RE: [PATCH] firmware: dmi: Stop decoding on broken entry

[Michael Kelley]

From: Jean Delvare <jdelvare@xxxxxxx> Sent: Wednesday, April 17, 2024 8:34 AM
> 
> If a DMI table entry is shorter than 4 bytes, it is invalid. Due to
> how DMI table parsing works, it is impossible to safely recover from
> such an error, so we have to stop decoding the table.
> 
> Signed-off-by: Jean Delvare <jdelvare@xxxxxxx>
> Link: https://lore.kernel.org/linux-kernel/Zh2K3-HLXOesT_vZ@liuwe-devbox-debian-v2/T/
> ---
> Michael, can you please test this patch and confirm that it prevents
> the early oops?
> 
> The root cause of the DMI table corruption still needs to be
> investigated.
> 
>  drivers/firmware/dmi_scan.c |   11 +++++++++++
>  1 file changed, 11 insertions(+)
> 
> --- linux-6.8.orig/drivers/firmware/dmi_scan.c
> +++ linux-6.8/drivers/firmware/dmi_scan.c
> @@ -102,6 +102,17 @@ static void dmi_decode_table(u8 *buf,
>  		const struct dmi_header *dm = (const struct dmi_header *)data;
> 
>  		/*
> +		 * If a short entry is found (less than 4 bytes), not only it
> +		 * is invalid, but we cannot reliably locate the next entry.
> +		 */
> +		if (dm->length < sizeof(struct dmi_header)) {
> +			pr_warn(FW_BUG
> +				"Corrupted DMI table (only %d entries processed)\n",
> +				i);

It would be useful to also output the three header fields: type, handle, and length,
and perhaps also the offset of the header in the DMI blob (i.e., "data - buf").
When looking at the error reported by user space dmidecode, the first thing
I did was add those fields to the error message.

Michael 

> +			break;
> +		}
> +
> +		/*
>  		 *  We want to know the total length (formatted area and
>  		 *  strings) before decoding to make sure we won't run off the
>  		 *  table in dmi_decode or dmi_string
> 
> --
> Jean Delvare
> SUSE L3 Support