freebsd-nq/sys/amd64/pci/pci_bus.c
John Baldwin e706f7f0c7 Revamp the MSI/MSI-X code a bit to achieve two main goals:
- Simplify the amount of work that has be done for each architecture by
  pushing more of the truly MI code down into the PCI bus driver.
- Don't bind MSI-X indicies to IRQs so that we can allow a driver to map
  multiple MSI-X messages into a single IRQ when handling a message
  shortage.

The changes include:
- Add a new pcib_if method: PCIB_MAP_MSI() which is called by the PCI bus
  to calculate the address and data values for a given MSI/MSI-X IRQ.
  The x86 nexus drivers map this into a call to a new 'msi_map()' function
  in msi.c that does the mapping.
- Retire the pcib_if method PCIB_REMAP_MSIX() and remove the 'index'
  parameter from PCIB_ALLOC_MSIX().  MD code no longer has any knowledge
  of the MSI-X index for a given MSI-X IRQ.
- The PCI bus driver now stores more MSI-X state in a child's ivars.
  Specifically, it now stores an array of IRQs (called "message vectors" in
  the code) that have associated address and data values, and a small
  virtual version of the MSI-X table that specifies the message vector
  that a given MSI-X table entry uses.  Sparse mappings are permitted in
  the virtual table.
- The PCI bus driver now configures the MSI and MSI-X address/data
  registers directly via custom bus_setup_intr() and bus_teardown_intr()
  methods.  pci_setup_intr() invokes PCIB_MAP_MSI() to determine the
  address and data values for a given message as needed.  The MD code
  no longer has to call back down into the PCI bus code to set these
  values from the nexus' bus_setup_intr() handler.
- The PCI bus code provides a callout (pci_remap_msi_irq()) that the MD
  code can call to force the PCI bus to re-invoke PCIB_MAP_MSI() to get
  new values of the address and data fields for a given IRQ.  The x86
  MSI code uses this when an MSI IRQ is moved to a different CPU, requiring
  a new value of the 'address' field.
- The x86 MSI psuedo-driver loses a lot of code, and in fact the separate
  MSI/MSI-X pseudo-PICs are collapsed down into a single MSI PIC driver
  since the only remaining diff between the two is a substring in a
  bootverbose printf.
- The PCI bus driver will now restore MSI-X state (including programming
  entries in the MSI-X table) on device resume.
- The interface for pci_remap_msix() has changed.  Instead of accepting
  indices for the allocated vectors, it accepts a mini-virtual table
  (with a new length parameter).  This table is an array of u_ints, where
  each value specifies which allocated message vector to use for the
  corresponding MSI-X message.  A vector of 0 forces a message to not
  have an associated IRQ.  The device may choose to only use some of the
  IRQs assigned, in which case the unused IRQs must be at the "end" and
  will be released back to the system.  This allows a driver to use the
  same remap table for different shortage values.  For example, if a driver
  wants 4 messages, it can use the same remap table (which only uses the
  first two messages) for the cases when it only gets 2 or 3 messages and
  in the latter case the PCI bus will release the 3rd IRQ back to the
  system.

MFC after:	1 month
2007-05-02 17:50:36 +00:00

418 lines
11 KiB
C

/*-
* Copyright (c) 1997, Stefan Esser <se@freebsd.org>
* All rights reserved.
*
* 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 unmodified, 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.
*
* 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.
