diff mbox series

[v2,3/9] crypto: aead - Add ctx helpers with DMA alignment

Message ID E1oyQS1-000djS-65@formenos.hmeau.com (mailing list archive)
State Accepted
Delegated to: Herbert Xu
Headers show
Series crypto: Add helpers for allocating with DMA alignment | expand

Commit Message

Herbert Xu Nov. 25, 2022, 4:36 a.m. UTC
This patch adds helpers to access the aead context structure and
request context structure with an added alignment for DMA access.

Signed-off-by: Herbert Xu <herbert@gondor.apana.org.au>
---

 include/crypto/internal/aead.h |   22 ++++++++++++++++++++++
 1 file changed, 22 insertions(+)

Comments

Catalin Marinas Nov. 25, 2022, 12:24 p.m. UTC | #1
On Fri, Nov 25, 2022 at 12:36:33PM +0800, Herbert Xu wrote:
> diff --git a/include/crypto/internal/aead.h b/include/crypto/internal/aead.h
> index d482017f3e20..cd8cb1e921b7 100644
> --- a/include/crypto/internal/aead.h
> +++ b/include/crypto/internal/aead.h
> @@ -39,6 +39,11 @@ static inline void *crypto_aead_ctx(struct crypto_aead *tfm)
>  	return crypto_tfm_ctx(&tfm->base);
>  }
>  
> +static inline void *crypto_aead_ctx_dma(struct crypto_aead *tfm)
> +{
> +	return crypto_tfm_ctx_dma(&tfm->base);
> +}

I wonder, if drivers end up calling this, isn't it easier to do the
alignment in crypto_aead_ctx() directly? There are over 300 callers (not
sure about the padding though, it may need still driver changes unless
we can add it to something like crypto_tfm_alg_alignmask()). Or is the
expectation that not all drivers need a DMA alignment?
Herbert Xu Nov. 28, 2022, 4:06 a.m. UTC | #2
On Fri, Nov 25, 2022 at 12:24:31PM +0000, Catalin Marinas wrote:
>
> I wonder, if drivers end up calling this, isn't it easier to do the
> alignment in crypto_aead_ctx() directly? There are over 300 callers (not
> sure about the padding though, it may need still driver changes unless
> we can add it to something like crypto_tfm_alg_alignmask()). Or is the
> expectation that not all drivers need a DMA alignment?

I'm only changing the specific drivers that map things from the ctx
structures for DMA.  So it's not as bad as a simple grep would suggest.

Cheers,
diff mbox series

Patch

diff --git a/include/crypto/internal/aead.h b/include/crypto/internal/aead.h
index d482017f3e20..cd8cb1e921b7 100644
--- a/include/crypto/internal/aead.h
+++ b/include/crypto/internal/aead.h
@@ -39,6 +39,11 @@  static inline void *crypto_aead_ctx(struct crypto_aead *tfm)
 	return crypto_tfm_ctx(&tfm->base);
 }
 
+static inline void *crypto_aead_ctx_dma(struct crypto_aead *tfm)
+{
+	return crypto_tfm_ctx_dma(&tfm->base);
+}
+
 static inline struct crypto_instance *aead_crypto_instance(
 	struct aead_instance *inst)
 {
@@ -65,6 +70,16 @@  static inline void *aead_request_ctx(struct aead_request *req)
 	return req->__ctx;
 }
 
+static inline void *aead_request_ctx_dma(struct aead_request *req)
+{
+	unsigned int align = crypto_dma_align();
+
+	if (align <= crypto_tfm_ctx_alignment())
+		align = 1;
+
+	return PTR_ALIGN(aead_request_ctx(req), align);
+}
+
 static inline void aead_request_complete(struct aead_request *req, int err)
 {
 	req->base.complete(&req->base, err);
@@ -108,6 +123,13 @@  static inline void crypto_aead_set_reqsize(struct crypto_aead *aead,
 	aead->reqsize = reqsize;
 }
 
+static inline void crypto_aead_set_reqsize_dma(struct crypto_aead *aead,
+					       unsigned int reqsize)
+{
+	reqsize += crypto_dma_align() & ~(crypto_tfm_ctx_alignment() - 1);
+	aead->reqsize = reqsize;
+}
+
 static inline void aead_init_queue(struct aead_queue *queue,
 				   unsigned int max_qlen)
 {