linux/arch/powerpc/sysdev/mpic.c

1574 lines
39 KiB
C
Raw Normal View History

/*
* arch/powerpc/kernel/mpic.c
*
* Driver for interrupt controllers following the OpenPIC standard, the
* common implementation beeing IBM's MPIC. This driver also can deal
* with various broken implementations of this HW.
*
* Copyright (C) 2004 Benjamin Herrenschmidt, IBM Corp.
*
* This file is subject to the terms and conditions of the GNU General Public
* License. See the file COPYING in the main directory of this archive
* for more details.
*/
#undef DEBUG
#undef DEBUG_IPI
#undef DEBUG_IRQ
#undef DEBUG_LOW
#include <linux/types.h>
#include <linux/kernel.h>
#include <linux/init.h>
#include <linux/irq.h>
#include <linux/smp.h>
#include <linux/interrupt.h>
#include <linux/bootmem.h>
#include <linux/spinlock.h>
#include <linux/pci.h>
#include <asm/ptrace.h>
#include <asm/signal.h>
#include <asm/io.h>
#include <asm/pgtable.h>
#include <asm/irq.h>
#include <asm/machdep.h>
#include <asm/mpic.h>
#include <asm/smp.h>
#include "mpic.h"
#ifdef DEBUG
#define DBG(fmt...) printk(fmt)
#else
#define DBG(fmt...)
#endif
static struct mpic *mpics;
static struct mpic *mpic_primary;
static DEFINE_SPINLOCK(mpic_lock);
#ifdef CONFIG_PPC32 /* XXX for now */
#ifdef CONFIG_IRQ_ALL_CPUS
#define distribute_irqs (1)
#else
#define distribute_irqs (0)
#endif
#endif
#ifdef CONFIG_MPIC_WEIRD
static u32 mpic_infos[][MPIC_IDX_END] = {
[0] = { /* Original OpenPIC compatible MPIC */
MPIC_GREG_BASE,
MPIC_GREG_FEATURE_0,
MPIC_GREG_GLOBAL_CONF_0,
MPIC_GREG_VENDOR_ID,
MPIC_GREG_IPI_VECTOR_PRI_0,
MPIC_GREG_IPI_STRIDE,
MPIC_GREG_SPURIOUS,
MPIC_GREG_TIMER_FREQ,
MPIC_TIMER_BASE,
MPIC_TIMER_STRIDE,
MPIC_TIMER_CURRENT_CNT,
MPIC_TIMER_BASE_CNT,
MPIC_TIMER_VECTOR_PRI,
MPIC_TIMER_DESTINATION,
MPIC_CPU_BASE,
MPIC_CPU_STRIDE,
MPIC_CPU_IPI_DISPATCH_0,
MPIC_CPU_IPI_DISPATCH_STRIDE,
MPIC_CPU_CURRENT_TASK_PRI,
MPIC_CPU_WHOAMI,
MPIC_CPU_INTACK,
MPIC_CPU_EOI,
MPIC_IRQ_BASE,
MPIC_IRQ_STRIDE,
MPIC_IRQ_VECTOR_PRI,
MPIC_VECPRI_VECTOR_MASK,
MPIC_VECPRI_POLARITY_POSITIVE,
MPIC_VECPRI_POLARITY_NEGATIVE,
MPIC_VECPRI_SENSE_LEVEL,
MPIC_VECPRI_SENSE_EDGE,
MPIC_VECPRI_POLARITY_MASK,
MPIC_VECPRI_SENSE_MASK,
MPIC_IRQ_DESTINATION
},
[1] = { /* Tsi108/109 PIC */
TSI108_GREG_BASE,
TSI108_GREG_FEATURE_0,
TSI108_GREG_GLOBAL_CONF_0,
TSI108_GREG_VENDOR_ID,
TSI108_GREG_IPI_VECTOR_PRI_0,
TSI108_GREG_IPI_STRIDE,
TSI108_GREG_SPURIOUS,
TSI108_GREG_TIMER_FREQ,
TSI108_TIMER_BASE,
TSI108_TIMER_STRIDE,
TSI108_TIMER_CURRENT_CNT,
TSI108_TIMER_BASE_CNT,
TSI108_TIMER_VECTOR_PRI,
TSI108_TIMER_DESTINATION,
TSI108_CPU_BASE,
TSI108_CPU_STRIDE,
TSI108_CPU_IPI_DISPATCH_0,
TSI108_CPU_IPI_DISPATCH_STRIDE,
TSI108_CPU_CURRENT_TASK_PRI,
TSI108_CPU_WHOAMI,
TSI108_CPU_INTACK,
TSI108_CPU_EOI,
TSI108_IRQ_BASE,
TSI108_IRQ_STRIDE,
TSI108_IRQ_VECTOR_PRI,
TSI108_VECPRI_VECTOR_MASK,
TSI108_VECPRI_POLARITY_POSITIVE,
TSI108_VECPRI_POLARITY_NEGATIVE,
TSI108_VECPRI_SENSE_LEVEL,
TSI108_VECPRI_SENSE_EDGE,
TSI108_VECPRI_POLARITY_MASK,
TSI108_VECPRI_SENSE_MASK,
TSI108_IRQ_DESTINATION
},
};
#define MPIC_INFO(name) mpic->hw_set[MPIC_IDX_##name]
#else /* CONFIG_MPIC_WEIRD */
#define MPIC_INFO(name) MPIC_##name
#endif /* CONFIG_MPIC_WEIRD */
/*
* Register accessor functions
*/
static inline u32 _mpic_read(enum mpic_reg_type type,
struct mpic_reg_bank *rb,
unsigned int reg)
{
switch(type) {
#ifdef CONFIG_PPC_DCR
case mpic_access_dcr:
return dcr_read(rb->dhost,
rb->dbase + reg + rb->doff);
#endif
case mpic_access_mmio_be:
return in_be32(rb->base + (reg >> 2));
case mpic_access_mmio_le:
default:
return in_le32(rb->base + (reg >> 2));
}
}
static inline void _mpic_write(enum mpic_reg_type type,
struct mpic_reg_bank *rb,
unsigned int reg, u32 value)
{
switch(type) {
#ifdef CONFIG_PPC_DCR
case mpic_access_dcr:
return dcr_write(rb->dhost,
rb->dbase + reg + rb->doff, value);
#endif
case mpic_access_mmio_be:
return out_be32(rb->base + (reg >> 2), value);
case mpic_access_mmio_le:
default:
return out_le32(rb->base + (reg >> 2), value);
}
}
static inline u32 _mpic_ipi_read(struct mpic *mpic, unsigned int ipi)
{
enum mpic_reg_type type = mpic->reg_type;
unsigned int offset = MPIC_INFO(GREG_IPI_VECTOR_PRI_0) +
(ipi * MPIC_INFO(GREG_IPI_STRIDE));
if ((mpic->flags & MPIC_BROKEN_IPI) && type == mpic_access_mmio_le)
type = mpic_access_mmio_be;
return _mpic_read(type, &mpic->gregs, offset);
}
static inline void _mpic_ipi_write(struct mpic *mpic, unsigned int ipi, u32 value)
{
unsigned int offset = MPIC_INFO(GREG_IPI_VECTOR_PRI_0) +
(ipi * MPIC_INFO(GREG_IPI_STRIDE));
_mpic_write(mpic->reg_type, &mpic->gregs, offset, value);
}
static inline u32 _mpic_cpu_read(struct mpic *mpic, unsigned int reg)
{
unsigned int cpu = 0;
if (mpic->flags & MPIC_PRIMARY)
cpu = hard_smp_processor_id();
return _mpic_read(mpic->reg_type, &mpic->cpuregs[cpu], reg);
}
static inline void _mpic_cpu_write(struct mpic *mpic, unsigned int reg, u32 value)
{
unsigned int cpu = 0;
if (mpic->flags & MPIC_PRIMARY)
cpu = hard_smp_processor_id();
_mpic_write(mpic->reg_type, &mpic->cpuregs[cpu], reg, value);
}
static inline u32 _mpic_irq_read(struct mpic *mpic, unsigned int src_no, unsigned int reg)
{
unsigned int isu = src_no >> mpic->isu_shift;
unsigned int idx = src_no & mpic->isu_mask;
return _mpic_read(mpic->reg_type, &mpic->isus[isu],
reg + (idx * MPIC_INFO(IRQ_STRIDE)));
}
static inline void _mpic_irq_write(struct mpic *mpic, unsigned int src_no,
unsigned int reg, u32 value)
{
unsigned int isu = src_no >> mpic->isu_shift;
