Borislav Petkov <bp@xxxxxxxxx> writes:
> On Thu, Mar 19, 2020 at 06:25:49PM +0100, Thomas Gleixner wrote:
>> TBH, I don't see how
>>
>> if (force_dma_decrypted(dev))
>> set_memory_encrypted((unsigned long)cpu_addr, 1 << page_order);
>>
>> makes more sense than the above. It's both non-sensical unless there is
>
> 9087c37584fb ("dma-direct: Force unencrypted DMA under SME for certain DMA masks")
Reading the changelog again...
I have to say that force_dma_unencrypted() makes way more sense in that
context than force_dma_decrypted(). It still wants a comment.
Linguistical semantics and correctness matters a lot. Consistency is
required as well, but not for the price of ambiguous wording.
Thanks,
tglx
