| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 1 | Kernel driver i2c-i801 | 
|  | 2 |  | 
|  | 3 | Supported adapters: | 
|  | 4 | * Intel 82801AA and 82801AB (ICH and ICH0 - part of the | 
|  | 5 | '810' and '810E' chipsets) | 
|  | 6 | * Intel 82801BA (ICH2 - part of the '815E' chipset) | 
|  | 7 | * Intel 82801CA/CAM (ICH3) | 
|  | 8 | * Intel 82801DB (ICH4) (HW PEC supported, 32 byte buffer not supported) | 
|  | 9 | * Intel 82801EB/ER (ICH5) (HW PEC supported, 32 byte buffer not supported) | 
|  | 10 | * Intel 6300ESB | 
|  | 11 | * Intel 82801FB/FR/FW/FRW (ICH6) | 
| Jason Gaston | a980a99 | 2006-12-10 21:21:31 +0100 | [diff] [blame] | 12 | * Intel 82801G (ICH7) | 
|  | 13 | * Intel 631xESB/632xESB (ESB2) | 
|  | 14 | * Intel 82801H (ICH8) | 
|  | 15 | * Intel ICH9 | 
| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 16 | Datasheets: Publicly available at the Intel website | 
|  | 17 |  | 
|  | 18 | Authors: | 
|  | 19 | Frodo Looijaard <frodol@dds.nl>, | 
|  | 20 | Philip Edelbrock <phil@netroedge.com>, | 
|  | 21 | Mark Studebaker <mdsxyz123@yahoo.com> | 
|  | 22 |  | 
|  | 23 |  | 
|  | 24 | Module Parameters | 
|  | 25 | ----------------- | 
|  | 26 |  | 
| Jean Delvare | d8db8f9 | 2006-06-12 21:50:11 +0200 | [diff] [blame] | 27 | None. | 
| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 28 |  | 
|  | 29 |  | 
|  | 30 | Description | 
|  | 31 | ----------- | 
|  | 32 |  | 
|  | 33 | The ICH (properly known as the 82801AA), ICH0 (82801AB), ICH2 (82801BA), | 
|  | 34 | ICH3 (82801CA/CAM) and later devices are Intel chips that are a part of | 
|  | 35 | Intel's '810' chipset for Celeron-based PCs, '810E' chipset for | 
|  | 36 | Pentium-based PCs, '815E' chipset, and others. | 
|  | 37 |  | 
|  | 38 | The ICH chips contain at least SEVEN separate PCI functions in TWO logical | 
|  | 39 | PCI devices. An output of lspci will show something similar to the | 
|  | 40 | following: | 
|  | 41 |  | 
|  | 42 | 00:1e.0 PCI bridge: Intel Corporation: Unknown device 2418 (rev 01) | 
|  | 43 | 00:1f.0 ISA bridge: Intel Corporation: Unknown device 2410 (rev 01) | 
|  | 44 | 00:1f.1 IDE interface: Intel Corporation: Unknown device 2411 (rev 01) | 
|  | 45 | 00:1f.2 USB Controller: Intel Corporation: Unknown device 2412 (rev 01) | 
|  | 46 | 00:1f.3 Unknown class [0c05]: Intel Corporation: Unknown device 2413 (rev 01) | 
|  | 47 |  | 
|  | 48 | The SMBus controller is function 3 in device 1f. Class 0c05 is SMBus Serial | 
|  | 49 | Controller. | 
|  | 50 |  | 
| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 51 | The ICH chips are quite similar to Intel's PIIX4 chip, at least in the | 
|  | 52 | SMBus controller. | 
|  | 53 |  | 
| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 54 |  | 
|  | 55 | Process Call Support | 
|  | 56 | -------------------- | 
|  | 57 |  | 
|  | 58 | Not supported. | 
|  | 59 |  | 
|  | 60 |  | 
|  | 61 | I2C Block Read Support | 
|  | 62 | ---------------------- | 
|  | 63 |  | 
|  | 64 | Not supported at the moment. | 
|  | 65 |  | 
|  | 66 |  | 
|  | 67 | SMBus 2.0 Support | 
|  | 68 | ----------------- | 
|  | 69 |  | 
|  | 70 | The 82801DB (ICH4) and later chips support several SMBus 2.0 features. | 
|  | 71 |  | 
| Jean Delvare | 099ab11 | 2007-02-13 22:09:00 +0100 | [diff] [blame] | 72 |  | 
|  | 73 | Hidden ICH SMBus | 
|  | 74 | ---------------- | 
|  | 75 |  | 
|  | 76 | If your system has an Intel ICH south bridge, but you do NOT see the | 
|  | 77 | SMBus device at 00:1f.3 in lspci, and you can't figure out any way in the | 
