Bug 3421

This is caused by the same symptom that causes bug#1951 - the free installs create an initrd which does not contain the firmware from the nonfree radeon-firmware package.

However, in the case of this chip the consequences are considerably more severe, for two reasons.  

First, this isn't just an eventually-terminating loop that eventually ends in the DM coming up.  It causes a hard freeze at about the time getty is being launched.  Hard, as in no magic keys and time to hit the power button.

Second, this does not appear to be related to using the X server, as I have recreated the hang after using a rescue disk to set the runlevel in inittab to 3.  Just in case systemd was ignoring inittab, I tried with init=/sbin/init as a kernel parm with the same result.

If I'm interpreting this correctly, the we have a major problem.

The lack-of-firmware problem exhibits with a range of ATI chips, and in the case of this chip, it manifests with a different and more serious symptom in text console mode without X involved.  As this chip is relatively new compared to the chips that just exhibited the loop, it is likely that other more current ATI chips will exhibit the more serious behavior.  Or, possibly the behavior is random based on the chip characteristics, and these are only two examples.

So the usual mantra of let-them-use-VESA until they enable nonfree on their own isn't going to work.  They'll never get to X on their first boot.

The machine involved is an Acer Aspire One 722 acquired from Walmart.  I have a test partition available, and if anyone would like tests run, just post them here.