Re: git: 4fadf2466468 - main - tcp_close: Use in6_pcbdisconnect for INET6 sockets
- In reply to: John Baldwin : "Re: git: 4fadf2466468 - main - tcp_close: Use in6_pcbdisconnect for INET6 sockets"
- Go to: [ bottom of page ] [ top of archives ] [ this month ]
Date: Tue, 14 Apr 2026 19:53:50 UTC
On Tue, Apr 14, 2026 at 03:42:03PM -0400, John Baldwin wrote:
J> > J> URL: https://cgit.FreeBSD.org/src/commit/?id=4fadf2466468dd6dcb6cf9e3739ed696a18c1bb4
J> > J>
J> > J> commit 4fadf2466468dd6dcb6cf9e3739ed696a18c1bb4
J> > J> Author: John Baldwin <jhb@FreeBSD.org>
J> > J> AuthorDate: 2026-04-14 18:07:51 +0000
J> > J> Commit: John Baldwin <jhb@FreeBSD.org>
J> > J> CommitDate: 2026-04-14 18:07:51 +0000
J> > J>
J> > J> tcp_close: Use in6_pcbdisconnect for INET6 sockets
J> > J>
J> > J> This also fixes the LINT-NOINET builds.
J> > J>
J> > J> Fixes: 40dbb06fa73c ("inpcb: retire INP_DROPPED and in_pcbdrop()")
J> >
J> > Thanks! I wonder how did it pass the tests. Looks like we don't have
J> > any UDPv6 tests?
J>
J> If you look at in_pcbdisconnect and in6_pcbdisconnect they are mostly the same and
J> given that it was a close operation about to discard the pcb anyway, the differences
J> probably didn't matter in practice. I only noticed due to make tinderbox being broken.
Indeed, since after this path socket can't be reused assignment to IPv4 part of the
union was harmless for IPv6.
--
Gleb Smirnoff