After upgrading to Leopard, SmartCard reader stops working. It should ask for PIN1 instead of the key (see attached image). It worked with Tiger. Model of a SmartCard reader: SCM Microsystems SCR3310. NB! Works with Firefox 3 Beta 2.
Created attachment 18132 [details] Safari wants to sign using key ... in your keychain
How does a web browser interface with a smartcard reader? Is there software that's installed with the reader? This is probably either a third-party issue or a non-WebKit issue.
Browser should prompt user for the PIN1 code as used for authentication, e.g. https://www.seb.ee/cgi-bin/unet3.sh/un3.w?sesskey=&act=LOGIN&lang=ENG It worked smoothly in Tiger and due to the fact that it works properly in Firefox it makes sense that the issue relies in WebKit. It might of course be some core issue of PC/SC in Leopard that triggers this but even so you guys probably have access to debugging it or at least raising a subject.
<rdar://problem/5664161>
If this works in Tiger but not Leopard, it's probably *not* a WebKit issue since the latest Tiger and Leopard have the same version of WebKit. I'd guess it's a lower level issue, probably in CFNetwork, which is much changed from Tiger to Leopard.
Yet the latest Firefox 3 Beta 2 works smoothly. Does Gecko have the stack bundled in opposite to the WebKit's SmartCard support that relies on Mac OS?
(In reply to comment #6) > Yet the latest Firefox 3 Beta 2 works smoothly. Does Gecko have the stack > bundled in opposite to the WebKit's SmartCard support that relies on Mac OS? Firefox and Safari use different HTTP/network stack implementations. I'm not sure what you mean by WebKit's "SmartCard support"--I don't think any such thing exists.
Regressions are P1.
Works now in the latest nightly!
Broken again in the latest nightly.
That's weird! Could you please provide revision numbers for these two nightlies?
Ain, please use the WebKitTools/Scripts/bisect-builds script to narrow down the revisions that work and don't work. If you don't have the WebKit source checked out, you may download a copy of the script from this page (use the links below "Download in other formats" at the bottom of the page to save the script as a file): http://trac.webkit.org/projects/webkit/browser/trunk/WebKitTools/Scripts/bisect-builds
I don't know if this is related to the original bug, but there is a known issue with smartcards under Leopard, not related to Webkit. I manage several Macs with SCR readers attached and cards that used to work under Tiger do not now work under Leopard. The cards do not appear in the system Keychain at all, which they should. The cards MOST affected are the Obethur One cards (current interim DoD issue cards) in conjunction with the SCR readers, although other cards have been affected. Switching to the OMNIKEY 3021 reader solved this problem on my single test case, I'm in the process of ordering more. Check your keychain, if the card doesn't show up there then you'll know it's not a Webkit issue.
(In reply to comment #12) > Ain, please use the WebKitTools/Scripts/bisect-builds script to narrow down the > revisions that work and don't work. > > If you don't have the WebKit source checked out, you may download a copy of the > script from this page (use the links below "Download in other formats" at the > bottom of the page to save the script as a file): > > http://trac.webkit.org/projects/webkit/browser/trunk/WebKitTools/Scripts/bisect-builds > The issue is very controversial. I upgrade my WebKit nightly on a regular basis almost always daily. And while I'm not changing anything it occasionally works and always breaks after the Mac wakes up from sleep. This was not case on Tiger and is not the case with Firefox.
I can now confirm the behaviour: 1. SmartCard works with WebKit right after a restart; 2. It stops working after coming up from a sleep.
WebKit itself does not interact with smart cards. If you are still experiencing issues I would recommend filing a bug report at <http://bugreport.apple.com>, and including detailed information about how to reproduce the problem you are seeing. That would include any steps that must be taken to trigger the problem, information about the hardware and type of smart card you're using. This bug will be closed as INVALID per our policy for handling of bugs that are outside of WebKit.