Marvell 88SX5081 SATA.

[fossala]

I've got this raid controller and I am trying to work (just as a sata controller).
Here Is the controller from dmesg

"Marvell 88SX5081 SATA" rev 0x03 at pci3 dev 2 function 0 not configured

I've got 2xWD 1TB in at the moment and non are showing up.
I really dont't know where to go from here so any help would be fantastic.

[jggimi]

When your dmesg shows "not configured", this means the device was recognized by the kernel, but there is no matching driver available.

[shep]

I took a quick glance at the supported hardware for openbsd and did not see the Marvell SATA controllers listed. I recently tried to get a Via VT 8237S controller to support more than one SATA drive in OpenBSD without sucess.

I had better luck putting FreeBSD on the machine and it looks like FreeBSD added Marvell SATA support recently (8,1 release)
http://forums.freebsd.org/showthread.php?t=9533

[fossala]

Any chance this will work? I found it here. http://www.webservertalk.com/message2133676.html

Index: dev/pci/ahci.c
 ========================================
===========================
RCS file: /cvs/src/sys/dev/pci/ahci.c,v
retrieving revision 1.130
diff -u -p -r1.130 ahci.c
--- dev/pci/ahci.c	2007/10/27 10:51:21	1.130
+++ dev/pci/ahci.c	2007/10/31 15:56:27
@@ -34,7 +34,7 @@
#include <dev/ata/atascsi.h>

/* change to AHCI_DEBUG for dmesg spam */
-#define NO_AHCI_DEBUG
+#define AHCI_DEBUG

#ifdef AHCI_DEBUG
#define DPRINTF(m, f...) do { if ((ahcidebug & (m)) == (m)) printf(f); 
} \
@@ -387,8 +387,9 @@ struct ahci_softc {
bus_dma_tag_t		sc_dmat;

int			sc_flags;
-#define AHCI_F_NO_NCQ			(1<<0)
-#define AHCI_F_NO_FER			(1<<1)
+#define AHCI_F_NO_NCQ		(1<<0)
+#define AHCI_F_NO_FR		(1<<1)
+#define AHCI_F_NO_CR		(1<<2)

u_int			sc_ncmds;

@@ -419,12 +420,28 @@ int			ahci_vt8251_attach(struct ahci_sof
struct pci_attach_args *);
int			ahci_ati_ixp600_attach(struct ahci_softc *,
struct pci_attach_args *);
+int			ahci_88se61xx_attach(struct ahci_softc *,
+			    struct pci_attach_args *);

static const struct ahci_device ahci_devices[] = {
{  PCI_VENDOR_VIATECH,	PCI_PRODUCT_VIATECH_
VT8251_SATA,
ahci_no_match,	ahci_vt8251_attach },
{  PCI_VENDOR_ATI,	PCI_PRODUCT_ATI_IXP_SATA
_600,
-	    NULL,		ahci_ati_ixp600_attach }
+	    NULL,		ahci_ati_ixp600_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6111,
+	    NULL,		ahci_88se61xx_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6120,
+	    NULL,		ahci_88se61xx_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6121,
+	    NULL,		ahci_88se61xx_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6122,
+	    NULL,		ahci_88se61xx_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6140,
+	    NULL,		ahci_88se61xx_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6141,
+	    NULL,		ahci_88se61xx_attach },
+	{  PCI_VENDOR_MARVELL,	PCI_PRODUCT_MARVELL_
88SE6145,
+	    NULL,		ahci_88se61xx_attach }
};

int			ahci_pci_match(struct device *, void *, void *);
@@ -554,8 +571,16 @@ ahci_vt8251_attach(struct ahci_softc *sc

int
ahci_ati_ixp600_attach(struct ahci_softc *sc, struct pci_attach_args *pa)
+{
+	sc->sc_flags |= AHCI_F_NO_FR;
+
+	return (0);
+}
+
+ int
+ahci_88se61xx_attach(struct ahci_softc *sc, struct pci_attach_args *pa)
{
-	sc->sc_flags |= AHCI_F_NO_FER;
+	sc->sc_flags |= AHCI_F_NO_FR | AHCI_F_NO_CR | AHCI_F_NO_NCQ;

return (0);
}
@@ -911,6 +936,8 @@ ahci_port_alloc(struct ahci_softc *sc, u
goto freeport;
}

+	printf("preg=0x%x\n", cmd);
+
/* Allocate a CCB for each command slot */
ap->ap_ccbs = malloc(sizeof(struct ahci_ccb) * sc->sc_ncmds, M_DEVBUF,
M_NOWAIT | M_ZERO);
@@ -1014,6 +1041,9 @@ nomem:
DPRINTF(AHCI_D_VERBOSE, "%s: detected device on port %d\n",
DEVNAME(sc), port);

