[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Xen-devel] [PATCH 2/9] xen: arm: parse modules from DT during early boot.



On Mon, 10 Dec 2012, Ian Campbell wrote:
> On Fri, 2012-12-07 at 17:30 +0000, Stefano Stabellini wrote:
> > On Thu, 6 Dec 2012, Ian Campbell wrote:
> > > The bootloader should populate /chosen/modules/module@<N>/ for each
> > > module it wishes to pass to the hypervisor. The content of these nodes
> > > is described in docs/misc/arm/device-tree/booting.txt
> > > 
> > > The hypervisor parses for 2 types of module, linux zImages and linux
> > > initrds. Currently we don't do anything with them.
> > > 
> > > Signed-off-by: Ian Campbell <ian.campbell@xxxxxxxxxx>
> > > ---
> > > v4: Use /chosen/modules/module@N
> > >     Identify module type by compatible property not number.
> > > v3: Use a reg = < > property for the module address/length.
> > > v2: Reserve the zeroeth module for Xen itself (not used yet)
> > >     Use a more idiomatic DT layout
> > >     Document said layout
> > > ---
> > >  docs/misc/arm/device-tree/booting.txt |   25 ++++++++++
> > >  xen/common/device_tree.c              |   86 
> > > +++++++++++++++++++++++++++++++++
> > >  xen/include/xen/device_tree.h         |   14 +++++
> > >  3 files changed, 125 insertions(+), 0 deletions(-)
> > >  create mode 100644 docs/misc/arm/device-tree/booting.txt
> > > 
> > > diff --git a/docs/misc/arm/device-tree/booting.txt 
> > > b/docs/misc/arm/device-tree/booting.txt
> > > new file mode 100644
> > > index 0000000..94cd3f1
> > > --- /dev/null
> > > +++ b/docs/misc/arm/device-tree/booting.txt
> > > @@ -0,0 +1,25 @@
> > > +Xen is passed the dom0 kernel and initrd via a reference in the /chosen
> > > +node of the device tree.
> > > +
> > > +Each node has the form /chosen/modules/module@<N> and contains the 
> > > following
> > > +properties:
> > > +
> > > +- compatible
> > > +
> > > + Must be:
> > > +
> > > +         "xen,<type>", "xen,multiboot-module"
> > > +
> > > + where <type> must be one of:
> > > +
> > > + - "linux-zimage" -- the dom0 kernel
> > > + - "linux-initrd" -- the dom0 ramdisk
> > > +
> > > +- reg
> > > +
> > > + Specifies the physical address of the module in RAM and the
> > > + length of the module.
> > > +
> > > +- bootargs (optional)
> > > +
> > > + Command line associated with this module
> > > diff --git a/xen/common/device_tree.c b/xen/common/device_tree.c
> > > index da0af77..4bb640e 100644
> > > --- a/xen/common/device_tree.c
> > > +++ b/xen/common/device_tree.c
> > > @@ -270,6 +270,90 @@ static void __init process_cpu_node(const void *fdt, 
> > > int node,
> > >      cpumask_set_cpu(start, &cpu_possible_map);
> > >  }
> > >  
> > > +static int __init process_chosen_modules_node(const void *fdt, int node,
> > > +                                              const char *name, int 
> > > *depth,
> > > +                                              u32 address_cells, u32 
> > > size_cells)
> > > +{
> > > +    const struct fdt_property *prop;
> > > +    const u32 *cell;
> > > +    int nr, nr_modules = 0;
> > > +    struct dt_mb_module *mod;
> > > +    int len;
> > > +
> > > +    for ( *depth = 1;
> > > +          *depth >= 1;
> > > +          node = fdt_next_node(fdt, node, depth) )
> > > +    {
> > > +        name = fdt_get_name(fdt, node, NULL);
> > > +        if ( strncmp(name, "module@", strlen("module@")) == 0 ) {
> > > +
> > > +            if ( fdt_node_check_compatible(fdt, node,
> > > +                                           "xen,multiboot-module" ) != 0 
> > > )
> > > +                early_panic("%s not a compatible module node\n", name);
> > > +
> > > +            if ( fdt_node_check_compatible(fdt, node,
> > > +                                           "xen,linux-zimage") == 0 )
> > > +                nr = 1;
> > > +            else if ( fdt_node_check_compatible(fdt, node,
> > > +                                                "xen,linux-initrd") == 0)
> > > +                nr = 2;
> > > +            else
> > > +                early_panic("%s not a known xen multiboot byte\n");
