Jim Bernard wrote: >On Thu, Apr 06, 2017 at 09:54:47AM +0100, Robert Swindells wrote: >> >> ... or I can comment it out of the kernel config. >> >> That isn't an answer to the question that I am asking. >> >> At some point when I am doing cycles of taking out other unused device >> entries from the config and trying out the resulting kernel the console >> will stop displaying text. I can't see a pattern to what causes the >> change. >> >> I just wanted to know if anyone else was seeing similar behaviour. > >Yes, I've been having the same problem, also with -current amd64 kernels. >It occurs on both a Dell Optiplex 960 with an Intel E8400 cpu and an Nvidia >GeForce 9300 GE card that nouveau doesn't handle (so I use no drm in the >kernel) and on a Lenovo T61 laptop with an Intel T9300 cpu and i965GM >graphics that drmkms doesn't handle (so I use legacy drm in the kernel). >On both machines, there is text from the boot code, but none from the kernel >or any userland programs that run during multiuser startup, including getty. >There is a cursor, and it moves occasionally, but no text is printed. If I >type blindly, I can log in (still no text printed) and start X, which does >correctly generate its usual display. > >FWIW, there is text printed by drmkms kernels on both machines, but both >machines fail to finish booting with drmkms. (That failure, of course, is >a different issue.) > >The most-recent no-drm kernel I have that works on the Optiplex was built >from -current updated 201702110025Z (7.99.59), and the oldest no-drm kernel >I have that prints no text was built from -current updated 201702130335Z >(7.99.59). So there's about a 2-day window during which that particular >problem must have been introduced. > >I do get successful boots with normal console text printout on a Lenovo >Yoga 460 laptop with -current GENERIC amd64 kernels, but X doesn't work >at all on that machine.
Thanks, both for confirming that I wasn't going crazy and for providing the time window during which it broke. The cause looks to be this: <http://mail-index.netbsd.org/source-changes/2017/02/11/msg081926.html> Backing out the change restores the console for me. It explains why progressively taking things out of a kernel config would break at a seemingly random point. Robert Swindells