+	printf("preg=0x%x\n",
+	    ahci_pread(ap, AHCI_PREG_CMD) & ~AHCI_PREG_CMD_ICC);
+
/* Enable command transfers on port */
if (ahci_port_start(ap, 0)) {
printf("%s: failed to start command DMA on port %d, "
@@ -1084,8 +1114,7 @@ ahci_port_start(struct ahci_port *ap, in

/* Turn on FRE (and ST) */
r = ahci_pread(ap, AHCI_PREG_CMD) & ~AHCI_PREG_CMD_ICC;
-	if (!(ap->ap_sc->sc_flags & AHCI_F_NO_FER))
-		r |= AHCI_PREG_CMD_FRE;
+	r |= AHCI_PREG_CMD_FRE;
if (!fre_only)
r |= AHCI_PREG_CMD_ST;
ahci_pwrite(ap, AHCI_PREG_CMD, r);
@@ -1098,16 +1127,22 @@ ahci_port_start(struct ahci_port *ap, in
ap->ap_sc->sc_ccc_ports_cur);
}
#endif
+
+	printf("preg=0x%x\n",
+	    ahci_pread(ap, AHCI_PREG_CMD) & ~AHCI_PREG_CMD_ICC);

-	if (!(ap->ap_sc->sc_flags & AHCI_F_NO_FER)) {
+	if (!(ap->ap_sc->sc_flags & AHCI_F_NO_FR)) {
/* Wait for FR to come on */
if (ahci_pwait_set(ap, AHCI_PREG_CMD, AHCI_PREG_CMD_FR))
return (2);
}

-	/* Wait for CR to come on */
-	if (!fre_only && ahci_pwait_set(ap, AHCI_PREG_CMD, AHCI_PREG_CMD_CR))
-		return (1);
+	if (!(ap->ap_sc->sc_flags & AHCI_F_NO_CR)) {
+		/* Wait for CR to come on */
+		if (!fre_only && ahci_pwait_set(ap, AHCI_PREG_CMD,
+		    AHCI_PREG_CMD_CR))
+			return (1);
+	}

return (0);
}

[BSDfan666]

That is a mailing list post (tech@) archived on a forum, did you read it? the changes add support for 88SE* based devices.. not 88SX*.

If the controller in question does support AHCI (..but messed up the standardized way of indicating this) it may be possible to force the driver to attach in a similar fashion.

I'd recommend checking the BIOS and seeing if the mode of the controller can be toggled, i.e: AHCI vs. IDE vs RAID mode.. if it's in RAID mode, OpenBSD usually won't attach to it.

[jggimi]

There was no follow up posting to this 3.5 year old tech@ thread. More importantly the Marvell product attachment code was never added to sys/dev/pci/ahci.c.

To put this into historical perspective, the request for testers was posted four weeks before 4.3 was locked down for -release testing.

[fossala]

Originally Posted by BSDfan666

I'd recommend checking the BIOS and seeing if the mode of the controller can be toggled, i.e: AHCI vs. IDE vs RAID mode.. if it's in RAID mode, OpenBSD usually won't attach to it.

The board doesn't do any form of hardware (fake or real) RAID. It's just a controller. I don't want to give up but it doesn't sound like I have any options.

I also tried it with FreeBSD and it kernel panics with the card in. I tried FreeNAS and that just corrupts my gpt on install then won't boot (on my scsi RAID card).

Starting to look like I will have to put debian

[BSDfan666]

Doing some digging it seems these SATA controllers are proprietary and don't support the standard AHCI/PCIIDE interfaces at all and requires a "native" driver.

You can try adding the following to OpenBSD's pciide.c and hoping some basic semblance of SATA support is to be found.. but it's highly unlikely.

const struct pciide_product_desc pciide_marvell_products[] = {
    { PCI_PRODUCT_MARVELL_88SX5081,
      0,
      sata_chip_map
    },
};

FreeBSD indeed has a driver for this family of cards, mvs(4), but it would appear you've run into other problems.

[fossala]

Thanks I'm trying that now. Not getting my hopes up though.
I cannot afford to get one now but can anyone recommend me a pci-x card for the future?

[fossala]

Nope no luck.

[BSDfan666]

When you buy a card, look for one that mentions PCIIDE/AHCI support.. or perhaps has marketing phrases like "plug and play" or "no drivers required", supporting Mac OS X may also be a good indicator.

You can get some simple cards for less then $20, and most places have return policies, you might have to buy a few different cards until you find one that works.

[fossala]

http://mail-index.netbsd.org/netbsd-...8/14/0010.html
This makes me think I could use netbsd but when I load it up its not detected. Is there any other steps I would need to take. Sorry I have never used NetBSD before.

[BSDfan666]

No.. OpenBSD also has a similar file for PCI product ID's, it is required for more "humanized" output of detected devices.

That does not indicate support.