*/
#include <sys/cdefs.h>
__FBSDID("$FreeBSD$");
#include "opt_cpu.h"
#include <sys/param.h>
#include <sys/systm.h>
#include <sys/bus.h>
#include <sys/kernel.h>
#include <sys/malloc.h>
#include <sys/module.h>
#include <sys/sysctl.h>
#include <dev/pci/pcivar.h>
#include <dev/pci/pcireg.h>
#include <dev/pci/pcib_private.h>
#include <isa/isavar.h>
#include <machine/legacyvar.h>
#include <machine/pci_cfgreg.h>
#include <machine/resource.h>
#include "pcib_if.h"
int
legacy_pcib_maxslots(device_t dev)
{
return 31;
}
/* read configuration space register */
u_int32_t
legacy_pcib_read_config(device_t dev, int bus, int slot, int func,
int reg, int bytes)
{
return(pci_cfgregread(bus, slot, func, reg, bytes));
}
/* write configuration space register */
void
legacy_pcib_write_config(device_t dev, int bus, int slot, int func,
int reg, u_int32_t data, int bytes)
{
pci_cfgregwrite(bus, slot, func, reg, data, bytes);
}
/* route interrupt */
static int
legacy_pcib_route_interrupt(device_t pcib, device_t dev, int pin)
{
/* No routing possible */
return (PCI_INVALID_IRQ);
}
/* Pass MSI requests up to the nexus. */
static int
legacy_pcib_alloc_msi(device_t pcib, device_t dev, int count, int maxcount,
int *irqs)
{
device_t bus;
bus = device_get_parent(pcib);
return (PCIB_ALLOC_MSI(device_get_parent(bus), dev, count, maxcount,
irqs));
}
static int
legacy_pcib_alloc_msix(device_t pcib, device_t dev, int *irq)
{
device_t bus;
bus = device_get_parent(pcib);
return (PCIB_ALLOC_MSIX(device_get_parent(bus), dev, irq));
}
static int
legacy_pcib_map_msi(device_t pcib, device_t dev, int irq, uint64_t *addr,
uint32_t *data)
{
device_t bus;
bus = device_get_parent(pcib);
return (PCIB_MAP_MSI(device_get_parent(bus), dev, irq, addr, data));
}
static const char *
legacy_pcib_is_host_bridge(int bus, int slot, int func,
uint32_t id, uint8_t class, uint8_t subclass,
uint8_t *busnum)
{
const char *s = NULL;
*busnum = 0;
if (class == PCIC_BRIDGE && subclass == PCIS_BRIDGE_HOST)
s = "Host to PCI bridge";
return s;
}
/*
* Scan the first pci bus for host-pci bridges and add pcib instances
* to the nexus for each bridge.
*/
static void
legacy_pcib_identify(driver_t *driver, device_t parent)
{
int bus, slot, func;
u_int8_t hdrtype;
int found = 0;
int pcifunchigh;
int found824xx = 0;
int found_orion = 0;
device_t child;
devclass_t pci_devclass;
if (pci_cfgregopen() == 0)
return;
/*
* Check to see if we haven't already had a PCI bus added
* via some other means. If we have, bail since otherwise
* we're going to end up duplicating it.
*/
if ((pci_devclass = devclass_find("pci")) &&
devclass_get_device(pci_devclass, 0))
return;
bus = 0;
retry:
for (slot = 0; slot <= PCI_SLOTMAX; slot++) {
func = 0;
hdrtype = legacy_pcib_read_config(0, bus, slot, func,
PCIR_HDRTYPE, 1);
/*
* When enumerating bus devices, the standard says that
* one should check the header type and ignore the slots whose
* header types that the software doesn't know about. We use
* this to filter out devices.
*/
if ((hdrtype & PCIM_HDRTYPE) > PCI_MAXHDRTYPE)
continue;
if ((hdrtype & PCIM_MFDEV) &&
(!found_orion || hdrtype != 0xff))
pcifunchigh = PCI_FUNCMAX;
else
pcifunchigh = 0;
for (func = 0; func <= pcifunchigh; func++) {
/*
* Read the IDs and class from the device.
