GNOME Bugzilla – Bug 373556
ability to interrupt pam auth
Last modified: 2006-11-13 18:37:39 UTC
Ray has a patch that adds the ability to interrupt the PAM authentication. This is important for things like smart-card support. I think this is the latest one: http://cvs.fedora.redhat.com/viewcvs/devel/gnome-screensaver/gnome-screensaver-2.17.1-better-pam-integration.patch?rev=1.2&view=auto The async stuff seems reasonable. But I don't like the way it is hiding the prompt. It makes the dialog resize and jump around and stuff. Maybe add a placeholder or something. Longer term I think its worth considering making GDM and g-s use the same auth code possibly via PolicyKit. I'll create a new bug for this...
We could just desensitize the entry instead of hiding it. One thing we have to be careful about though, is making sure that key presses pressed before the dialog comes up still get propagated.
Note the gnome-screensaver-2.17.1-better-pam-integration.patch didn't actually work. When gnome-screensaver got rebased to 2.17.1 in rawhide a bit of the patch got dropped off. gnome-screensaver-2.17.2-better-pam-integration.patch is the latest working version.
Created attachment 76421 [details] [review] what will be in tomorrows rawhide this patch is what will be in tomorrows rawhide. It desensitizes the entry instead of hiding it.
Cool, I've commited this to HEAD with a few changes. * Don't show/hide the status/pam messages label since this causes the dialog to resize * Fixed conflicts * Change all tab indents to spaces * Use braces on all blocks * Various spacing nits Thanks Ray!
Hi Jon, So one problem that the current patch has I believe, is that queued keystrokes don't work properly. I mean if the user just starts typing away when their screen saver is active then the password entry in the dialog that pops up wont have everything they typed. I haven't had a chance to debug that yet.
So it is. Still having this is CVS is good. I'll reopen this to keep it on the map.
Sounds good.
Created attachment 76505 [details] [review] patch to fix key forwarding This seems to work for me so I've committed it to HEAD. Strictly speaking I think we could use map rather than show but this should be fine. I suppose at some point we should test that when no input is needed for authentication that the dialog doesn't even appear.