Re: [PATCH] hw/mips/malta: Fix the malta machine on big endian hosts

[Thomas Huth <thuth@redhat.com>]

On 30/03/2023 17.33, Peter Maydell wrote:
> On Thu, 30 Mar 2023 at 16:27, Thomas Huth <thuth@redhat.com> wrote:
>>
>> Booting a Linux kernel with the malta machine is currently broken
>> on big endian hosts. The cpu_to_gt32 macro wants to byteswap a value
>> for little endian targets only, but uses the wrong way to do this:
>> cpu_to_[lb]e32 works the other way round on big endian hosts! Fix
>> it by using the same ways on both, big and little endian hosts.
>>
>> Fixes: 0c8427baf0 ("hw/mips/malta: Use bootloader helper to set BAR registers")
>> Signed-off-by: Thomas Huth <thuth@redhat.com>
>> ---
>>   I've checked that both, the kernel from
>>   https://landley.net/toybox/downloads/binaries/mkroot/0.8.9/mipsel.tgz
>>   and the kernel from
>>   https://landley.net/toybox/downloads/binaries/mkroot/0.8.9/mips.tgz
>>   now boot fine on both, a little endian (x86) and a big endian (s390x) host.
>>
>>   hw/mips/malta.c | 4 ++--
>>   1 file changed, 2 insertions(+), 2 deletions(-)
>>
>> diff --git a/hw/mips/malta.c b/hw/mips/malta.c
>> index af9021316d..b26ed1fc9a 100644
>> --- a/hw/mips/malta.c
>> +++ b/hw/mips/malta.c
>> @@ -629,9 +629,9 @@ static void bl_setup_gt64120_jump_kernel(void **p, uint64_t run_addr,
>>
>>       /* Bus endianess is always reversed */
>>   #if TARGET_BIG_ENDIAN
>> -#define cpu_to_gt32 cpu_to_le32
>> +#define cpu_to_gt32(x) (x)
>>   #else
>> -#define cpu_to_gt32 cpu_to_be32
>> +#define cpu_to_gt32(x) bswap32(x)
>>   #endif
> 
> So if we:
>   * do nothing to the value on a BE host
>   * swap the value on an LE host
> 
> isn't that the same as cpu_to_be32() in both cases?

No, it's about the *target*, not the host:

* do nothing to the value for a BE *target*
* swap the value for LE *targets*

It's quite weird and it also took me a while to understand this, but this 
seems to be the way it's working right with all combinations.

  Thomas