[PATCH 1/3] efi_loader: efi_tcg2_register returns appropriate error

[Ilias Apalodimas]

On Fri, Dec 03, 2021 at 12:58:13PM +0900, Masahisa Kojima wrote:
> This commit modify efi_tcg2_register() to return the
> appropriate error.
> With this fix, sandbox will not boot because efi_tcg2_register()
> fails due to some missing feature in GetCapabilities.
> So disable sandbox if EFI_TCG2_PROTOCOL is enabled.
> 
> UEFI secure boot variable measurement is not directly related
> to TCG2 protocol installation, tcg2_measure_secure_boot_variable()
> is moved to the separate function.
> 
> Signed-off-by: Masahisa Kojima <masahisa.kojima at linaro.org>
> ---
>  include/efi_loader.h       |  2 ++
>  lib/efi_loader/Kconfig     |  2 ++
>  lib/efi_loader/efi_setup.c |  2 ++
>  lib/efi_loader/efi_tcg2.c  | 65 +++++++++++++++++++++++++++-----------
>  4 files changed, 53 insertions(+), 18 deletions(-)
> 
> diff --git a/include/efi_loader.h b/include/efi_loader.h
> index 67c40ca57a..f4860e87fc 100644
> --- a/include/efi_loader.h
> +++ b/include/efi_loader.h
> @@ -525,6 +525,8 @@ efi_status_t efi_disk_register(void);
>  efi_status_t efi_rng_register(void);
>  /* Called by efi_init_obj_list() to install EFI_TCG2_PROTOCOL */
>  efi_status_t efi_tcg2_register(void);
> +/* Called by efi_init_obj_list() to do initial measurement */
> +efi_status_t efi_tcg2_do_initial_measurement(void);
>  /* measure the pe-coff image, extend PCR and add Event Log */
>  efi_status_t tcg2_measure_pe_image(void *efi, u64 efi_size,
>  				   struct efi_loaded_image_obj *handle,
> diff --git a/lib/efi_loader/Kconfig b/lib/efi_loader/Kconfig
> index 700dc838dd..24f9a2bb75 100644
> --- a/lib/efi_loader/Kconfig
> +++ b/lib/efi_loader/Kconfig
> @@ -308,6 +308,8 @@ config EFI_TCG2_PROTOCOL
>  	bool "EFI_TCG2_PROTOCOL support"
>  	default y
>  	depends on TPM_V2
> +	# Sandbox TPM currently fails on GetCapabilities needed for TCG2
> +	depends on !SANDBOX
>  	select SHA1
>  	select SHA256
>  	select SHA384
> diff --git a/lib/efi_loader/efi_setup.c b/lib/efi_loader/efi_setup.c
> index 1aba71cd96..f58a4afa7f 100644
> --- a/lib/efi_loader/efi_setup.c
> +++ b/lib/efi_loader/efi_setup.c
> @@ -241,6 +241,8 @@ efi_status_t efi_init_obj_list(void)
>  		ret = efi_tcg2_register();
>  		if (ret != EFI_SUCCESS)
>  			goto out;
> +
> +		efi_tcg2_do_initial_measurement();
>  	}
>  
>  	/* Secure boot */
> diff --git a/lib/efi_loader/efi_tcg2.c b/lib/efi_loader/efi_tcg2.c
> index 5f71b188a0..6dbdd35f29 100644
> --- a/lib/efi_loader/efi_tcg2.c
> +++ b/lib/efi_loader/efi_tcg2.c
> @@ -153,6 +153,15 @@ static u16 alg_to_len(u16 hash_alg)
>  	return 0;
>  }
>  
> +static bool is_tcg2_protocol_installed(void)
> +{
> +	struct efi_handler *handler;
> +	efi_status_t ret;
> +
> +	ret = efi_search_protocol(efi_root, &efi_guid_tcg2_protocol, &handler);
> +	return ret == EFI_SUCCESS;
> +}
> +
>  static u32 tcg_event_final_size(struct tpml_digest_values *digest_list)
>  {
>  	u32 len;
> @@ -1664,6 +1673,14 @@ void tcg2_uninit(void)
>  	event_log.buffer = NULL;
>  	efi_free_pool(event_log.final_buffer);
>  	event_log.final_buffer = NULL;
> +
> +	if (!is_tcg2_protocol_installed())
> +		return;
> +
> +	ret = efi_remove_protocol(efi_root, &efi_guid_tcg2_protocol,
> +				  (void *)&efi_tcg2_protocol);
> +	if (ret != EFI_SUCCESS)
> +		log_err("Failed to remove EFI TCG2 protocol\n");
>  }
>  
>  /**
> @@ -2345,12 +2362,37 @@ error:
>  	return ret;
>  }
>  
> +/**
> + * efi_tcg2_do_initial_measurement() - do initial measurement
> + *
> + * Return:	status code
> + */
> +efi_status_t efi_tcg2_do_initial_measurement(void)
> +{
> +	efi_status_t ret;
> +	struct udevice *dev;
> +
> +	if (!is_tcg2_protocol_installed())
> +		return EFI_SUCCESS;
> +
> +	ret = platform_get_tpm2_device(&dev);
> +	if (ret != EFI_SUCCESS)
> +		goto out;
> +

Would it make more sense to return a security violation here and treat this
error similarly to patch [3/3]?

> +	ret = tcg2_measure_secure_boot_variable(dev);
> +	if (ret != EFI_SUCCESS)
> +		goto out;
> +
> +out:
> +	return ret;
> +}
> +
>  /**
>   * efi_tcg2_register() - register EFI_TCG2_PROTOCOL
>   *
>   * If a TPM2 device is available, the TPM TCG2 Protocol is registered
>   *
> - * Return:	An error status is only returned if adding the protocol fails.
> + * Return:	status code
>   */
>  efi_status_t efi_tcg2_register(void)
>  {
> @@ -2373,8 +2415,10 @@ efi_status_t efi_tcg2_register(void)
>  	}
>  
>  	ret = efi_init_event_log();
> -	if (ret != EFI_SUCCESS)
> +	if (ret != EFI_SUCCESS) {
> +		tcg2_uninit();
>  		goto fail;
> +	}
>  
>  	ret = efi_add_protocol(efi_root, &efi_guid_tcg2_protocol,
>  			       (void *)&efi_tcg2_protocol);
> @@ -2391,24 +2435,9 @@ efi_status_t efi_tcg2_register(void)
>  		goto fail;
>  	}
>  
> -	ret = tcg2_measure_secure_boot_variable(dev);
> -	if (ret != EFI_SUCCESS) {
> -		tcg2_uninit();
> -		goto fail;
> -	}
> -
>  	return ret;
>  
>  fail:
>  	log_err("Cannot install EFI_TCG2_PROTOCOL\n");
> -	/*
> -	 * Return EFI_SUCCESS and don't stop the EFI subsystem.
> -	 * That's done for 2 reasons
> -	 * - If the protocol is not installed the PCRs won't be extended.  So
> -	 *   someone later in the boot flow will notice that and take the
> -	 *   necessary actions.
> -	 * - The TPM sandbox is limited and we won't be able to run any efi
> -	 *   related tests with TCG2 enabled
> -	 */
> -	return EFI_SUCCESS;
> +	return ret;
>  }
> -- 
> 2.17.1
> 

Cheers
/Ilias