35a17eb6a8
This is kind of hokey, we could use the hardware provided facilities much better. MSIs are assosciated with MSI Queues. MSI Queues generate interrupts when any MSI assosciated with it is signalled. This suggests a two-tiered IRQ dispatch scheme: MSI Queue interrupt --> queue interrupt handler MSI dispatch --> driver interrupt handler But we just get one-level under Linux currently. What I'd like to do is possibly stick the IRQ actions into a per-MSI-Queue data structure, and dispatch them form there, but the generic IRQ layer doesn't provide a way to do that right now. So, the current kludge is to "ACK" the interrupt by processing the MSI Queue data structures and ACK'ing them, then we run the actual handler like normal. We are wasting a lot of useful information, for example the MSI data and address are provided with ever MSI, as well as a system tick if available. If we could pass this into the IRQ handler it could help with certain things, in particular for PCI-Express error messages. The MSI entries on sparc64 also tell you exactly which bus/device/fn sent the MSI, which would be great for error handling when no registered IRQ handler can service the interrupt. We override the disable/enable IRQ chip methods in sun4v_msi, so we have to call {mask,unmask}_msi_irq() directly from there. This is another ugly wart. Signed-off-by: David S. Miller <davem@davemloft.net>
270 lines
7.7 KiB
C
270 lines
7.7 KiB
C
/* $Id: pbm.h,v 1.27 2001/08/12 13:18:23 davem Exp $
|
|
* pbm.h: UltraSparc PCI controller software state.
|
|
*
|
|
* Copyright (C) 1997, 1998, 1999 David S. Miller (davem@redhat.com)
|
|
*/
|
|
|
|
#ifndef __SPARC64_PBM_H
|
|
#define __SPARC64_PBM_H
|
|
|
|
#include <linux/types.h>
|
|
#include <linux/pci.h>
|
|
#include <linux/ioport.h>
|
|
#include <linux/spinlock.h>
|
|
#include <linux/msi.h>
|
|
|
|
#include <asm/io.h>
|
|
#include <asm/page.h>
|
|
#include <asm/oplib.h>
|
|
#include <asm/prom.h>
|
|
#include <asm/of_device.h>
|
|
#include <asm/iommu.h>
|
|
|
|
/* The abstraction used here is that there are PCI controllers,
|
|
* each with one (Sabre) or two (PSYCHO/SCHIZO) PCI bus modules
|
|
* underneath. Each PCI bus module uses an IOMMU (shared by both
|
|
* PBMs of a controller, or per-PBM), and if a streaming buffer
|
|
* is present, each PCI bus module has it's own. (ie. the IOMMU
|
|
* might be shared between PBMs, the STC is never shared)
|
|
* Furthermore, each PCI bus module controls it's own autonomous
|
|
* PCI bus.
|
|
*/
|
|
|
|
struct pci_controller_info;
|
|
|
|
/* This contains the software state necessary to drive a PCI
|
|
* controller's IOMMU.
|
|
*/
|
|
struct pci_iommu_arena {
|
|
unsigned long *map;
|
|
unsigned int hint;
|
|
unsigned int limit;
|
|
};
|
|
|
|
struct pci_iommu {
|
|
/* This protects the controller's IOMMU and all
|
|
* streaming buffers underneath.
|
|
*/
|
|
spinlock_t lock;
|
|
|
|
struct pci_iommu_arena arena;
|
|
|
|
/* IOMMU page table, a linear array of ioptes. */
|
|
iopte_t *page_table; /* The page table itself. */
|
|
|
|
/* Base PCI memory space address where IOMMU mappings
|
|
* begin.
|
|
*/
|
|
u32 page_table_map_base;
|
|
|
|
/* IOMMU Controller Registers */
|
|
unsigned long iommu_control; /* IOMMU control register */
|
|
unsigned long iommu_tsbbase; /* IOMMU page table base register */
|
|
unsigned long iommu_flush; /* IOMMU page flush register */
|
|
unsigned long iommu_ctxflush; /* IOMMU context flush register */
|
|
|
|
/* This is a register in the PCI controller, which if
|
|
* read will have no side-effects but will guarantee
|
|
* completion of all previous writes into IOMMU/STC.
|
|
*/
|
|
unsigned long write_complete_reg;
|
|
|
|
/* In order to deal with some buggy third-party PCI bridges that
|
|
* do wrong prefetching, we never mark valid mappings as invalid.
