[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <877e56ux2g.fsf@mpe.ellerman.id.au>
Date: Tue, 15 Oct 2019 21:23:19 +1100
From: Michael Ellerman <mpe@...erman.id.au>
To: Nayna Jain <nayna@...ux.ibm.com>, linuxppc-dev@...abs.org,
linux-efi@...r.kernel.org, linux-integrity@...r.kernel.org
Cc: linux-kernel@...r.kernel.org,
Benjamin Herrenschmidt <benh@...nel.crashing.org>,
Paul Mackerras <paulus@...ba.org>,
Ard Biesheuvel <ard.biesheuvel@...aro.org>,
Jeremy Kerr <jk@...abs.org>,
Matthew Garret <matthew.garret@...ula.com>,
Mimi Zohar <zohar@...ux.ibm.com>,
Greg Kroah-Hartman <gregkh@...uxfoundation.org>,
Claudio Carvalho <cclaudio@...ux.ibm.com>,
George Wilson <gcwilson@...ux.ibm.com>,
Elaine Palmer <erpalmer@...ibm.com>,
Eric Ricther <erichte@...ux.ibm.com>,
Oliver O'Halloran <oohall@...il.com>,
Nayna Jain <nayna@...ux.ibm.com>
Subject: Re: [PATCH v7 3/8] powerpc: detect the trusted boot state of the system
Nayna Jain <nayna@...ux.ibm.com> writes:
> PowerNV systems enables the IMA measurement rules only if the
> trusted boot is enabled on the system.
That confused me a lot. But the key is the distinction between appraisal
rules vs measurement rules, right?
I think it would be clearer if it was phrased as a positive statement, eg:
On PowerNV systems when trusted boot is enabled, additional IMA rules
are enabled to implement measurement.
Or something like that.
> This patch adds the function to detect if the system has trusted
> boot enabled.
It would probably help people to briefly explain the difference between
secure vs trusted boot.
> diff --git a/arch/powerpc/include/asm/secure_boot.h b/arch/powerpc/include/asm/secure_boot.h
> index 23d2ef2f1f7b..ecd08515e301 100644
> --- a/arch/powerpc/include/asm/secure_boot.h
> +++ b/arch/powerpc/include/asm/secure_boot.h
> @@ -12,6 +12,7 @@
>
> bool is_powerpc_os_secureboot_enabled(void);
> struct device_node *get_powerpc_os_sb_node(void);
> +bool is_powerpc_trustedboot_enabled(void);
>
> #else
>
> @@ -25,5 +26,10 @@ static inline struct device_node *get_powerpc_os_sb_node(void)
> return NULL;
> }
>
> +static inline bool is_powerpc_os_trustedboot_enabled(void)
That has an extra "_os" in it.
> +{
> + return false;
> +}
> +
> #endif
> #endif
> diff --git a/arch/powerpc/kernel/secure_boot.c b/arch/powerpc/kernel/secure_boot.c
> index 0488dbcab6b9..9d5ac1b39e46 100644
> --- a/arch/powerpc/kernel/secure_boot.c
> +++ b/arch/powerpc/kernel/secure_boot.c
> @@ -7,6 +7,27 @@
> #include <linux/of.h>
> #include <asm/secure_boot.h>
>
> +static const char * const fwsecureboot_compat[] = {
> + "ibm,secureboot-v1",
> + "ibm,secureboot-v2",
> + NULL,
> +};
> +
> +static struct device_node *get_powerpc_fw_sb_node(void)
> +{
> + struct device_node *node;
> + int i;
> +
> + for (i = 0; i < ARRAY_SIZE(fwsecureboot_compat); ++i) {
> + node = of_find_compatible_node(NULL, NULL,
> + fwsecureboot_compat[i]);
> + if (node)
> + return node;
> + }
> +
> + return NULL;
> +}
You shouldn't need to do that by hand, instead use
of_find_matching_node(), eg:
static struct device_node *get_powerpc_fw_sb_node(void)
{
static const struct of_device_id ids[] = {
{ .compatible = "ibm,secureboot-v1", },
{ .compatible = "ibm,secureboot-v2", },
{},
};
return of_find_matching_node(NULL, ids);
}
> @@ -40,3 +61,17 @@ bool is_powerpc_os_secureboot_enabled(void)
> pr_info("secureboot mode disabled\n");
> return false;
> }
> +
> +bool is_powerpc_trustedboot_enabled(void)
> +{
> + struct device_node *node;
> +
> + node = get_powerpc_fw_sb_node();
> + if (node && (of_find_property(node, "trusted-enabled", NULL))) {
Again this can use of_property_read_bool(), which copes with a NULL node
also, so just:
+ if (of_property_read_bool(node, "trusted-enabled"))) {
> + pr_info("trustedboot mode enabled\n");
> + return true;
> + }
> +
> + pr_info("trustedboot mode disabled\n");
> + return false;
> +}
> --
> 2.20.1
cheers
Powered by blists - more mailing lists