Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 1 Aug 2018 14:50:41 +0000 (UTC)
From:      Justin Hibbits <jhibbits@FreeBSD.org>
To:        src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   svn commit: r337043 - head/sys/dev/sound/pci/hda
Message-ID:  <201808011450.w71EofqW098435@repo.freebsd.org>

next in thread | raw e-mail | index | archive | help
Author: jhibbits
Date: Wed Aug  1 14:50:41 2018
New Revision: 337043
URL: https://svnweb.freebsd.org/changeset/base/337043

Log:
  snd_hda: Synchronize DMA buffers for the control path
  
  Make sure both sides of the DMA buffer memory accesses for the CORB and RIRB
  (control buffers) in snd_hda (device and CPU) can see coherent memory.  This
  is needed on weakly ordered architectures including PowerPC and ARM.  Patch
  originally by mmel, with small changes.
  
  This does not cover the data path of snd_hda.  We don't have sync operations
  for in-progress DMA buffers, to sync ranges of a map.
  
  Reviewed By: mmel
  Differential Revision: https://reviews.freebsd.org/D16517

Modified:
  head/sys/dev/sound/pci/hda/hdaa.c
  head/sys/dev/sound/pci/hda/hdac.c

Modified: head/sys/dev/sound/pci/hda/hdaa.c
==============================================================================
--- head/sys/dev/sound/pci/hda/hdaa.c	Wed Aug  1 13:13:43 2018	(r337042)
+++ head/sys/dev/sound/pci/hda/hdaa.c	Wed Aug  1 14:50:41 2018	(r337043)
@@ -1750,7 +1750,8 @@ hdaa_channel_init(kobj_t obj, void *data, struct snd_d
 	hdaa_unlock(devinfo);
 
 	if (sndbuf_alloc(ch->b, bus_get_dma_tag(devinfo->dev),
-	    hda_get_dma_nocache(devinfo->dev) ? BUS_DMA_NOCACHE : 0,
+	    hda_get_dma_nocache(devinfo->dev) ? BUS_DMA_NOCACHE :
+	    BUS_DMA_COHERENT,
 	    pdevinfo->chan_size) != 0)
 		return (NULL);
 

Modified: head/sys/dev/sound/pci/hda/hdac.c
==============================================================================
--- head/sys/dev/sound/pci/hda/hdac.c	Wed Aug  1 13:13:43 2018	(r337042)
+++ head/sys/dev/sound/pci/hda/hdac.c	Wed Aug  1 14:50:41 2018	(r337043)
@@ -580,7 +580,8 @@ hdac_dma_alloc(struct hdac_softc *sc, struct hdac_dma 
 	 */
 	result = bus_dmamem_alloc(dma->dma_tag, (void **)&dma->dma_vaddr,
 	    BUS_DMA_NOWAIT | BUS_DMA_ZERO |
-	    ((sc->flags & HDAC_F_DMA_NOCACHE) ? BUS_DMA_NOCACHE : 0),
+	    ((sc->flags & HDAC_F_DMA_NOCACHE) ? BUS_DMA_NOCACHE :
+	     BUS_DMA_COHERENT),
 	    &dma->dma_map);
 	if (result != 0) {
 		device_printf(sc->dev, "%s: bus_dmamem_alloc failed (%d)\n",
@@ -627,11 +628,9 @@ static void
 hdac_dma_free(struct hdac_softc *sc, struct hdac_dma *dma)
 {
 	if (dma->dma_paddr != 0) {
-#if 0
 		/* Flush caches */
 		bus_dmamap_sync(dma->dma_tag, dma->dma_map,
 		    BUS_DMASYNC_POSTREAD | BUS_DMASYNC_POSTWRITE);
-#endif
 		bus_dmamap_unload(dma->dma_tag, dma->dma_map);
 		dma->dma_paddr = 0;
 	}
@@ -854,7 +853,6 @@ hdac_rirb_init(struct hdac_softc *sc)
 	HDAC_WRITE_1(&sc->mem, HDAC_RIRBCTL, HDAC_RIRBCTL_RINTCTL);
 #endif
 
-#if 0
 	/*
 	 * Make sure that the Host CPU cache doesn't contain any dirty
 	 * cache lines that falls in the rirb. If I understood correctly, it
@@ -863,7 +861,6 @@ hdac_rirb_init(struct hdac_softc *sc)
 	 */
 	bus_dmamap_sync(sc->rirb_dma.dma_tag, sc->rirb_dma.dma_map,
 	    BUS_DMASYNC_PREREAD);
-#endif
 }
 
 /****************************************************************************
@@ -907,10 +904,8 @@ hdac_rirb_flush(struct hdac_softc *sc)
 
 	rirb_base = (struct hdac_rirb *)sc->rirb_dma.dma_vaddr;
 	rirbwp = HDAC_READ_1(&sc->mem, HDAC_RIRBWP);
-#if 0
 	bus_dmamap_sync(sc->rirb_dma.dma_tag, sc->rirb_dma.dma_map,
 	    BUS_DMASYNC_POSTREAD);
-#endif
 
 	ret = 0;
 	while (sc->rirb_rp != rirbwp) {
@@ -934,6 +929,9 @@ hdac_rirb_flush(struct hdac_softc *sc)
 		}
 		ret++;
 	}
+
+	bus_dmamap_sync(sc->rirb_dma.dma_tag, sc->rirb_dma.dma_map,
+	    BUS_DMASYNC_PREREAD);
 	return (ret);
 }
 
@@ -983,15 +981,11 @@ hdac_send_command(struct hdac_softc *sc, nid_t cad, ui
 	sc->corb_wp++;
 	sc->corb_wp %= sc->corb_size;
 	corb = (uint32_t *)sc->corb_dma.dma_vaddr;
-#if 0
 	bus_dmamap_sync(sc->corb_dma.dma_tag,
 	    sc->corb_dma.dma_map, BUS_DMASYNC_PREWRITE);
-#endif
 	corb[sc->corb_wp] = htole32(verb);
-#if 0
 	bus_dmamap_sync(sc->corb_dma.dma_tag,
 	    sc->corb_dma.dma_map, BUS_DMASYNC_POSTWRITE);
-#endif
 	HDAC_WRITE_2(&sc->mem, HDAC_CORBWP, sc->corb_wp);
 
 	timeout = 10000;
@@ -1937,6 +1931,9 @@ hdac_stream_start(device_t dev, device_t child,
 		addr += blksz;
 	}
 
+	bus_dmamap_sync(sc->streams[ss].bdl.dma_tag,
+	    sc->streams[ss].bdl.dma_map, BUS_DMASYNC_PREWRITE);
+
 	off = ss << 5;
 	HDAC_WRITE_4(&sc->mem, off + HDAC_SDCBL, blksz * blkcnt);
 	HDAC_WRITE_2(&sc->mem, off + HDAC_SDLVI, blkcnt - 1);
@@ -1984,6 +1981,9 @@ hdac_stream_stop(device_t dev, device_t child, int dir
 	ss = hdac_find_stream(sc, dir, stream);
 	KASSERT(ss >= 0,
 	    ("Stop for not allocated stream (%d/%d)\n", dir, stream));
+
+	bus_dmamap_sync(sc->streams[ss].bdl.dma_tag,
+	    sc->streams[ss].bdl.dma_map, BUS_DMASYNC_POSTWRITE);
 
 	off = ss << 5;
 	ctl = HDAC_READ_1(&sc->mem, off + HDAC_SDCTL0);



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201808011450.w71EofqW098435>