unsigned int idx = src_no & mpic->isu_mask;
_mpic_write(mpic->reg_type, &mpic->isus[isu],
reg + (idx * MPIC_INFO(IRQ_STRIDE)), value);
}
#define mpic_read(b,r) _mpic_read(mpic->reg_type,&(b),(r))
#define mpic_write(b,r,v) _mpic_write(mpic->reg_type,&(b),(r),(v))
#define mpic_ipi_read(i) _mpic_ipi_read(mpic,(i))
#define mpic_ipi_write(i,v) _mpic_ipi_write(mpic,(i),(v))
#define mpic_cpu_read(i) _mpic_cpu_read(mpic,(i))
#define mpic_cpu_write(i,v) _mpic_cpu_write(mpic,(i),(v))
#define mpic_irq_read(s,r) _mpic_irq_read(mpic,(s),(r))
#define mpic_irq_write(s,r,v) _mpic_irq_write(mpic,(s),(r),(v))
/*
* Low level utility functions
*/
static void _mpic_map_mmio(struct mpic *mpic, unsigned long phys_addr,
struct mpic_reg_bank *rb, unsigned int offset,
unsigned int size)
{
rb->base = ioremap(phys_addr + offset, size);
BUG_ON(rb->base == NULL);
}
#ifdef CONFIG_PPC_DCR
static void _mpic_map_dcr(struct mpic *mpic, struct mpic_reg_bank *rb,
unsigned int offset, unsigned int size)
{
rb->dbase = mpic->dcr_base;
rb->doff = offset;
rb->dhost = dcr_map(mpic->of_node, rb->dbase + rb->doff, size);
BUG_ON(!DCR_MAP_OK(rb->dhost));
}
static inline void mpic_map(struct mpic *mpic, unsigned long phys_addr,
struct mpic_reg_bank *rb, unsigned int offset,
unsigned int size)
{
if (mpic->flags & MPIC_USES_DCR)
_mpic_map_dcr(mpic, rb, offset, size);
else
_mpic_map_mmio(mpic, phys_addr, rb, offset, size);
}
#else /* CONFIG_PPC_DCR */
#define mpic_map(m,p,b,o,s) _mpic_map_mmio(m,p,b,o,s)
#endif /* !CONFIG_PPC_DCR */
/* Check if we have one of those nice broken MPICs with a flipped endian on
* reads from IPI registers
*/
static void __init mpic_test_broken_ipi(struct mpic *mpic)
{
u32 r;
mpic_write(mpic->gregs, MPIC_INFO(GREG_IPI_VECTOR_PRI_0), MPIC_VECPRI_MASK);
r = mpic_read(mpic->gregs, MPIC_INFO(GREG_IPI_VECTOR_PRI_0));
if (r == le32_to_cpu(MPIC_VECPRI_MASK)) {
printk(KERN_INFO "mpic: Detected reversed IPI registers\n");
mpic->flags |= MPIC_BROKEN_IPI;
}
}
#ifdef CONFIG_MPIC_U3_HT_IRQS
/* Test if an interrupt is sourced from HyperTransport (used on broken U3s)
* to force the edge setting on the MPIC and do the ack workaround.
*/
static inline int mpic_is_ht_interrupt(struct mpic *mpic, unsigned int source)
{
if (source >= 128 || !mpic->fixups)
return 0;
return mpic->fixups[source].base != NULL;
}
static inline void mpic_ht_end_irq(struct mpic *mpic, unsigned int source)
{
struct mpic_irq_fixup *fixup = &mpic->fixups[source];
if (fixup->applebase) {
unsigned int soff = (fixup->index >> 3) & ~3;
unsigned int mask = 1U << (fixup->index & 0x1f);
writel(mask, fixup->applebase + soff);
} else {
spin_lock(&mpic->fixup_lock);
writeb(0x11 + 2 * fixup->index, fixup->base + 2);
writel(fixup->data, fixup->base + 4);
spin_unlock(&mpic->fixup_lock);
}
}
static void mpic_startup_ht_interrupt(struct mpic *mpic, unsigned int source,
unsigned int irqflags)
{
struct mpic_irq_fixup *fixup = &mpic->fixups[source];
unsigned long flags;
u32 tmp;
if (fixup->base == NULL)
return;
DBG("startup_ht_interrupt(0x%x, 0x%x) index: %d\n",
source, irqflags, fixup->index);
spin_lock_irqsave(&mpic->fixup_lock, flags);
/* Enable and configure */
writeb(0x10 + 2 * fixup->index, fixup->base + 2);
tmp = readl(fixup->base + 4);
tmp &= ~(0x23U);
if (irqflags & IRQ_LEVEL)
tmp |= 0x22;
writel(tmp, fixup->base + 4);
spin_unlock_irqrestore(&mpic->fixup_lock, flags);
#ifdef CONFIG_PM
/* use the lowest bit inverted to the actual HW,
* set if this fixup was enabled, clear otherwise */
mpic->save_data[source].fixup_data = tmp | 1;
#endif
}
static void mpic_shutdown_ht_interrupt(struct mpic *mpic, unsigned int source,
unsigned int irqflags)
{
struct mpic_irq_fixup *fixup = &mpic->fixups[source];
unsigned long flags;
u32 tmp;
if (fixup->base == NULL)
return;
DBG("shutdown_ht_interrupt(0x%x, 0x%x)\n", source, irqflags);
/* Disable */
spin_lock_irqsave(&mpic->fixup_lock, flags);
writeb(0x10 + 2 * fixup->index, fixup->base + 2);
tmp = readl(fixup->base + 4);
tmp |= 1;
writel(tmp, fixup->base + 4);
spin_unlock_irqrestore(&mpic->fixup_lock, flags);
#ifdef CONFIG_PM
/* use the lowest bit inverted to the actual HW,
* set if this fixup was enabled, clear otherwise */
mpic->save_data[source].fixup_data = tmp & ~1;
#endif
}
#ifdef CONFIG_PCI_MSI
static void __init mpic_scan_ht_msi(struct mpic *mpic, u8 __iomem *devbase,
unsigned int devfn)
{
u8 __iomem *base;
u8 pos, flags;
u64 addr = 0;
for (pos = readb(devbase + PCI_CAPABILITY_LIST); pos != 0;
pos = readb(devbase + pos + PCI_CAP_LIST_NEXT)) {
u8 id = readb(devbase + pos + PCI_CAP_LIST_ID);
if (id == PCI_CAP_ID_HT) {
id = readb(devbase + pos + 3);
if ((id & HT_5BIT_CAP_MASK) == HT_CAPTYPE_MSI_MAPPING)
break;
}
}
if (pos == 0)
return;
base = devbase + pos;
flags = readb(base + HT_MSI_FLAGS);
if (!(flags & HT_MSI_FLAGS_FIXED)) {
addr = readl(base + HT_MSI_ADDR_LO) & HT_MSI_ADDR_LO_MASK;
addr = addr | ((u64)readl(base + HT_MSI_ADDR_HI) << 32);
}
printk(KERN_DEBUG "mpic: - HT:%02x.%x %s MSI mapping found @ 0x%lx\n",
PCI_SLOT(devfn), PCI_FUNC(devfn),
flags & HT_MSI_FLAGS_ENABLE ? "enabled" : "disabled", addr);
if (!(flags & HT_MSI_FLAGS_ENABLE))
writeb(flags | HT_MSI_FLAGS_ENABLE, base + HT_MSI_FLAGS);
}
#else
static void __init mpic_scan_ht_msi(struct mpic *mpic, u8 __iomem *devbase,
unsigned int devfn)
{
return;
}
#endif
static void __init mpic_scan_ht_pic(struct mpic *mpic, u8 __iomem *devbase,
unsigned int devfn, u32 vdid)
{
int i, irq, n;
u8 __iomem *base;
