WSL2-Linux-Kernel/arch/powerpc/kernel/isa-bridge.c

270 строки
7.3 KiB
C
Исходник Обычный вид История

[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
/*
* Routines for tracking a legacy ISA bridge
*
* Copyrigh 2007 Benjamin Herrenschmidt <benh@kernel.crashing.org>, IBM Corp.
*
* Some bits and pieces moved over from pci_64.c
*
* Copyrigh 2003 Anton Blanchard <anton@au.ibm.com>, IBM Corp.
*
* 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.
*/
#define DEBUG
#include <linux/kernel.h>
#include <linux/pci.h>
#include <linux/string.h>
#include <linux/export.h>
[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
#include <linux/init.h>
#include <linux/mm.h>
#include <linux/notifier.h>
#include <asm/processor.h>
#include <asm/io.h>
#include <asm/prom.h>
#include <asm/pci-bridge.h>
#include <asm/machdep.h>
#include <asm/ppc-pci.h>
#include <asm/firmware.h>
unsigned long isa_io_base; /* NULL if no ISA bus */
EXPORT_SYMBOL(isa_io_base);
/* Cached ISA bridge dev. */
static struct device_node *isa_bridge_devnode;
struct pci_dev *isa_bridge_pcidev;
EXPORT_SYMBOL_GPL(isa_bridge_pcidev);
#define ISA_SPACE_MASK 0x1
#define ISA_SPACE_IO 0x1
static void __devinit pci_process_ISA_OF_ranges(struct device_node *isa_node,
unsigned long phb_io_base_phys)
{
/* We should get some saner parsing here and remove these structs */
struct pci_address {
u32 a_hi;
u32 a_mid;
u32 a_lo;
};
struct isa_address {
u32 a_hi;
u32 a_lo;
};
struct isa_range {
struct isa_address isa_addr;
struct pci_address pci_addr;
unsigned int size;
};
const struct isa_range *range;
unsigned long pci_addr;
unsigned int isa_addr;
unsigned int size;
int rlen = 0;
range = of_get_property(isa_node, "ranges", &rlen);
if (range == NULL || (rlen < sizeof(struct isa_range)))
goto inval_range;
/* From "ISA Binding to 1275"
* The ranges property is laid out as an array of elements,
* each of which comprises:
* cells 0 - 1: an ISA address
* cells 2 - 4: a PCI address
* (size depending on dev->n_addr_cells)
* cell 5: the size of the range
*/
if ((range->isa_addr.a_hi & ISA_SPACE_MASK) != ISA_SPACE_IO) {
[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
range++;
rlen -= sizeof(struct isa_range);
if (rlen < sizeof(struct isa_range))
goto inval_range;
}
if ((range->isa_addr.a_hi & ISA_SPACE_MASK) != ISA_SPACE_IO)
[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
goto inval_range;
isa_addr = range->isa_addr.a_lo;
pci_addr = (unsigned long) range->pci_addr.a_mid << 32 |
range->pci_addr.a_lo;
/* Assume these are both zero. Note: We could fix that and
* do a proper parsing instead ... oh well, that will do for
* now as nobody uses fancy mappings for ISA bridges
*/
if ((pci_addr != 0) || (isa_addr != 0)) {
printk(KERN_ERR "unexpected isa to pci mapping: %s\n",
__func__);
[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
return;
}
/* Align size and make sure it's cropped to 64K */
size = PAGE_ALIGN(range->size);
if (size > 0x10000)
size = 0x10000;
__ioremap_at(phb_io_base_phys, (void *)ISA_IO_BASE,
size, _PAGE_NO_CACHE|_PAGE_GUARDED);
return;
inval_range:
printk(KERN_ERR "no ISA IO ranges or unexpected isa range, "
[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
"mapping 64k\n");
__ioremap_at(phb_io_base_phys, (void *)ISA_IO_BASE,
0x10000, _PAGE_NO_CACHE|_PAGE_GUARDED);
}
/**
* isa_bridge_find_early - Find and map the ISA IO space early before
* main PCI discovery. This is optionally called by
* the arch code when adding PCI PHBs to get early
* access to ISA IO ports
*/
void __init isa_bridge_find_early(struct pci_controller *hose)
{
struct device_node *np, *parent = NULL, *tmp;
/* If we already have an ISA bridge, bail off */
if (isa_bridge_devnode != NULL)
return;
/* For each "isa" node in the system. Note : we do a search by
* type and not by name. It might be better to do by name but that's
* what the code used to do and I don't want to break too much at
* once. We can look into changing that separately
*/
for_each_node_by_type(np, "isa") {
/* Look for our hose being a parent */
for (parent = of_get_parent(np); parent;) {
if (parent == hose->dn) {
[POWERPC] Rewrite IO allocation & mapping on powerpc64 This rewrites pretty much from scratch the handling of MMIO and PIO space allocations on powerpc64. The main goals are: - Get rid of imalloc and use more common code where possible - Simplify the current mess so that PIO space is allocated and mapped in a single place for PCI bridges - Handle allocation constraints of PIO for all bridges including hot plugged ones within the 2GB space reserved for IO ports, so that devices on hotplugged busses will now work with drivers that assume IO ports fit in an int. - Cleanup and separate tracking of the ISA space in the reserved low 64K of IO space. No ISA -> Nothing mapped there. I booted a cell blade with IDE on PIO and MMIO and a dual G5 so far, that's it :-) With