Date: Sun, 25 Apr 2010 21:49:47 -0400 From: Michael Butler <imb@protected-networks.net> To: Alexander Motin <mav@FreeBSD.org> Cc: FreeBSD-Current <freebsd-current@freebsd.org>, Rui Paulo <rpaulo@freebsd.org> Subject: Re: SVN rev 206755 breakage Message-ID: <4BD4F13B.4080205@protected-networks.net> In-Reply-To: <4BCBF87C.7020400@FreeBSD.org> References: <4BCA325A.1060600@protected-networks.net> <mailpost.1271550720.7316110.69309.mailing.freebsd.current@FreeBSD.cs.nctu.edu.tw> <4BCA9F44.50002@FreeBSD.org> <987831E7-4893-4F2F-B96F-A1E25BD9BCA0@freebsd.org> <4BCB03AA.1050405@FreeBSD.org> <9B82ED79-C168-482C-A3A9-26D71060BB0F@freebsd.org> <4BCBF87C.7020400@FreeBSD.org>
next in thread | previous in thread | raw e-mail | index | archive | help
This is an OpenPGP/MIME signed message (RFC 2440 and 3156)
--------------enig04D5146E391C2A25641DA674
Content-Type: multipart/mixed; boundary="------------070506060909040809040701"
This is a multi-part message in MIME format.
--------------070506060909040809040701
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: quoted-printable
On 04/19/10 02:30, Alexander Motin wrote:
> Rui Paulo wrote:
>> On 18 Apr 2010, at 14:05, Alexander Motin wrote:
>>> Most of AHCI controllers could also work as usual PCI ATA, but not ev=
ery
>>> PCI ATA could work as AHCI. It would be nice to compare `pciconf -lvb=
c`
>>> output in both working (Rui) and not working (Michael) cases.
>>
>> ahci0@pci0:0:31:2: class=3D0x01018f card=3D0x72708086 chip=3D0x27c4808=
6 rev=3D0x02 hdr=3D0x00
>> vendor =3D 'Intel Corporation'
>> device =3D '82801GBM/GHM (ICH7-M Family) Serial ATA Storage Co=
ntroller'
>> class =3D mass storage
>> subclass =3D ATA
>=20
> ^^^
> It doesn't report itself as AHCI.
>=20
>> bar [10] =3D type I/O Port, range 32, base 0x20d8, size 8, enab=
led
>> bar [14] =3D type I/O Port, range 32, base 0x20fc, size 4, enab=
led
>> bar [18] =3D type I/O Port, range 32, base 0x20d0, size 8, enab=
led
>> bar [1c] =3D type I/O Port, range 32, base 0x20f8, size 4, enab=
led
>> bar [20] =3D type I/O Port, range 32, base 0x2020, size 16, enab=
led
>> bar [24] =3D type Memory, range 32, base 0x90445000, size 1024, =
enabled
>=20
> This resource (BAR(5)) is absent on Michael's system. It is needed to
> work in AHCI mode, but not required for legacy PCI ATA. Probably some
> kind of BIOS magic in your case makes it appear in this mode with this
> chip ID.
More for my own amusement than anything, I came up with an _horrible_
patch to force my ICH7M into AHCI mode (attached). It has two issues:
1) I haven't figured out how to automagically determine which
address(es) I can use without colliding with anything else. Simply
letting bus_allocate_any() decide where to point BAR(5) doesn't appear
to work. I suspect I have to dig through the SMAP stuff to find out what
the BIOS has already claimed and use something outside of those ranges.
2) Since my laptop has both a SATA drive and a PATA DVD-R/W, the
manufacturer commissioned a BIOS which brings the ICH7M up in "combined
mode" with D31-F1 completely disabled. Since it can't (per Intel spec)
be re-enabled without a "platform reset", flipping into AHCI mode
effectively removes the DVD.
