freebsd-dev/sys/powerpc/mpc85xx/pci_mpc85xx.c

829 lines
22 KiB
C
Raw Normal View History

/*-
* Copyright 2006-2007 by Juniper Networks.
* Copyright 2008 Semihalf.
* Copyright 2010 The FreeBSD Foundation
* All rights reserved.
*
* Portions of this software were developed by Semihalf
* under sponsorship from the FreeBSD Foundation.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR
* IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
* OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
* IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT,
* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING,
* BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
* LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
* AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
* OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* From: FreeBSD: src/sys/powerpc/mpc85xx/pci_ocp.c,v 1.9 2010/03/23 23:46:28 marcel
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/ktr.h>
#include <sys/sockio.h>
#include <sys/mbuf.h>
#include <sys/malloc.h>
#include <sys/kernel.h>
#include <sys/module.h>
#include <sys/socket.h>
#include <sys/queue.h>
#include <sys/bus.h>
#include <sys/lock.h>
#include <sys/mutex.h>
#include <sys/rman.h>
#include <sys/endian.h>
#include <vm/vm.h>
#include <vm/pmap.h>
#include <dev/ofw/ofw_pci.h>
#include <dev/ofw/ofw_bus.h>
#include <dev/ofw/ofw_bus_subr.h>
#include <dev/pci/pcivar.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcib_private.h>
#include <powerpc/ofw/ofw_pci.h>
#include "ofw_bus_if.h"
#include "pcib_if.h"
#include <machine/resource.h>
#include <machine/bus.h>
#include <machine/intr_machdep.h>
#include <powerpc/mpc85xx/mpc85xx.h>
#define REG_CFG_ADDR 0x0000
#define CONFIG_ACCESS_ENABLE 0x80000000
#define REG_CFG_DATA 0x0004
#define REG_INT_ACK 0x0008
#define REG_POTAR(n) (0x0c00 + 0x20 * (n))
#define REG_POTEAR(n) (0x0c04 + 0x20 * (n))
#define REG_POWBAR(n) (0x0c08 + 0x20 * (n))
#define REG_POWAR(n) (0x0c10 + 0x20 * (n))
#define REG_PITAR(n) (0x0e00 - 0x20 * (n))
#define REG_PIWBAR(n) (0x0e08 - 0x20 * (n))
#define REG_PIWBEAR(n) (0x0e0c - 0x20 * (n))
#define REG_PIWAR(n) (0x0e10 - 0x20 * (n))
#define REG_PEX_MES_DR 0x0020
#define REG_PEX_MES_IER 0x0028
#define REG_PEX_ERR_DR 0x0e00
#define REG_PEX_ERR_EN 0x0e08
#define PCIR_LTSSM 0x404
#define LTSSM_STAT_L0 0x16
#define DEVFN(b, s, f) ((b << 16) | (s << 8) | f)
struct fsl_pcib_softc {
struct ofw_pci_softc pci_sc;
device_t sc_dev;
int sc_iomem_target;
bus_addr_t sc_iomem_alloc, sc_iomem_start, sc_iomem_end;
int sc_ioport_target;
bus_addr_t sc_ioport_alloc, sc_ioport_start, sc_ioport_end;
struct resource *sc_res;
bus_space_handle_t sc_bsh;
bus_space_tag_t sc_bst;
int sc_rid;
int sc_busnr;
int sc_pcie;
uint8_t sc_pcie_capreg; /* PCI-E Capability Reg Set */
/* Devices that need special attention. */
int sc_devfn_tundra;
int sc_devfn_via_ide;
};
/* Local forward declerations. */
static uint32_t fsl_pcib_cfgread(struct fsl_pcib_softc *, u_int, u_int, u_int,
u_int, int);
static void fsl_pcib_cfgwrite(struct fsl_pcib_softc *, u_int, u_int, u_int,
u_int, uint32_t, int);
static int fsl_pcib_decode_win(phandle_t, struct fsl_pcib_softc *);
static void fsl_pcib_err_init(device_t);
static void fsl_pcib_inbound(struct fsl_pcib_softc *, int, int, u_long,
u_long, u_long);
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
static int fsl_pcib_init(struct fsl_pcib_softc *, int, int);
static void fsl_pcib_outbound(struct fsl_pcib_softc *, int, int, u_long,
u_long, u_long);
/* Forward declerations. */
static int fsl_pcib_attach(device_t);
static int fsl_pcib_detach(device_t);
static int fsl_pcib_probe(device_t);
static int fsl_pcib_maxslots(device_t);
static uint32_t fsl_pcib_read_config(device_t, u_int, u_int, u_int, u_int, int);
static void fsl_pcib_write_config(device_t, u_int, u_int, u_int, u_int,
uint32_t, int);
/* Configuration r/w mutex. */
struct mtx pcicfg_mtx;
static int mtx_initialized = 0;
/*
* Bus interface definitions.
