remarkable-linux/arch/frv/mb93090-mb00/pci-vdk.c
Bjorn Helgaas 237865f195 PCI: Revert "PCI: Call pci_read_bridge_bases() from core instead of arch code"
Revert dff22d2054 ("PCI: Call pci_read_bridge_bases() from core instead
of arch code").

Reading PCI bridge windows is not arch-specific in itself, but there is PCI
core code that doesn't work correctly if we read them too early.  For
example, Hannes found this case on an ARM Freescale i.mx6 board:

  pci_bus 0000:00: root bus resource [mem 0x01000000-0x01efffff]
  pci 0000:00:00.0: PCI bridge to [bus 01-ff]
  pci 0000:00:00.0: BAR 8: no space for [mem size 0x01000000] (mem window)
  pci 0000:01:00.0: BAR 2: failed to assign [mem size 0x00200000]
  pci 0000:01:00.0: BAR 1: failed to assign [mem size 0x00004000]
  pci 0000:01:00.0: BAR 0: failed to assign [mem size 0x00000100]

The 00:00.0 mem window needs to be at least 3MB: the 01:00.0 device needs
0x204100 of space, and mem windows are megabyte-aligned.

Bus sizing can increase a bridge window size, but never *decrease* it (see
d65245c329 ("PCI: don't shrink bridge resources")).  Prior to
dff22d2054, ARM didn't read bridge windows at all, so the "original size"
was zero, and we assigned a 3MB window.

After dff22d2054, we read the bridge windows before sizing the bus.  The
firmware programmed a 16MB window (size 0x01000000) in 00:00.0, and since
we never decrease the size, we kept 16MB even though we only needed 3MB.
But 16MB doesn't fit in the host bridge aperture, so we failed to assign
space for the window and the downstream devices.

I think this is a defect in the PCI core: we shouldn't rely on the firmware
to assign sensible windows.

Ray reported a similar problem, also on ARM, with Broadcom iProc.

Issues like this are too hard to fix right now, so revert dff22d2054.

Reported-by: Hannes <oe5hpm@gmail.com>
Reported-by: Ray Jui <rjui@broadcom.com>
Link: http://lkml.kernel.org/r/CAAa04yFQEUJm7Jj1qMT57-LG7ZGtnhNDBe=PpSRa70Mj+XhW-A@mail.gmail.com
Link: http://lkml.kernel.org/r/55F75BB8.4070405@broadcom.com
Signed-off-by: Bjorn Helgaas <bhelgaas@google.com>
Acked-by: Yinghai Lu <yinghai@kernel.org>
Acked-by: Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>
2015-09-15 13:18:04 -05:00

