coreboot-kgpe-d16/src/southbridge/intel/lynxpoint/serialio.c

281 lines
7.7 KiB
C
Raw Normal View History

lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/*
* This file is part of the coreboot project.
*
* Copyright 2013 Google Inc.
*
* This program is free software; you can redistribute it and/or
* modify it under the terms of the GNU General Public License as
* published by the Free Software Foundation; version 2 of
* the License.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*/
#include <arch/io.h>
#include <cbmem.h>
#include <console/console.h>
#include <device/device.h>
#include <device/pci.h>
#include <device/pciexp.h>
#include <device/pci_ids.h>
#include <stdlib.h>
#include "pch.h"
#include "nvs.h"
/* Enable clock in PCI mode */
static void serialio_enable_clock(struct resource *bar0)
{
u32 reg32 = read32(res2mmio(bar0, SIO_REG_PPR_CLOCK, 0));
reg32 |= SIO_REG_PPR_CLOCK_EN;
write32(res2mmio(bar0, SIO_REG_PPR_CLOCK, 0), reg32);
}
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* Put Serial IO D21:F0-F6 device into desired mode. */
static void serialio_d21_mode(int sio_index, int int_pin, int acpi_mode)
{
u32 portctrl = SIO_IOBP_PORTCTRL_PM_CAP_PRSNT;
/* Snoop select 1. */
portctrl |= SIO_IOBP_PORTCTRL_SNOOP_SELECT(1);
/* Set interrupt pin. */
portctrl |= SIO_IOBP_PORTCTRL_INT_PIN(int_pin);
if (acpi_mode) {
/* Enable ACPI interrupt mode. */
portctrl |= SIO_IOBP_PORTCTRL_ACPI_IRQ_EN;
/* Disable PCI config space. */
portctrl |= SIO_IOBP_PORTCTRL_PCI_CONF_DIS;
}
pch_iobp_update(SIO_IOBP_PORTCTRLX(sio_index), 0, portctrl);
}
/* Put Serial IO D23:F0 device into desired mode. */
static void serialio_d23_mode(int acpi_mode)
{
u32 portctrl = 0;
/* Snoop select 1. */
pch_iobp_update(SIO_IOBP_PORTCTRL1, 0,
SIO_IOBP_PORTCTRL1_SNOOP_SELECT(1));
if (acpi_mode) {
/* Enable ACPI interrupt mode. */
portctrl |= SIO_IOBP_PORTCTRL0_ACPI_IRQ_EN;
/* Disable PCI config space. */
portctrl |= SIO_IOBP_PORTCTRL0_PCI_CONF_DIS;
}
pch_iobp_update(SIO_IOBP_PORTCTRL0, 0, portctrl);
}
/* Enable LTR Auto Mode for D21:F1-F6. */
static void serialio_d21_ltr(struct resource *bar0)
{
u32 reg;
/* 1. Program BAR0 + 808h[2] = 0b */
reg = read32(res2mmio(bar0, SIO_REG_PPR_GEN, 0));
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
reg &= ~SIO_REG_PPR_GEN_LTR_MODE_MASK;
write32(res2mmio(bar0, SIO_REG_PPR_GEN, 0), reg);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* 2. Program BAR0 + 804h[1:0] = 00b */
reg = read32(res2mmio(bar0, SIO_REG_PPR_RST, 0));
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
reg &= ~SIO_REG_PPR_RST_ASSERT;
write32(res2mmio(bar0, SIO_REG_PPR_RST, 0), reg);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* 3. Program BAR0 + 804h[1:0] = 11b */
reg = read32(res2mmio(bar0, SIO_REG_PPR_RST, 0));
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
reg |= SIO_REG_PPR_RST_ASSERT;
write32(res2mmio(bar0, SIO_REG_PPR_RST, 0), reg);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* 4. Program BAR0 + 814h[31:0] = 00000000h */
write32(res2mmio(bar0, SIO_REG_AUTO_LTR, 0), 0);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
}
/* Enable LTR Auto Mode for D23:F0. */
static void serialio_d23_ltr(struct resource *bar0)
{
u32 reg;
/* Program BAR0 + 1008h[2] = 1b */
reg = read32(res2mmio(bar0, SIO_REG_SDIO_PPR_GEN, 0));
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
reg |= SIO_REG_PPR_GEN_LTR_MODE_MASK;
