Re: drivers/nvme/host/auth.c:950 nvme_auth_init_ctrl() warn: missing error code? 'ret'

From: Sagi Grimberg
Date: Sun Dec 25 2022 - 05:36:38 EST




On 12/23/22 17:47, Christoph Hellwig wrote:
Based on the code in nvme_auth_generate_key I assume this is intentional,
but the code looks really confusing.

Hannes, Sagi, what do you think of something like this:


diff --git a/drivers/nvme/common/auth.c b/drivers/nvme/common/auth.c
index d90e4f0c08b7b9..a07eb4cd9ce173 100644
--- a/drivers/nvme/common/auth.c
+++ b/drivers/nvme/common/auth.c
@@ -455,28 +455,18 @@ int nvme_auth_gen_shared_secret(struct crypto_kpp *dh_tfm,
}
EXPORT_SYMBOL_GPL(nvme_auth_gen_shared_secret);
-int nvme_auth_generate_key(u8 *secret, struct nvme_dhchap_key **ret_key)
+struct nvme_dhchap_key *nvme_auth_generate_key(u8 *secret)
{
- struct nvme_dhchap_key *key;
u8 key_hash;
- if (!secret) {
- *ret_key = NULL;
- return 0;
- }
+ if (!secret)
+ return NULL;
if (sscanf(secret, "DHHC-1:%hhd:%*s:", &key_hash) != 1)
- return -EINVAL;
+ return ERR_PTR(-EINVAL);
/* Pass in the secret without the 'DHHC-1:XX:' prefix */
- key = nvme_auth_extract_key(secret + 10, key_hash);
- if (IS_ERR(key)) {
- *ret_key = NULL;
- return PTR_ERR(key);
- }
-
- *ret_key = key;
- return 0;
+ return nvme_auth_extract_key(secret + 10, key_hash);
}
EXPORT_SYMBOL_GPL(nvme_auth_generate_key);
diff --git a/drivers/nvme/host/auth.c b/drivers/nvme/host/auth.c
index bb0abbe4491cdc..c808652966a94f 100644
--- a/drivers/nvme/host/auth.c
+++ b/drivers/nvme/host/auth.c
@@ -943,16 +943,19 @@ int nvme_auth_init_ctrl(struct nvme_ctrl *ctrl)
INIT_WORK(&ctrl->dhchap_auth_work, nvme_ctrl_auth_work);
if (!ctrl->opts)
return 0;
- ret = nvme_auth_generate_key(ctrl->opts->dhchap_secret,
- &ctrl->host_key);
- if (ret)
- return ret;
- ret = nvme_auth_generate_key(ctrl->opts->dhchap_ctrl_secret,
- &ctrl->ctrl_key);
- if (ret)
+
+ ctrl->host_key = nvme_auth_generate_key(ctrl->opts->dhchap_secret);
+ if (IS_ERR(ctrl->host_key)) {

nvme_auth_generate_key can return NULL, so in this case we should avoid
calling it if the secret is null here.

Other than that, this looks good.
Although I think that for this specific report, we should do a simple
fix and then make the code look better.