Re: [PATCH] hw/arm/boot: set initrd parameters to 64bit in fdt

[Alex Bennée]

Schspa Shi <schspa@gmail.com> writes:

> We use 32bit value for linux,initrd-[start/end], when we have
> loader_start > 4GB, there will be a wrong initrd_start passed
> to the kernel, and the kernel will report the following warning.
>
> [    0.000000] ------------[ cut here ]------------
> [    0.000000] initrd not fully accessible via the linear mapping -- please 
> check your bootloader ...
> [    0.000000] WARNING: CPU: 0 PID: 0 at arch/arm64/mm/init.c:355 
> arm64_memblock_init+0x158/0x244
> [    0.000000] Modules linked in:
> [    0.000000] CPU: 0 PID: 0 Comm: swapper Tainted: G        W          
> 6.1.0-rc3-13250-g30a0b95b1335-dirty #28
> [    0.000000] Hardware name: Horizon Sigi Virtual development board
> (DT)

Is this an out-of-tree board model?

> [    0.000000] pstate: 600000c5 (nZCv daIF -PAN -UAO -TCO -DIT -SSBS BTYPE=--)
> [    0.000000] pc : arm64_memblock_init+0x158/0x244
> [    0.000000] lr : arm64_memblock_init+0x158/0x244
> [    0.000000] sp : ffff800009273df0
> [    0.000000] x29: ffff800009273df0 x28: 0000001000cc0010 x27: 
> 0000800000000000
> [    0.000000] x26: 000000000050a3e2 x25: ffff800008b46000 x24: 
> ffff800008b46000
> [    0.000000] x23: ffff800008a53000 x22: ffff800009420000 x21: 
> ffff800008a53000
> [    0.000000] x20: 0000000004000000 x19: 0000000004000000 x18: 
> 00000000ffff1020
> [    0.000000] x17: 6568632065736165 x16: 6c70202d2d20676e x15: 
> 697070616d207261
> [    0.000000] x14: 656e696c20656874 x13: 0a2e2e2e20726564 x12: 
> 0000000000000000
> [    0.000000] x11: 0000000000000000 x10: 00000000ffffffff x9 : 
> 0000000000000000
> [    0.000000] x8 : 0000000000000000 x7 : 796c6c756620746f x6 : 
> 6e20647274696e69
> [    0.000000] x5 : ffff8000093c7c47 x4 : ffff800008a2102f x3 : 
> ffff800009273a88
> [    0.000000] x2 : 80000000fffff038 x1 : 00000000000000c0 x0 : 
> 0000000000000056
> [    0.000000] Call trace:
> [    0.000000]  arm64_memblock_init+0x158/0x244
> [    0.000000]  setup_arch+0x164/0x1cc
> [    0.000000]  start_kernel+0x94/0x4ac
> [    0.000000]  __primary_switched+0xb4/0xbc
> [    0.000000] ---[ end trace 0000000000000000 ]---
> [    0.000000] Zone ranges:
> [    0.000000]   DMA      [mem 0x0000001000000000-0x0000001007ffffff]
>
> To fix it, we can change it to u64 type.
>
> Signed-off-by: Schspa Shi <schspa@gmail.com>
> ---
>  hw/arm/boot.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
>
> diff --git a/hw/arm/boot.c b/hw/arm/boot.c
> index 57efb61ee419..da719a4f8874 100644
> --- a/hw/arm/boot.c
> +++ b/hw/arm/boot.c
> @@ -638,14 +638,14 @@ int arm_load_dtb(hwaddr addr, const struct 
> arm_boot_info *binfo,
>      }
>  
>      if (binfo->initrd_size) {
> -        rc = qemu_fdt_setprop_cell(fdt, "/chosen", "linux,initrd-start",
> +        rc = qemu_fdt_setprop_u64(fdt, "/chosen", "linux,initrd-start",
>                                     binfo->initrd_start);
>          if (rc < 0) {
>              fprintf(stderr, "couldn't set /chosen/linux,initrd-start\n");
>              goto fail;
>          }
>  
> -        rc = qemu_fdt_setprop_cell(fdt, "/chosen", "linux,initrd-end",
> +        rc = qemu_fdt_setprop_u64(fdt, "/chosen", "linux,initrd-end",
>                                     binfo->initrd_start + binfo->initrd_size);
>          if (rc < 0) {
>              fprintf(stderr, "couldn't set /chosen/linux,initrd-end\n");

On the face of things this seems fine because unlike the other linux
properties they are not specified to be "expressed in #address-cells and
#size-cells" but I do wonder how we got into the situation where the
kernel and initrd ended up so high in the physical address space.

There is a whole comment in boot.c talking about keeping initrd within
lowmem:

    /*
     * We want to put the initrd far enough into RAM that when the
     * kernel is uncompressed it will not clobber the initrd. However
     * on boards without much RAM we must ensure that we still leave
     * enough room for a decent sized initrd, and on boards with large
     * amounts of RAM we must avoid the initrd being so far up in RAM
     * that it is outside lowmem and inaccessible to the kernel.
     * So for boards with less  than 256MB of RAM we put the initrd
     * halfway into RAM, and for boards with 256MB of RAM or more we put
     * the initrd at 128MB.
     * We also refuse to put the initrd somewhere that will definitely
     * overlay the kernel we just loaded, though for kernel formats which
     * don't tell us their exact size (eg self-decompressing 32-bit kernels)
     * we might still make a bad choice here.
     */

Is this just because the base RAM address of the board is outside of the
32 bit address range?

-- 
Alex Bennée