> > > +
> > > +            if ( nr > nr_modules )
> > > +                nr_modules = nr;
> > > +
> > > +            mod = &early_info.modules.module[nr];
> > > +
> > > +            prop = fdt_get_property(fdt, node, "reg", NULL);
> > > +            if ( !prop )
> > > +                early_panic("node %s missing `reg' property\n", name);
> > > +
> > > +            cell = (const u32 *)prop->data;
> > > +            device_tree_get_reg(&cell, address_cells, size_cells,
> > > +                                &mod->start, &mod->size);
> > > +
> > > +            prop = fdt_get_property(fdt, node, "bootargs", &len);
> > > +            if ( prop )
> > > +            {
> > > +                if ( len > sizeof(mod->cmdline) )
> > > +                    early_panic("module %d command line too long\n", nr);
> > > +
> > > +                safe_strcpy(mod->cmdline, prop->data);
> > > +            }
> > > +            else
> > > +                mod->cmdline[0] = 0;
> > > +        }
> > > +    }
> > > +
> > > +    for ( nr = 1 ; nr < nr_modules ; nr++ )
> > > +    {
> > > +        mod = &early_info.modules.module[nr];
> > > +        if ( !mod->start || !mod->size )
> > > +            early_panic("module %d  missing / invalid\n", nr);
> > > +    }
> > > +
> > > +    early_info.modules.nr_mods = nr_modules;
> > > +    return node;
> > > +}
> > > +
> > > +static void __init process_chosen_node(const void *fdt, int node,
> > > +                                       const char *name,
> > > +                                       u32 address_cells, u32 size_cells)
> > > +{
> > > +    int depth;
> > > +
> > > +    for ( depth = 0;
> > > +          depth >= 0;
> > > +          node = fdt_next_node(fdt, node, &depth) )
> > > +    {
> > > +        name = fdt_get_name(fdt, node, NULL);
> > > +        if ( depth == 1 && strcmp(name, "modules") == 0 )
> > > +            node = process_chosen_modules_node(fdt, node, name, &depth,
> > > +                                               address_cells, 
> > > size_cells);
> > > +    }
> > > +}
> > > +
> > >  static int __init early_scan_node(const void *fdt,
> > >                                    int node, const char *name, int depth,
> > >                                    u32 address_cells, u32 size_cells,
> > > @@ -279,6 +363,8 @@ static int __init early_scan_node(const void *fdt,
> > >          process_memory_node(fdt, node, name, address_cells, size_cells);
> > >      else if ( device_tree_type_matches(fdt, node, "cpu") )
> > >          process_cpu_node(fdt, node, name, address_cells, size_cells);
> > > +    else if ( device_tree_node_matches(fdt, node, "chosen") )
> > > +        process_chosen_node(fdt, node, name, address_cells, size_cells);
> > >  
> > >      return 0;
> > >  }
> > 
> > You have really written a lot of code here!
> > I would have thought that just matching on the compatible string would
> > be enough:
> > 
> > else if ( device_tree_node_matches(fdt, node, "linux-zimage") )
> >      process_linuxzimage_node(fdt, node, name, address_cells, size_cells);
> > else if ( device_tree_node_matches(fdt, node, "linux-initrd") )
> >      process_linuxinitrd_node(fdt, node, name, address_cells, size_cells);
> > 
> > so that your process_linuxzimage_node and process_linuxinitrd_node start
> > from the right node and have everything they need to parse it
> 
> Is the tree structure of Device Tree meaningless? I'd have thought that
> a compatible node would only have meaning at a particular place in the
> tree. Granted compatible nodes are often pretty specific and precise,
> but is that inherent enough in DT that we can rely on it?

I don't know if it is entirely meaningless, but surely the compatible
string is regarded as a much more reliable way to identify a node AFAIK.
More often than not Linux drivers just use of_find_compatible_node to
find their DT node.

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel


 


Rackspace

Lists.xenproject.org is hosted with RackSpace, monitoring our
servers 24x7x365 and backed by RackSpace's Fanatical Support®.