RE: [PATCH v2 1/3] crypto: permit users to specify numa node of acomp hardware

-----Original Message-----
From: Herbert Xu [mailto:herbert@gondor.apana.org.au]
Sent: Friday, July 3, 2020 4:11 PM
To: Song Bao Hua (Barry Song) <redacted>
Cc: davem@davemloft.net; Wangzhou (B) <wangzhou1@hisilicon.com>;
Jonathan Cameron [off-list ref];
akpm@linux-foundation.org; linux-crypto@vger.kernel.org;
linux-mm@kvack.org; linux-kernel@vger.kernel.org; Linuxarm
[off-list ref]; Seth Jennings [off-list ref]; Dan
Streetman [off-list ref]; Vitaly Wool [off-list ref]
Subject: Re: [PATCH v2 1/3] crypto: permit users to specify numa node of
acomp hardware

On Tue, Jun 23, 2020 at 04:16:08PM +1200, Barry Song wrote:

quoted

-void *crypto_create_tfm(struct crypto_alg *alg,
-			const struct crypto_type *frontend)
+void *crypto_create_tfm_node(struct crypto_alg *alg,
+			const struct crypto_type *frontend,
+			int node)
 {
 	char *mem;
 	struct crypto_tfm *tfm = NULL;

@@ -451,6 +452,7 @@ void *crypto_create_tfm(struct crypto_alg *alg,

 	tfm = (struct crypto_tfm *)(mem + tfmsize);
 	tfm->__crt_alg = alg;
+	tfm->node = node;

Should the kzalloc also use node?

Yes, it would be nice since the tfm will mainly be accessed by CPU on the specific node.

Thanks,
--
Email: Herbert Xu [off-list ref]
Home Page: http://gondor.apana.org.au/~herbert/
PGP Key: http://gondor.apana.org.au/~herbert/pubkey.txt

Thanks
Barry