420 lines
11 KiB
C

/* pci-vdk.c: MB93090-MB00 (VDK) PCI support
*
* Copyright (C) 2003, 2004 Red Hat, Inc. All Rights Reserved.
* Written by David Howells (dhowells@redhat.com)
*
* 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; either version
* 2 of the License, or (at your option) any later version.
*/
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/pci.h>
#include <linux/init.h>
#include <linux/ioport.h>
#include <linux/delay.h>
#include <asm/segment.h>
#include <asm/io.h>
#include <asm/mb-regs.h>
#include <asm/mb86943a.h>
#include "pci-frv.h"
unsigned int __nongpreldata pci_probe = 1;
struct pci_ops *__nongpreldata pci_root_ops;
/*
* The accessible PCI window does not cover the entire CPU address space, but
* there are devices we want to access outside of that window, so we need to
* insert specific PCI bus resources instead of using the platform-level bus
* resources directly for the PCI root bus.
*
* These are configured and inserted by pcibios_init() and are attached to the
* root bus by pcibios_fixup_bus().
*/
static struct resource pci_ioport_resource = {
.name = "PCI IO",
.start = 0,
.end = IO_SPACE_LIMIT,
.flags = IORESOURCE_IO,
};
static struct resource pci_iomem_resource = {
.name = "PCI mem",
.start = 0,
.end = -1,
.flags = IORESOURCE_MEM,
};
/*
* Functions for accessing PCI configuration space
*/
#define CONFIG_CMD(bus, dev, where) \
(0x80000000 | (bus->number << 16) | (devfn << 8) | (where & ~3))
#define __set_PciCfgAddr(A) writel((A), (volatile void __iomem *) __region_CS1 + 0x80)
#define __get_PciCfgDataB(A) readb((volatile void __iomem *) __region_CS1 + 0x88 + ((A) & 3))
#define __get_PciCfgDataW(A) readw((volatile void __iomem *) __region_CS1 + 0x88 + ((A) & 2))
#define __get_PciCfgDataL(A) readl((volatile void __iomem *) __region_CS1 + 0x88)
#define __set_PciCfgDataB(A,V) \
writeb((V), (volatile void __iomem *) __region_CS1 + 0x88 + (3 - ((A) & 3)))
#define __set_PciCfgDataW(A,V) \
writew((V), (volatile void __iomem *) __region_CS1 + 0x88 + (2 - ((A) & 2)))
#define __set_PciCfgDataL(A,V) \
writel((V), (volatile void __iomem *) __region_CS1 + 0x88)
#define __get_PciBridgeDataB(A) readb((volatile void __iomem *) __region_CS1 + 0x800 + (A))
#define __get_PciBridgeDataW(A) readw((volatile void __iomem *) __region_CS1 + 0x800 + (A))
#define __get_PciBridgeDataL(A) readl((volatile void __iomem *) __region_CS1 + 0x800 + (A))
#define __set_PciBridgeDataB(A,V) writeb((V), (volatile void __iomem *) __region_CS1 + 0x800 + (A))
#define __set_PciBridgeDataW(A,V) writew((V), (volatile void __iomem *) __region_CS1 + 0x800 + (A))
#define __set_PciBridgeDataL(A,V) writel((V), (volatile void __iomem *) __region_CS1 + 0x800 + (A))
static inline int __query(const struct pci_dev *dev)
{
// return dev->bus->number==0 && (dev->devfn==PCI_DEVFN(0,0));
// return dev->bus->number==1;
// return dev->bus->number==0 &&
// (dev->devfn==PCI_DEVFN(2,0) || dev->devfn==PCI_DEVFN(3,0));
return 0;
}
/*****************************************************************************/
/*
*
*/
static int pci_frv_read_config(struct pci_bus *bus, unsigned int devfn, int where, int size,
u32 *val)
{
u32 _value;
if (bus->number == 0 && devfn == PCI_DEVFN(0, 0)) {
_value = __get_PciBridgeDataL(where & ~3);
}
else {
__set_PciCfgAddr(CONFIG_CMD(bus, devfn, where));
_value = __get_PciCfgDataL(where & ~3);
}
switch (size) {
case 1:
_value = _value >> ((where & 3) * 8);
break;
case 2:
_value = _value >> ((where & 2) * 8);
break;
case 4:
break;
default:
BUG();
}
*val = _value;
return PCIBIOS_SUCCESSFUL;
}
static int pci_frv_write_config(struct pci_bus *bus, unsigned int devfn, int where, int size,
u32 value)
{
switch (size) {
case 1:
if (bus->number == 0 && devfn == PCI_DEVFN(0, 0)) {
