atf: refer to `common.mk` using `SRCTOP`
This helps make the dependent Makefile (and the values contained within)
easier to catch when upgrading atf in the future.
MFC after: 1 week
(cherry picked from commit 36c4912749ab1d12692302386050e25ce8b1d9de)
atf: refer to `common.mk` using `SRCTOP`
This helps make the dependent Makefile (and the values contained within)
easier to catch when upgrading atf in the future.
MFC after: 1 week
(cherry picked from commit 36c4912749ab1d12692302386050e25ce8b1d9de)
src.opts.mk: force the Dtrace tests off if ${MK_DTRACE} == no
Prior to this change Dtrace's tests could be enabled even if
`${MK_DTRACE}` == no. Allowing this doesn't make sense so remove the
tests if `${MK_DTRACE}` is also disabled.
MFC after: 2 weeks
(cherry picked from commit f6db138f97b78e641b07a7c3695ccf62b0f42f28)
ip6_forward: Do not reference the nexthop
ip6_forward() looked the route up with NHR_REF and released the nexthop
before returning. It runs in the network epoch and uses the nexthop
only until it returns: nd6_output_ifp() gets the interface, and
icmp6_redirect_output() runs before the return. The epoch keeps the
nexthop valid that long, which ip6_tryforward() already relies on.
Every packet that takes this path wrote twice to the nexthop reference
count, which all CPUs forwarding to that nexthop share. Forwarded
packets take it whenever IPsec has a security policy, and today also
whenever net.inet6.ip6.redirect is set, which is the default.
Look the nexthop up without a reference, and assert that the caller is
in the network epoch.
No functional change intended.
Reviewed by: pouria
[3 lines not shown]
iwx: unbreak monitor mode support
A bunch of changes that turns it into a working mode (at least for me)
1) Call iwx_auth instead of iwx_run on -> RUN transition in monitor mode.
That's how it is intended to work because we never transition to the
AUTH state in this mode and current implementation of iwx_run just
results in endless 'firmware error' -> 'restart' loop.
2) Use ic_curchan instead of ic_bsschan while initializing NIC state;
the last one contains ANYC channel in monitor mode.
3) Do not print 'input_all returned -1' message - ieee80211_input_mimo_all
always returns -1 in monitor mode.
4) Always set IEEE80211_RADIOTAP_F_FCS in radiotap flags while in monitor mode.
While header documentation states it is controlled by IWX_MAC_FILTER_IN_CRC32
flag in practice IWX_MAC_FILTER_IN_PROMISC appends it as well - and it is
required since this is the only flag that allows to receive data and
control frames. While it alone is enough I left other flags in place for
smaller diff with other drivers.
[6 lines not shown]