Re: KCSAN: data-race in glue_cbc_decrypt_req_128bit / glue_cbc_decrypt_req_128bit

[Herbert Xu]

On Wed, Apr 01, 2020 at 09:20:28AM -0700, Eric Biggers wrote:
>
> The issue is that fixing it would require adding READ_ONCE() / WRITE_ONCE() in
> hundreds of different places, affecting most crypto-related .c files.

I don't think we should be doing that.  This is exactly the same
as using sendfile(2) and modifying the data during the send.  As
long as you don't trigger behaviours such as crashes or uncontrolled
execution then it's fine.  The output is simply undefined.

Cheers,
-- 
Email: Herbert Xu <herbert@xxxxxxxxxxxxxxxxxxx>
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt