[U-Boot] [PATCH V4 2/2] bootm: Handle kernel_noload on arm64

[Bin Chen]

On 13 June 2018 at 14:13, Marek Vasut <marek.vasut at gmail.com> wrote:

> The ARM64 has 2 MiB alignment requirement for the kernel. When using
> fitImage, this requirement may by violated, the kernel will thus be
> executed from unaligned address and fail to boot. Do what booti does
> and run booti_setup() for kernel_noload images on arm64 to obtain a
> suitable aligned address to which the image shall be relocated.
>
> Signed-off-by: Marek Vasut <marek.vasut+renesas at gmail.com>
> Cc: Bin Chen <bin.chen at linaro.org>
> Cc: Masahiro Yamada <yamada.masahiro at socionext.com>
> Cc: Tom Rini <trini at konsulko.com>
> ---
> V2: Protect the ARM64 booti bit with if IS_ENABLED(CMD_BOOTI)
> V3: Use if() instead of #ifdef
> V4: Switch force_reloc to bool
> ---
>  common/bootm.c | 19 +++++++++++++++++--
>  1 file changed, 17 insertions(+), 2 deletions(-)
>
> diff --git a/common/bootm.c b/common/bootm.c
> index e789f6818a..e517d9f118 100644
> --- a/common/bootm.c
> +++ b/common/bootm.c
> @@ -202,8 +202,23 @@ static int bootm_find_os(cmd_tbl_t *cmdtp, int flag,
> int argc,
>         }
>
>         if (images.os.type == IH_TYPE_KERNEL_NOLOAD) {
> -               images.os.load = images.os.image_start;
> -               images.ep += images.os.load;
> +               if (CONFIG_IS_ENABLED(CMD_BOOTI) &&
> +                   images.os.arch == IH_ARCH_ARM64) {
> +                       ulong image_addr;
> +                       ulong image_size;
> +
> +                       ret = booti_setup(images.os.image_start,
> &image_addr,
> +                                         &image_size, true);
>

Is it guaranteed to be conflict free (with other images) by always moving
the kernel to the start of the RAM?



> +                       if (ret != 0)
> +                               return 1;
>

Do you think it helps to add some debug/error message here as in other path
that returns 1?


> +
> +                       images.os.type = IH_TYPE_KERNEL;
> +                       images.os.load = image_addr;
> +                       images.ep = image_addr;
> +               } else {
> +                       images.os.load = images.os.image_start;
> +                       images.ep += images.os.image_start;
>

I know this is same the orinigal code. I'm not famaliar the history/purpse
of type
IH_TYPE_KERNEL_NOLOAD (compared with IH_TYPE_KERNEL), so just for my
understanding,
why in this case we have to increment the images.ep by images.os.load, not
just set images.ep to
images.os.load.


+               }
>         }
>
>         images.os.start = map_to_sysmem(os_hdr);
> --
> 2.17.1
>
>


-- 
Regards,
Bin