Re: [PATCH v1 4/8] xen/riscv: introduce sbi call to putchar to console

[Oleksii <oleksii.kurochko@xxxxxxxxx>]

On Fri, 2023-01-06 at 17:16 +0000, Andrew Cooper wrote:
> On 20/12/2022 6:23 am, Bobby Eshleman wrote:
> > On Fri, Jan 06, 2023 at 03:14:25PM +0200, Oleksii Kurochko wrote:
> > > The patch introduce sbi_putchar() SBI call which is necessary
> > > to implement initial early_printk
> > > 
> > I think that it might be wise to start off with an alternative to
> > sbi_putchar() since it is already planned for deprecation. I
> > realize
> > that this will require rework, but it is almost guaranteed that
> > early_printk() will break on future SBI implementations if using
> > this
> > SBI call. IIRC, Xen/ARM's early printk looked like a reasonable
> > analogy
> > for how it could work on RISC-V, IMHO.
> 
> Hmm, we're using it as a stopgap right now in CI so we can break
> "upstream RISC-V support" into manageable chunks.
> 
> So perhaps we want to forgo plumbing sbi_putchar() into
> early_printk() 
> (to avoid giving the impression that this will stay around in the
> future) and use sbi_putchar() directly for the hello world print.
> 
> Next, we focus on getting the real uart driver working along with
> real
> printk (rather than focusing on exceptions which was going to be the
> next task), and we can drop sbi_putchar() entirely.
> 
> Thoughts?
> 
I think it would be better to remain it as it is done now and add TODO
comment "TODO: rewrite early_printk() to use uart directly as
sbi_putchar() is already planned for deprecation" and update the commit
message too.
After real uart will be ready it will be easy to remove sbi_putchar()
and do something similar to ARM early printk implementation.
> ~Andrew