*/
static device_method_t fsl_pcib_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, fsl_pcib_probe),
DEVMETHOD(device_attach, fsl_pcib_attach),
DEVMETHOD(device_detach, fsl_pcib_detach),
/* pcib interface */
DEVMETHOD(pcib_maxslots, fsl_pcib_maxslots),
DEVMETHOD(pcib_read_config, fsl_pcib_read_config),
DEVMETHOD(pcib_write_config, fsl_pcib_write_config),
DEVMETHOD_END
};
static devclass_t fsl_pcib_devclass;
DEFINE_CLASS_1(pcib, fsl_pcib_driver, fsl_pcib_methods,
sizeof(struct fsl_pcib_softc), ofw_pci_driver);
DRIVER_MODULE(pcib, nexus, fsl_pcib_driver, fsl_pcib_devclass, 0, 0);
static int
fsl_pcib_probe(device_t dev)
{
if (ofw_bus_get_type(dev) == NULL ||
strcmp(ofw_bus_get_type(dev), "pci") != 0)
return (ENXIO);
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
if (!(ofw_bus_is_compatible(dev, "fsl,mpc8540-pci") ||
ofw_bus_is_compatible(dev, "fsl,mpc8540-pcie") ||
ofw_bus_is_compatible(dev, "fsl,mpc8548-pcie")))
return (ENXIO);
device_set_desc(dev, "Freescale Integrated PCI/PCI-E Controller");
return (BUS_PROBE_DEFAULT);
}
static int
fsl_pcib_attach(device_t dev)
{
struct fsl_pcib_softc *sc;
phandle_t node;
uint32_t cfgreg;
int maxslot, error;
uint8_t ltssm, capptr;
sc = device_get_softc(dev);
sc->sc_dev = dev;
sc->sc_rid = 0;
sc->sc_res = bus_alloc_resource_any(dev, SYS_RES_MEMORY, &sc->sc_rid,
RF_ACTIVE);
if (sc->sc_res == NULL) {
device_printf(dev, "could not map I/O memory\n");
return (ENXIO);
}
sc->sc_bst = rman_get_bustag(sc->sc_res);
sc->sc_bsh = rman_get_bushandle(sc->sc_res);
sc->sc_busnr = 0;
if (!mtx_initialized) {
mtx_init(&pcicfg_mtx, "pcicfg", NULL, MTX_SPIN);
mtx_initialized = 1;
}
cfgreg = fsl_pcib_cfgread(sc, 0, 0, 0, PCIR_VENDOR, 2);
if (cfgreg != 0x1057 && cfgreg != 0x1957)
goto err;
capptr = fsl_pcib_cfgread(sc, 0, 0, 0, PCIR_CAP_PTR, 1);
while (capptr != 0) {
cfgreg = fsl_pcib_cfgread(sc, 0, 0, 0, capptr, 2);
switch (cfgreg & 0xff) {
case PCIY_PCIX:
break;
case PCIY_EXPRESS:
sc->sc_pcie = 1;
sc->sc_pcie_capreg = capptr;
break;
}
capptr = (cfgreg >> 8) & 0xff;
}
node = ofw_bus_get_node(dev);
/*
* Initialize generic OF PCI interface (ranges, etc.)
