Just try with "console=ttyS0". systemd reads this parameter from the kernel command line and sets up the proper getty.
The systemd documentation has lots of information on that. > On Jul 26, 2016, at 11:37 PM, [email protected] wrote: > >> On 2016-07-26 17:31, [email protected] wrote: >>> On 2016-07-26 17:25, [email protected] wrote: >>> On 2016-07-26 17:15, John Paul Adrian Glaubitz wrote: >>> Perhaps both are panicing? In this situation it seems that systemd, >>> initramfs, etc... are all on one console and the kernel messages are >>> on another. I.E. kernel messages are on ttyS0 and everything else is >>> on tty0 (whatever tty0 is). >>> Come to think of it, my t5120 shows a similar behavior. Booting >>> without console=ttyS0 i can see all kernel messages and systemd / >>> initramfs messages, but no login prompt. If i boot with console=ttyS0 >>> i can see kernel messages, and eventually get a login prompt, but all >>> systemd / initramfs messages are missing. >>> It seems the same issue is being experienced here, but init is dying >>> before the login prompt can start. With console=ttyS0 we loose the >>> systemd messages, and without it systemd seems to move to the wrong >>> tty before panicing as we loose all output before a panic. >>> The real question is why initramfs / systemd aren't sending output to >>> ttyS0 when console=ttyS0 is specified. Is... >>> [244919.084723] Warning: unable to open an initial console. >>> indicating that systemd can't attach to ttyS0 for whatever reason? >>> And further more, what is tty0? Why is systemd starting getty with no >>> video card. Sun4U machines do not start getty; does the ldom provide a >>> virtual vga console? >> Meh, i take that back. Sun4U boxes do start getty. The question is >> what is tty0 and why is systemd attaching to it over ttyS0 where as on >> sun4u it chooses to attach to ttyS0. > > I need to lay off the coffee, i sometimes type too fast for my own good... > > <snip> and why is getty attaching to it over ttyS0 </snip>
