Parse DTB during startup, allowing memory banks and reserved
memory regions to be set up, along with early device tree node
( chosen, "xen,domain", "reserved-memory", etc  ) handling.

Signed-off-by: Oleksii Kurochko <oleksii.kurochko@gmail.com>
---
 xen/arch/riscv/setup.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/xen/arch/riscv/setup.c b/xen/arch/riscv/setup.c
index 39375b3366..a8d8ef793d 100644
--- a/xen/arch/riscv/setup.c
+++ b/xen/arch/riscv/setup.c
@@ -28,6 +28,7 @@ void __init noreturn start_xen(unsigned long bootcpu_id,
                                paddr_t dtb_addr)
 {
     struct bootmodule *xen_bootmodule;
+    size_t fdt_size;
 
     remove_identity_mapping();
 
@@ -54,6 +55,9 @@ void __init noreturn start_xen(unsigned long bootcpu_id,
 
     BUG_ON(!xen_bootmodule);
 
+    fdt_size = boot_fdt_info(device_tree_flattened, dtb_addr);
+    BUG_ON(!fdt_size);
+
     printk("All set up\n");
 
     machine_halt();
-- 
2.46.1

On 30.09.2024 17:13, Oleksii Kurochko wrote:
> --- a/xen/arch/riscv/setup.c
> +++ b/xen/arch/riscv/setup.c
> @@ -28,6 +28,7 @@ void __init noreturn start_xen(unsigned long bootcpu_id,
>                                 paddr_t dtb_addr)
>  {
>      struct bootmodule *xen_bootmodule;
> +    size_t fdt_size;
>  
>      remove_identity_mapping();
>  
> @@ -54,6 +55,9 @@ void __init noreturn start_xen(unsigned long bootcpu_id,
>  
>      BUG_ON(!xen_bootmodule);
>  
> +    fdt_size = boot_fdt_info(device_tree_flattened, dtb_addr);
> +    BUG_ON(!fdt_size);
> +
>      printk("All set up\n");
>  
>      machine_halt();

Looks plausible, and judging from Arm code there'll be an actual use of fdt_size
down the road. Or else I would have asked what use the local variable is. In
fact, again from looking at Arm code, I now question the need for the
xen_bootmodule local var - that has no further use in Arm, and hence that's
likely going to be the case for RISC-V, too.

However, may I ask that you limit the dependency trees in what you submit? This
series depends on another series, which in turn depends on yet something else,
all not yet committed. That's getting unwieldy, I'm afraid.

Jan

On Tue, 2024-10-01 at 17:54 +0200, Jan Beulich wrote:
> On 30.09.2024 17:13, Oleksii Kurochko wrote:
> > --- a/xen/arch/riscv/setup.c
> > +++ b/xen/arch/riscv/setup.c
> > @@ -28,6 +28,7 @@ void __init noreturn start_xen(unsigned long
> > bootcpu_id,
> >                                 paddr_t dtb_addr)
> >  {
> >      struct bootmodule *xen_bootmodule;
> > +    size_t fdt_size;
> >  
> >      remove_identity_mapping();
> >  
> > @@ -54,6 +55,9 @@ void __init noreturn start_xen(unsigned long
> > bootcpu_id,
> >  
> >      BUG_ON(!xen_bootmodule);
> >  
> > +    fdt_size = boot_fdt_info(device_tree_flattened, dtb_addr);
> > +    BUG_ON(!fdt_size);
> > +
> >      printk("All set up\n");
> >  
> >      machine_halt();
> 
> Looks plausible, and judging from Arm code there'll be an actual use
> of fdt_size
> down the road. Or else I would have asked what use the local variable
> is. In
> fact, again from looking at Arm code, I now question the need for the
> xen_bootmodule local var - that has no further use in Arm, and hence
> that's
> likely going to be the case for RISC-V, too.
fdt_size and xen_bootmodule are used only for BUG_ON() now and in
future ( the same as you mentioned as in Arm ). They could be dropped.
I'll do that in the next patch version.

> 
> However, may I ask that you limit the dependency trees in what you
> submit? This
> series depends on another series, which in turn depends on yet
> something else,
> all not yet committed. That's getting unwieldy, I'm afraid.
Sure, sorry for that. I will limit the depedency trees in future.

~ Oleksii