[PATCH] efi_loader: EFI TCG2 free efi memory on protocol failure

[Heinrich Schuchardt <xypron.glpk@gmx.de>]

On 24.03.21 15:23, Ilias Apalodimas wrote:
> Current code doesn't free the efi allocated memory in case the protocol
> failed to install
>
> Fixes: c8d0fd582576 ("efi_loader: Introduce eventlog support for TCG2_PROTOCOL")
> Signed-off-by: Ilias Apalodimas <ilias.apalodimas@linaro.org>
> ---
>  lib/efi_loader/efi_tcg2.c | 21 +++++++++++++++++----
>  1 file changed, 17 insertions(+), 4 deletions(-)
>
> diff --git a/lib/efi_loader/efi_tcg2.c b/lib/efi_loader/efi_tcg2.c
> index 797d6eb134f6..02de63808f9a 100644
> --- a/lib/efi_loader/efi_tcg2.c
> +++ b/lib/efi_loader/efi_tcg2.c
> @@ -988,6 +988,7 @@ static efi_status_t create_final_event(void)
>
>  	return EFI_SUCCESS;
>  out:
> +	efi_free_pool(event_log.final_buffer);
>  	return ret;
>  }
>
> @@ -1041,8 +1042,12 @@ static efi_status_t efi_init_event_log(void)
>  	event_log.last_event_size = event_log.pos;
>
>  	ret = create_final_event();

Shouldn't create_final_event() should free event_log.final_buffer if
efi_install_configuration_table() fails?

> +	if (ret != EFI_SUCCESS)
> +		goto out;
>
> +	return EFI_SUCCESS;
>  out:
> +	efi_free_pool(event_log.buffer);

We must set event_log.buffer to NULL to avoid double free?

>  	return ret;
>  }
>
> @@ -1055,23 +1060,31 @@ out:
>   */
>  efi_status_t efi_tcg2_register(void)
>  {
> -	efi_status_t ret;
> +	efi_status_t ret = EFI_SUCCESS;
>  	struct udevice *dev;
>
>  	ret = platform_get_tpm2_device(&dev);
>  	if (ret != EFI_SUCCESS) {
>  		log_warning("Unable to find TPMv2 device\n");
> -		return EFI_SUCCESS;
> +		ret = EFI_SUCCESS;
> +		goto out;
>  	}
>
>  	ret = efi_init_event_log();
>  	if (ret != EFI_SUCCESS)
> -		return ret;
> +		goto fail;
>
>  	ret = efi_add_protocol(efi_root, &efi_guid_tcg2_protocol,
>  			       (void *)&efi_tcg2_protocol);
> -	if (ret != EFI_SUCCESS)
> +	if (ret != EFI_SUCCESS) {
>  		log_err("Cannot install EFI_TCG2_PROTOCOL\n");
> +		goto fail;
> +	}
>
> +out:
> +	return ret;
> +fail:

In create_final_event() you have installed a configuration table
EFI_TCG2_FINAL_EVENTS_TABLE_GUID. You should not free the memory of the
configuration table without uninstalling the table.

Should we create separate uninit function which takes care of the logic?

Best regards

Heinrich

> +	efi_free_pool(event_log.final_buffer);
> +	efi_free_pool(event_log.buffer);
>  	return ret;
>  }
>