|
|
* Instead we point them at this dummy page.
|
|
*/
|
|
unsigned long dummy_page;
|
|
unsigned long dummy_page_pa;
|
|
|
|
/* CTX allocation. */
|
|
unsigned long ctx_lowest_free;
|
|
unsigned long ctx_bitmap[IOMMU_NUM_CTXS / (sizeof(unsigned long) * 8)];
|
|
|
|
/* Here a PCI controller driver describes the areas of
|
|
* PCI memory space where DMA to/from physical memory
|
|
* are addressed. Drivers interrogate the PCI layer
|
|
* if their device has addressing limitations. They
|
|
* do so via pci_dma_supported, and pass in a mask of
|
|
* DMA address bits their device can actually drive.
|
|
*
|
|
* The test for being usable is:
|
|
* (device_mask & dma_addr_mask) == dma_addr_mask
|
|
*/
|
|
u32 dma_addr_mask;
|
|
};
|
|
|
|
extern void pci_iommu_table_init(struct pci_iommu *iommu, int tsbsize, u32 dma_offset, u32 dma_addr_mask);
|
|
|
|
/* This describes a PCI bus module's streaming buffer. */
|
|
struct pci_strbuf {
|
|
int strbuf_enabled; /* Present and using it? */
|
|
|
|
/* Streaming Buffer Control Registers */
|
|
unsigned long strbuf_control; /* STC control register */
|
|
unsigned long strbuf_pflush; /* STC page flush register */
|
|
unsigned long strbuf_fsync; /* STC flush synchronization reg */
|
|
unsigned long strbuf_ctxflush; /* STC context flush register */
|
|
unsigned long strbuf_ctxmatch_base; /* STC context flush match reg */
|
|
unsigned long strbuf_flushflag_pa; /* Physical address of flush flag */
|
|
volatile unsigned long *strbuf_flushflag; /* The flush flag itself */
|
|
|
|
/* And this is the actual flush flag area.
|
|
* We allocate extra because the chips require
|
|
* a 64-byte aligned area.
|
|
*/
|
|
volatile unsigned long __flushflag_buf[(64 + (64 - 1)) / sizeof(long)];
|
|
};
|
|
|
|
#define PCI_STC_FLUSHFLAG_INIT(STC) \
|
|
(*((STC)->strbuf_flushflag) = 0UL)
|
|
#define PCI_STC_FLUSHFLAG_SET(STC) \
|
|
(*((STC)->strbuf_flushflag) != 0UL)
|
|
|
|
/* There can be quite a few ranges and interrupt maps on a PCI
|
|
* segment. Thus...
|
|
*/
|
|
#define PROM_PCIRNG_MAX 64
|
|
#define PROM_PCIIMAP_MAX 64
|
|
|
|
struct pci_pbm_info {
|
|
/* PCI controller we sit under. */
|
|
struct pci_controller_info *parent;
|
|
|
|
/* Physical address base of controller registers. */
|
|
unsigned long controller_regs;
|
|
|
|
/* Physical address base of PBM registers. */
|
|
unsigned long pbm_regs;
|
|
|
|
/* Physical address of DMA sync register, if any. */
|
|
unsigned long sync_reg;
|
|
|
|
/* Opaque 32-bit system bus Port ID. */
|
|
u32 portid;
|
|
|
|
/* Opaque 32-bit handle used for hypervisor calls. */
|
|
u32 devhandle;
|
|
|
|
/* Chipset version information. */
|
|
int chip_type;
|
|
#define PBM_CHIP_TYPE_SABRE 1
|
|
#define PBM_CHIP_TYPE_PSYCHO 2
|
|
#define PBM_CHIP_TYPE_SCHIZO 3
|
|
#define PBM_CHIP_TYPE_SCHIZO_PLUS 4
|
|
#define PBM_CHIP_TYPE_TOMATILLO 5
|
|
int chip_version;
|
|
int chip_revision;
|
|
|
|
/* Name used for top-level resources. */
|
|
char *name;
|
|
|
|
/* OBP specific information. */
|
|
struct device_node *prom_node;
|
|
struct linux_prom_pci_ranges *pbm_ranges;
|
|
int num_pbm_ranges;
|
|