u32 tmp;
u8 pos;
for (pos = readb(devbase + PCI_CAPABILITY_LIST); pos != 0;
pos = readb(devbase + pos + PCI_CAP_LIST_NEXT)) {
u8 id = readb(devbase + pos + PCI_CAP_LIST_ID);
if (id == PCI_CAP_ID_HT) {
id = readb(devbase + pos + 3);
if ((id & HT_5BIT_CAP_MASK) == HT_CAPTYPE_IRQ)
break;
}
}
if (pos == 0)
return;
base = devbase + pos;
writeb(0x01, base + 2);
n = (readl(base + 4) >> 16) & 0xff;
printk(KERN_INFO "mpic: - HT:%02x.%x [0x%02x] vendor %04x device %04x"
" has %d irqs\n",
devfn >> 3, devfn & 0x7, pos, vdid & 0xffff, vdid >> 16, n + 1);
for (i = 0; i <= n; i++) {
writeb(0x10 + 2 * i, base + 2);
tmp = readl(base + 4);
irq = (tmp >> 16) & 0xff;
DBG("HT PIC index 0x%x, irq 0x%x, tmp: %08x\n", i, irq, tmp);
/* mask it , will be unmasked later */
tmp |= 0x1;
writel(tmp, base + 4);
mpic->fixups[irq].index = i;
mpic->fixups[irq].base = base;
/* Apple HT PIC has a non-standard way of doing EOIs */
if ((vdid & 0xffff) == 0x106b)
mpic->fixups[irq].applebase = devbase + 0x60;
else
mpic->fixups[irq].applebase = NULL;
writeb(0x11 + 2 * i, base + 2);
mpic->fixups[irq].data = readl(base + 4) | 0x80000000;
}
}
static void __init mpic_scan_ht_pics(struct mpic *mpic)
{
unsigned int devfn;
u8 __iomem *cfgspace;
printk(KERN_INFO "mpic: Setting up HT PICs workarounds for U3/U4\n");
/* Allocate fixups array */
mpic->fixups = alloc_bootmem(128 * sizeof(struct mpic_irq_fixup));
BUG_ON(mpic->fixups == NULL);
memset(mpic->fixups, 0, 128 * sizeof(struct mpic_irq_fixup));
/* Init spinlock */
spin_lock_init(&mpic->fixup_lock);
/* Map U3 config space. We assume all IO-APICs are on the primary bus
* so we only need to map 64kB.
*/
cfgspace = ioremap(0xf2000000, 0x10000);
BUG_ON(cfgspace == NULL);
/* Now we scan all slots. We do a very quick scan, we read the header
* type, vendor ID and device ID only, that's plenty enough
*/
for (devfn = 0; devfn < 0x100; devfn++) {
u8 __iomem *devbase = cfgspace + (devfn << 8);
u8 hdr_type = readb(devbase + PCI_HEADER_TYPE);
u32 l = readl(devbase + PCI_VENDOR_ID);
u16 s;
DBG("devfn %x, l: %x\n", devfn, l);
/* If no device, skip */
if (l == 0xffffffff || l == 0x00000000 ||
l == 0x0000ffff || l == 0xffff0000)
goto next;
/* Check if is supports capability lists */
s = readw(devbase + PCI_STATUS);
if (!(s & PCI_STATUS_CAP_LIST))
goto next;
mpic_scan_ht_pic(mpic, devbase, devfn, l);
mpic_scan_ht_msi(mpic, devbase, devfn);
next:
/* next device, if function 0 */
if (PCI_FUNC(devfn) == 0 && (hdr_type & 0x80) == 0)
devfn += 7;
}
}
#else /* CONFIG_MPIC_U3_HT_IRQS */
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
static inline int mpic_is_ht_interrupt(struct mpic *mpic, unsigned int source)
{
return 0;
}
static void __init mpic_scan_ht_pics(struct mpic *mpic)
{
}
#endif /* CONFIG_MPIC_U3_HT_IRQS */
2006-07-03 19:36:01 +08:00
#define mpic_irq_to_hw(virq) ((unsigned int)irq_map[virq].hwirq)
/* Find an mpic associated with a given linux interrupt */
static struct mpic *mpic_find(unsigned int irq, unsigned int *is_ipi)
{
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
struct mpic *mpic;
2006-07-03 19:36:01 +08:00
if (irq < NUM_ISA_INTERRUPTS)
return NULL;
mpic = irq_desc[irq].chip_data;
2006-07-03 19:36:01 +08:00
if (is_ipi)
*is_ipi = (src >= mpic->ipi_vecs[0] &&
src <= mpic->ipi_vecs[3]);
2006-07-03 19:36:01 +08:00
return mpic;
}
/* Convert a cpu mask from logical to physical cpu numbers. */
static inline u32 mpic_physmask(u32 cpumask)
{
int i;
u32 mask = 0;
for (i = 0; i < NR_CPUS; ++i, cpumask >>= 1)
mask |= (cpumask & 1) << get_hard_smp_processor_id(i);
return mask;
}
#ifdef CONFIG_SMP
/* Get the mpic structure from the IPI number */
static inline struct mpic * mpic_from_ipi(unsigned int ipi)
{
return irq_desc[ipi].chip_data;
}
#endif
/* Get the mpic structure from the irq number */
static inline struct mpic * mpic_from_irq(unsigned int irq)
{
return irq_desc[irq].chip_data;
}
/* Send an EOI */
static inline void mpic_eoi(struct mpic *mpic)
{
mpic_cpu_write(MPIC_INFO(CPU_EOI), 0);
(void)mpic_cpu_read(MPIC_INFO(CPU_WHOAMI));
}
#ifdef CONFIG_SMP
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 21:55:46 +08:00
static irqreturn_t mpic_ipi_action(int irq, void *dev_id)
{
struct mpic *mpic;
mpic = mpic_find(irq, NULL);
smp_message_recv(mpic_irq_to_hw(irq) - mpic->ipi_vecs[0]);
return IRQ_HANDLED;
}
#endif /* CONFIG_SMP */
/*
* Linux descriptor level callbacks
*/
void mpic_unmask_irq(unsigned int irq)
{
unsigned int loops = 100000;
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
DBG("%p: %s: enable_irq: %d (src %d)\n", mpic, mpic->name, irq, src);
mpic_irq_write(src, MPIC_INFO(IRQ_VECTOR_PRI),
mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI)) &
~MPIC_VECPRI_MASK);
/* make sure mask gets to controller before we return to user */
do {
if (!loops--) {
printk(KERN_ERR "mpic_enable_irq timeout\n");
break;
}
} while(mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI)) & MPIC_VECPRI_MASK);
}
void mpic_mask_irq(unsigned int irq)
{
unsigned int loops = 100000;
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
DBG("%s: disable_irq: %d (src %d)\n", mpic->name, irq, src);
mpic_irq_write(src, MPIC_INFO(IRQ_VECTOR_PRI),
mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI)) |
MPIC_VECPRI_MASK);
/* make sure mask gets to controller before we return to user */
do {
if (!loops--) {
printk(KERN_ERR "mpic_enable_irq timeout\n");
break;
}
} while(!(mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI)) & MPIC_VECPRI_MASK));
