OpenCloudOS-Kernel/drivers/pcmcia/hd64465_ss.c

938 lines
24 KiB
C
Raw Normal View History

/*
* $Id: hd64465_ss.c,v 1.7 2003/07/06 14:42:50 lethal Exp $
*
* Device driver for the PCMCIA controller module of the
* Hitachi HD64465 handheld companion chip.
*
* Note that the HD64465 provides a very thin PCMCIA host bridge
* layer, requiring a lot of the work of supporting cards to be
* performed by the processor. For example: mapping of card
* interrupts to processor IRQs is done by IRQ demuxing software;
* IO and memory mappings are fixed; setting voltages according
* to card Voltage Select pins etc is done in software.
*
* Note also that this driver uses only the simple, fixed,
* 16MB, 16-bit wide mappings to PCMCIA spaces defined by the
* HD64465. Larger mappings, smaller mappings, or mappings of
* different width to the same socket, are all possible only by
* involving the SH7750's MMU, which is considered unnecessary here.
* The downside is that it may be possible for some drivers to
* break because they need or expect 8-bit mappings.
*
* This driver currently supports only the following configuration:
* SH7750 CPU, HD64465, TPS2206 voltage control chip.
*
* by Greg Banks <gbanks@pocketpenguins.com>
* (c) 2000 PocketPenguins Inc
*/
#include <linux/types.h>
#include <linux/module.h>
#include <linux/init.h>
#include <linux/string.h>
#include <linux/kernel.h>
#include <linux/ioport.h>
#include <linux/mm.h>
#include <linux/vmalloc.h>
#include <asm/errno.h>
#include <linux/irq.h>
#include <linux/interrupt.h>
#include <linux/platform_device.h>
#include <asm/io.h>
#include <asm/hd64465/hd64465.h>
#include <asm/hd64465/io.h>
#include <pcmcia/cs_types.h>
#include <pcmcia/cs.h>
#include <pcmcia/cistpl.h>
#include <pcmcia/ds.h>
#include <pcmcia/ss.h>
#include <pcmcia/bulkmem.h>
#include "cs_internal.h"
#define MODNAME "hd64465_ss"
/* #define HD64465_DEBUG 1 */
#if HD64465_DEBUG
#define DPRINTK(args...) printk(MODNAME ": " args)
#else
#define DPRINTK(args...)
#endif
extern int hd64465_io_debug;
extern void * p3_ioremap(unsigned long phys_addr, unsigned long size, unsigned long flags);
extern void p3_iounmap(void *addr);
/*============================================================*/
#define HS_IO_MAP_SIZE (64*1024)
typedef struct hs_socket_t
{
unsigned int number;
u_int irq;
u_long mem_base;
void *io_base;
u_long mem_length;
u_int ctrl_base;
socket_state_t state;
pccard_io_map io_maps[MAX_IO_WIN];
pccard_mem_map mem_maps[MAX_WIN];
struct pcmcia_socket socket;
} hs_socket_t;
#define HS_MAX_SOCKETS 2
static hs_socket_t hs_sockets[HS_MAX_SOCKETS];
#define hs_in(sp, r) inb((sp)->ctrl_base + (r))
#define hs_out(sp, v, r) outb(v, (sp)->ctrl_base + (r))
/* translate a boolean value to a bit in a register */
#define bool_to_regbit(sp, r, bi, bo) \
do { \
unsigned short v = hs_in(sp, r); \
if (bo) \
v |= (bi); \
else \
v &= ~(bi); \
hs_out(sp, v, r); \
} while(0)
/* register offsets from HD64465_REG_PCC[01]ISR */
#define ISR 0x0
#define GCR 0x2
#define CSCR 0x4
#define CSCIER 0x6
#define SCR 0x8
/* Mask and values for CSCIER register */
#define IER_MASK 0x80
#define IER_ON 0x3f /* interrupts on */
#define IER_OFF 0x00 /* interrupts off */
/*============================================================*/
#if HD64465_DEBUG > 10
static void cis_hex_dump(const unsigned char *x, int len)
{
int i;
for (i=0 ; i<len ; i++)
{
if (!(i & 0xf))
printk("\n%08x", (unsigned)(x + i));
printk(" %02x", *(volatile unsigned short*)x);
x += 2;
}
printk("\n");
}
#endif
/*============================================================*/
/*
* This code helps create the illusion that the IREQ line from
* the PC card is mapped to one of the CPU's IRQ lines by the
* host bridge hardware (which is how every host bridge *except*
* the HD64465 works). In particular, it supports enabling
* and disabling the IREQ line by code which knows nothing
* about the host bridge (e.g. device drivers, IDE code) using
* the request_irq(), free_irq(), probe_irq_on() and probe_irq_off()
* functions. Also, it supports sharing the mapped IRQ with
* real hardware IRQs from the -IRL0-3 lines.
