Re: [PATCH v11 01/21] target/s390x: Fix wrong address handling in address loops

[Harald Freudenberger <freude@linux.ibm.com>]

On 2026-07-08 13:18, Michael Tokarev wrote:
> On 08.07.2026 13:37, Michael Tokarev wrote:
>> On 07.07.2026 19:17, Harald Freudenberger wrote:
>>> The loop increments addr by the element stride (+= 4) before calling
>>> wrap_address, but then overwrites the loop addr with the wrapped
>>> value. On the next iteration the stride is applied to the wrapped
>>> address of the previous element, not to the original unwrapped
>>> address. This results in every element after the first is read from a
>>> wrong (wrapped) address.
>>> 
>>> Fixes: 9f17bfdab4 ("target/s390x: support SHA-512 extensions")
>>> Signed-off-by: Harald Freudenberger <freude@linux.ibm.com>
>>> ---
>>>   target/s390x/tcg/crypto_helper.c | 12 ++++--------
>>>   1 file changed, 4 insertions(+), 8 deletions(-)
>> 
>> This feels like qemu-stable material.
>> I'm picking this change up for currently active stable qemu series.
>> Please let me know if I shouldn't.
> 
> Um, this was a bit premature.
> 
> This particular commit requires a previous commit in this area,
> de96db79 "target/s390x: Compile crypto_helper.c as common unit".
> It's easy to back-port across, but since this change ("Fix wrong
> address handling...") hasn't been marked as for-stable, I'm *not*
> doing that.  If this change is actually needed in -stable, please
> let me know.  Also, is it okay to pick up the previous change
> ("compile as common unit") to -stable too, to avoid fixing up
> subsequent changes.
> 
> Thanks,
> 
> /mjt

Fine with me. However, Connie replied to my v10 series that she
wants to pick this patch. However, it is yet not on the qemu
master branch.