}
void mpic_end_irq(unsigned int irq)
{
struct mpic *mpic = mpic_from_irq(irq);
#ifdef DEBUG_IRQ
DBG("%s: end_irq: %d\n", mpic->name, irq);
#endif
/* We always EOI on end_irq() even for edge interrupts since that
* should only lower the priority, the MPIC should have properly
* latched another edge interrupt coming in anyway
*/
mpic_eoi(mpic);
}
#ifdef CONFIG_MPIC_U3_HT_IRQS
static void mpic_unmask_ht_irq(unsigned int irq)
{
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
mpic_unmask_irq(irq);
if (irq_desc[irq].status & IRQ_LEVEL)
mpic_ht_end_irq(mpic, src);
}
static unsigned int mpic_startup_ht_irq(unsigned int irq)
{
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
mpic_unmask_irq(irq);
mpic_startup_ht_interrupt(mpic, src, irq_desc[irq].status);
return 0;
}
static void mpic_shutdown_ht_irq(unsigned int irq)
{
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
mpic_shutdown_ht_interrupt(mpic, src, irq_desc[irq].status);
mpic_mask_irq(irq);
}
static void mpic_end_ht_irq(unsigned int irq)
{
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
#ifdef DEBUG_IRQ
DBG("%s: end_irq: %d\n", mpic->name, irq);
#endif
/* We always EOI on end_irq() even for edge interrupts since that
* should only lower the priority, the MPIC should have properly
* latched another edge interrupt coming in anyway
*/
if (irq_desc[irq].status & IRQ_LEVEL)
mpic_ht_end_irq(mpic, src);
mpic_eoi(mpic);
}
#endif /* !CONFIG_MPIC_U3_HT_IRQS */
#ifdef CONFIG_SMP
static void mpic_unmask_ipi(unsigned int irq)
{
struct mpic *mpic = mpic_from_ipi(irq);
unsigned int src = mpic_irq_to_hw(irq) - mpic->ipi_vecs[0];
DBG("%s: enable_ipi: %d (ipi %d)\n", mpic->name, irq, src);
mpic_ipi_write(src, mpic_ipi_read(src) & ~MPIC_VECPRI_MASK);
}
static void mpic_mask_ipi(unsigned int irq)
{
/* NEVER disable an IPI... that's just plain wrong! */
}
static void mpic_end_ipi(unsigned int irq)
{
struct mpic *mpic = mpic_from_ipi(irq);
/*
* IPIs are marked IRQ_PER_CPU. This has the side effect of
* preventing the IRQ_PENDING/IRQ_INPROGRESS logic from
* applying to them. We EOI them late to avoid re-entering.
* We mark IPI's with IRQF_DISABLED as they must run with
* irqs disabled.
*/
mpic_eoi(mpic);
}
#endif /* CONFIG_SMP */
static void mpic_set_affinity(unsigned int irq, cpumask_t cpumask)
{
struct mpic *mpic = mpic_from_irq(irq);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
cpumask_t tmp;
cpus_and(tmp, cpumask, cpu_online_map);
mpic_irq_write(src, MPIC_INFO(IRQ_DESTINATION),
mpic_physmask(cpus_addr(tmp)[0]));
}
static unsigned int mpic_type_to_vecpri(struct mpic *mpic, unsigned int type)
2006-07-03 19:36:01 +08:00
{
/* Now convert sense value */
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
switch(type & IRQ_TYPE_SENSE_MASK) {
2006-07-03 19:36:01 +08:00
case IRQ_TYPE_EDGE_RISING:
return MPIC_INFO(VECPRI_SENSE_EDGE) |
MPIC_INFO(VECPRI_POLARITY_POSITIVE);
2006-07-03 19:36:01 +08:00
case IRQ_TYPE_EDGE_FALLING:
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
case IRQ_TYPE_EDGE_BOTH:
return MPIC_INFO(VECPRI_SENSE_EDGE) |
MPIC_INFO(VECPRI_POLARITY_NEGATIVE);
2006-07-03 19:36:01 +08:00
case IRQ_TYPE_LEVEL_HIGH:
return MPIC_INFO(VECPRI_SENSE_LEVEL) |
MPIC_INFO(VECPRI_POLARITY_POSITIVE);
2006-07-03 19:36:01 +08:00
case IRQ_TYPE_LEVEL_LOW:
default:
return MPIC_INFO(VECPRI_SENSE_LEVEL) |
MPIC_INFO(VECPRI_POLARITY_NEGATIVE);
2006-07-03 19:36:01 +08:00
}
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
}
int mpic_set_irq_type(unsigned int virq, unsigned int flow_type)
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
{
struct mpic *mpic = mpic_from_irq(virq);
unsigned int src = mpic_irq_to_hw(virq);
struct irq_desc *desc = get_irq_desc(virq);
unsigned int vecpri, vold, vnew;
DBG("mpic: set_irq_type(mpic:@%p,virq:%d,src:0x%x,type:0x%x)\n",
mpic, virq, src, flow_type);
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
if (src >= mpic->irq_count)
return -EINVAL;
if (flow_type == IRQ_TYPE_NONE)
if (mpic->senses && src < mpic->senses_count)
flow_type = mpic->senses[src];
if (flow_type == IRQ_TYPE_NONE)
flow_type = IRQ_TYPE_LEVEL_LOW;
desc->status &= ~(IRQ_TYPE_SENSE_MASK | IRQ_LEVEL);
desc->status |= flow_type & IRQ_TYPE_SENSE_MASK;
if (flow_type & (IRQ_TYPE_LEVEL_HIGH | IRQ_TYPE_LEVEL_LOW))
desc->status |= IRQ_LEVEL;
if (mpic_is_ht_interrupt(mpic, src))
vecpri = MPIC_VECPRI_POLARITY_POSITIVE |
MPIC_VECPRI_SENSE_EDGE;
else
vecpri = mpic_type_to_vecpri(mpic, flow_type);
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
vold = mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI));
vnew = vold & ~(MPIC_INFO(VECPRI_POLARITY_MASK) |
MPIC_INFO(VECPRI_SENSE_MASK));
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
vnew |= vecpri;
if (vold != vnew)
mpic_irq_write(src, MPIC_INFO(IRQ_VECTOR_PRI), vnew);
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
return 0;
2006-07-03 19:36:01 +08:00
}
static struct irq_chip mpic_irq_chip = {
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
.mask = mpic_mask_irq,
.unmask = mpic_unmask_irq,
.eoi = mpic_end_irq,
.set_type = mpic_set_irq_type,
};
#ifdef CONFIG_SMP
static struct irq_chip mpic_ipi_chip = {
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
.mask = mpic_mask_ipi,
.unmask = mpic_unmask_ipi,
.eoi = mpic_end_ipi,
};
#endif /* CONFIG_SMP */
#ifdef CONFIG_MPIC_U3_HT_IRQS
static struct irq_chip mpic_irq_ht_chip = {
.startup = mpic_startup_ht_irq,
.shutdown = mpic_shutdown_ht_irq,