*/
u_int32_t id;
u_int8_t class, subclass, busnum;
const char *s;
device_t *devs;
int ndevs, i;
id = legacy_pcib_read_config(0, bus, slot, func,
PCIR_DEVVENDOR, 4);
if (id == -1)
continue;
class = legacy_pcib_read_config(0, bus, slot, func,
PCIR_CLASS, 1);
subclass = legacy_pcib_read_config(0, bus, slot, func,
PCIR_SUBCLASS, 1);
s = legacy_pcib_is_host_bridge(bus, slot, func,
id, class, subclass,
&busnum);
if (s == NULL)
continue;
/*
* Check to see if the physical bus has already
* been seen. Eg: hybrid 32 and 64 bit host
* bridges to the same logical bus.
*/
if (device_get_children(parent, &devs, &ndevs) == 0) {
for (i = 0; s != NULL && i < ndevs; i++) {
if (strcmp(device_get_name(devs[i]),
"pcib") != 0)
continue;
if (legacy_get_pcibus(devs[i]) == busnum)
s = NULL;
}
free(devs, M_TEMP);
}
if (s == NULL)
continue;
/*
* Add at priority 100 to make sure we
* go after any motherboard resources
*/
child = BUS_ADD_CHILD(parent, 100,
"pcib", busnum);
device_set_desc(child, s);
legacy_set_pcibus(child, busnum);
found = 1;
if (id == 0x12258086)
found824xx = 1;
if (id == 0x84c48086)
found_orion = 1;
}
}
if (found824xx && bus == 0) {
bus++;
goto retry;
}
/*
* Make sure we add at least one bridge since some old
* hardware doesn't actually have a host-pci bridge device.
* Note that pci_cfgregopen() thinks we have PCI devices..
*/
if (!found) {
if (bootverbose)
printf(
"legacy_pcib_identify: no bridge found, adding pcib0 anyway\n");
child = BUS_ADD_CHILD(parent, 100, "pcib", 0);
legacy_set_pcibus(child, 0);
}
}
static int
legacy_pcib_probe(device_t dev)
{
if (pci_cfgregopen() == 0)
return ENXIO;
return -100;
}
static int
legacy_pcib_attach(device_t dev)
{
int bus;
bus = pcib_get_bus(dev);
device_add_child(dev, "pci", bus);
return bus_generic_attach(dev);
}
int
legacy_pcib_read_ivar(device_t dev, device_t child, int which,
uintptr_t *result)
{
switch (which) {
case PCIB_IVAR_BUS:
*result = legacy_get_pcibus(dev);
return 0;
}
return ENOENT;
}
int
legacy_pcib_write_ivar(device_t dev, device_t child, int which,
uintptr_t value)
{
switch (which) {
case PCIB_IVAR_BUS:
legacy_set_pcibus(dev, value);
return 0;
}
return ENOENT;
}
SYSCTL_DECL(_hw_pci);
static unsigned long legacy_host_mem_start = 0x80000000;
TUNABLE_ULONG("hw.pci.host_mem_start", &legacy_host_mem_start);
SYSCTL_ULONG(_hw_pci, OID_AUTO, host_mem_start, CTLFLAG_RDTUN,
&legacy_host_mem_start, 0x80000000,
"Limit the host bridge memory to being above this address. Must be\n\
set at boot via a tunable.");
struct resource *
legacy_pcib_alloc_resource(device_t dev, device_t child, int type, int *rid,
u_long start, u_long end, u_long count, u_int flags)
{
/*
* If no memory preference is given, use upper 32MB slot most
* bioses use for their memory window. Typically other bridges
* before us get in the way to assert their preferences on memory.
* Hardcoding like this sucks, so a more MD/MI way needs to be
* found to do it. This is typically only used on older laptops
* that don't have pci busses behind pci bridge, so assuming > 32MB
* is liekly OK.
*
* However, this can cause problems for other chipsets, so we make
* this tunable by hw.pci.host_mem_start.
