Re: [PATCH v11 09/15] arm64: kexec_file: load initrd and device-tree

From: James Morse
Date: Tue Jul 17 2018 - 12:57:14 EST


Hi Akashi,

On 11/07/18 08:41, AKASHI Takahiro wrote:
> load_other_segments() is expected to allocate and place all the necessary
> memory segments other than kernel, including initrd and device-tree
> blob (and elf core header for crash).
> While most of the code was borrowed from kexec-tools' counterpart,
> users may not be allowed to specify dtb explicitly, instead, the dtb
> presented by the original boot loader is reused.
>
> arch_kimage_kernel_post_load_cleanup() is responsible for freeing arm64-
> specific data allocated in load_other_segments().

> diff --git a/arch/arm64/kernel/machine_kexec_file.c b/arch/arm64/kernel/machine_kexec_file.c
> index c38a8048ed00..ca00681c25c6 100644
> --- a/arch/arm64/kernel/machine_kexec_file.c
> +++ b/arch/arm64/kernel/machine_kexec_file.c

> +int arch_kimage_file_post_load_cleanup(struct kimage *image)
> +{
> + vfree(image->arch.dtb_buf);
> + image->arch.dtb_buf = NULL;
> +
> + return kexec_image_post_load_cleanup_default(image);
> +}

A nit from sparse:
| warning: symbol 'arch_kimage_file_post_load_cleanup' was not declared

Can we add a definition for this to a header file somewhere. asm/kexec.h is
probably the best bet.


> +static int setup_dtb(struct kimage *image,
> + unsigned long initrd_load_addr, unsigned long initrd_len,
> + char *cmdline, unsigned long cmdline_len,
> + char **dtb_buf, size_t *dtb_buf_len)
> +{

> + /* add initrd-* */
> + if (initrd_load_addr) {
> + value = cpu_to_fdt64(initrd_load_addr);
> + ret = fdt_setprop_u64(buf, nodeoffset, "linux,initrd-start",
> + value);

fdt_setprop_u64() already does the endian conversion.

>From scripts/dtc/libfdt/libfdt.h, its implemented as:
| fdt64_t tmp = cpu_to_fdt64(val);
| return fdt_setprop(fdt, nodeoffset, name, &tmp, sizeof(tmp));

(I think you were using setprop directly in an older version)


This leads to:
| ------------[ cut here ]------------
| initrd not fully accessible via the linear mapping -- please check your
| bootloader ...
| WARNING: CPU: 0 PID: 0 at ../arch/arm64/mm/init.c:429
| arm64_memblock_init+0x150/0x3d8
| Modules linked in:
| CPU: 0 PID: 0 Comm: swapper Not tainted 4.18.0-rc5-00015-g95b5c843d0da #10150
| Hardware name: AMD Seattle (Rev.B0) Development Board (Overdrive) (DT)
| pstate: 60000085 (nZCv daIf -PAN -UAO)
| pc : arm64_memblock_init+0x150/0x3d8
| lr : arm64_memblock_init+0x150/0x3d8

| Call trace:
| arm64_memblock_init+0x150/0x3d8
| setup_arch+0x1c0/0x510
| start_kernel+0x80/0x418
| random: get_random_bytes called from print_oops_end_marker+0x4c/0x68 with
| crng_init=0
| ---[ end trace 0000000000000000 ]---


Which is caused by the values being miles outside ram due to the extra byte
swapping:
| morse@frikadeller:~$ sudo dtc -I dtb -O dts /sys/firmware/fdt | grep initrd
| linux,initrd-end = <0x900b6c05 0x80000000>;
| linux,initrd-start = <0x906a04 0x80000000>;


With the two extra cpu_to_fdt64() calls removed:
Reviewed-by: James Morse <james.morse@xxxxxxx>


Thanks,

James