.mask = mpic_mask_irq,
.unmask = mpic_unmask_ht_irq,
.eoi = mpic_end_ht_irq,
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
.set_type = mpic_set_irq_type,
};
#endif /* CONFIG_MPIC_U3_HT_IRQS */
2006-07-03 19:36:01 +08:00
static int mpic_host_match(struct irq_host *h, struct device_node *node)
{
struct mpic *mpic = h->host_data;
/* Exact match, unless mpic node is NULL */
return mpic->of_node == NULL || mpic->of_node == node;
}
static int mpic_host_map(struct irq_host *h, unsigned int virq,
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
irq_hw_number_t hw)
2006-07-03 19:36:01 +08:00
{
struct mpic *mpic = h->host_data;
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
struct irq_chip *chip;
2006-07-03 19:36:01 +08:00
DBG("mpic: map virq %d, hwirq 0x%lx\n", virq, hw);
2006-07-03 19:36:01 +08:00
if (hw == mpic->spurious_vec)
2006-07-03 19:36:01 +08:00
return -EINVAL;
2006-07-03 19:36:01 +08:00
#ifdef CONFIG_SMP
else if (hw >= mpic->ipi_vecs[0]) {
2006-07-03 19:36:01 +08:00
WARN_ON(!(mpic->flags & MPIC_PRIMARY));
DBG("mpic: mapping as IPI\n");
2006-07-03 19:36:01 +08:00
set_irq_chip_data(virq, mpic);
set_irq_chip_and_handler(virq, &mpic->hc_ipi,
handle_percpu_irq);
return 0;
}
#endif /* CONFIG_SMP */
if (hw >= mpic->irq_count)
return -EINVAL;
mpic_msi_reserve_hwirq(mpic, hw);
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
/* Default chip */
2006-07-03 19:36:01 +08:00
chip = &mpic->hc_irq;
#ifdef CONFIG_MPIC_U3_HT_IRQS
2006-07-03 19:36:01 +08:00
/* Check for HT interrupts, override vecpri */
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
if (mpic_is_ht_interrupt(mpic, hw))
2006-07-03 19:36:01 +08:00
chip = &mpic->hc_ht_irq;
#endif /* CONFIG_MPIC_U3_HT_IRQS */
2006-07-03 19:36:01 +08:00
DBG("mpic: mapping to irq chip @%p\n", chip);
2006-07-03 19:36:01 +08:00
set_irq_chip_data(virq, mpic);
set_irq_chip_and_handler(virq, chip, handle_fasteoi_irq);
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
/* Set default irq type */
set_irq_type(virq, IRQ_TYPE_NONE);
2006-07-03 19:36:01 +08:00
return 0;
}
static int mpic_host_xlate(struct irq_host *h, struct device_node *ct,
u32 *intspec, unsigned int intsize,
irq_hw_number_t *out_hwirq, unsigned int *out_flags)
{
static unsigned char map_mpic_senses[4] = {
IRQ_TYPE_EDGE_RISING,
IRQ_TYPE_LEVEL_LOW,
IRQ_TYPE_LEVEL_HIGH,
IRQ_TYPE_EDGE_FALLING,
};
*out_hwirq = intspec[0];
if (intsize > 1) {
u32 mask = 0x3;
/* Apple invented a new race of encoding on machines with
* an HT APIC. They encode, among others, the index within
* the HT APIC. We don't care about it here since thankfully,
* it appears that they have the APIC already properly
* configured, and thus our current fixup code that reads the
* APIC config works fine. However, we still need to mask out
* bits in the specifier to make sure we only get bit 0 which
* is the level/edge bit (the only sense bit exposed by Apple),
* as their bit 1 means something else.
*/
if (machine_is(powermac))
mask = 0x1;
*out_flags = map_mpic_senses[intspec[1] & mask];
} else
2006-07-03 19:36:01 +08:00
*out_flags = IRQ_TYPE_NONE;
DBG("mpic: xlate (%d cells: 0x%08x 0x%08x) to line 0x%lx sense 0x%x\n",
intsize, intspec[0], intspec[1], *out_hwirq, *out_flags);
2006-07-03 19:36:01 +08:00
return 0;
}
static struct irq_host_ops mpic_host_ops = {
.match = mpic_host_match,
.map = mpic_host_map,
.xlate = mpic_host_xlate,
};
/*
* Exported functions
*/
2006-07-03 19:36:01 +08:00
struct mpic * __init mpic_alloc(struct device_node *node,
phys_addr_t phys_addr,
unsigned int flags,
unsigned int isu_size,
unsigned int irq_count,
const char *name)
{
struct mpic *mpic;
u32 reg;
const char *vers;
int i;
int intvec_top;
u64 paddr = phys_addr;
mpic = alloc_bootmem(sizeof(struct mpic));
if (mpic == NULL)
return NULL;
memset(mpic, 0, sizeof(struct mpic));
mpic->name = name;
mpic->of_node = of_node_get(node);
mpic->irqhost = irq_alloc_host(IRQ_HOST_MAP_LINEAR, isu_size,
2006-07-03 19:36:01 +08:00
&mpic_host_ops,
flags & MPIC_LARGE_VECTORS ? 2048 : 256);
2006-07-03 19:36:01 +08:00
if (mpic->irqhost == NULL) {
of_node_put(node);
return NULL;
}
mpic->irqhost->host_data = mpic;
mpic->hc_irq = mpic_irq_chip;
mpic->hc_irq.typename = name;
if (flags & MPIC_PRIMARY)
mpic->hc_irq.set_affinity = mpic_set_affinity;
#ifdef CONFIG_MPIC_U3_HT_IRQS
mpic->hc_ht_irq = mpic_irq_ht_chip;
mpic->hc_ht_irq.typename = name;
if (flags & MPIC_PRIMARY)
mpic->hc_ht_irq.set_affinity = mpic_set_affinity;
#endif /* CONFIG_MPIC_U3_HT_IRQS */
#ifdef CONFIG_SMP
mpic->hc_ipi = mpic_ipi_chip;
2006-07-03 19:36:01 +08:00
mpic->hc_ipi.typename = name;
#endif /* CONFIG_SMP */
mpic->flags = flags;
mpic->isu_size = isu_size;
mpic->irq_count = irq_count;
mpic->num_sources = 0; /* so far */
if (flags & MPIC_LARGE_VECTORS)
intvec_top = 2047;
else
intvec_top = 255;
mpic->timer_vecs[0] = intvec_top - 8;
mpic->timer_vecs[1] = intvec_top - 7;
mpic->timer_vecs[2] = intvec_top - 6;
mpic->timer_vecs[3] = intvec_top - 5;
mpic->ipi_vecs[0] = intvec_top - 4;
mpic->ipi_vecs[1] = intvec_top - 3;
mpic->ipi_vecs[2] = intvec_top - 2;
mpic->ipi_vecs[3] = intvec_top - 1;
mpic->spurious_vec = intvec_top;
/* Check for "big-endian" in device-tree */
if (node && of_get_property(node, "big-endian", NULL) != NULL)
mpic->flags |= MPIC_BIG_ENDIAN;
#ifdef CONFIG_MPIC_WEIRD
mpic->hw_set = mpic_infos[MPIC_GET_REGSET(flags)];
#endif
/* default register type */
mpic->reg_type = (flags & MPIC_BIG_ENDIAN) ?
