summaryrefslogtreecommitdiff
path: root/sys/dev
diff options
context:
space:
mode:
authorcgd <cgd@NetBSD.org>1998-03-06 17:41:59 +0000
committercgd <cgd@NetBSD.org>1998-03-06 17:41:59 +0000
commitf528463b81ff4d392241fbe431a1b2f7b4cf8bca (patch)
tree4f619e95996f49e8e8be23907be21b617bca99fa /sys/dev
parenteed5e964106890c4e0d76553cf32192961f2a094 (diff)
despite the spec, some people map the bus master IDE registers into
memory space. Note that in a comment, but don't try to fix it (for now).
Diffstat (limited to 'sys/dev')
-rw-r--r--sys/dev/pci/pciide.c9
1 files changed, 8 insertions, 1 deletions
diff --git a/sys/dev/pci/pciide.c b/sys/dev/pci/pciide.c
index b3f8a3a4d31..4eea8582764 100644
--- a/sys/dev/pci/pciide.c
+++ b/sys/dev/pci/pciide.c
@@ -1,4 +1,4 @@
-/* $NetBSD: pciide.c,v 1.3 1998/03/04 19:19:21 cgd Exp $ */
+/* $NetBSD: pciide.c,v 1.4 1998/03/06 17:41:59 cgd Exp $ */
/*
* Copyright (c) 1996, 1998 Christopher G. Demetriou. All rights reserved.
@@ -189,6 +189,13 @@ pciide_attach(parent, self, aux)
* it'll be zero. I.e., sc_dma_ioh_valid will only be non-zero
* if the interface supports DMA and the registers could be
* mapped.
+ *
+ * XXX Note that despite the fact that the Bus Master IDE specs
+ * XXX say that "The bus master IDE functoin uses 16 bytes of IO
+ * XXX space," some controllers (at least the United
+ * XXX Microelectronics UM8886BF) place it in memory space.
+ * XXX eventually, we should probably read the register and check
+ * XXX which type it is. Either that or 'quirk' certain devices.
*/
if (interface & PCIIDE_INTERFACE_BUS_MASTER_DMA) {
sc->sc_dma_ioh_valid = (pci_mapreg_map(pa,