*/
error = ofw_pci_init(dev);
if (error)
return (error);
/*
* Configure decode windows for PCI(E) access.
*/
if (fsl_pcib_decode_win(node, sc) != 0)
goto err;
cfgreg = fsl_pcib_cfgread(sc, 0, 0, 0, PCIR_COMMAND, 2);
cfgreg |= PCIM_CMD_SERRESPEN | PCIM_CMD_BUSMASTEREN | PCIM_CMD_MEMEN |
PCIM_CMD_PORTEN;
fsl_pcib_cfgwrite(sc, 0, 0, 0, PCIR_COMMAND, cfgreg, 2);
sc->sc_devfn_tundra = -1;
sc->sc_devfn_via_ide = -1;
/*
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
* Scan bus using firmware configured, 0 based bus numbering.
*/
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
sc->sc_busnr = 0;
maxslot = (sc->sc_pcie) ? 0 : PCI_SLOTMAX;
fsl_pcib_init(sc, sc->sc_busnr, maxslot);
if (sc->sc_pcie) {
ltssm = fsl_pcib_cfgread(sc, 0, 0, 0, PCIR_LTSSM, 1);
if (ltssm < LTSSM_STAT_L0) {
if (bootverbose)
printf("PCI %d: no PCIE link, skipping\n",
device_get_unit(dev));
return (0);
}
}
fsl_pcib_err_init(dev);
return (ofw_pci_attach(dev));
err:
return (ENXIO);
}
static uint32_t
fsl_pcib_cfgread(struct fsl_pcib_softc *sc, u_int bus, u_int slot, u_int func,
u_int reg, int bytes)
{
uint32_t addr, data;
if (bus == sc->sc_busnr - 1)
bus = 0;
addr = CONFIG_ACCESS_ENABLE;
addr |= (bus & 0xff) << 16;
addr |= (slot & 0x1f) << 11;
addr |= (func & 0x7) << 8;
addr |= reg & 0xfc;
if (sc->sc_pcie)
addr |= (reg & 0xf00) << 16;
mtx_lock_spin(&pcicfg_mtx);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_CFG_ADDR, addr);
switch (bytes) {
case 1:
data = bus_space_read_1(sc->sc_bst, sc->sc_bsh,
REG_CFG_DATA + (reg & 3));
break;
case 2:
data = le16toh(bus_space_read_2(sc->sc_bst, sc->sc_bsh,
REG_CFG_DATA + (reg & 2)));
break;
case 4:
data = le32toh(bus_space_read_4(sc->sc_bst, sc->sc_bsh,
REG_CFG_DATA));
break;
default:
data = ~0;
break;
}
mtx_unlock_spin(&pcicfg_mtx);
return (data);
}
static void
fsl_pcib_cfgwrite(struct fsl_pcib_softc *sc, u_int bus, u_int slot, u_int func,
u_int reg, uint32_t data, int bytes)
{
uint32_t addr;
if (bus == sc->sc_busnr - 1)
bus = 0;
addr = CONFIG_ACCESS_ENABLE;
addr |= (bus & 0xff) << 16;
addr |= (slot & 0x1f) << 11;
addr |= (func & 0x7) << 8;
addr |= reg & 0xfc;
if (sc->sc_pcie)
addr |= (reg & 0xf00) << 16;
mtx_lock_spin(&pcicfg_mtx);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_CFG_ADDR, addr);
switch (bytes) {
case 1:
bus_space_write_1(sc->sc_bst, sc->sc_bsh,
REG_CFG_DATA + (reg & 3), data);
break;
case 2:
bus_space_write_2(sc->sc_bst, sc->sc_bsh,
REG_CFG_DATA + (reg & 2), htole16(data));
break;
case 4:
bus_space_write_4(sc->sc_bst, sc->sc_bsh,
REG_CFG_DATA, htole32(data));
break;
}
mtx_unlock_spin(&pcicfg_mtx);
}
#if 0
static void
dump(struct fsl_pcib_softc *sc)
{
unsigned int i;
#define RD(o) bus_space_read_4(sc->sc_bst, sc->sc_bsh, o)
for (i = 0; i < 5; i++) {
printf("POTAR%u =0x%08x\n", i, RD(REG_POTAR(i)));
printf("POTEAR%u =0x%08x\n", i, RD(REG_POTEAR(i)));
printf("POWBAR%u =0x%08x\n", i, RD(REG_POWBAR(i)));
printf("POWAR%u =0x%08x\n", i, RD(REG_POWAR(i)));
}
printf("\n");
for (i = 1; i < 4; i++) {
printf("PITAR%u =0x%08x\n", i, RD(REG_PITAR(i)));
printf("PIWBAR%u =0x%08x\n", i, RD(REG_PIWBAR(i)));
printf("PIWBEAR%u=0x%08x\n", i, RD(REG_PIWBEAR(i)));
printf("PIWAR%u =0x%08x\n", i, RD(REG_PIWAR(i)));
}
printf("\n");
#undef RD
for (i = 0; i < 0x48; i += 4) {
printf("cfg%02x=0x%08x\n", i, fsl_pcib_cfgread(sc, 0, 0, 0,