mpic_access_mmio_be : mpic_access_mmio_le;
/* If no physical address is passed in, a device-node is mandatory */
BUG_ON(paddr == 0 && node == NULL);
/* If no physical address passed in, check if it's dcr based */
if (paddr == 0 && of_get_property(node, "dcr-reg", NULL) != NULL)
mpic->flags |= MPIC_USES_DCR;
#ifdef CONFIG_PPC_DCR
if (mpic->flags & MPIC_USES_DCR) {
const u32 *dbasep;
dbasep = of_get_property(node, "dcr-reg", NULL);
BUG_ON(dbasep == NULL);
mpic->dcr_base = *dbasep;
mpic->reg_type = mpic_access_dcr;
}
#else
BUG_ON (mpic->flags & MPIC_USES_DCR);
#endif /* CONFIG_PPC_DCR */
/* If the MPIC is not DCR based, and no physical address was passed
* in, try to obtain one
*/
if (paddr == 0 && !(mpic->flags & MPIC_USES_DCR)) {
const u32 *reg;
reg = of_get_property(node, "reg", NULL);
BUG_ON(reg == NULL);
paddr = of_translate_address(node, reg);
BUG_ON(paddr == OF_BAD_ADDR);
}
/* Map the global registers */
mpic_map(mpic, paddr, &mpic->gregs, MPIC_INFO(GREG_BASE), 0x1000);
mpic_map(mpic, paddr, &mpic->tmregs, MPIC_INFO(TIMER_BASE), 0x1000);
/* Reset */
if (flags & MPIC_WANTS_RESET) {
mpic_write(mpic->gregs, MPIC_INFO(GREG_GLOBAL_CONF_0),
mpic_read(mpic->gregs, MPIC_INFO(GREG_GLOBAL_CONF_0))
| MPIC_GREG_GCONF_RESET);
while( mpic_read(mpic->gregs, MPIC_INFO(GREG_GLOBAL_CONF_0))
& MPIC_GREG_GCONF_RESET)
mb();
}
/* Read feature register, calculate num CPUs and, for non-ISU
* MPICs, num sources as well. On ISU MPICs, sources are counted
* as ISUs are added
*/
reg = mpic_read(mpic->gregs, MPIC_INFO(GREG_FEATURE_0));
mpic->num_cpus = ((reg & MPIC_GREG_FEATURE_LAST_CPU_MASK)
>> MPIC_GREG_FEATURE_LAST_CPU_SHIFT) + 1;
if (isu_size == 0)
mpic->num_sources = ((reg & MPIC_GREG_FEATURE_LAST_SRC_MASK)
>> MPIC_GREG_FEATURE_LAST_SRC_SHIFT) + 1;
/* Map the per-CPU registers */
for (i = 0; i < mpic->num_cpus; i++) {
mpic_map(mpic, paddr, &mpic->cpuregs[i],
MPIC_INFO(CPU_BASE) + i * MPIC_INFO(CPU_STRIDE),
0x1000);
}
/* Initialize main ISU if none provided */
if (mpic->isu_size == 0) {
mpic->isu_size = mpic->num_sources;
mpic_map(mpic, paddr, &mpic->isus[0],
MPIC_INFO(IRQ_BASE), MPIC_INFO(IRQ_STRIDE) * mpic->isu_size);
}
mpic->isu_shift = 1 + __ilog2(mpic->isu_size - 1);
mpic->isu_mask = (1 << mpic->isu_shift) - 1;
/* Display version */
switch (reg & MPIC_GREG_FEATURE_VERSION_MASK) {
case 1:
vers = "1.0";
break;
case 2:
vers = "1.2";
break;
case 3:
vers = "1.3";
break;
default:
vers = "<unknown>";
break;
}
printk(KERN_INFO "mpic: Setting up MPIC \"%s\" version %s at %llx,"
" max %d CPUs\n",
name, vers, (unsigned long long)paddr, mpic->num_cpus);
printk(KERN_INFO "mpic: ISU size: %d, shift: %d, mask: %x\n",
mpic->isu_size, mpic->isu_shift, mpic->isu_mask);
mpic->next = mpics;
mpics = mpic;
2006-07-03 19:36:01 +08:00
if (flags & MPIC_PRIMARY) {
mpic_primary = mpic;
2006-07-03 19:36:01 +08:00
irq_set_default_host(mpic->irqhost);
}
return mpic;
}
void __init mpic_assign_isu(struct mpic *mpic, unsigned int isu_num,
phys_addr_t paddr)
{
unsigned int isu_first = isu_num * mpic->isu_size;
BUG_ON(isu_num >= MPIC_MAX_ISU);
mpic_map(mpic, paddr, &mpic->isus[isu_num], 0,
MPIC_INFO(IRQ_STRIDE) * mpic->isu_size);
if ((isu_first + mpic->isu_size) > mpic->num_sources)
mpic->num_sources = isu_first + mpic->isu_size;
}
2006-07-03 19:36:01 +08:00
void __init mpic_set_default_senses(struct mpic *mpic, u8 *senses, int count)
{
mpic->senses = senses;
mpic->senses_count = count;
}
void __init mpic_init(struct mpic *mpic)
{
int i;
BUG_ON(mpic->num_sources == 0);
printk(KERN_INFO "mpic: Initializing for %d sources\n", mpic->num_sources);
/* Set current processor priority to max */
mpic_cpu_write(MPIC_INFO(CPU_CURRENT_TASK_PRI), 0xf);
/* Initialize timers: just disable them all */
for (i = 0; i < 4; i++) {
mpic_write(mpic->tmregs,
i * MPIC_INFO(TIMER_STRIDE) +
MPIC_INFO(TIMER_DESTINATION), 0);
mpic_write(mpic->tmregs,
i * MPIC_INFO(TIMER_STRIDE) +
MPIC_INFO(TIMER_VECTOR_PRI),
MPIC_VECPRI_MASK |
(mpic->timer_vecs[0] + i));
}
/* Initialize IPIs to our reserved vectors and mark them disabled for now */
mpic_test_broken_ipi(mpic);
for (i = 0; i < 4; i++) {
mpic_ipi_write(i,
MPIC_VECPRI_MASK |
(10 << MPIC_VECPRI_PRIORITY_SHIFT) |
(mpic->ipi_vecs[0] + i));
}
/* Initialize interrupt sources */
if (mpic->irq_count == 0)
mpic->irq_count = mpic->num_sources;
/* Do the HT PIC fixups on U3 broken mpic */
DBG("MPIC flags: %x\n", mpic->flags);
if ((mpic->flags & MPIC_U3_HT_IRQS) && (mpic->flags & MPIC_PRIMARY)) {
mpic_scan_ht_pics(mpic);
mpic_u3msi_init(mpic);
}
for (i = 0; i < mpic->num_sources; i++) {
/* start with vector = source number, and masked */
[PATCH] powerpc: fix trigger handling in the new irq code This patch slightly reworks the new irq code to fix a small design error. I removed the passing of the trigger to the map() calls entirely, it was not a good idea to have one call do two different things. It also fixes a couple of corner cases. Mapping a linux virtual irq to a physical irq now does only that. Setting the trigger is a different action which has a different call. The main changes are: - I no longer call host->ops->map() for an already mapped irq, I just return the virtual number that was already mapped. It was called before to give an opportunity to change the trigger, but that was causing issues as that could happen while the interrupt was in use by a device, and because of the trigger change, map would potentially muck around with things in a racy way. That was causing much burden on a given's controller implementation of map() to get it right. This is much