[U-Boot] [PATCH] armv8: aarch64: Fix the warning about x1-x3 nonzero issue

[Alexander Graf]

On 01/13/2017 10:36 AM, Alison Wang wrote:
>> On 01/13/2017 09:23 AM, Alison Wang wrote:
>>>>> Am 13.01.2017 um 08:50 schrieb Alison Wang <b18965 at freescale.com>:
>>>>>
>>>>> For 64-bit kernel, there is a warning about x1-x3 nonzero in
>>>> violation
>>>>> of boot protocol. x3 should be reset to zero before jumping to the
>>>>> kernel.
>>>>>
>>>>> This patch will adjust the parameters to transfer and make sure x3
>>>>> is zero.
>>>>>
>>>>> Signed-off-by: Alison Wang <alison.wang at nxp.com>
>>>> If x3 is part of the boot protocol, please treat it as function
>>>> argument rather than hard code it to 0.
>>>>
>>> [Alison Wang] I think x3 need to set to 0. Refer to the codes for the
>>> warning in kernel,
>>>
>>>           if (boot_args[1] || boot_args[2] || boot_args[3]) {
>>>                   pr_err("WARNING: x1-x3 nonzero in violation of boot
>> protocol:\n"
>> "\tx1: %016llx\n\tx2: %016llx\n\tx3: %016llx\n"
>>>                           "This indicates a broken bootloader or old
>> kernel\n",
>>>                           boot_args[1], boot_args[2], boot_args[3]);
>>>           }
>> Yes, that merely means that the kernel doesn't take 3 arguments, it
>> takes 4. Today args[1..3] have to be 0, but I don't want to iron that
>> out in u-boot code.
>>
>> So in pseudo-code, what we do today is we call
>>
>>     entry(dt_addr, 0, 0);
>>
>> and instead we should be calling
>>
>>     entry(dt_addr, 0, 0, 0);
>>
>> So in U-Boot code that means we should change the prototype for our
>> switch function from
>>
>> void armv8_switch_to_el2(u64 args, u64 mach_nr, u64 fdt_addr,
>>                            u64 entry_point, u64 es_flag);
>>
>> to
>>
>> void armv8_switch_to_el2(u64 args, u64 mach_nr, u64 fdt_addr,
>>                            u64 arg4, u64 entry_point, u64 es_flag);
>>
>> and adapt the macros accordingly to use x5/x6 instead of x4/5.
>>
> [Alison Wang] I understand your meaning. Actually, I thought that method too. Well, my concern is that there are more changes. Not only armv8_switch_to_el2 and armv8_switch_to_el1, all the places calling them need to change too. It may cause more problems. With this concern, I choose a simple way to fix this issue before.

I think there'll be enough people testing Linux boot :), let's try to 
fix it properly right away.


Alex