write32(res2mmio(bar0, SIO_REG_SDIO_PPR_GEN, 0), reg);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* Program BAR0 + 1010h = 0x00000000 */
write32(res2mmio(bar0, SIO_REG_SDIO_PPR_SW_LTR, 0), 0);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* Program BAR0 + 3Ch[30] = 1b */
reg = read32(res2mmio(bar0, SIO_REG_SDIO_PPR_CMD12, 0));
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
reg |= SIO_REG_SDIO_PPR_CMD12_B30;
write32(res2mmio(bar0, SIO_REG_SDIO_PPR_CMD12, 0), reg);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
}
/* Select I2C voltage of 1.8V or 3.3V. */
static void serialio_i2c_voltage_sel(struct resource *bar0, u8 voltage)
{
u32 reg32 = read32(res2mmio(bar0, SIO_REG_PPR_GEN, 0));
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
reg32 &= ~SIO_REG_PPR_GEN_VOLTAGE_MASK;
reg32 |= SIO_REG_PPR_GEN_VOLTAGE(voltage);
write32(res2mmio(bar0, SIO_REG_PPR_GEN, 0), reg32);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
}
/* Init sequence to be run once, done as part of D21:F0 (SDMA) init. */
static void serialio_init_once(int acpi_mode)
{
if (acpi_mode) {
/* Enable ACPI IRQ for IRQ13, IRQ7, IRQ6, IRQ5 in RCBA. */
RCBA32_OR(ACPIIRQEN, (1 << 13)|(1 << 7)|(1 << 6)|(1 << 5));
}
/* Program IOBP CB000154h[12,9:8,4:0] = 1001100011111b. */
pch_iobp_update(SIO_IOBP_GPIODF, ~0x0000131f, 0x0000131f);
/* Program IOBP CB000180h[5:0] = 111111b (undefined register) */
pch_iobp_update(0xcb000180, ~0x0000003f, 0x0000003f);
}
static void serialio_init(struct device *dev)
{
struct southbridge_intel_lynxpoint_config *config = dev->chip_info;
struct resource *bar0, *bar1;
int sio_index = -1;
u32 reg32;
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
printk(BIOS_DEBUG, "Initializing Serial IO device\n");
/* Ensure memory and bus master are enabled */
reg32 = pci_read_config32(dev, PCI_COMMAND);
reg32 |= PCI_COMMAND_MASTER | PCI_COMMAND_MEMORY;
pci_write_config32(dev, PCI_COMMAND, reg32);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
/* Find BAR0 and BAR1 */
bar0 = find_resource(dev, PCI_BASE_ADDRESS_0);
if (!bar0)
return;
bar1 = find_resource(dev, PCI_BASE_ADDRESS_1);
if (!bar1)
return;
if (!config->sio_acpi_mode)
serialio_enable_clock(bar0);
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
switch (dev->path.pci.devfn) {
case PCI_DEVFN(21, 0): /* SDMA */
sio_index = SIO_ID_SDMA;
serialio_init_once(config->sio_acpi_mode);
serialio_d21_mode(sio_index, SIO_PIN_INTB,
config->sio_acpi_mode);
break;
case PCI_DEVFN(21, 1): /* I2C0 */
sio_index = SIO_ID_I2C0;
serialio_d21_ltr(bar0);
serialio_i2c_voltage_sel(bar0, config->sio_i2c0_voltage);
serialio_d21_mode(sio_index, SIO_PIN_INTC,
config->sio_acpi_mode);
break;
case PCI_DEVFN(21, 2): /* I2C1 */
sio_index = SIO_ID_I2C1;
serialio_d21_ltr(bar0);
serialio_i2c_voltage_sel(bar0, config->sio_i2c1_voltage);
serialio_d21_mode(sio_index, SIO_PIN_INTC,
config->sio_acpi_mode);
break;
case PCI_DEVFN(21, 3): /* SPI0 */
sio_index = SIO_ID_SPI0;
serialio_d21_ltr(bar0);
serialio_d21_mode(sio_index, SIO_PIN_INTC,
config->sio_acpi_mode);
break;
case PCI_DEVFN(21, 4): /* SPI1 */
sio_index = SIO_ID_SPI1;
serialio_d21_ltr(bar0);
serialio_d21_mode(sio_index, SIO_PIN_INTC,
config->sio_acpi_mode);
break;
case PCI_DEVFN(21, 5): /* UART0 */
sio_index = SIO_ID_UART0;
serialio_d21_ltr(bar0);
serialio_d21_mode(sio_index, SIO_PIN_INTD,
config->sio_acpi_mode);
break;