i, 4));
}
}
#endif
static int
fsl_pcib_maxslots(device_t dev)
{
struct fsl_pcib_softc *sc = device_get_softc(dev);
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
return ((sc->sc_pcie) ? 0 : PCI_SLOTMAX);
}
static uint32_t
fsl_pcib_read_config(device_t dev, u_int bus, u_int slot, u_int func,
u_int reg, int bytes)
{
struct fsl_pcib_softc *sc = device_get_softc(dev);
u_int devfn;
if (bus == sc->sc_busnr && !sc->sc_pcie && slot < 10)
return (~0);
devfn = DEVFN(bus, slot, func);
if (devfn == sc->sc_devfn_tundra)
return (~0);
if (devfn == sc->sc_devfn_via_ide && reg == PCIR_INTPIN)
return (1);
return (fsl_pcib_cfgread(sc, bus, slot, func, reg, bytes));
}
static void
fsl_pcib_write_config(device_t dev, u_int bus, u_int slot, u_int func,
u_int reg, uint32_t val, int bytes)
{
struct fsl_pcib_softc *sc = device_get_softc(dev);
if (bus == sc->sc_busnr && !sc->sc_pcie && slot < 10)
return;
fsl_pcib_cfgwrite(sc, bus, slot, func, reg, val, bytes);
}
static void
fsl_pcib_init_via(struct fsl_pcib_softc *sc, uint16_t device, int bus,
int slot, int fn)
{
if (device == 0x0686) {
fsl_pcib_write_config(sc->sc_dev, bus, slot, fn, 0x52, 0x34, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, fn, 0x77, 0x00, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, fn, 0x83, 0x98, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, fn, 0x85, 0x03, 1);
} else if (device == 0x0571) {
sc->sc_devfn_via_ide = DEVFN(bus, slot, fn);
fsl_pcib_write_config(sc->sc_dev, bus, slot, fn, 0x40, 0x0b, 1);
}
}
static int
fsl_pcib_init_bar(struct fsl_pcib_softc *sc, int bus, int slot, int func,
int barno)
{
bus_addr_t *allocp;
uint32_t addr, mask, size;
int reg, width;
reg = PCIR_BAR(barno);
if (DEVFN(bus, slot, func) == sc->sc_devfn_via_ide) {
switch (barno) {
case 0: addr = 0x1f0; break;
case 1: addr = 0x3f4; break;
case 2: addr = 0x170; break;
case 3: addr = 0x374; break;
case 4: addr = 0xcc0; break;
default: return (1);
}
fsl_pcib_write_config(sc->sc_dev, bus, slot, func, reg, addr, 4);
return (1);
}
fsl_pcib_write_config(sc->sc_dev, bus, slot, func, reg, ~0, 4);
size = fsl_pcib_read_config(sc->sc_dev, bus, slot, func, reg, 4);
if (size == 0)
return (1);
width = ((size & 7) == 4) ? 2 : 1;
if (size & 1) { /* I/O port */
allocp = &sc->sc_ioport_alloc;
size &= ~3;
if ((size & 0xffff0000) == 0)
size |= 0xffff0000;
} else { /* memory */
allocp = &sc->sc_iomem_alloc;
size &= ~15;
}
mask = ~size;
size = mask + 1;
/* Sanity check (must be a power of 2). */
if (size & mask)
return (width);
addr = (*allocp + mask) & ~mask;
*allocp = addr + size;
if (bootverbose)
printf("PCI %u:%u:%u:%u: reg %x: size=%08x: addr=%08x\n",
device_get_unit(sc->sc_dev), bus, slot, func, reg,
size, addr);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func, reg, addr, 4);
if (width == 2)
fsl_pcib_write_config(sc->sc_dev, bus, slot, func, reg + 4,
0, 4);
return (width);
}
static int
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
fsl_pcib_init(struct fsl_pcib_softc *sc, int bus, int maxslot)
{
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
int secbus;
int old_pribus, old_secbus, old_subbus;
int new_pribus, new_secbus, new_subbus;
int slot, func, maxfunc;
int bar, maxbar;
uint16_t vendor, device;
uint8_t command, hdrtype, class, subclass;
