On 12/23/25 11:26, Stefano Garzarella wrote:
> On Tue, Dec 23, 2025 at 10:15:28AM +0100, Michal Luczaj wrote:
...
>> diff --git a/net/vmw_vsock/af_vsock.c b/net/vmw_vsock/af_vsock.c
>> index adcba1b7bf74..c093db8fec2d 100644
>> --- a/net/vmw_vsock/af_vsock.c
>> +++ b/net/vmw_vsock/af_vsock.c
>> @@ -1787,6 +1787,7 @@ static int vsock_accept(struct socket *sock, struct
>> socket *newsock,
>> } else {
>> newsock->state = SS_CONNECTED;
>> sock_graft(connected, newsock);
>> + set_bit(SOCK_CUSTOM_SOCKOPT, &newsock->flags);
>
> I was a bit confused about next lines calling set_bit on
> `connected->sk_socket->flags`, but after `sock_graft(connected,
> newsock)` they are equivalent.
>
> So, maybe I would move the new line before the sock_graft() call or use
> `connected->sk_socket->flags` if you want to keep it after it.
...
>> if (vsock_msgzerocopy_allow(vconnected->transport))
>> set_bit(SOCK_SUPPORT_ZC,
>> &connected->sk_socket->flags);
Hmm, isn't using both `connected->sk_socket->flags` and `newsock->flags` a
bit confusing? `connected->sk_socket->flags` feels unnecessary long to me.
So how about a not-so-minimal-patch to have
newsock->state = SS_CONNECTED;
set_bit(SOCK_CUSTOM_SOCKOPT, &newsock->flags);
if (vsock_msgzerocopy_allow(vconnected->transport))
set_bit(SOCK_SUPPORT_ZC, &newsock->flags);
sock_graft(connected, newsock);
?
