Re: kern/53189 (reproducable kernel assertion)

[maxv%NetBSD.org@localhost]

Synopsis: reproducable kernel assertion

State-Changed-From-To: open->analyzed
State-Changed-By: maxv%NetBSD.org@localhost
State-Changed-When: Tue, 17 Apr 2018 15:16:56 +0000
State-Changed-Why:
There are many places that actually put M_PKTHDR on secondary mbufs. In
your case, it was in tcp_build_datapkt().

I've disabled the check, and it will probably remain disabled. It is
difficult to know whether such chains can trigger bugs; they certainly
can, but only few places may be affected. So far, I identified only one
(IPsec), and I fixed it.

For now I believe the assumption should be:

	"M_PKTHDR should not be set on secondary mbufs. However,
	 everybody should be ready to handle this case."