Re: [08/36] AArch64: Kernel booting and initialisation

[Catalin Marinas]

On Tue, Jul 24, 2012 at 07:53:16PM +0100, Arnd Bergmann wrote:
> On Tuesday 24 July 2012, Catalin Marinas wrote:
> > On Mon, Jul 23, 2012 at 09:52:22PM +0100, Christopher Covington wrote:
> > > Catalin Marinas - July 8, 2012, 9:18 a.m.
> > > > On Fri, Jul 06, 2012 at 10:32:54PM +0100, Stephen Warren wrote:
> > > > 
> > > >> Also, on Tegra at least and perhaps OMAP too, there are a few parts
> > > >> of earlyprintk serial port setup (i.e. port selection) which happen
> > > >> as part of the decompressor rather than the main kernel image, so
> > > >> we'd have to re-jig that too.
> > > > 
> > > > Whatever we did on AArch32, if it was useful and we want to follow a
> > > > similar model it needs porting given that the code cannot be shared. So
> > > > there is nothing to re-jig but rather implement.
> > > > 
> > > > As for earlyprintk, you can initialise it in the decompressed kernel
> > > > anyway. But even though it's a useful debugging tool, it goes against
> > > > the single Image aim (at least the current printascii/addruart
> > > > implementation). I don't have a proper implementation yet.
> > > 
> > > Would DCC [1] be a reasonably commonly available early printk mechanism
> > > for the single image to provide?
> > 
> > Yes, this can be implemented but I think it has its own restrictions,
> > requiring a JTAG connection.
> 
> Right, which means that it's not generic and we still need to make
> a decision for a how we want to solve the underlying problem. The
> options I see are:
> 
> 1. Have no earlyprintk before loading the console drivers

It may be a long time and wouldn't help with early bugs.

> 2. Make the series from Domenico architecture independent code

I haven't seen this yet.

> 3. Reimplement a layer like that, differently

Any of the two above would be fine.

> 4. Make it a compile-time option and allow this only for
>    low level debugging purposes but disable it in distro kernels.

That's what I currently have (though not included in this post) but I
don't particularly the approach. There are advantages in being able to
call printascii from assembly early on but if we don't have a standard
UART than it goes against single Image file.

My proposal is to allow DCC very early on for development kernels
(possibly called from assembly) with a timeout if there is no JTAG
connection. Immediately after setup_machine_fdt() switch to a UART early
console if there is one. But we retrieve this information from the FDT.

I would say the early_printk console should not bother with UART
initialisation (done by the boot loader) but just the actual character
sending. It should also be placed somewhere under drivers/ as it may
be shared by other architectures (it's just specific to the UART chip).

-- 
Catalin
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/