Re: [PATCH v8 3/6] kho: persist blob size in KHO FDT

[Pratyush Yadav <pratyush@kernel.org>]

On Mon, Mar 09 2026, Breno Leitao wrote:

> kho_add_subtree() accepts a size parameter but only forwards it to
> debugfs. The size is not persisted in the KHO FDT, so it is lost across
> kexec. This makes it impossible for the incoming kernel to determine the
> blob size without understanding the blob format.
>
> Store the blob size as a "blob-size" property in the KHO FDT alongside
> the "preserved-data" physical address. This allows the receiving kernel
> to recover the size for any blob regardless of format.
>
> Also extend kho_retrieve_subtree() with an optional size output
> parameter so callers can learn the blob size without needing to
> understand the blob format. Update all callers to pass NULL for the
> new parameter.
>
> Signed-off-by: Breno Leitao <leitao@debian.org>
> ---
[...]
> diff --git a/kernel/liveupdate/kexec_handover.c b/kernel/liveupdate/kexec_handover.c
> index 54fe59fe43acd..1f22705d5d246 100644
> --- a/kernel/liveupdate/kexec_handover.c
> +++ b/kernel/liveupdate/kexec_handover.c
> @@ -768,6 +768,7 @@ int kho_add_subtree(const char *name, void *blob, size_t size)
>  {
>  	phys_addr_t phys = virt_to_phys(blob);
>  	void *root_fdt = kho_out.fdt;
> +	u64 size_u64 = size;
>  	int err = -ENOMEM;
>  	int off, fdt_err;
>  
> @@ -784,11 +785,16 @@ int kho_add_subtree(const char *name, void *blob, size_t size)
>  		goto out_pack;
>  	}
>  
> -	err = fdt_setprop(root_fdt, off, KHO_FDT_SUB_TREE_PROP_NAME,
> +	err = fdt_setprop(root_fdt, off, KHO_SUB_TREE_PROP_NAME,
>  			  &phys, sizeof(phys));
>  	if (err < 0)
>  		goto out_pack;
>  
> +	err = fdt_setprop(root_fdt, off, KHO_SUB_TREE_SIZE_PROP_NAME,
> +			  &size_u64, sizeof(size_u64));
> +	if (err < 0)
> +		goto out_pack;
> +

I noticed that the error handling here is a bit broken. We open the
subnode for the subtree, but then if we fail to add the "preserved-data"
property, we don't remove the subnode. So the next kernel gets an
invalid FDT (per KHO ABI) and might as well refuse to parse it.

Similarly here, the FDT might also be missing the size and then the next
kernel might reject the FDT.

Also, we directly return the FDT error code to the caller, which
wouldn't make sense since it probably expects -errno.

Not something this patchset has to fix, but I am pointing this out in
case someone (possibly also future me) is interested in fixing this up.

>  	WARN_ON_ONCE(kho_debugfs_blob_add(&kho_out.dbg, name, blob,
>  					  size, false));
>  
> @@ -817,7 +823,7 @@ void kho_remove_subtree(void *blob)
>  		const u64 *val;
>  		int len;
>  
> -		val = fdt_getprop(root_fdt, off, KHO_FDT_SUB_TREE_PROP_NAME, &len);
> +		val = fdt_getprop(root_fdt, off, KHO_SUB_TREE_PROP_NAME, &len);
>  		if (!val || len != sizeof(phys_addr_t))
>  			continue;
>  
> @@ -1314,13 +1320,14 @@ EXPORT_SYMBOL_GPL(is_kho_boot);
>   * kho_retrieve_subtree - retrieve a preserved sub blob by its name.
>   * @name: the name of the sub blob passed to kho_add_subtree().
>   * @phys: if found, the physical address of the sub blob is stored in @phys.
> + * @size: if not NULL and found, the size of the sub blob is stored in @size.
>   *
>   * Retrieve a preserved sub blob named @name and store its physical
> - * address in @phys.
> + * address in @phys and optionally its size in @size.
>   *
>   * Return: 0 on success, error code on failure
>   */
> -int kho_retrieve_subtree(const char *name, phys_addr_t *phys)
> +int kho_retrieve_subtree(const char *name, phys_addr_t *phys, size_t *size)
>  {
>  	const void *fdt = kho_get_fdt();
>  	const u64 *val;
> @@ -1336,12 +1343,21 @@ int kho_retrieve_subtree(const char *name, phys_addr_t *phys)
>  	if (offset < 0)
>  		return -ENOENT;
>  
> -	val = fdt_getprop(fdt, offset, KHO_FDT_SUB_TREE_PROP_NAME, &len);
> +	val = fdt_getprop(fdt, offset, KHO_SUB_TREE_PROP_NAME, &len);
>  	if (!val || len != sizeof(*val))
>  		return -EINVAL;
>  
>  	*phys = (phys_addr_t)*val;
>  
> +	if (size) {
> +		val = fdt_getprop(fdt, offset, KHO_SUB_TREE_SIZE_PROP_NAME,
> +				  &len);
> +		if (val && len == sizeof(*val))
> +			*size = (size_t)*val;
> +		else
> +			*size = 0;

If the size property is invalid, is it a good idea to ignore it? Should
we instead consider the subnode to be broken and reject it entirely with
an error message? Because if a caller expects a blob of 16 bytes but
gets one with 0 bytes, it will likely error out anyway.

> +	}
> +
>  	return 0;
>  }
>  EXPORT_SYMBOL_GPL(kho_retrieve_subtree);
[...]

-- 
Regards,
Pratyush Yadav