*/
#define HS_NUM_MAPPED_IRQS 16 /* Limitation of the PCMCIA code */
static struct
{
/* index is mapped irq number */
hs_socket_t *sock;
hw_irq_controller *old_handler;
} hs_mapped_irq[HS_NUM_MAPPED_IRQS];
static void hs_socket_enable_ireq(hs_socket_t *sp)
{
unsigned short cscier;
DPRINTK("hs_socket_enable_ireq(sock=%d)\n", sp->number);
cscier = hs_in(sp, CSCIER);
cscier &= ~HD64465_PCCCSCIER_PIREQE_MASK;
cscier |= HD64465_PCCCSCIER_PIREQE_LEVEL;
hs_out(sp, cscier, CSCIER);
}
static void hs_socket_disable_ireq(hs_socket_t *sp)
{
unsigned short cscier;
DPRINTK("hs_socket_disable_ireq(sock=%d)\n", sp->number);
cscier = hs_in(sp, CSCIER);
cscier &= ~HD64465_PCCCSCIER_PIREQE_MASK;
hs_out(sp, cscier, CSCIER);
}
static unsigned int hs_startup_irq(unsigned int irq)
{
hs_socket_enable_ireq(hs_mapped_irq[irq].sock);
hs_mapped_irq[irq].old_handler->startup(irq);
return 0;
}
static void hs_shutdown_irq(unsigned int irq)
{
hs_socket_disable_ireq(hs_mapped_irq[irq].sock);
hs_mapped_irq[irq].old_handler->shutdown(irq);
}
static void hs_enable_irq(unsigned int irq)
{
hs_socket_enable_ireq(hs_mapped_irq[irq].sock);
hs_mapped_irq[irq].old_handler->enable(irq);
}
static void hs_disable_irq(unsigned int irq)
{
hs_socket_disable_ireq(hs_mapped_irq[irq].sock);
hs_mapped_irq[irq].old_handler->disable(irq);
}
extern struct hw_interrupt_type no_irq_type;
static void hs_mask_and_ack_irq(unsigned int irq)
{
hs_socket_disable_ireq(hs_mapped_irq[irq].sock);
/* ack_none() spuriously complains about an unexpected IRQ */
if (hs_mapped_irq[irq].old_handler != &no_irq_type)
hs_mapped_irq[irq].old_handler->ack(irq);
}
static void hs_end_irq(unsigned int irq)
{
hs_socket_enable_ireq(hs_mapped_irq[irq].sock);
hs_mapped_irq[irq].old_handler->end(irq);
}
static struct hw_interrupt_type hd64465_ss_irq_type = {
.typename = "PCMCIA-IRQ",
.startup = hs_startup_irq,
.shutdown = hs_shutdown_irq,
.enable = hs_enable_irq,
.disable = hs_disable_irq,
.ack = hs_mask_and_ack_irq,
.end = hs_end_irq
};
/*
* This function should only ever be called with interrupts disabled.
*/
static void hs_map_irq(hs_socket_t *sp, unsigned int irq)
{
DPRINTK("hs_map_irq(sock=%d irq=%d)\n", sp->number, irq);
if (irq >= HS_NUM_MAPPED_IRQS)
return;
hs_mapped_irq[irq].sock = sp;
/* insert ourselves as the irq controller */
[PATCH] genirq: rename desc->handler to desc->chip This patch-queue improves the generic IRQ layer to be truly generic, by adding various abstractions and features to it, without impacting existing functionality. While the queue can be best described as "fix and improve everything in the generic IRQ layer that we could think of", and thus it consists of many smaller features and lots of cleanups, the one feature that stands out most is the new 'irq chip' abstraction. The irq-chip abstraction is about describing and coding and IRQ controller driver by mapping its raw hardware capabilities [and quirks, if needed] in a straightforward way, without having to think about "IRQ flow" (level/edge/etc.) type of details. This stands in contrast with the current 'irq-type' model of genirq architectures, which 'mixes' raw hardware capabilities with 'flow' details. The patchset supports both types of irq controller designs at once, and converts i386 and x86_64 to the new irq-chip design. As a bonus side-effect of the irq-chip approach, chained interrupt controllers (master/slave PIC constructs, etc.) are now supported by design as well. The end result of this patchset intends to be simpler architecture-level code and more consolidation between architectures. We reused many bits of code and many concepts from Russell King's ARM IRQ layer, the merging of which was one of the motivations for this patchset. This patch: rename desc->handler to desc->chip. Originally i did not want to do this, because it's a big patch. But having both "desc->handler", "desc->handle_irq" and "action->handler" caused a large degree of confusion and made the code appear alot less clean than it truly is. I have also attempted a dual approach as well by introducing a desc->chip alias - but that just wasnt robust enough and broke frequently. So lets get over with this quickly. The conversion was done automatically via scripts and converts all the code in the kernel. This renaming patch is the first one amongst the patches, so that the remaining patches can stay flexible and can be merged and split up without having some big monolithic patch act as a merge barrier. [akpm@osdl.org: build fix] [akpm@osdl.org: another build fix] Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-29 17:24:36 +08:00
hs_mapped_irq[irq].old_handler = irq_desc[irq].chip;
irq_desc[irq].chip = &hd64465_ss_irq_type;
}
/*
* This function should only ever be called with interrupts disabled.
