diff options
| author | Patrick Wildt <patrick@cvs.openbsd.org> | 2016-10-02 18:51:45 +0000 |
|---|---|---|
| committer | Patrick Wildt <patrick@cvs.openbsd.org> | 2016-10-02 18:51:45 +0000 |
| commit | 1fd447882b95ebf3b29bca73039223b116c4db2d (patch) | |
| tree | 125c25ae356e09207256fa5dae58c33d7de68d0f /sys | |
| parent | e2e68a301f2e4c3117bde6fefec46090196ef00c (diff) | |
Contrary to e. g. the xHCI specification, the AHCI rev. 1.3 spec
does not say anything about which type (snoopable/uncached etc.) of
bus transactions are issued by HBAs on accesses to the descriptors.
Thus, the right assumption would be "classical" concurrent accesses
by both CPU/driver as well as hardware to the descriptors and their
DMA memory backings respectively, which is also confirmed by actual
testing. Consequently, switch to BUS_DMA_COHERENT mappings for said
DMA memory as otherwise corruption of descriptors is seen.
From Marius Strobl
tested by awolk (amd64), bluhm (amd64, i386), myself (amd64, armv7)
ok jmatthew@
Diffstat (limited to 'sys')
| -rw-r--r-- | sys/dev/ic/ahci.c | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/sys/dev/ic/ahci.c b/sys/dev/ic/ahci.c index 0188ab32010..1026e1e9046 100644 --- a/sys/dev/ic/ahci.c +++ b/sys/dev/ic/ahci.c @@ -1,4 +1,4 @@ -/* $OpenBSD: ahci.c,v 1.24 2016/03/10 13:56:14 krw Exp $ */ +/* $OpenBSD: ahci.c,v 1.25 2016/10/02 18:51:44 patrick Exp $ */ /* * Copyright (c) 2006 David Gwynne <dlg@openbsd.org> @@ -2610,7 +2610,7 @@ ahci_dmamem_alloc(struct ahci_softc *sc, size_t size) goto destroy; if (bus_dmamem_map(sc->sc_dmat, &adm->adm_seg, nsegs, size, - &adm->adm_kva, BUS_DMA_NOWAIT) != 0) + &adm->adm_kva, BUS_DMA_NOWAIT | BUS_DMA_COHERENT) != 0) goto free; if (bus_dmamap_load(sc->sc_dmat, adm->adm_map, adm->adm_kva, size, |