On 5/31/22 11:51, Jan Beulich wrote:
> On 31.05.2022 17:08, Daniel P. Smith wrote:
>> It is possible to select a few different build configurations that results in
>> the unnecessary walking of the boot module list looking for a policy module.
>> This specifically occurs when the flask policy is enabled but either the
>> dummy
>> or the SILO policy is selected as the enforcing policy. This is not ideal for
>> configurations like hyperlaunch and dom0less when there could be a number of
>> modules to be walked or doing an unnecessary device tree lookup.
>>
>> This patch introduces the policy_file_required flag for tracking when an XSM
>> policy module requires a policy file. Only when the policy_file_required flag
>> is set to true, will XSM search the boot modules for a policy file.
>>
>> Signed-off-by: Daniel P. Smith <dpsm...@apertussolutions.com>
>
> Looks technically okay, so
> Reviewed-by: Jan Beulich <jbeul...@suse.com>
> but couldn't you ...
>
>> @@ -148,7 +160,7 @@ int __init xsm_multiboot_init(
>>
>> printk("XSM Framework v" XSM_FRAMEWORK_VERSION " initialized\n");
>>
>> - if ( XSM_MAGIC )
>> + if ( policy_file_required && XSM_MAGIC )
>> {
>> ret = xsm_multiboot_policy_init(module_map, mbi, &policy_buffer,
>> &policy_size);
>> @@ -176,7 +188,7 @@ int __init xsm_dt_init(void)
>>
>> printk("XSM Framework v" XSM_FRAMEWORK_VERSION " initialized\n");
>>
>> - if ( XSM_MAGIC )
>> + if ( policy_file_required && XSM_MAGIC )
>> {
>> ret = xsm_dt_policy_init(&policy_buffer, &policy_size);
>> if ( ret )
>
> ... drop the two "&& XSM_MAGIC" here at this time? Afaict policy_file_required
> cannot be true when XSM_MAGIC is zero.
I was on the fence about this, as it should be rendered as redundant as
you point out. I am good with dropping on next spin.
v/r,
dps