*/
static void hs_unmap_irq(hs_socket_t *sp, unsigned int irq)
{
DPRINTK("hs_unmap_irq(sock=%d irq=%d)\n", sp->number, irq);
if (irq >= HS_NUM_MAPPED_IRQS)
return;
/* restore the original irq controller */
[PATCH] genirq: rename desc->handler to desc->chip This patch-queue improves the generic IRQ layer to be truly generic, by adding various abstractions and features to it, without impacting existing functionality. While the queue can be best described as "fix and improve everything in the generic IRQ layer that we could think of", and thus it consists of many smaller features and lots of cleanups, the one feature that stands out most is the new 'irq chip' abstraction. The irq-chip abstraction is about describing and coding and IRQ controller driver by mapping its raw hardware capabilities [and quirks, if needed] in a straightforward way, without having to think about "IRQ flow" (level/edge/etc.) type of details. This stands in contrast with the current 'irq-type' model of genirq architectures, which 'mixes' raw hardware capabilities with 'flow' details. The patchset supports both types of irq controller designs at once, and converts i386 and x86_64 to the new irq-chip design. As a bonus side-effect of the irq-chip approach, chained interrupt controllers (master/slave PIC constructs, etc.) are now supported by design as well. The end result of this patchset intends to be simpler architecture-level code and more consolidation between architectures. We reused many bits of code and many concepts from Russell King's ARM IRQ layer, the merging of which was one of the motivations for this patchset. This patch: rename desc->handler to desc->chip. Originally i did not want to do this, because it's a big patch. But having both "desc->handler", "desc->handle_irq" and "action->handler" caused a large degree of confusion and made the code appear alot less clean than it truly is. I have also attempted a dual approach as well by introducing a desc->chip alias - but that just wasnt robust enough and broke frequently. So lets get over with this quickly. The conversion was done automatically via scripts and converts all the code in the kernel. This renaming patch is the first one amongst the patches, so that the remaining patches can stay flexible and can be merged and split up without having some big monolithic patch act as a merge barrier. [akpm@osdl.org: build fix] [akpm@osdl.org: another build fix] Signed-off-by: Ingo Molnar <mingo@elte.hu> Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-06-29 17:24:36 +08:00
irq_desc[irq].chip = hs_mapped_irq[irq].old_handler;
}
/*============================================================*/
/*
* Set Vpp and Vcc (in tenths of a Volt). Does not
* support the hi-Z state.
*
* Note, this assumes the board uses a TPS2206 chip to control
* the Vcc and Vpp voltages to the hs_sockets. If your board
* uses the MIC2563 (also supported by the HD64465) then you
* will have to modify this function.
*/
/* 0V 3.3V 5.5V */
static const u_char hs_tps2206_avcc[3] = { 0x00, 0x04, 0x08 };
static const u_char hs_tps2206_bvcc[3] = { 0x00, 0x80, 0x40 };
static int hs_set_voltages(hs_socket_t *sp, int Vcc, int Vpp)
{
u_int psr;
u_int vcci = 0;
u_int sock = sp->number;
DPRINTK("hs_set_voltage(%d, %d, %d)\n", sock, Vcc, Vpp);
switch (Vcc)
{
case 0: vcci = 0; break;
case 33: vcci = 1; break;
case 50: vcci = 2; break;
default: return 0;
}
/* Note: Vpp = 120 not supported -- Greg Banks */
if (Vpp != 0 && Vpp != Vcc)
return 0;
/* The PSR register holds 8 of the 9 bits which control
* the TPS2206 via its serial interface.