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
secbus = bus;
for (slot = 0; slot <= maxslot; slot++) {
maxfunc = 0;
for (func = 0; func <= maxfunc; func++) {
hdrtype = fsl_pcib_read_config(sc->sc_dev, bus, slot,
func, PCIR_HDRTYPE, 1);
if ((hdrtype & PCIM_HDRTYPE) > PCI_MAXHDRTYPE)
continue;
if (func == 0 && (hdrtype & PCIM_MFDEV))
maxfunc = PCI_FUNCMAX;
vendor = fsl_pcib_read_config(sc->sc_dev, bus, slot,
func, PCIR_VENDOR, 2);
device = fsl_pcib_read_config(sc->sc_dev, bus, slot,
func, PCIR_DEVICE, 2);
if (vendor == 0x1957 && device == 0x3fff) {
sc->sc_devfn_tundra = DEVFN(bus, slot, func);
continue;
}
command = fsl_pcib_read_config(sc->sc_dev, bus, slot,
func, PCIR_COMMAND, 1);
command &= ~(PCIM_CMD_MEMEN | PCIM_CMD_PORTEN);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_COMMAND, command, 1);
if (vendor == 0x1106)
fsl_pcib_init_via(sc, device, bus, slot, func);
/* Program the base address registers. */
maxbar = (hdrtype & PCIM_HDRTYPE) ? 1 : 6;
bar = 0;
while (bar < maxbar)
bar += fsl_pcib_init_bar(sc, bus, slot, func,
bar);
/* Put a placeholder interrupt value */
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_INTLINE, PCI_INVALID_IRQ, 1);
command |= PCIM_CMD_MEMEN | PCIM_CMD_PORTEN;
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_COMMAND, command, 1);
/*
* Handle PCI-PCI bridges
*/
class = fsl_pcib_read_config(sc->sc_dev, bus, slot,
func, PCIR_CLASS, 1);
subclass = fsl_pcib_read_config(sc->sc_dev, bus, slot,
func, PCIR_SUBCLASS, 1);
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
/* Allow only proper PCI-PCI briges */
if (class != PCIC_BRIDGE)
continue;
if (subclass != PCIS_BRIDGE_PCI)
continue;
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
secbus++;
/* Program I/O decoder. */
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_IOBASEL_1, sc->sc_ioport_start >> 8, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_IOLIMITL_1, sc->sc_ioport_end >> 8, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_IOBASEH_1, sc->sc_ioport_start >> 16, 2);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_IOLIMITH_1, sc->sc_ioport_end >> 16, 2);
/* Program (non-prefetchable) memory decoder. */
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_MEMBASE_1, sc->sc_iomem_start >> 16, 2);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_MEMLIMIT_1, sc->sc_iomem_end >> 16, 2);
/* Program prefetchable memory decoder. */
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_PMBASEL_1, 0x0010, 2);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_PMLIMITL_1, 0x000f, 2);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_PMBASEH_1, 0x00000000, 4);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_PMLIMITH_1, 0x00000000, 4);
/* Read currect bus register configuration */
old_pribus = fsl_pcib_read_config(sc->sc_dev, bus,
slot, func, PCIR_PRIBUS_1, 1);
old_secbus = fsl_pcib_read_config(sc->sc_dev, bus,
slot, func, PCIR_SECBUS_1, 1);
old_subbus = fsl_pcib_read_config(sc->sc_dev, bus,
slot, func, PCIR_SUBBUS_1, 1);
if (bootverbose)
printf("PCI: reading firmware bus numbers for "
"secbus = %d (bus/sec/sub) = (%d/%d/%d)\n",
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
secbus, old_pribus, old_secbus, old_subbus);
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
