New X.Org Server 1.13 RC Bumps The ABI

[uid313]

The real showstopper at the moment is the lack of revoke() support in the kernel. Without that, your input devices are wide open to pretty much anyone at all times.

Why isn't revoke() in the kernel?
Have anyone talked about it on LKML?
Has anyone submitted a patch that implements revoke()?

Alan's already done a hell of a lot for X (and most of it utterly thankless yet totally essential work) over the last god-knows-how-many years, so I'm not sure why you're heckling him at all. What have you done to improve X/Wayland?

I didn't mean to hack on Alan.
I haven't contributed anything, he have contributed and that is cool!

[elg2001]

Foot-in-mouth syndrome sucks, doesn't it?

Thanks to both of you for making the Linux desktop what it is today. And I'm sure a lot of other work that may not be flashy but is important and formative. We are truly standing on the shoulders of giants.

[daniels]

Why isn't revoke() in the kernel?
Have anyone talked about it on LKML?
Has anyone submitted a patch that implements revoke()?

Yeah, it's been discussed on and off for several years, but is really hard to implement, especially in the presence of mmap()ed files. At the moment the current best suggestion is an input-specific revoke(), but no-one's managed to write and submit a patch implementing that which would be accepted.

[uid313]

Yeah, it's been discussed on and off for several years, but is really hard to implement, especially in the presence of mmap()ed files. At the moment the current best suggestion is an input-specific revoke(), but no-one's managed to write and submit a patch implementing that which would be accepted.

Then how did other operating systems with support for revoke() manage to implement it?
I assume Solaris support revoke()?

[daniels]

Then how did other operating systems with support for revoke() manage to implement it?
I assume Solaris support revoke()?

From the looks of things, Solaris supports vhangup(), which is TTY-specific, and it doesn't have a general-purpose revoke() either.

[uid313]

From the looks of things, Solaris supports vhangup(), which is TTY-specific, and it doesn't have a general-purpose revoke() either.

Why hasn't this been fixed?
Someone must been able to fix it, Linux kernel have so many smart developers.

How about OpenBSD, does it support revoke()?
Does OpenBSD run X.org without superuser privileges?