Re: [PATCH] crypto: mxs-dcp: handle zero-length skcipher requests

[Linmao Li]

在 2026/8/31 23:01, Frank Li 写道:
> On Mon, Aug 31, 2026 at 08:42:20PM +0800, Linmao Li wrote:
> > [You don't often get email from lilinmao@xxxxxxxxxx. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]
> >
> > 在 2026/8/29 3:25, Frank Li 写道:
> > > On Fri, Aug 28, 2026 at 05:32:09PM +0800, Linmao Li wrote:
> > > > [You don't often get email from lilinmao@xxxxxxxxxx. Learn why this is important at https://aka.ms/LearnAboutSenderIdentification ]
> > > >
> > > > Zero-length skcipher requests are valid no-ops, but MXS-DCP queues them.
> > > Any document show it? if return 0 when cryptelen, it is async, no
> > > call crypto_request_complete.
> >    Yes. CRYPTO_ALG_ASYNC means that the implementation may complete a
> >    request asynchronously, not that every request must do so.
> >
> >    A return value of 0 indicates synchronous completion.  Only
> >    -EINPROGRESS and -EBUSY cause crypto_wait_req() to wait for the
> >    completion callback.  The existing fallback path relies on the same
> >    semantics.
> >
> >    There is also a direct precedent in commit 8a4e047c6cc0
> >    ("crypto: marvell/cesa - Handle zero-length skcipher requests").
> >    Marvell CESA registers asynchronous skcipher algorithms but handles a
> >    zero-length request with:
> >
> >            if (!req->cryptlen)
> >                    return 0;
> Thanks, but still not answer my question,
>
> Any document show "Zero-length skcipher requests are valid no-ops", why
> not API direct return 0, instead of pass to every driver to check it.
>
> I suppose it should be checked before call driver's callback.
  You are right; I could not find an explicit Crypto API document defining
  the zero-length behavior, so the wording "valid no-ops" was too broad.

  With the current API, simply returning 0 in the core for every
  zero-length request would change XTS semantics, since the generic XTS
  implementation returns -EINVAL.  MXS-DCP only implements ECB and CBC
  here, whose generic implementations currently return 0.  Handling it in
  the driver therefore matches existing behavior; centralizing this would
  require algorithm-specific length information.

  I agree that the commit message should describe this as existing ECB/CBC
  behavior rather than a general skcipher API rule.

  Thanks,
  Linmao
> Frank
>
> > Thanks,
> > Linmao
> >
> > > Frank
> > >
> > > > When such a request reaches the worker, last_out_len remains zero. The
> > > > CBC completion path then subtracts AES_BLOCK_SIZE from this unsigned
> > > > value when updating the IV, causing the offset to underflow. On
> > > > decryption, the resulting source address precedes aes_in_buf.
> > > >
> > > > Return success before enqueueing zero-length requests. This avoids the
> > > > invalid source access and leaves the IV unchanged for a no-op.
> > > >
> > > > Fixes: fadd7a6e616b ("crypto: mxs-dcp - Fix AES issues")
> > > > Signed-off-by: Linmao Li <lilinmao@xxxxxxxxxx>
> > > > ---
> > > >    drivers/crypto/mxs-dcp.c | 3 +++
> > > >    1 file changed, 3 insertions(+)
> > > >
> > > > diff --git a/drivers/crypto/mxs-dcp.c b/drivers/crypto/mxs-dcp.c
> > > > index 133ebc998236..60794b4d49fa 100644
> > > > --- a/drivers/crypto/mxs-dcp.c
> > > > +++ b/drivers/crypto/mxs-dcp.c
> > > > @@ -473,6 +473,9 @@ static int mxs_dcp_aes_enqueue(struct skcipher_request *req, int enc, int ecb)
> > > >           struct dcp_aes_req_ctx *rctx = skcipher_request_ctx(req);
> > > >           int ret;
> > > >
> > > > +       if (!req->cryptlen)
> > > > +               return 0;
> > > > +
> > > >           if (unlikely(actx->key_len != AES_KEYSIZE_128 && !actx->key_referenced))
> > > >                   return mxs_dcp_block_fallback(req, enc);
> > > >
> > > >
> > > > base-commit: 8d3ae59288f1e7d58d76558a6ee96d533bc5019f
> > > > --
> > > > 2.25.1