*/
psr = inw(HD64465_REG_PCCPSR);
switch (sock)
{
case 0:
psr &= 0x0f;
psr |= hs_tps2206_avcc[vcci];
psr |= (Vpp == 0 ? 0x00 : 0x02);
break;
case 1:
psr &= 0xf0;
psr |= hs_tps2206_bvcc[vcci];
psr |= (Vpp == 0 ? 0x00 : 0x20);
break;
};
outw(psr, HD64465_REG_PCCPSR);
return 1;
}
/*============================================================*/
/*
* Drive the RESET line to the card.
*/
static void hs_reset_socket(hs_socket_t *sp, int on)
{
unsigned short v;
v = hs_in(sp, GCR);
if (on)
v |= HD64465_PCCGCR_PCCR;
else
v &= ~HD64465_PCCGCR_PCCR;
hs_out(sp, v, GCR);
}
/*============================================================*/
static int hs_init(struct pcmcia_socket *s)
{
hs_socket_t *sp = container_of(s, struct hs_socket_t, socket);
DPRINTK("hs_init(%d)\n", sp->number);
return 0;
}
/*============================================================*/
static int hs_get_status(struct pcmcia_socket *s, u_int *value)
{
hs_socket_t *sp = container_of(s, struct hs_socket_t, socket);
unsigned int isr;
u_int status = 0;
isr = hs_in(sp, ISR);
/* Card is seated and powered when *both* CD pins are low */
if ((isr & HD64465_PCCISR_PCD_MASK) == 0)
{
status |= SS_DETECT; /* card present */
switch (isr & HD64465_PCCISR_PBVD_MASK)
{
case HD64465_PCCISR_PBVD_BATGOOD:
break;
case HD64465_PCCISR_PBVD_BATWARN:
status |= SS_BATWARN;
break;
default:
status |= SS_BATDEAD;
break;
}
if (isr & HD64465_PCCISR_PREADY)
status |= SS_READY;
if (isr & HD64465_PCCISR_PMWP)
status |= SS_WRPROT;
/* Voltage Select pins interpreted as per Table 4-5 of the std.
* Assuming we have the TPS2206, the socket is a "Low Voltage
* key, 3.3V and 5V available, no X.XV available".
*/
switch (isr & (HD64465_PCCISR_PVS2|HD64465_PCCISR_PVS1))
{
case HD64465_PCCISR_PVS1:
printk(KERN_NOTICE MODNAME ": cannot handle X.XV card, ignored\n");
status = 0;
break;
case 0:
case HD64465_PCCISR_PVS2:
/* 3.3V */
status |= SS_3VCARD;
break;
case HD64465_PCCISR_PVS2|HD64465_PCCISR_PVS1:
/* 5V */
break;
}
/* TODO: SS_POWERON */
/* TODO: SS_STSCHG */
}
DPRINTK("hs_get_status(%d) = %x\n", sock, status);
*value = status;
return 0;
}
/*============================================================*/
static int hs_set_socket(struct pcmcia_socket *s, socket_state_t *state)
{
hs_socket_t *sp = container_of(s, struct hs_socket_t, socket);
u_long flags;
u_int changed;
unsigned short cscier;
DPRINTK("hs_set_socket(sock=%d, flags=%x, csc_mask=%x, Vcc=%d, Vpp=%d, io_irq=%d)\n",
sock, state->flags, state->csc_mask, state->Vcc, state->Vpp, state->io_irq);
local_irq_save(flags); /* Don't want interrupts happening here */
if (state->Vpp != sp->state.Vpp ||
state->Vcc != sp->state.Vcc) {
if (!hs_set_voltages(sp, state->Vcc, state->Vpp)) {
local_irq_restore(flags);
return -EINVAL;
}
}
/* hd64465_io_debug = 1; */
/*
* Handle changes in the Card Status Change mask,
* by propagating to the CSCR register
*/
changed = sp->state.csc_mask ^ state->csc_mask;
cscier = hs_in(sp, CSCIER);
if (changed & SS_DETECT) {
if (state->csc_mask & SS_DETECT)
cscier |= HD64465_PCCCSCIER_PCDE;
else
cscier &= ~HD64465_PCCCSCIER_PCDE;
}
if (changed & SS_READY) {
if (state->csc_mask & SS_READY)
cscier |= HD64465_PCCCSCIER_PRE;
else
cscier &= ~HD64465_PCCCSCIER_PRE;
}
if (changed & SS_BATDEAD) {
if (state->csc_mask & SS_BATDEAD)
cscier |= HD64465_PCCCSCIER_PBDE;
else
cscier &= ~HD64465_PCCCSCIER_PBDE;
}
if (changed & SS_BATWARN) {
if (state->csc_mask & SS_BATWARN)
cscier |= HD64465_PCCCSCIER_PBWE;
else
cscier &= ~HD64465_PCCCSCIER_PBWE;
}
if (changed & SS_STSCHG) {
if (state->csc_mask & SS_STSCHG)
cscier |= HD64465_PCCCSCIER_PSCE;
else
cscier &= ~HD64465_PCCCSCIER_PSCE;
}
hs_out(sp, cscier, CSCIER);
if (sp->state.io_irq && !state->io_irq)
hs_unmap_irq(sp, sp->state.io_irq);
else if (!sp->state.io_irq && state->io_irq)
hs_map_irq(sp, state->io_irq);
/*
* Handle changes in the flags field,
* by propagating to config registers.
