From: Matthew Dillon <dillon@apollo.backplane.com>
Date: Tue, 3 Jan 2012 17:18:28 +0000 (-0800)
Subject: kernel - Fix a case in if_re which could lockup the system
X-Git-Tag: v3.0.0~191
X-Git-Url: https://gitweb.dragonflybsd.org/dragonfly.git/commitdiff_plain/3c7587607776f143f9793ad05356ba5ac0941a0e?hp=e43987e0d95a259c9e35a3d70b8875d290b89af6

kernel - Fix a case in if_re which could lockup the system

* if_re's re_start function was not always properly setting IFF_OACTIVE
  in situations where packets would be left on the if_snd queue, causing
  if_start_dispatch() to get into an endless message loop.

* Add debugging to device_printf() when the case occurs.
---

diff --git a/sys/dev/netif/re/if_re.c b/sys/dev/netif/re/if_re.c
index 6231be43c8..34d7b9a9a4 100644
--- a/sys/dev/netif/re/if_re.c
+++ b/sys/dev/netif/re/if_re.c
@@ -2468,6 +2468,21 @@ re_start(struct ifnet *ifp)
 		ETHER_BPF_MTAP(ifp, m_head);
 	}
 
+	/*
+	 * If sc->re_ldata.re_tx_mbuf[idx] is not NULL it is possible
+	 * for IFF_OACTIVE to not be properly set when we also do not
+	 * have sufficient free tx descriptors, leaving packet in
+	 * ifp->if_send.  This can cause if_start_dispatch() to loop
+	 * infinitely so make sure IFF_OACTIVE is set properly.
+	 */
+	if (sc->re_ldata.re_tx_free <= RE_TXDESC_SPARE) {
+		if ((ifp->if_flags & IFF_OACTIVE) == 0) {
+			device_printf(sc->re_dev,
+				      "Debug: IFF_OACTIVE was not set when"
+				      " re_tx_free was below minimum!\n");
+			ifp->if_flags |= IFF_OACTIVE;
+		}
+	}
 	if (!need_trans)
 		return;