However - on the "up side", I now get NCQ ;-)
ahci0: <Intel ICH7M AHCI SATA controller> port
0x1f0-0x1f7,0x3f6,0x170-0x177,0x376,0x18b0-0x18bf
at device 31.2 on pci0
ahci0: BAR(5): 0xf0d44400 AHCI_CAP: 0xdf12ff03 PI: 0x1
pcib0: matched entry for 0.31.INTB
pcib0: slot 31 INTB hardwired to IRQ 19
ahci0: [MPSAFE]
ahci0: [ITHREAD]
ahci0: AHCI v1.10 with 4 1.5Gbps ports, Port Multiplier supported
ahci0: Caps: 64bit NCQ MPS SS ALP AL CLO 1.5Gbps PM PMD SSC PSC
32cmd 4ports
ahci0: Caps2:
ahcich0: <AHCI channel> at channel 0 on ahci0
ahcich0: [MPSAFE]
ahcich0: [ITHREAD]
ahcich0: Caps:
[ .. ]
ada0 at ahcich0 bus 0 scbus1 target 0 lun 0
ada0: <FUJITSU MHZ2320BJ G2 0000001E> ATA-8 SATA 2.x device
ada0: Serial Number K82BT89256VDGEOM: new disk ada0
ada0: 150.000MB/s transfers (SATA 1.x, UDMA5, PIO 8192bytes)
ada0: Command Queueing enabled
ada0: 305245MB (625142448 512 byte sectors: 16H 63S/T 16383C)
--------------070506060909040809040701
Content-Type: text/plain;
name="ahci-a105-patch"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: attachment;
filename="ahci-a105-patch"
*** sys/dev/ahci/ahci.c.orig Sat Apr 24 21:36:42 2010
--- sys/dev/ahci/ahci.c Sun Apr 25 21:30:57 2010
***************
*** 126,131 ****
--- 126,132 ----
{0x26838086, 0x00, "Intel ESB2", 0},
{0x27c18086, 0x00, "Intel ICH7", 0},
{0x27c38086, 0x00, "Intel ICH7", 0},
+ {0x27c48086, 0x00, "Intel ICH7M", 0},
{0x27c58086, 0x00, "Intel ICH7M", 0},
{0x27c68086, 0x00, "Intel ICH7M", 0},
{0x28218086, 0x00, "Intel ICH8", 0},
***************
*** 321,331 ****
--- 322,353 ----
ctlr->quirks =3D ahci_ids[i].quirks;
resource_int_value(device_get_name(dev),
device_get_unit(dev), "ccc", &ctlr->ccc);
+=20
+ #define AHCI_MEM_HACK 0xF0D44400 /* 0xf0d443ff is the last used by othe=
rs on Toshiba A105 */
+=20
+ /* Need to set the SCRAE bit and ensure SCRD not set */
+ pci_write_config(dev, 0x94, (pci_read_config(dev, 0x94, 4) | 0x200) & =
~0x4000, 4);
+ /* enable MSE */
+ pci_write_config(dev, 0x4, (pci_read_config(dev, 0x4, 2) | 2), 2);
+ pci_write_config(dev, 0x24, AHCI_MEM_HACK, 4);=09
+ pci_write_config(dev, 0x90, 0x40, 1); /* AHCI + non-combined */
+=20
+ /* allocate a free memory window for BAR(5) */
+ ctlr->r_rid =3D PCIR_BAR(5);
+ bus_set_resource(dev, SYS_RES_MEMORY, ctlr->r_rid, AHCI_MEM_HACK, 0x40=
0);
+=20
/* if we have a memory BAR(5) we are likely on an AHCI part */
ctlr->r_rid =3D PCIR_BAR(5);
if (!(ctlr->r_mem =3D bus_alloc_resource_any(dev, SYS_RES_MEMORY,
&ctlr->r_rid, RF_ACTIVE)))
return ENXIO;
+=20
+ /* enable ICH7M ports in AHCI space */
+ ATA_OUTL(ctlr->r_mem, AHCI_PI, ATA_INL(ctlr->r_mem, AHCI_PI) | 5);
+ #if 0
+ device_printf(dev, "BAR(5): 0x%lx AHCI_CAP: 0x%lx PI: 0x%lx\n", (unsig=
ned long)pci_read_config(dev, 0x24, 4),
+ (unsigned long)ATA_INL(ctlr->r_mem, 0), (unsigned long)ATA_INL(ctlr->=
r_mem, AHCI_PI));
+ #endif
/* Setup our own memory management for channels. */
ctlr->sc_iomem.rm_type =3D RMAN_ARRAY;
ctlr->sc_iomem.rm_descr =3D "I/O memory addresses";
--------------070506060909040809040701--
--------------enig04D5146E391C2A25641DA674
Content-Type: application/pgp-signature; name="signature.asc"
Content-Description: OpenPGP digital signature
Content-Disposition: attachment; filename="signature.asc"
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (FreeBSD)
iEYEARECAAYFAkvU8T8ACgkQQv9rrgRC1JL4bACfcP2qWfDsoYSiAr2i02/4Nc7Z
8/sAoMPWiQ0lvK/Ka/0ZucaoP22cxAfa
=Hb7M
-----END PGP SIGNATURE-----
--------------enig04D5146E391C2A25641DA674--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4BD4F13B.4080205>