[PATCH 2/3] add command line boot support

[Jean-Christophe PLAGNIOL-VILLARD]

On 14:10 Thu 27 Jan     , Sascha Hauer wrote:
> On Thu, Jan 27, 2011 at 12:55:31PM +0100, Jean-Christophe PLAGNIOL-VILLARD wrote:
> > Hi Sascha,
> > 
> > On 10:20 Thu 27 Jan     , Sascha Hauer wrote:
> > > Hi J,
> > > 
> > > On Wed, Jan 26, 2011 at 05:46:03PM +0100, Jean-Christophe PLAGNIOL-VILLARD wrote:
> > > > for now just support static boot command support
> > > > 
> > > > Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj at jcrosoft.com>
> > > > ---
> > > >  arch/arm/lib/barebox.lds.S                |    4 +
> > > >  arch/blackfin/boards/ipe337/barebox.lds.S |    4 +
> > > >  arch/ppc/boards/pcm030/barebox.lds.S      |    4 +
> > > >  arch/sandbox/board/barebox.lds.S          |    4 +
> > > >  arch/sandbox/lib/barebox.lds.S            |    4 +
> > > >  arch/x86/lib/barebox.lds.S                |    9 ++-
> > > >  common/Kconfig                            |   11 ++
> > > >  common/Makefile                           |    1 +
> > > >  common/params.c                           |  158 +++++++++++++++++++++++++++++
> > > >  common/startup.c                          |   74 +++++++++++++-
> > > >  include/asm-generic/barebox.lds.h         |    2 +
> > > >  include/init.h                            |   28 +++++
> > > >  12 files changed, 301 insertions(+), 2 deletions(-)
> > > >  create mode 100644 common/params.c
> > > 
> > > Except for the inlined comments I'm generally ok with this patch.
> > > The more interesting part will be how the calling convention for
> > > barebox as a second stage loader is and how you preserve the
> > > registers used for commandline/atag passing during startup.
> > I take a look and no need to do something special we just have to put the data
> > in the .data section as we will supposed to be call from memmory directly
> > evenif we may ned to relocate our self
> 
> Nothing special? Then how does barebox know where to find the command
> line?
yeah I agree I'll send a patch i'll be more clear
> 
> > > 
> > > An idea which comes to my mind is that we could introduce a
> > > CONFIG_BAREBOX_SECOND_STAGE which switches to a completely different
> > > startup process. This startup process could then assume that
> > > sdram is already initialized and that we do not have to call
> > > board_init_lowlevel.
> > I get in mind to keep it maybe as we can run the lowlevel init from cache as a
> > peekpoke table to reconfigure the board if needed
> > > Another interesting thing is how will barebox behave if called
> > > as a second stage loader but without valid atags?
> > I think to use in this case the default config and let the boards to get a
> > callback to overwrite it
> > I'll try to push patch soon to show the solution
> > I get in mind to maybe use the device tree also not sure yet
> > > 
> > > > 
> > > >  
> > > > +config BOOT_CMDLINE
> > > > +	bool "barebox boot command string"
> > > > +	help
> > > 
> > > 	  This is useful if you intend to use barebox as a second stage bootloader
> > > 	  and want to pass kernel like command line parameters to barebox. Otherwise
> > > 	  say no here.
> > 	as you can have built_in cmdline you can use it also as a first stage
> 
> Yes, but this really serves no purpose, does it?
yeah it's as you can dynamse the board via the static cmdline
and other part without modifying the code
and you will be able to dynamise the shell as I'm prepaaring a patch to export
them in the shell too

Best Regards,
J.