Re: tcp.c: leverage MSG_PEEK with offset kernel capability when available

On Wed, 6 Dec 2023 15:59:40 +0100 Stefano Brivio wrote:

[...]

but on a kernel with your patch, I get ENOTCONN on recvmsg(). If I replace that by a simple recv():

sendto(5, "ab", 2, 0, NULL, 0) = 2 recvfrom(6, "ab", 10, 0, NULL, NULL) = 2

...so I don't think it's a fundamental issue with this approach, rather something with your patch, but I'm not yet sure what. :)

Oops, my bad, I got the order of fields in struct msghdr wrong. New version attached, this one works. -- Stefano