case PCI_DEVFN(21, 6): /* UART1 */
sio_index = SIO_ID_UART1;
serialio_d21_ltr(bar0);
serialio_d21_mode(sio_index, SIO_PIN_INTD,
config->sio_acpi_mode);
break;
case PCI_DEVFN(23, 0): /* SDIO */
sio_index = SIO_ID_SDIO;
serialio_d23_ltr(bar0);
serialio_d23_mode(config->sio_acpi_mode);
break;
default:
return;
}
if (config->sio_acpi_mode) {
global_nvs_t *gnvs;
/* Find ACPI NVS to update BARs */
gnvs = (global_nvs_t *)cbmem_find(CBMEM_ID_ACPI_GNVS);
if (!gnvs) {
printk(BIOS_ERR, "Unable to locate Global NVS\n");
return;
}
/* Save BAR0 and BAR1 to ACPI NVS */
gnvs->s0b[sio_index] = (u32)bar0->base;
gnvs->s1b[sio_index] = (u32)bar1->base;
}
}
static void serialio_set_subsystem(device_t dev, unsigned vendor,
unsigned device)
{
if (!vendor || !device) {
pci_write_config32(dev, PCI_SUBSYSTEM_VENDOR_ID,
pci_read_config32(dev, PCI_VENDOR_ID));
} else {
pci_write_config32(dev, PCI_SUBSYSTEM_VENDOR_ID,
((device & 0xffff) << 16) | (vendor & 0xffff));
}
}
static struct pci_operations pci_ops = {
.set_subsystem = serialio_set_subsystem,
};
static struct device_operations device_ops = {
.read_resources = pci_dev_read_resources,
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
.set_resources = pci_dev_set_resources,
.enable_resources = pci_dev_enable_resources,
lynxpoint: Basic configuration of SerialIO devices This adds configuration of SerialIO devices in the Lynxpoint-LP chipset. This includes DMA, I2C, SPI, UART, and SDIO controllers. There is assorted magic setup necessary for the devices and while it is similar for each device there are subtle differences in some register settings. These devices must be put into "ACPI Mode" in order to take advantage of S0ix. When in ACPI mode the allocated PCI BARs must be passed to ACPI so it can be relayed to the OS. When the devices are in ACPI mode BAR0+BAR1 is saved into ACPI NVS and then updated and returned when the OS calls _CRS. Note that is is not entirely complete yet. We need to update the IASL compiler in our build environment to support ACPI 5.0 in order to be able to pass the FixedDMA entries to the kernel. There are also no ACPI methods defined yet to do D0->D3->D0 transitions for actually entering/exiting S0ix states. This is hard to test right now because our kernel does not support any of these devices in ACPI mode. I was able to build and test the upstream bleeding-edge branch of the linux-pm git tree. With that tree I was able to enumerate and load the driver for the DesignWare I2C driver and attempt to probe the I2C bus -- although there are no devices attatched. I am also able to see the resources from ACPI in /proc/iomem get reserved properly in the kernel. Change-Id: Ie311addd6a25f3b7edf3388fe68c1cd691a0a500 Signed-off-by: Duncan Laurie <dlaurie@chromium.org> Reviewed-on: http://review.coreboot.org/2971 Tested-by: build bot (Jenkins) Reviewed-by: Ronald G. Minnich <rminnich@gmail.com>
2013-03-22 19:21:14 +01:00
.init = serialio_init,
.ops_pci = &pci_ops,
};
static const unsigned short pci_device_ids[] = {
0x9c60, /* 0:15.0 - SDMA */
0x9c61, /* 0:15.1 - I2C0 */
0x9c62, /* 0:15.2 - I2C1 */
0x9c65, /* 0:15.3 - SPI0 */
0x9c66, /* 0:15.4 - SPI1 */
0x9c63, /* 0:15.5 - UART0 */
0x9c64, /* 0:15.6 - UART1 */
0x9c35, /* 0:17.0 - SDIO */
0
};
static const struct pci_driver pch_pcie __pci_driver = {
.ops = &device_ops,
.vendor = PCI_VENDOR_ID_INTEL,
.devices = pci_device_ids,
};