new_pribus = bus;
new_secbus = secbus;
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
secbus = fsl_pcib_init(sc, secbus,
(subclass == PCIS_BRIDGE_PCI) ? PCI_SLOTMAX : 0);
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
new_subbus = secbus;
if (bootverbose)
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
printf("PCI: translate firmware bus numbers "
"for secbus %d (%d/%d/%d) -> (%d/%d/%d)\n",
secbus, old_pribus, old_secbus, old_subbus,
new_pribus, new_secbus, new_subbus);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_PRIBUS_1, new_pribus, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_SECBUS_1, new_secbus, 1);
fsl_pcib_write_config(sc->sc_dev, bus, slot, func,
PCIR_SUBBUS_1, new_subbus, 1);
}
}
Fix the interrupt code, broken 7 months ago. The interrupt framework already supported nested PICs, but was limited to having a nested AT-PIC only. With G5 support the need for nested OpenPIC controllers needed to be added. This was done the wrong way and broke the MPC8555 eval system in the process. OFW, as well as FDT, describe the interrupt routing in terms of a controller and an interrupt pin on it. This needs to be mapped to a flat and global resource: the IRQ. The IRQ is the same as the PCI intline and as such needs to be representable in 8 bits. Secondly, ISA support pretty much dictates that IRQ 0-15 should be reserved for ISA interrupts, because of the internal workins of south bridges. Both were broken. This change reverts revision 209298 for a big part and re-implements it simpler. In particular: o The id() method of the PIC I/F is removed again. It's not needed. o The openpic_attach() function has been changed to take the OFW or FDT phandle of the controller as a second argument. All bus attachments that previously used openpic_attach() as the attach method of the device I/F now implement as bus-specific method and pass the phandle_t to the renamed openpic_attach(). o Change powerpc_register_pic() to take a few more arguments. In particular: - Pass the number of IPIs specificly. The number of IRQs carved out for a PIC is the sum of the number of int. pins and IPIs. - Pass a flag indicating whether the PIC is an AT-PIC or not. This tells the interrupt framework whether to assign IRQ 0-15 or some other range. o Until we implement proper multi-pass bus enumeration, we have to handle the case where we need to map from PIC+pin to IRQ *before* the PIC gets registered. This is done in a similar way as before, but rather than carving out 256 IRQs per PIC, we carve out 128 IRQs (124 pins + 4 IPIs). This is supposed to handle the G5 case, but should really be fixed properly using multiple passes. o Have the interrupt framework set root_pic in most cases and not put that burden in PIC drivers (for the most part). o Remove powerpc_ign_lookup() and replace it with powerpc_get_irq(). Remove IGN_SHIFT, INTR_INTLINE and INTR_IGN. Related to the above, fix the Freescale PCI controller driver, broken by the FDT code. Besides not attaching properly, bus numbers were assigned improperly and enumeration was broken in general. This prevented the AT PIC from being discovered and interrupt routing to work properly. Consequently, the ata(4) controller stopped functioning. Fix the driver, and FDT PCI support, enough to get the MPC8555CDS going again. The FDT PCI code needs a whole lot more work. No breakages are expected, but lackiong G5 hardware, it's possible that there are unpleasant side-effects. At least MPC85xx support is back to where it was 7 months ago -- it's amazing how badly support can be broken in just 7 months... Sponsored by: Juniper Networks
