Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 23 Aug 2014 18:55:52 +0000 (UTC)
From:      Adrian Chadd <adrian@FreeBSD.org>
To:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   svn commit: r270430 - head/sys/dev/ath
Message-ID:  <201408231855.s7NItqOM082671@svn.freebsd.org>

Next in thread | Raw E-Mail | Index | Archive | Help
Author: adrian
Date: Sat Aug 23 18:55:51 2014
New Revision: 270430
URL: http://svnweb.freebsd.org/changeset/base/270430

Log:
  Shut down RX before TX - in theory, this should make the chip less likely
  to get upset.
  
  The Qualcomm Atheros reference design code goes through significant
  hacks to shut down RX before TX.  It doesn't even try do do it in the
  driver - it actually makes the DMA stop routines in the HAL shut down
  RX before shutting down TX.
  
  So, to make this work for chips that aren't the AR9380 and later, do
  it in the driver.  Shuffle the TX stop/drain HAL calls to be called
  *after* the RX stop HAL call.
  
  Tested:
  
  * AR5413 (STA)
  * AR5212 (STA)
  * AR5416 (STA)
  * AR9380 (STA)
  * AR9331 (AP)
  * AR9341 (AP)
  
  TODO:
  
  * test ar92xx series NIC and the AR5210/AR5211, in case there's something
    even odder about those.

Modified:
  head/sys/dev/ath/if_ath.c
  head/sys/dev/ath/if_ath_beacon.c

Modified: head/sys/dev/ath/if_ath.c
==============================================================================
--- head/sys/dev/ath/if_ath.c	Sat Aug 23 18:11:54 2014	(r270429)
+++ head/sys/dev/ath/if_ath.c	Sat Aug 23 18:55:51 2014	(r270430)
@@ -1599,9 +1599,9 @@ ath_vap_delete(struct ieee80211vap *vap)
 		 * the vap state by any frames pending on the tx queues.
 		 */
 		ath_hal_intrset(ah, 0);		/* disable interrupts */
-		ath_draintxq(sc, ATH_RESET_DEFAULT);		/* stop hw xmit side */
 		/* XXX Do all frames from all vaps/nodes need draining here? */
 		ath_stoprecv(sc, 1);		/* stop recv side */
+		ath_draintxq(sc, ATH_RESET_DEFAULT);		/* stop hw xmit side */
 	}
 
 	/* .. leave the hardware awake for now. */
@@ -2503,12 +2503,13 @@ ath_stop_locked(struct ifnet *ifp)
 			}
 			ath_hal_intrset(ah, 0);
 		}
-		ath_draintxq(sc, ATH_RESET_DEFAULT);
+		/* XXX we should stop RX regardless of whether it's valid */
 		if (!sc->sc_invalid) {
 			ath_stoprecv(sc, 1);
 			ath_hal_phydisable(ah);
 		} else
 			sc->sc_rxlink = NULL;
+		ath_draintxq(sc, ATH_RESET_DEFAULT);
 		ath_beacon_free(sc);	/* XXX not needed */
 	}
 
@@ -2710,13 +2711,6 @@ ath_reset(struct ifnet *ifp, ATH_RESET_T
 	ATH_PCU_UNLOCK(sc);
 
 	/*
-	 * Should now wait for pending TX/RX to complete
-	 * and block future ones from occuring. This needs to be
-	 * done before the TX queue is drained.
-	 */
-	ath_draintxq(sc, reset_type);	/* stop xmit side */
-
-	/*
 	 * Regardless of whether we're doing a no-loss flush or
 	 * not, stop the PCU and handle what's in the RX queue.
 	 * That way frames aren't dropped which shouldn't be.
@@ -2724,6 +2718,13 @@ ath_reset(struct ifnet *ifp, ATH_RESET_T
 	ath_stoprecv(sc, (reset_type != ATH_RESET_NOLOSS));
 	ath_rx_flush(sc);
 
+	/*
+	 * Should now wait for pending TX/RX to complete
+	 * and block future ones from occuring. This needs to be
+	 * done before the TX queue is drained.
+	 */
+	ath_draintxq(sc, reset_type);	/* stop xmit side */
+
 	ath_settkipmic(sc);		/* configure TKIP MIC handling */
 	/* NB: indicate channel change so we do a full reset */
 	ath_update_chainmasks(sc, ic->ic_curchan);

Modified: head/sys/dev/ath/if_ath_beacon.c
==============================================================================
--- head/sys/dev/ath/if_ath_beacon.c	Sat Aug 23 18:11:54 2014	(r270429)
+++ head/sys/dev/ath/if_ath_beacon.c	Sat Aug 23 18:55:51 2014	(r270430)
@@ -749,6 +749,11 @@ ath_beacon_generate(struct ath_softc *sc
 			 *
 			 * More thought is required here.
 			 */
+			/*
+			 * XXX can we even stop TX DMA here? Check what the
+			 * reference driver does for cabq for beacons, given
+			 * that stopping TX requires RX is paused.
+			 */
 			ath_tx_draintxq(sc, cabq);
 		}
 	}



Want to link to this message? Use this URL: <http://docs.FreeBSD.org/cgi/mid.cgi?201408231855.s7NItqOM082671>