struct linux_prom_pci_intmap *pbm_intmap;
|
|
int num_pbm_intmap;
|
|
struct linux_prom_pci_intmask *pbm_intmask;
|
|
u64 ino_bitmap;
|
|
|
|
/* PBM I/O and Memory space resources. */
|
|
struct resource io_space;
|
|
struct resource mem_space;
|
|
|
|
/* Base of PCI Config space, can be per-PBM or shared. */
|
|
unsigned long config_space;
|
|
|
|
/* State of 66MHz capabilities on this PBM. */
|
|
int is_66mhz_capable;
|
|
int all_devs_66mhz;
|
|
|
|
#ifdef CONFIG_PCI_MSI
|
|
/* MSI info. */
|
|
u32 msiq_num;
|
|
u32 msiq_ent_count;
|
|
u32 msiq_first;
|
|
u32 msiq_first_devino;
|
|
u32 msi_num;
|
|
u32 msi_first;
|
|
u32 msi_data_mask;
|
|
u32 msix_data_width;
|
|
u64 msi32_start;
|
|
u64 msi64_start;
|
|
u32 msi32_len;
|
|
u32 msi64_len;
|
|
void *msi_queues;
|
|
unsigned long *msi_bitmap;
|
|
#endif /* !(CONFIG_PCI_MSI) */
|
|
|
|
/* This PBM's streaming buffer. */
|
|
struct pci_strbuf stc;
|
|
|
|
/* IOMMU state, potentially shared by both PBM segments. */
|
|
struct pci_iommu *iommu;
|
|
|
|
/* PCI slot mapping. */
|
|
unsigned int pci_first_slot;
|
|
|
|
/* Now things for the actual PCI bus probes. */
|
|
unsigned int pci_first_busno;
|
|
unsigned int pci_last_busno;
|
|
struct pci_bus *pci_bus;
|
|
};
|
|
|
|
struct pci_controller_info {
|
|
/* List of all PCI controllers. */
|
|
struct pci_controller_info *next;
|
|
|
|
/* Each controller gets a unique index, used mostly for
|
|
* error logging purposes.
|
|
*/
|
|
int index;
|
|
|
|
/* Do the PBMs both exist in the same PCI domain? */
|
|
int pbms_same_domain;
|
|
|
|
/* The PCI bus modules controlled by us. */
|
|
struct pci_pbm_info pbm_A;
|
|
struct pci_pbm_info pbm_B;
|
|
|
|
/* Operations which are controller specific. */
|
|
void (*scan_bus)(struct pci_controller_info *);
|
|
void (*base_address_update)(struct pci_dev *, int);
|
|
void (*resource_adjust)(struct pci_dev *, struct resource *, struct resource *);
|
|
|
|
#ifdef CONFIG_PCI_MSI
|
|
int (*setup_msi_irq)(unsigned int *virt_irq_p, struct pci_dev *pdev,
|
|
struct msi_desc *entry);
|
|
void (*teardown_msi_irq)(unsigned int virt_irq, struct pci_dev *pdev);
|
|
#endif
|
|
|
|
/* Now things for the actual PCI bus probes. */
|
|
struct pci_ops *pci_ops;
|
|
unsigned int pci_first_busno;
|
|
unsigned int pci_last_busno;
|
|
};
|
|
|
|
/* PCI devices which are not bridges have this placed in their pci_dev
|
|
* sysdata member. This makes OBP aware PCI device drivers easier to
|
|
* code.
|
|
*/
|
|
struct pcidev_cookie {
|
|
struct pci_pbm_info *pbm;
|
|
struct device_node *prom_node;
|
|
struct of_device *op;
|
|
struct linux_prom_pci_registers prom_regs[PROMREG_MAX];
|
|
int num_prom_regs;
|
|
struct linux_prom_pci_registers prom_assignments[PROMREG_MAX];
|
|
int num_prom_assignments;
|
|
#ifdef CONFIG_PCI_MSI
|
|
unsigned int msi_num;
|
|
#endif
|
|
};
|
|
|
|
/* Currently these are the same across all PCI controllers
|
|
* we support. Someday they may not be...
|
|
*/
|
|
#define PCI_IRQ_IGN 0x000007c0 /* Interrupt Group Number */
|
|
#define PCI_IRQ_INO 0x0000003f /* Interrupt Number */
|
|
|
|
#endif /* !(__SPARC64_PBM_H) */
|