2011-01-29 20:58:38 +00:00
return (secbus);
}
static void
fsl_pcib_inbound(struct fsl_pcib_softc *sc, int wnd, int tgt, u_long start,
u_long size, u_long pci_start)
{
uint32_t attr, bar, tar;
KASSERT(wnd > 0, ("%s: inbound window 0 is invalid", __func__));
switch (tgt) {
/* XXX OCP85XX_TGTIF_RAM2, OCP85XX_TGTIF_RAM_INTL should be handled */
case OCP85XX_TGTIF_RAM1:
attr = 0xa0f55000 | (ffsl(size) - 2);
break;
default:
attr = 0;
break;
}
tar = start >> 12;
bar = pci_start >> 12;
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PITAR(wnd), tar);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PIWBEAR(wnd), 0);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PIWBAR(wnd), bar);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PIWAR(wnd), attr);
}
static void
fsl_pcib_outbound(struct fsl_pcib_softc *sc, int wnd, int res, u_long start,
u_long size, u_long pci_start)
{
uint32_t attr, bar, tar;
switch (res) {
case SYS_RES_MEMORY:
attr = 0x80044000 | (ffsl(size) - 2);
break;
case SYS_RES_IOPORT:
attr = 0x80088000 | (ffsl(size) - 2);
break;
default:
attr = 0x0004401f;
break;
}
bar = start >> 12;
tar = pci_start >> 12;
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_POTAR(wnd), tar);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_POTEAR(wnd), 0);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_POWBAR(wnd), bar);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_POWAR(wnd), attr);
}
static void
fsl_pcib_err_init(device_t dev)
{
struct fsl_pcib_softc *sc;
uint16_t sec_stat, dsr;
uint32_t dcr, err_en;
sc = device_get_softc(dev);
sec_stat = fsl_pcib_cfgread(sc, 0, 0, 0, PCIR_SECSTAT_1, 2);
if (sec_stat)
fsl_pcib_cfgwrite(sc, 0, 0, 0, PCIR_SECSTAT_1, 0xffff, 2);
if (sc->sc_pcie) {
/* Clear error bits */
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PEX_MES_IER,
0xffffffff);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PEX_MES_DR,
0xffffffff);
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PEX_ERR_DR,
0xffffffff);
dsr = fsl_pcib_cfgread(sc, 0, 0, 0,
sc->sc_pcie_capreg + PCIER_DEVICE_STA, 2);
if (dsr)
fsl_pcib_cfgwrite(sc, 0, 0, 0,
sc->sc_pcie_capreg + PCIER_DEVICE_STA,
0xffff, 2);
/* Enable all errors reporting */
err_en = 0x00bfff00;
bus_space_write_4(sc->sc_bst, sc->sc_bsh, REG_PEX_ERR_EN,
err_en);
/* Enable error reporting: URR, FER, NFER */
dcr = fsl_pcib_cfgread(sc, 0, 0, 0,
sc->sc_pcie_capreg + PCIER_DEVICE_CTL, 4);
dcr |= PCIEM_CTL_URR_ENABLE | PCIEM_CTL_FER_ENABLE |
PCIEM_CTL_NFER_ENABLE;
fsl_pcib_cfgwrite(sc, 0, 0, 0,
sc->sc_pcie_capreg + PCIER_DEVICE_CTL, dcr, 4);
}
}
static int
fsl_pcib_detach(device_t dev)
{
if (mtx_initialized) {
mtx_destroy(&pcicfg_mtx);
mtx_initialized = 0;
}
return (bus_generic_detach(dev));
}
static int
fsl_pcib_decode_win(phandle_t node, struct fsl_pcib_softc *sc)
{
device_t dev;
int error, i, trgt;
dev = sc->sc_dev;
fsl_pcib_outbound(sc, 0, -1, 0, 0, 0);
/*
* Configure LAW decode windows.