__set_PciBridgeDataB(where, value);
}
else {
__set_PciCfgAddr(CONFIG_CMD(bus, devfn, where));
__set_PciCfgDataB(where, value);
}
break;
case 2:
if (bus->number == 0 && devfn == PCI_DEVFN(0, 0)) {
__set_PciBridgeDataW(where, value);
}
else {
__set_PciCfgAddr(CONFIG_CMD(bus, devfn, where));
__set_PciCfgDataW(where, value);
}
break;
case 4:
if (bus->number == 0 && devfn == PCI_DEVFN(0, 0)) {
__set_PciBridgeDataL(where, value);
}
else {
__set_PciCfgAddr(CONFIG_CMD(bus, devfn, where));
__set_PciCfgDataL(where, value);
}
break;
default:
BUG();
}
return PCIBIOS_SUCCESSFUL;
}
static struct pci_ops pci_direct_frv = {
.read = pci_frv_read_config,
.write = pci_frv_write_config,
};
/*
* Before we decide to use direct hardware access mechanisms, we try to do some
* trivial checks to ensure it at least _seems_ to be working -- we just test
* whether bus 00 contains a host bridge (this is similar to checking
* techniques used in XFree86, but ours should be more reliable since we
* attempt to make use of direct access hints provided by the PCI BIOS).
*
* This should be close to trivial, but it isn't, because there are buggy
* chipsets (yes, you guessed it, by Intel and Compaq) that have no class ID.
*/
static int __init pci_sanity_check(struct pci_ops *o)
{
struct pci_bus bus; /* Fake bus and device */
u32 id;
bus.number = 0;
if (o->read(&bus, 0, PCI_VENDOR_ID, 4, &id) == PCIBIOS_SUCCESSFUL) {
printk("PCI: VDK Bridge device:vendor: %08x\n", id);
if (id == 0x200e10cf)
return 1;
}
printk("PCI: VDK Bridge: Sanity check failed\n");
return 0;
}
static struct pci_ops * __init pci_check_direct(void)
{
unsigned long flags;
local_irq_save(flags);
/* check if access works */
if (pci_sanity_check(&pci_direct_frv)) {
local_irq_restore(flags);
printk("PCI: Using configuration frv\n");
// request_mem_region(0xBE040000, 256, "FRV bridge");
// request_mem_region(0xBFFFFFF4, 12, "PCI frv");
return &pci_direct_frv;
}
local_irq_restore(flags);
return NULL;
}
/*
* Exceptions for specific devices. Usually work-arounds for fatal design flaws.
*/
static void __init pci_fixup_umc_ide(struct pci_dev *d)
{
/*
* UM8886BF IDE controller sets region type bits incorrectly,
* therefore they look like memory despite of them being I/O.
*/
int i;
printk("PCI: Fixing base address flags for device %s\n", pci_name(d));
for(i=0; i<4; i++)
d->resource[i].flags |= PCI_BASE_ADDRESS_SPACE_IO;
}
static void pci_fixup_ide_bases(struct pci_dev *d)
{
int i;
/*
* PCI IDE controllers use non-standard I/O port decoding, respect it.
*/
if ((d->class >> 8) != PCI_CLASS_STORAGE_IDE)
return;
printk("PCI: IDE base address fixup for %s\n", pci_name(d));
for(i=0; i<4; i++) {
struct resource *r = &d->resource[i];
if ((r->start & ~0x80) == 0x374) {
r->start |= 2;
r->end = r->start;
}
}
}
static void pci_fixup_ide_trash(struct pci_dev *d)
{
int i;
/*
* There exist PCI IDE controllers which have utter garbage
* in first four base registers. Ignore that.
*/
printk("PCI: IDE base address trash cleared for %s\n", pci_name(d));
for(i=0; i<4; i++)
d->resource[i].start = d->resource[i].end = d->resource[i].flags = 0;
}
static void pci_fixup_latency(struct pci_dev *d)
{
/*
* SiS 5597 and 5598 chipsets require latency timer set to
* at most 32 to avoid lockups.
*/
DBG("PCI: Setting max latency to 32\n");
pcibios_max_latency = 32;
}
DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_UMC, PCI_DEVICE_ID_UMC_UM8886BF, pci_fixup_umc_ide);
DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_SI, PCI_DEVICE_ID_SI_5513, pci_fixup_ide_trash);
DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_SI, PCI_DEVICE_ID_SI_5597, pci_fixup_latency);
DECLARE_PCI_FIXUP_HEADER(PCI_VENDOR_ID_SI, PCI_DEVICE_ID_SI_5598, pci_fixup_latency);
