting owned
- # [00:46] <Hixie> they are highly insecure
- # [00:46] <Hixie> i'm shocked that anyone at mozilla considers them even a plausible security measure
- # [00:46] <smaug> I understand whitelisting is what should be done
- # [00:46] <smaug> but that isn't what all the site necessarily do
- # [00:46] <Hixie> sure, some do neither whitelisting nor blacklisting
- # [01:08] * Quits: adrianba (adrianba@131.107.0.80) (Quit: leaving)
- # [01:33] * Joins: shepazu (schepers@128.30.52.30)
- # [01:34] <smaug> shepazu is alive
- # [01:35] <shepazu> smaug: barely!
- # [01:35] <shepazu> I can finally breathe, in about 4 days... I'll be back from Japan, with no crazy deadlines looming over my head
- # [01:36] * Quits: heycam (cam@130.194.72.84) (Quit: bye)
- # [01:52] <MikeSmith> shepazu: you need to get moving if you are going to get to Shibuya
- # [01:52] <shepazu> yeah, leaving right now
- # [01:59] * Quits: shepazu (schepers@128.30.52.30) (Ping timeout)
- # [04:25] * Quits: maxf (maxf@84.202.168.250) (Ping timeout)
- # [04:25] * Joins: maxf (maxf@84.202.168.250)
- # [04:57] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [05:55] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [07:59] * Quits: aroben (aroben@71.58.97.175) (Quit: Leaving)
- # [08:35] * Joins: heycam (cam@210.84.56.87)
- # [08:52] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [09:29] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [11:06] * Joins: arve (arve@213.236.208.22)
- # [12:50] * Joins: ArtB (ce846302@128.30.52.43)
- # [13:07] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [13:08] * Joins: MikeSmith (MikeSmith@mcclure.w3.org)
- # [13:12] * Quits: MikeSmith (MikeSmith@mcclure.w3.org) (Quit: sex break)
- # [14:46] * Quits: arve (arve@213.236.208.22) (Quit: Leaving)
- # [15:20] * Quits: ArtB (ce846302@128.30.52.43) (Quit: CGI:IRC)
- # [15:43] * Joins: ArtB (ce846302@128.30.52.43)
- # [16:06] * Joins: aroben (aroben@71.58.97.175)
- # [17:10] * Quits: smaug (chatzilla@82.181.141.13) (Client exited)
- # [17:13] * Quits: gavin (gavin@63.245.208.169) (Ping timeout)
- # [17:14] * Joins: smaug (chatzilla@82.181.141.13)
- # [17:49] * Joins: gavin (gavin@63.245.208.169)
- # [18:20] * Joins: shepazu (schepers@128.30.52.30)
- # [18:49] * Quits: shepazu (schepers@128.30.52.30) (Quit: shepazu)
- # [20:01] * Quits: sicking (chatzilla@63.245.220.242) (Client exited)
- # [21:43] * Joins: arve (arve@80.202.65.163)
- # [22:06] * Quits: ArtB (ce846302@128.30.52.43) (Quit: CGI:IRC)
- # [23:40] * Joins: sicking (chatzilla@63.245.220.242)
- # Session Close: Sat Nov 15 00:00:00 2008
The end :)