Re: [PATCH v6 050/103] tests/tcg/multiarch/sha1.c: fix big endian implementation

On 28/7/26 09:56, Pierrick Bouvier wrote:
> On 7/28/26 12:36 AM, Philippe Mathieu-Daudé wrote:
>> On 28/7/26 06:41, Pierrick Bouvier wrote:
>>> All big endian targets were returning a wrong value due to this.
>>>
>>> Tested-by: Alex Bennée <[email protected]>
>>> Signed-off-by: Pierrick Bouvier <[email protected]>
>>> ---
>>>    tests/tcg/multiarch/sha1.c | 4 ++--
>>>    1 file changed, 2 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/tests/tcg/multiarch/sha1.c b/tests/tcg/multiarch/sha1.c
>>> index 0081bd76577..6cebad2951d 100644
>>> --- a/tests/tcg/multiarch/sha1.c
>>> +++ b/tests/tcg/multiarch/sha1.c
>>> @@ -48,10 +48,10 @@ void SHA1Final(unsigned char digest[20], 
>>> SHA1_CTX* context);
>>>    /* blk0() and blk() perform the initial expand. */
>>>    /* I got the idea of expanding during the round function from 
>>> SSLeay */
>>> -#if BYTE_ORDER == LITTLE_ENDIAN
>>> +#if __BYTE_ORDER__ == __ORDER_LITTLE_ENDIAN__
>>>    #define blk0(i) (block->l[i] = (rol(block->l[i],24)&0xFF00FF00) \
>>>        |(rol(block->l[i],8)&0x00FF00FF))
>>> -#elif BYTE_ORDER == BIG_ENDIAN
>>> +#elif __BYTE_ORDER__ == __ORDER_BIG_ENDIAN__
>>>    #define blk0(i) block->l[i]
>>>    #else
>>>    #error "Endianness not defined!"
>>
>> Orthogonal but what about tests/tcg/multiarch/sigbus.c?
> 
> That's a good point.
> Taking a look, the original issue with the current test is that #include 
> <endian.h> was missing (for some wrong reason, I thought it was defined 
> by qemu in compiler.h), letting the defines undefined.
> sigbus is correct since it has the right include.
> I'll change the current fix to just add the right header.

Then simply including <endian.h> in sha1.c:
Reviewed-by: Philippe Mathieu-Daudé <[email protected]>