DECLARE_PCI_FIXUP_HEADER(PCI_ANY_ID, PCI_ANY_ID, pci_fixup_ide_bases);
/*
* Called after each bus is probed, but before its children
* are examined.
*/
void pcibios_fixup_bus(struct pci_bus *bus)
{
#if 0
printk("### PCIBIOS_FIXUP_BUS(%d)\n",bus->number);
#endif
pci_read_bridge_bases(bus);
if (bus->number == 0) {
struct pci_dev *dev;
list_for_each_entry(dev, &bus->devices, bus_list) {
if (dev->devfn == 0) {
dev->resource[0].start = 0;
dev->resource[0].end = 0;
}
}
}
}
/*
* Initialization. Try all known PCI access methods. Note that we support
* using both PCI BIOS and direct access: in such cases, we use I/O ports
* to access config space, but we still keep BIOS order of cards to be
* compatible with 2.0.X. This should go away some day.
*/
int __init pcibios_init(void)
{
struct pci_bus *bus;
struct pci_ops *dir = NULL;
LIST_HEAD(resources);
if (!mb93090_mb00_detected)
return -ENXIO;
__reg_MB86943_sl_ctl |= MB86943_SL_CTL_DRCT_MASTER_SWAP | MB86943_SL_CTL_DRCT_SLAVE_SWAP;
__reg_MB86943_ecs_base(1) = ((__region_CS2 + 0x01000000) >> 9) | 0x08000000;
__reg_MB86943_ecs_base(2) = ((__region_CS2 + 0x00000000) >> 9) | 0x08000000;
*(volatile uint32_t *) (__region_CS1 + 0x848) = 0xe0000000;
*(volatile uint32_t *) (__region_CS1 + 0x8b8) = 0x00000000;
__reg_MB86943_sl_pci_io_base = (__region_CS2 + 0x04000000) >> 9;
__reg_MB86943_sl_pci_mem_base = (__region_CS2 + 0x08000000) >> 9;
__reg_MB86943_pci_sl_io_base = __region_CS2 + 0x04000000;
__reg_MB86943_pci_sl_mem_base = __region_CS2 + 0x08000000;
mb();
/* enable PCI arbitration */
__reg_MB86943_pci_arbiter = MB86943_PCIARB_EN;
pci_ioport_resource.start = (__reg_MB86943_sl_pci_io_base << 9) & 0xfffffc00;
pci_ioport_resource.end = (__reg_MB86943_sl_pci_io_range << 9) | 0x3ff;
pci_ioport_resource.end += pci_ioport_resource.start;
printk("PCI IO window: %08llx-%08llx\n",
(unsigned long long) pci_ioport_resource.start,
(unsigned long long) pci_ioport_resource.end);
pci_iomem_resource.start = (__reg_MB86943_sl_pci_mem_base << 9) & 0xfffffc00;
pci_iomem_resource.end = (__reg_MB86943_sl_pci_mem_range << 9) | 0x3ff;
pci_iomem_resource.end += pci_iomem_resource.start;
/* Reserve somewhere to write to flush posted writes. This is used by
* __flush_PCI_writes() from asm/io.h to force the write FIFO in the
* CPU-PCI bridge to flush as this doesn't happen automatically when a
* read is performed on the MB93090 development kit motherboard.
*/
pci_iomem_resource.start += 0x400;
printk("PCI MEM window: %08llx-%08llx\n",
(unsigned long long) pci_iomem_resource.start,
(unsigned long long) pci_iomem_resource.end);
printk("PCI DMA memory: %08lx-%08lx\n",
dma_coherent_mem_start, dma_coherent_mem_end);
if (insert_resource(&iomem_resource, &pci_iomem_resource) < 0)
panic("Unable to insert PCI IOMEM resource\n");
if (insert_resource(&ioport_resource, &pci_ioport_resource) < 0)
panic("Unable to insert PCI IOPORT resource\n");
if (!pci_probe)
return -ENXIO;
dir = pci_check_direct();
if (dir)
pci_root_ops = dir;
else {
printk("PCI: No PCI bus detected\n");
return -ENXIO;
}
printk("PCI: Probing PCI hardware\n");
pci_add_resource(&resources, &pci_ioport_resource);
pci_add_resource(&resources, &pci_iomem_resource);
bus = pci_scan_root_bus(NULL, 0, pci_root_ops, NULL, &resources);
pcibios_irq_init();
pcibios_fixup_irqs();
pcibios_resource_survey();
if (!bus)
return 0;
pci_bus_add_devices(bus);
return 0;
}
arch_initcall(pcibios_init);
char * __init pcibios_setup(char *str)
{
if (!strcmp(str, "off")) {
pci_probe = 0;
return NULL;
}
return str;
}
int pcibios_enable_device(struct pci_dev *dev, int mask)
{
int err;
if ((err = pci_enable_resources(dev, mask)) < 0)
return err;
if (!dev->msi_enabled)
pcibios_enable_irq(dev);
return 0;
}