netdev - Re: [PATCH 5/8] platform/x86/intel/sdsi: Add in-band BIOS lock support

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Message-ID: <4135a390-22c9-f483-2aca-b89439dc682c@linux.intel.com>
Date: Thu, 8 Feb 2024 15:52:59 +0200 (EET)
From: Ilpo Järvinen <ilpo.jarvinen@...ux.intel.com>
To: "David E. Box" <david.e.box@...ux.intel.com>
cc: Netdev <netdev@...r.kernel.org>, 
    sathyanarayanan.kuppuswamy@...ux.intel.com, 
    LKML <linux-kernel@...r.kernel.org>, platform-driver-x86@...r.kernel.org
Subject: Re: [PATCH 5/8] platform/x86/intel/sdsi: Add in-band BIOS lock
 support

On Wed, 31 Jan 2024, David E. Box wrote:

> From: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>
> 
> As per SDSi in-band interface specification, sec titled "BIOS lock for
> in-band provisioning", when IB_LOCK bit is set in control qword, the
> SDSI agent is only allowed to perform the read flow, but not allowed to
> provision license blob or license key. So add check for it in
> sdsi_provision().
> 
> Signed-off-by: Kuppuswamy Sathyanarayanan <sathyanarayanan.kuppuswamy@...ux.intel.com>
> Signed-off-by: David E. Box <david.e.box@...ux.intel.com>
> ---
>  drivers/platform/x86/intel/sdsi.c | 10 ++++++++++
>  1 file changed, 10 insertions(+)
> 
> diff --git a/drivers/platform/x86/intel/sdsi.c b/drivers/platform/x86/intel/sdsi.c
> index 14821fee249c..287780fe65bb 100644
> --- a/drivers/platform/x86/intel/sdsi.c
> +++ b/drivers/platform/x86/intel/sdsi.c
> @@ -61,6 +61,7 @@
>  #define CTRL_OWNER			GENMASK(5, 4)
>  #define CTRL_COMPLETE			BIT(6)
>  #define CTRL_READY			BIT(7)
> +#define CTRL_INBAND_LOCK		BIT(32)
>  #define CTRL_STATUS			GENMASK(15, 8)
>  #define CTRL_PACKET_SIZE		GENMASK(31, 16)
>  #define CTRL_MSG_SIZE			GENMASK(63, 48)
> @@ -331,12 +332,21 @@ static int sdsi_mbox_read(struct sdsi_priv *priv, struct sdsi_mbox_info *info, s
>  	return sdsi_mbox_cmd_read(priv, info, data_size);
>  }
>  
> +static bool sdsi_ib_locked(struct sdsi_priv *priv)
> +{
> +	return !!FIELD_GET(CTRL_INBAND_LOCK, readq(priv->control_addr));
> +}
> +
>  static ssize_t sdsi_provision(struct sdsi_priv *priv, char *buf, size_t count,
>  			      enum sdsi_command command)
>  {
>  	struct sdsi_mbox_info info = {};
>  	int ret;
>  
> +	/* Make sure In-band lock is not set */
> +	if (sdsi_ib_locked(priv))
> +		return -EPERM;
> +
>  	if (count > (SDSI_SIZE_WRITE_MSG - SDSI_SIZE_CMD))
>  		return -EOVERFLOW;

Any reason why this order was chosen? I'd prefer these checks be other way 
around (a stupid caller providing too long count gets notified of its 
stupidity regardless of the locked state).

-- 
 i.