Re: [GIT PULL] TTY/Serial driver updates for 6.3-rc1

[Al Viro]

On Fri, Feb 24, 2023 at 01:30:58PM -0800, Linus Torvalds wrote:
> On Fri, Feb 24, 2023 at 1:24 PM Linus Torvalds
> <torvalds@xxxxxxxxxxxxxxxxxxxx> wrote:
> >
> >         default SERIAL_8250
> 
> Hmm. Looking around, it seems to be a pattern. I'm not convinced any
> of them are really valid, except probably the ones that aren't about
> specific drivers, but about base support (like the "SERIAL_8250_DMA"
> one).
> 
> I also do get the feeling that the base "SERIAL_8250" feature might be
> better off as a _selected_ option, rather than a "ask if you want it".
> I think we've ended up with that SERIAL_8250 being a base option
> mainly for historical reasons (ie it was part of the original PC/AT
> base specs, and then later it has become a "base driver for a lot of
> random cards".
> 
> It's probably most legacy by now - it's a long time since I saw a
> serial port being used outside of management ports, and even those are
> often ethernet these days.

Serial console is hard to replace - sure, netconsole can take care of
the printk side of things, but the things like "boot with init=/bin/sh
and try to recover using that" are occasionally useful; same for "drop
into shell on initramfs and use that to poke around"...  What do you
use that kind of situations?  Not a rethorical question - I would be
really happy to know alternative variants for that; IME it's not
needed often (thankfully), but when it's needed it's really useful
to have...