*/
error = law_pci_target(sc->sc_res, &sc->sc_iomem_target,
&sc->sc_ioport_target);
if (error != 0) {
device_printf(dev, "could not retrieve PCI LAW target info\n");
return (error);
}
for (i = 0; i < sc->pci_sc.sc_nrange; i++) {
switch (sc->pci_sc.sc_range[i].pci_hi &
OFW_PCI_PHYS_HI_SPACEMASK) {
case OFW_PCI_PHYS_HI_SPACE_CONFIG:
continue;
case OFW_PCI_PHYS_HI_SPACE_IO:
trgt = sc->sc_ioport_target;
fsl_pcib_outbound(sc, 2, SYS_RES_IOPORT,
sc->pci_sc.sc_range[i].host,
sc->pci_sc.sc_range[i].size,
sc->pci_sc.sc_range[i].pci);
sc->sc_ioport_start = sc->pci_sc.sc_range[i].host;
sc->sc_ioport_end = sc->pci_sc.sc_range[i].host +
sc->pci_sc.sc_range[i].size;
sc->sc_ioport_alloc = 0x1000 + sc->pci_sc.sc_range[i].pci;
break;
case OFW_PCI_PHYS_HI_SPACE_MEM32:
case OFW_PCI_PHYS_HI_SPACE_MEM64:
trgt = sc->sc_iomem_target;
fsl_pcib_outbound(sc, 1, SYS_RES_MEMORY,
sc->pci_sc.sc_range[i].host,
sc->pci_sc.sc_range[i].size,
sc->pci_sc.sc_range[i].pci);
sc->sc_iomem_start = sc->pci_sc.sc_range[i].host;
sc->sc_iomem_end = sc->pci_sc.sc_range[i].host +
sc->pci_sc.sc_range[i].size;
sc->sc_iomem_alloc = sc->pci_sc.sc_range[i].pci;
break;
default:
panic("Unknown range type %#x\n",
sc->pci_sc.sc_range[i].pci_hi &
OFW_PCI_PHYS_HI_SPACEMASK);
}
error = law_enable(trgt, sc->pci_sc.sc_range[i].host,
sc->pci_sc.sc_range[i].size);
if (error != 0) {
device_printf(dev, "could not program LAW for range "
"%d\n", i);
return (error);
}
}
/*
* Set outbout and inbound windows.
*/
fsl_pcib_outbound(sc, 3, -1, 0, 0, 0);
fsl_pcib_outbound(sc, 4, -1, 0, 0, 0);
fsl_pcib_inbound(sc, 1, -1, 0, 0, 0);
fsl_pcib_inbound(sc, 2, -1, 0, 0, 0);
fsl_pcib_inbound(sc, 3, OCP85XX_TGTIF_RAM1, 0,
2U * 1024U * 1024U * 1024U, 0);
return (0);
}