|  | 78 | BIOS to enable it, it means it has been hidden by the BIOS code. Asus is | 
|  | 79 | well known for first doing this on their P4B motherboard, and many other | 
|  | 80 | boards after that. Some vendor machines are affected as well. | 
|  | 81 |  | 
|  | 82 | The first thing to try is the "i2c_ec" ACPI driver. It could be that the | 
|  | 83 | SMBus was hidden on purpose because it'll be driven by ACPI. If the | 
|  | 84 | i2c_ec driver works for you, just forget about the i2c-i801 driver and | 
|  | 85 | don't try to unhide the ICH SMBus. Even if i2c_ec doesn't work, you | 
|  | 86 | better make sure that the SMBus isn't used by the ACPI code. Try loading | 
|  | 87 | the "fan" and "thermal" drivers, and check in /proc/acpi/fan and | 
|  | 88 | /proc/acpi/thermal_zone. If you find anything there, it's likely that | 
|  | 89 | the ACPI is accessing the SMBus and it's safer not to unhide it. Only | 
|  | 90 | once you are certain that ACPI isn't using the SMBus, you can attempt | 
|  | 91 | to unhide it. | 
|  | 92 |  | 
|  | 93 | In order to unhide the SMBus, we need to change the value of a PCI | 
|  | 94 | register before the kernel enumerates the PCI devices. This is done in | 
|  | 95 | drivers/pci/quirks.c, where all affected boards must be listed (see | 
|  | 96 | function asus_hides_smbus_hostbridge.) If the SMBus device is missing, | 
|  | 97 | and you think there's something interesting on the SMBus (e.g. a | 
|  | 98 | hardware monitoring chip), you need to add your board to the list. | 
|  | 99 |  | 
|  | 100 | The motherboard is identified using the subvendor and subdevice IDs of the | 
|  | 101 | host bridge PCI device. Get yours with "lspci -n -v -s 00:00.0": | 
|  | 102 |  | 
|  | 103 | 00:00.0 Class 0600: 8086:2570 (rev 02) | 
|  | 104 | Subsystem: 1043:80f2 | 
|  | 105 | Flags: bus master, fast devsel, latency 0 | 
|  | 106 | Memory at fc000000 (32-bit, prefetchable) [size=32M] | 
|  | 107 | Capabilities: [e4] #09 [2106] | 
|  | 108 | Capabilities: [a0] AGP version 3.0 | 
|  | 109 |  | 
|  | 110 | Here the host bridge ID is 2570 (82865G/PE/P), the subvendor ID is 1043 | 
|  | 111 | (Asus) and the subdevice ID is 80f2 (P4P800-X). You can find the symbolic | 
|  | 112 | names for the bridge ID and the subvendor ID in include/linux/pci_ids.h, | 
|  | 113 | and then add a case for your subdevice ID at the right place in | 
|  | 114 | drivers/pci/quirks.c. Then please give it very good testing, to make sure | 
|  | 115 | that the unhidden SMBus doesn't conflict with e.g. ACPI. | 
|  | 116 |  | 
|  | 117 | If it works, proves useful (i.e. there are usable chips on the SMBus) | 
|  | 118 | and seems safe, please submit a patch for inclusion into the kernel. | 
|  | 119 |  | 
|  | 120 | Note: There's a useful script in lm_sensors 2.10.2 and later, named | 
|  | 121 | unhide_ICH_SMBus (in prog/hotplug), which uses the fakephp driver to | 
|  | 122 | temporarily unhide the SMBus without having to patch and recompile your | 
|  | 123 | kernel. It's very convenient if you just want to check if there's | 
|  | 124 | anything interesting on your hidden ICH SMBus. | 
|  | 125 |  | 
|  | 126 |  | 
| Linus Torvalds | 1da177e | 2005-04-16 15:20:36 -0700 | [diff] [blame] | 127 | ********************** | 
|  | 128 | The lm_sensors project gratefully acknowledges the support of Texas | 
|  | 129 | Instruments in the initial development of this driver. | 
|  | 130 |  | 
|  | 131 | The lm_sensors project gratefully acknowledges the support of Intel in the | 
|  | 132 | development of SMBus 2.0 / ICH4 features of this driver. |