simpler now. map() is only called on the initial mapping of an irq, meaning that you know that this irq is _not_ being used. You can initialize the hardware if you want (though you don't have to). - Controllers that can handle different type of triggers (level/edge/etc...) now implement the standard irq_chip->set_type() call as defined by the generic code. That means that you can use the standard set_irq_type() to configure an irq line manually if you wish or (though I don't like that interface), pass explicit trigger flags to request_irq() as defined by the generic kernel interfaces. Also, using those interfaces guarantees that your controller set_type callback is called with the descriptor lock held, thus providing locking against activity on the same interrupt (including mask/unmask/etc...) automatically. A result is that, for example, MPIC's own map() implementation calls irq_set_type(NONE) to configure the hardware to the default triggers. - To allow the above, the irq_map array entry for the new mapped interrupt is now set before map() callback is called for the controller. - The irq_create_of_mapping() (also used by irq_of_parse_and_map()) function for mapping interrupts from the device-tree now also call the separate set_irq_type(), and only does so if there is a change in the trigger type. - While I was at it, I changed pci_read_irq_line() (which is the helper I would expect most archs to use in their pcibios_fixup() to get the PCI interrupt routing from the device tree) to also handle a fallback when the DT mapping fails consisting of reading the PCI_INTERRUPT_PIN to know wether the device has an interrupt at all, and the the PCI_INTERRUPT_LINE to get an interrupt number from the device. That number is then mapped using the default controller, and the trigger is set to level low. That default behaviour works for several platforms that don't have a proper interrupt tree like Pegasos. If it doesn't work for your platform, then either provide a proper interrupt tree from the firmware so that fallback isn't needed, or don't call pci_read_irq_line() - Add back a bit that got dropped by my main rework patch for properly clearing pending IPIs on pSeries when using a kexec Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org> Cc: Paul Mackerras <paulus@samba.org> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-07-10 19:44:42 +08:00
u32 vecpri = MPIC_VECPRI_MASK | i |
(8 << MPIC_VECPRI_PRIORITY_SHIFT);
/* init hw */
mpic_irq_write(i, MPIC_INFO(IRQ_VECTOR_PRI), vecpri);
mpic_irq_write(i, MPIC_INFO(IRQ_DESTINATION),
1 << hard_smp_processor_id());
}
/* Init spurious vector */
mpic_write(mpic->gregs, MPIC_INFO(GREG_SPURIOUS), mpic->spurious_vec);
/* Disable 8259 passthrough, if supported */
if (!(mpic->flags & MPIC_NO_PTHROU_DIS))
mpic_write(mpic->gregs, MPIC_INFO(GREG_GLOBAL_CONF_0),
mpic_read(mpic->gregs, MPIC_INFO(GREG_GLOBAL_CONF_0))
| MPIC_GREG_GCONF_8259_PTHROU_DIS);
/* Set current processor priority to 0 */
mpic_cpu_write(MPIC_INFO(CPU_CURRENT_TASK_PRI), 0);
#ifdef CONFIG_PM
/* allocate memory to save mpic state */
mpic->save_data = alloc_bootmem(mpic->num_sources * sizeof(struct mpic_irq_save));
BUG_ON(mpic->save_data == NULL);
#endif
}
void __init mpic_set_clk_ratio(struct mpic *mpic, u32 clock_ratio)
{
u32 v;
v = mpic_read(mpic->gregs, MPIC_GREG_GLOBAL_CONF_1);
v &= ~MPIC_GREG_GLOBAL_CONF_1_CLK_RATIO_MASK;
v |= MPIC_GREG_GLOBAL_CONF_1_CLK_RATIO(clock_ratio);
mpic_write(mpic->gregs, MPIC_GREG_GLOBAL_CONF_1, v);
}
void __init mpic_set_serial_int(struct mpic *mpic, int enable)
{
unsigned long flags;
u32 v;
spin_lock_irqsave(&mpic_lock, flags);
v = mpic_read(mpic->gregs, MPIC_GREG_GLOBAL_CONF_1);
if (enable)
v |= MPIC_GREG_GLOBAL_CONF_1_SIE;
else
v &= ~MPIC_GREG_GLOBAL_CONF_1_SIE;
mpic_write(mpic->gregs, MPIC_GREG_GLOBAL_CONF_1, v);
spin_unlock_irqrestore(&mpic_lock, flags);
}
void mpic_irq_set_priority(unsigned int irq, unsigned int pri)
{
int is_ipi;
struct mpic *mpic = mpic_find(irq, &is_ipi);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
unsigned long flags;
u32 reg;
spin_lock_irqsave(&mpic_lock, flags);
if (is_ipi) {
reg = mpic_ipi_read(src - mpic->ipi_vecs[0]) &
~MPIC_VECPRI_PRIORITY_MASK;
mpic_ipi_write(src - mpic->ipi_vecs[0],
reg | (pri << MPIC_VECPRI_PRIORITY_SHIFT));
} else {
reg = mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI))
& ~MPIC_VECPRI_PRIORITY_MASK;
mpic_irq_write(src, MPIC_INFO(IRQ_VECTOR_PRI),
reg | (pri << MPIC_VECPRI_PRIORITY_SHIFT));
}
spin_unlock_irqrestore(&mpic_lock, flags);
}
unsigned int mpic_irq_get_priority(unsigned int irq)
{
int is_ipi;
struct mpic *mpic = mpic_find(irq, &is_ipi);
2006-07-03 19:36:01 +08:00
unsigned int src = mpic_irq_to_hw(irq);
unsigned long flags;
u32 reg;
spin_lock_irqsave(&mpic_lock, flags);
if (is_ipi)
reg = mpic_ipi_read(src = mpic->ipi_vecs[0]);
else
reg = mpic_irq_read(src, MPIC_INFO(IRQ_VECTOR_PRI));
spin_unlock_irqrestore(&mpic_lock, flags);
return (reg & MPIC_VECPRI_PRIORITY_MASK) >> MPIC_VECPRI_PRIORITY_SHIFT;
}
void mpic_setup_this_cpu(void)
{
#ifdef CONFIG_SMP
struct mpic *mpic = mpic_primary;
unsigned long flags;
u32 msk = 1 << hard_smp_processor_id();
unsigned int i;
BUG_ON(mpic == NULL);
DBG("%s: setup_this_cpu(%d)\n", mpic->name, hard_smp_processor_id());
spin_lock_irqsave(&mpic_lock, flags);
/* let the mpic know we want intrs. default affinity is 0xffffffff
* until changed via /proc. That's how it's done on x86. If we want
* it differently, then we should make sure we also change the default
* values of irq_desc[].affinity in irq.c.
