**** BEGIN LOGGING AT Wed Oct 17 03:00:01 2012 Oct 17 07:08:35 hi Oct 17 07:37:54 good morning Oct 17 08:38:48 morning all Oct 17 08:39:54 bluelightning: hi Oct 17 08:50:31 hi mckoan Oct 17 12:40:19 fun Oct 17 12:40:36 login.tinylogin allows passwordless, login.shadow does not Oct 17 12:43:04 hrw: does login.shadow also still blow up if you enter a space in the username? Oct 17 12:44:12 pb_: I have only root in images Oct 17 12:44:37 it doesn't matter what your password file says. it used to just crash if you typed "a b c" at the login prompt. Oct 17 12:44:56 and for Linaro images we have them autologin to root. works fine as long as login.tinylogin is used Oct 17 12:44:59 (irrespective of whether you actually had any users with spaces in or not) Oct 17 12:45:26 oh, it is been discussed here as well ;) Oct 17 12:45:55 the problem why login didn't work is that our serial console, ttyAMA0 is not in /etc/securetty Oct 17 12:47:17 ah. yes, it sucks that it is the year 2012 and there is still no unified naming for serial ports. Oct 17 12:47:53 yep Oct 17 12:48:47 of course, for those of us who are not using amba, there is effectively a unified naming because everything is a 16550 :-} Oct 17 12:49:24 it sucks that in year 2012 we think there is security value in allowing root login only over selected serial consoles Oct 17 12:49:52 well, I dunno, openssh also has an option to enable or disable root login. Oct 17 12:50:08 whether it's actually of real security value is perhaps debatable, but it doesn't seem totally unreasonable to have the option. Oct 17 12:50:42 as an option not enabled by default Oct 17 12:50:44 arguably, since "root" is an account which does almost always exist, allowing an attacker to try to login in as that name does give them an advantage compared to having to guess both a username and a password Oct 17 12:52:13 in reality, nobody is going to brute force you over serial console Oct 17 12:54:12 patch sent to oecore ml Oct 17 15:15:48 is ecore and rest of EFL in separate layer? Oct 17 15:16:13 meta-fsl in meta-openembedded Oct 17 15:16:16 erm, meta-efl Oct 17 15:16:22 see LayerIndex Oct 17 15:20:42 thx Oct 17 15:20:49 morning kergoth Oct 17 18:25:19 question Oct 17 18:27:17 I do some kernel image mangling in do_compile_append() but outside of ${S} and it's stored in deploy/images during do_deploy(). But with sstate-cache in use it is not there on second run. should I do something extra in do_install() step? **** ENDING LOGGING AT Thu Oct 18 03:00:01 2012