this patch, all allocations are done using the code in mm/vmalloc.c, though we use the low level __get_vm_area with explicit start/stop constraints in order to manage separate areas for vmalloc/vmap, ioremap, and PCI IOs. This greatly simplifies a lot of things, as you can see in the diffstat of that patch :-) A new pair of functions pcibios_map/unmap_io_space() now replace all of the previous code that used to manipulate PCI IOs space. The allocation is done at mapping time, which is now called from scan_phb's, just before the devices are probed (instead of after, which is by itself a bug fix). The only other caller is the PCI hotplug code for hot adding PCI-PCI bridges (slots). imalloc is gone, as is the "sub-allocation" thing, but I do beleive that hotplug should still work in the sense that the space allocation is always done by the PHB, but if you unmap a child bus of this PHB (which seems to be possible), then the code should properly tear down all the HPTE mappings for that area of the PHB allocated IO space. I now always reserve the first 64K of IO space for the bridge with the ISA bus on it. I have moved the code for tracking ISA in a separate file which should also make it smarter if we ever are capable of hot unplugging or re-plugging an ISA bridge. This should have a side effect on platforms like powermac where VGA IOs will no longer work. This is done on purpose though as they would have worked semi-randomly before. The idea at this point is to isolate drivers that might need to access those and fix them by providing a proper function to obtain an offset to the legacy IOs of a given bus. Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Signed-off-by: Paul Mackerras <paulus@samba.org>
2007-06-04 09:15:36 +04:00
of_node_put(parent);
break;
}
tmp = parent;
parent = of_get_parent(parent);
of_node_put(tmp);
}
if (parent != NULL)
break;
}
if (np == NULL)
return;
isa_bridge_devnode = np;
/* Now parse the "ranges" property and setup the ISA mapping */
pci_process_ISA_OF_ranges(np, hose->io_base_phys);
/* Set the global ISA io base to indicate we have an ISA bridge */
isa_io_base = ISA_IO_BASE;
pr_debug("ISA bridge (early) is %s\n", np->full_name);
}
/**
* isa_bridge_find_late - Find and map the ISA IO space upon discovery of
* a new ISA bridge
*/
static void __devinit isa_bridge_find_late(struct pci_dev *pdev,
struct device_node *devnode)
{
struct pci_controller *hose = pci_bus_to_host(pdev->bus);
/* Store ISA device node and PCI device */
isa_bridge_devnode = of_node_get(devnode);
isa_bridge_pcidev = pdev;
/* Now parse the "ranges" property and setup the ISA mapping */
pci_process_ISA_OF_ranges(devnode, hose->io_base_phys);
/* Set the global ISA io base to indicate we have an ISA bridge */
isa_io_base = ISA_IO_BASE;
pr_debug("ISA bridge (late) is %s on %s\n",
devnode->full_name, pci_name(pdev));
}
/**
* isa_bridge_remove - Remove/unmap an ISA bridge
*/
static void isa_bridge_remove(void)
{
pr_debug("ISA bridge removed !\n");
/* Clear the global ISA io base to indicate that we have no more
* ISA bridge. Note that drivers don't quite handle that, though
* we should probably do something about it. But do we ever really
* have ISA bridges being removed on machines using legacy devices ?
*/
isa_io_base = ISA_IO_BASE;
/* Clear references to the bridge */
of_node_put(isa_bridge_devnode);
isa_bridge_devnode = NULL;
isa_bridge_pcidev = NULL;
/* Unmap the ISA area */
__iounmap_at((void *)ISA_IO_BASE, 0x10000);
}
/**
* isa_bridge_notify - Get notified of PCI devices addition/removal
*/
static int __devinit isa_bridge_notify(struct notifier_block *nb,
unsigned long action, void *data)
{
struct device *dev = data;
struct pci_dev *pdev = to_pci_dev(dev);
struct device_node *devnode = pci_device_to_OF_node(pdev);
switch(action) {
case BUS_NOTIFY_ADD_DEVICE:
/* Check if we have an early ISA device, without PCI dev */
if (isa_bridge_devnode && isa_bridge_devnode == devnode &&
!isa_bridge_pcidev) {
pr_debug("ISA bridge PCI attached: %s\n",
pci_name(pdev));
isa_bridge_pcidev = pdev;
}
/* Check if we have no ISA device, and this happens to be one,
* register it as such if it has an OF device
*/
if (!isa_bridge_devnode && devnode && devnode->type &&
!strcmp(devnode->type, "isa"))
isa_bridge_find_late(pdev, devnode);
return 0;
case BUS_NOTIFY_DEL_DEVICE:
/* Check if this our existing ISA device */
if (pdev == isa_bridge_pcidev ||
(devnode && devnode == isa_bridge_devnode))
isa_bridge_remove();
return 0;
}
return 0;
}
static struct notifier_block isa_bridge_notifier = {
.notifier_call = isa_bridge_notify
};
/**
* isa_bridge_init - register to be notified of ISA bridge addition/removal
*
*/
static int __init isa_bridge_init(void)
{
if (firmware_has_feature(FW_FEATURE_ISERIES))
return 0;
bus_register_notifier(&pci_bus_type, &isa_bridge_notifier);
return 0;
}
arch_initcall(isa_bridge_init);