[PATCH 2/5] sandbox: do not register device before base is started

[Jean-Christophe PLAGNIOL-VILLARD]

On 19:41 Wed 19 Sep     , Sascha Hauer wrote:
> On Wed, Sep 19, 2012 at 04:47:55PM +0200, Jean-Christophe PLAGNIOL-VILLARD wrote:
> > This will crash when use registered bus with device registered to it.
> > 
> > Signed-off-by: Jean-Christophe PLAGNIOL-VILLARD <plagnioj at jcrosoft.com>
> > ---
> >  arch/sandbox/board/Makefile   |    1 +
> >  arch/sandbox/board/console.c  |    2 +-
> >  arch/sandbox/board/devices.c  |   35 +++++++++++++++++++++++++++++++++++
> >  arch/sandbox/board/hostfile.c |   19 +++++++++++++++++--
> >  4 files changed, 54 insertions(+), 3 deletions(-)
> >  create mode 100644 arch/sandbox/board/devices.c
> > 
> > diff --git a/arch/sandbox/board/Makefile b/arch/sandbox/board/Makefile
> > index 266c3a3..5104f5c 100644
> > --- a/arch/sandbox/board/Makefile
> > +++ b/arch/sandbox/board/Makefile
> > @@ -2,5 +2,6 @@ obj-y += board.o
> >  obj-y += clock.o
> >  obj-y += hostfile.o
> >  obj-y += console.o
> > +obj-y += devices.o
> >  
> >  extra-y += barebox.lds
> > diff --git a/arch/sandbox/board/console.c b/arch/sandbox/board/console.c
> > index 18b63e1..b0afa54 100644
> > --- a/arch/sandbox/board/console.c
> > +++ b/arch/sandbox/board/console.c
> > @@ -44,6 +44,6 @@ int barebox_register_console(char *name, int stdinfd, int stdoutfd)
> >  	data->stdoutfd = stdoutfd;
> >  	data->stdinfd  = stdinfd;
> >  
> > -	return register_device(dev);
> > +	return sandbox_add_device(dev);
> >  }
> >  
> > diff --git a/arch/sandbox/board/devices.c b/arch/sandbox/board/devices.c
> > new file mode 100644
> > index 0000000..d3fad78
> > --- /dev/null
> > +++ b/arch/sandbox/board/devices.c
> > @@ -0,0 +1,35 @@
> > +/*
> > + * Copyright (c) 2012 Jean-Christophe PLAGNIOL-VILLARD <plagnioj at jcrosoft.com>
> > + *
> > + * Under GPLv2 only
> > + */
> > +
> > +#include <common.h>
> > +#include <driver.h>
> > +#include <init.h>
> > +
> > +static LIST_HEAD(sandbox_device_list);
> > +
> > +int sandbox_add_device(struct device_d *dev)
> > +{
> > +	if (!dev)
> > +		return -EINVAL;
> 
> Can we please agree that everyone registering NULL pointers deserves a
> nice crash? I do not see a reason to check this. All these kinds of
> checks do is cover bugs and make tracking them down harder.
I agree but not on the crash we need to use BUG_ON

Best Regards,
J.