On 10.12.2024 15:45, Frediano Ziglio wrote:
> On Tue, Dec 10, 2024 at 10:38 AM Jan Beulich <jbeul...@suse.com> wrote:
>> On 22.11.2024 10:33, Frediano Ziglio wrote:
>>> switch ( magic )
>>> {
>>> case MULTIBOOT_BOOTLOADER_MAGIC:
>>> - return mbi_reloc(in, &ctx);
>>> + res = mbi_reloc(in, &ctx);
>>> + break;
>>>
>>> case MULTIBOOT2_BOOTLOADER_MAGIC:
>>> - return mbi2_reloc(in, &ctx);
>>> + res = mbi2_reloc(in, &ctx);
>>> + break;
>>>
>>> case XEN_HVM_START_MAGIC_VALUE:
>>> if ( IS_ENABLED(CONFIG_PVH_GUEST) )
>>> - return pvh_info_reloc(in, &ctx);
>>> + {
>>> + res = pvh_info_reloc(in, &ctx);
>>> + break;
>>> + }
>>> /* Fallthrough */
>>>
>>> default:
>>> /* Nothing we can do */
>>> - return NULL;
>>> + res = NULL;
>>
>> Simply keep returning here? No need to write the NULL when the variables
>> start out zeroed?
>>
>
> Yes, considering pvh_start_info_pa and multiboot_ptr should be already
> NULL it makes sense
>
>>> }
>>> +
>>> +#ifdef CONFIG_PVH_GUEST
>>> + if ( pvh_boot )
>>> + pvh_start_info_pa = (unsigned long)res;
>>> +#endif
>>> +
>>> + multiboot_ptr = (unsigned long)res;
>>
>> In the assembly original this is an "else" to the if().
>>
>
> I suppose the return change above would solve also this.
I'm not convinced of this.
Jan
