2021-02-04 08:58:38

by Zheng Yongjun

Subject: [PATCH net-next] net: Return the correct errno code

When kzalloc failed, should return ENOMEM rather than ENOBUFS.

Signed-off-by: Zheng Yongjun <[email protected]>
---
net/decnet/dn_dev.c | 6 +++---
1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/net/decnet/dn_dev.c b/net/decnet/dn_dev.c
index 15d42353f1a3..50e375dcd5bd 100644
--- a/net/decnet/dn_dev.c
+++ b/net/decnet/dn_dev.c
@@ -469,7 +469,7 @@ int dn_dev_ioctl(unsigned int cmd, void __user *arg)
case SIOCSIFADDR:
if (!ifa) {
if ((ifa = dn_dev_alloc_ifa()) == NULL) {
- ret = -ENOBUFS;
+ ret = -ENOMEM;
break;
}
memcpy(ifa->ifa_label, dev->name, IFNAMSIZ);
@@ -645,7 +645,7 @@ static int dn_nl_newaddr(struct sk_buff *skb, struct nlmsghdr *nlh,
}

if ((ifa = dn_dev_alloc_ifa()) == NULL)
- return -ENOBUFS;
+ return -ENOMEM;

if (tb[IFA_ADDRESS] == NULL)
tb[IFA_ADDRESS] = tb[IFA_LOCAL];
@@ -1088,7 +1088,7 @@ static struct dn_dev *dn_dev_create(struct net_device *dev, int *err)
if (i == DN_DEV_LIST_SIZE)
return NULL;

- *err = -ENOBUFS;
+ *err = -ENOMEM;
if ((dn_db = kzalloc(sizeof(struct dn_dev), GFP_ATOMIC)) == NULL)
return NULL;

--
2.22.0

2021-02-04 19:22:18

by Jesse Brandeburg

Subject: Re: [PATCH net-next] net: Return the correct errno code

Zheng Yongjun wrote:

> When kzalloc failed, should return ENOMEM rather than ENOBUFS.

All these patches have the same subject and description, couldn't they
just be part of a single series with a good cover letter?

I'm not saying make them a single patch, because that is bad for
bisection, but having them as a single series means we review related
changes at one time, and can comment on them as a group.

2021-02-06 19:23:01

by Jakub Kicinski

Subject: Re: [PATCH net-next] net: Return the correct errno code

On Thu, 4 Feb 2021 11:14:26 -0800 Jesse Brandeburg wrote:
> Zheng Yongjun wrote:
>
> > When kzalloc failed, should return ENOMEM rather than ENOBUFS.
>
> All these patches have the same subject and description, couldn't they
> just be part of a single series with a good cover letter?

Agreed. The patches seem to be lacking clear justification.
Cover letter would be good.

> I'm not saying make them a single patch, because that is bad for
> bisection, but having them as a single series means we review related
> changes at one time, and can comment on them as a group.