*/
if (distribute_irqs) {
for (i = 0; i < mpic->num_sources ; i++)
mpic_irq_write(i, MPIC_INFO(IRQ_DESTINATION),
mpic_irq_read(i, MPIC_INFO(IRQ_DESTINATION)) | msk);
}
/* Set current processor priority to 0 */
mpic_cpu_write(MPIC_INFO(CPU_CURRENT_TASK_PRI), 0);
spin_unlock_irqrestore(&mpic_lock, flags);
#endif /* CONFIG_SMP */
}
int mpic_cpu_get_priority(void)
{
struct mpic *mpic = mpic_primary;
return mpic_cpu_read(MPIC_INFO(CPU_CURRENT_TASK_PRI));
}
void mpic_cpu_set_priority(int prio)
{
struct mpic *mpic = mpic_primary;
prio &= MPIC_CPU_TASKPRI_MASK;
mpic_cpu_write(MPIC_INFO(CPU_CURRENT_TASK_PRI), prio);
}
/*
* XXX: someone who knows mpic should check this.
* do we need to eoi the ipi including for kexec cpu here (see xics comments)?
* or can we reset the mpic in the new kernel?
*/
void mpic_teardown_this_cpu(int secondary)
{
struct mpic *mpic = mpic_primary;
unsigned long flags;
u32 msk = 1 << hard_smp_processor_id();
unsigned int i;
BUG_ON(mpic == NULL);
DBG("%s: teardown_this_cpu(%d)\n", mpic->name, hard_smp_processor_id());
spin_lock_irqsave(&mpic_lock, flags);
/* let the mpic know we don't want intrs. */
for (i = 0; i < mpic->num_sources ; i++)
mpic_irq_write(i, MPIC_INFO(IRQ_DESTINATION),
mpic_irq_read(i, MPIC_INFO(IRQ_DESTINATION)) & ~msk);
/* Set current processor priority to max */
mpic_cpu_write(MPIC_INFO(CPU_CURRENT_TASK_PRI), 0xf);
spin_unlock_irqrestore(&mpic_lock, flags);
}
void mpic_send_ipi(unsigned int ipi_no, unsigned int cpu_mask)
{
struct mpic *mpic = mpic_primary;
BUG_ON(mpic == NULL);
#ifdef DEBUG_IPI
DBG("%s: send_ipi(ipi_no: %d)\n", mpic->name, ipi_no);
#endif
mpic_cpu_write(MPIC_INFO(CPU_IPI_DISPATCH_0) +
ipi_no * MPIC_INFO(CPU_IPI_DISPATCH_STRIDE),
mpic_physmask(cpu_mask & cpus_addr(cpu_online_map)[0]));
}
unsigned int mpic_get_one_irq(struct mpic *mpic)
{
2006-07-03 19:36:01 +08:00
u32 src;
src = mpic_cpu_read(MPIC_INFO(CPU_INTACK)) & MPIC_INFO(VECPRI_VECTOR_MASK);
#ifdef DEBUG_LOW
2006-07-03 19:36:01 +08:00
DBG("%s: get_one_irq(): %d\n", mpic->name, src);
#endif
if (unlikely(src == mpic->spurious_vec)) {
if (mpic->flags & MPIC_SPV_EOI)
mpic_eoi(mpic);
2006-07-03 19:36:01 +08:00
return NO_IRQ;
}
2006-07-03 19:36:01 +08:00
return irq_linear_revmap(mpic->irqhost, src);
}
unsigned int mpic_get_irq(void)
{
struct mpic *mpic = mpic_primary;
BUG_ON(mpic == NULL);
return mpic_get_one_irq(mpic);
}
#ifdef CONFIG_SMP
void mpic_request_ipis(void)
{
struct mpic *mpic = mpic_primary;
int i, err;
2006-07-03 19:36:01 +08:00
static char *ipi_names[] = {
"IPI0 (call function)",
"IPI1 (reschedule)",
"IPI2 (unused)",
"IPI3 (debugger break)",
};
BUG_ON(mpic == NULL);
2006-07-03 19:36:01 +08:00
printk(KERN_INFO "mpic: requesting IPIs ... \n");
for (i = 0; i < 4; i++) {
unsigned int vipi = irq_create_mapping(mpic->irqhost,
mpic->ipi_vecs[0] + i);
2006-07-03 19:36:01 +08:00
if (vipi == NO_IRQ) {
printk(KERN_ERR "Failed to map IPI %d\n", i);
break;
}
err = request_irq(vipi, mpic_ipi_action,
IRQF_DISABLED|IRQF_PERCPU,
ipi_names[i], mpic);
if (err) {
printk(KERN_ERR "Request of irq %d for IPI %d failed\n",
vipi, i);
break;
}
2006-07-03 19:36:01 +08:00
}
}
void smp_mpic_message_pass(int target, int msg)
{
/* make sure we're sending something that translates to an IPI */
if ((unsigned int)msg > 3) {
printk("SMP %d: smp_message_pass: unknown msg %d\n",
smp_processor_id(), msg);
return;
}
switch (target) {
case MSG_ALL:
mpic_send_ipi(msg, 0xffffffff);
break;
case MSG_ALL_BUT_SELF:
mpic_send_ipi(msg, 0xffffffff & ~(1 << smp_processor_id()));
break;
default:
mpic_send_ipi(msg, 1 << target);
break;
}
}
int __init smp_mpic_probe(void)
{
int nr_cpus;
DBG("smp_mpic_probe()...\n");
nr_cpus = cpus_weight(cpu_possible_map);
DBG("nr_cpus: %d\n", nr_cpus);
if (nr_cpus > 1)
mpic_request_ipis();
return nr_cpus;
}
void __devinit smp_mpic_setup_cpu(int cpu)
{
mpic_setup_this_cpu();
}
#endif /* CONFIG_SMP */
#ifdef CONFIG_PM
static int mpic_suspend(struct sys_device *dev, pm_message_t state)
{
struct mpic *mpic = container_of(dev, struct mpic, sysdev);
int i;
for (i = 0; i < mpic->num_sources; i++) {
mpic->save_data[i].vecprio =
mpic_irq_read(i, MPIC_INFO(IRQ_VECTOR_PRI));
mpic->save_data[i].dest =
mpic_irq_read(i, MPIC_INFO(IRQ_DESTINATION));
}
return 0;
}
static int mpic_resume(struct sys_device *dev)
{
struct mpic *mpic = container_of(dev, struct mpic, sysdev);
int i;
for (i = 0; i < mpic->num_sources; i++) {
mpic_irq_write(i, MPIC_INFO(IRQ_VECTOR_PRI),
mpic->save_data[i].vecprio);
mpic_irq_write(i, MPIC_INFO(IRQ_DESTINATION),
mpic->save_data[i].dest);
#ifdef CONFIG_MPIC_U3_HT_IRQS
{
struct mpic_irq_fixup *fixup = &mpic->fixups[i];
if (fixup->base) {
/* we use the lowest bit in an inverted meaning */
if ((mpic->save_data[i].fixup_data & 1) == 0)
continue;
/* Enable and configure */
writeb(0x10 + 2 * fixup->index, fixup->base + 2);
writel(mpic->save_data[i].fixup_data & ~1,
fixup->base + 4);
}
}
#endif
} /* end for loop */
return 0;
}
#endif
static struct sysdev_class mpic_sysclass = {
#ifdef CONFIG_PM
.resume = mpic_resume,
.suspend = mpic_suspend,
#endif
set_kset_name("mpic"),
};
static int mpic_init_sys(void)
{
struct mpic *mpic = mpics;
int error, id = 0;
error = sysdev_class_register(&mpic_sysclass);
while (mpic && !error) {
mpic->sysdev.cls = &mpic_sysclass;
mpic->sysdev.id = id++;
error = sysdev_register(&mpic->sysdev);
mpic = mpic->next;
}
return error;
}
device_initcall(mpic_init_sys);