Re: [microblaze-uclinux] Note to bootloader developers

[Yasushi SHOJI <yashi@xxxxxxxxxxxxxxxxx>]

At Mon, 31 May 2004 08:01:20 +1000,
John Williams wrote:
> 
> Hi Yashi,
> 
> Yasushi SHOJI wrote:
> 
> >>A note to the developers of these bootloaders, if you wish to add kernel 
> >>command line support, it's easy.  When you jump to the kernel start 
> >>address, just have r5 contain the address of a null-terminated string 
> >>that is the kernel command line.  This string could live anywhere - 
> >>BRAM, flash, doesn't matter. The kernel copies it away safely very early 
> >>in the piece, so you don't need to worry about it getting corrupted.
> > 
> > I have to object on this.
> > 
> > so as I (as a programmer working on microblaze port; not as an
> > employee of Atmark Techno) am an embedded system programmer, what we
> > need are:
> > 
> > 1. standard tag based kernel paramter passing scheme
> > 2. compile time parter passing scheme
> > 3. (if any one needs it) current string based simple parameter scheme.
> > 
> > comments welcome!
> 
> Feel free to object, and to implement the tag based scheme! :)

That's what we need. enough chatting, do the coding, right? ;)

# i've been writing too much docs these days.  need to do some _real_
# work ;p

>                                                                  I'll 
> freely admit that the current approach was a quick hack by myself and 
> Brett to get some form of command line params into the kernel. 
> Necessity is the mother of invention and all that.

yes. I understand that we needed to somehow pass params to the
kernel. What I object is to let others to use this (non-standard)
hack.

> However, from what I saw in the source, basically all the kernel expects 
> ultimately is a pointer to a string that it later parses.  So, is your 
> objection just the simplistic mechanism for forming that string (ie a 
> bootloader issue)?  Or is it something deeper than that, that would go 
> into setup_arch()?

I'm not exactly sure why others are not doing our simple way.  I just
thought there must be a reason.  tag base param passing is relatively
new. it's been introduced to the kernel when we moved to 2.4, IIRC.

can someone dig the lkml archives?
--
         yashi
___________________________
microblaze-uclinux mailing list
microblaze-uclinux@itee.uq.edu.au
Project Home Page : http://www.itee.uq.edu.au/~jwilliams/mblaze-uclinux
Mailing List Archive : http://www.itee.uq.edu.au/~listarch/microblaze-uclinux/