*/
if (type == SYS_RES_MEMORY && start == 0UL && end == ~0UL)
start = legacy_host_mem_start;
if (type == SYS_RES_IOPORT && start == 0UL && end == ~0UL)
start = 0x1000;
return (bus_generic_alloc_resource(dev, child, type, rid, start, end,
count, flags));
}
static device_method_t legacy_pcib_methods[] = {
/* Device interface */
DEVMETHOD(device_identify, legacy_pcib_identify),
DEVMETHOD(device_probe, legacy_pcib_probe),
DEVMETHOD(device_attach, legacy_pcib_attach),
DEVMETHOD(device_shutdown, bus_generic_shutdown),
DEVMETHOD(device_suspend, bus_generic_suspend),
DEVMETHOD(device_resume, bus_generic_resume),
/* Bus interface */
DEVMETHOD(bus_print_child, bus_generic_print_child),
DEVMETHOD(bus_read_ivar, legacy_pcib_read_ivar),
DEVMETHOD(bus_write_ivar, legacy_pcib_write_ivar),
DEVMETHOD(bus_alloc_resource, legacy_pcib_alloc_resource),
DEVMETHOD(bus_release_resource, bus_generic_release_resource),
DEVMETHOD(bus_activate_resource, bus_generic_activate_resource),
DEVMETHOD(bus_deactivate_resource, bus_generic_deactivate_resource),
DEVMETHOD(bus_setup_intr, bus_generic_setup_intr),
DEVMETHOD(bus_teardown_intr, bus_generic_teardown_intr),
/* pcib interface */
DEVMETHOD(pcib_maxslots, legacy_pcib_maxslots),
DEVMETHOD(pcib_read_config, legacy_pcib_read_config),
DEVMETHOD(pcib_write_config, legacy_pcib_write_config),
DEVMETHOD(pcib_route_interrupt, legacy_pcib_route_interrupt),
DEVMETHOD(pcib_alloc_msi, legacy_pcib_alloc_msi),
DEVMETHOD(pcib_release_msi, pcib_release_msi),
DEVMETHOD(pcib_alloc_msix, legacy_pcib_alloc_msix),
DEVMETHOD(pcib_release_msix, pcib_release_msix),
DEVMETHOD(pcib_map_msi, legacy_pcib_map_msi),
{ 0, 0 }
};
static devclass_t hostb_devclass;
DEFINE_CLASS_0(pcib, legacy_pcib_driver, legacy_pcib_methods, 1);
DRIVER_MODULE(pcib, legacy, legacy_pcib_driver, hostb_devclass, 0, 0);
/*
* Install placeholder to claim the resources owned by the
* PCI bus interface. This could be used to extract the
* config space registers in the extreme case where the PnP
* ID is available and the PCI BIOS isn't, but for now we just
* eat the PnP ID and do nothing else.
*
* XXX we should silence this probe, as it will generally confuse
* people.
*/
static struct isa_pnp_id pcibus_pnp_ids[] = {
{ 0x030ad041 /* PNP0A03 */, "PCI Bus" },
{ 0 }
};
static int
pcibus_pnp_probe(device_t dev)
{
int result;
if ((result = ISA_PNP_PROBE(device_get_parent(dev), dev, pcibus_pnp_ids)) <= 0)
device_quiet(dev);
return(result);
}
static int
pcibus_pnp_attach(device_t dev)
{
return(0);
}
static device_method_t pcibus_pnp_methods[] = {
/* Device interface */
DEVMETHOD(device_probe, pcibus_pnp_probe),
DEVMETHOD(device_attach, pcibus_pnp_attach),
DEVMETHOD(device_detach, bus_generic_detach),
DEVMETHOD(device_shutdown, bus_generic_shutdown),
DEVMETHOD(device_suspend, bus_generic_suspend),
DEVMETHOD(device_resume, bus_generic_resume),
{ 0, 0 }
};
static devclass_t pcibus_pnp_devclass;
DEFINE_CLASS_0(pcibus_pnp, pcibus_pnp_driver, pcibus_pnp_methods, 1);
DRIVER_MODULE(pcibus_pnp, isa, pcibus_pnp_driver, pcibus_pnp_devclass, 0, 0);