*/
changed = sp->state.flags ^ state->flags;
if (changed & SS_IOCARD) {
DPRINTK("card type: %s\n",
(state->flags & SS_IOCARD ? "i/o" : "memory" ));
bool_to_regbit(sp, GCR, HD64465_PCCGCR_PCCT,
state->flags & SS_IOCARD);
}
if (changed & SS_RESET) {
DPRINTK("%s reset card\n",
(state->flags & SS_RESET ? "start" : "stop"));
bool_to_regbit(sp, GCR, HD64465_PCCGCR_PCCR,
state->flags & SS_RESET);
}
if (changed & SS_OUTPUT_ENA) {
DPRINTK("%sabling card output\n",
(state->flags & SS_OUTPUT_ENA ? "en" : "dis"));
bool_to_regbit(sp, GCR, HD64465_PCCGCR_PDRV,
state->flags & SS_OUTPUT_ENA);
}
/* TODO: SS_SPKR_ENA */
/* hd64465_io_debug = 0; */
sp->state = *state;
local_irq_restore(flags);
#if HD64465_DEBUG > 10
if (state->flags & SS_OUTPUT_ENA)
cis_hex_dump((const unsigned char*)sp->mem_base, 0x100);
#endif
return 0;
}
/*============================================================*/
static int hs_set_io_map(struct pcmcia_socket *s, struct pccard_io_map *io)
{
hs_socket_t *sp = container_of(s, struct hs_socket_t, socket);
int map = io->map;
int sock = sp->number;
struct pccard_io_map *sio;
pgprot_t prot;
DPRINTK("hs_set_io_map(sock=%d, map=%d, flags=0x%x, speed=%dns, start=%#lx, stop=%#lx)\n",
sock, map, io->flags, io->speed, io->start, io->stop);
if (map >= MAX_IO_WIN)
return -EINVAL;
sio = &sp->io_maps[map];
/* check for null changes */
if (io->flags == sio->flags &&
io->start == sio->start &&
io->stop == sio->stop)
return 0;
if (io->flags & MAP_AUTOSZ)
prot = PAGE_KERNEL_PCC(sock, _PAGE_PCC_IODYN);
else if (io->flags & MAP_16BIT)
prot = PAGE_KERNEL_PCC(sock, _PAGE_PCC_IO16);
else
prot = PAGE_KERNEL_PCC(sock, _PAGE_PCC_IO8);
/* TODO: handle MAP_USE_WAIT */
if (io->flags & MAP_USE_WAIT)
printk(KERN_INFO MODNAME ": MAP_USE_WAIT unimplemented\n");
/* TODO: handle MAP_PREFETCH */
if (io->flags & MAP_PREFETCH)
printk(KERN_INFO MODNAME ": MAP_PREFETCH unimplemented\n");
/* TODO: handle MAP_WRPROT */
if (io->flags & MAP_WRPROT)
printk(KERN_INFO MODNAME ": MAP_WRPROT unimplemented\n");
/* TODO: handle MAP_0WS */
if (io->flags & MAP_0WS)
printk(KERN_INFO MODNAME ": MAP_0WS unimplemented\n");
if (io->flags & MAP_ACTIVE) {
unsigned long pstart, psize, paddrbase;
paddrbase = virt_to_phys((void*)(sp->mem_base + 2 * HD64465_PCC_WINDOW));
pstart = io->start & PAGE_MASK;
psize = ((io->stop + PAGE_SIZE) & PAGE_MASK) - pstart;
/*
* Change PTEs in only that portion of the mapping requested
* by the caller. This means that most of the time, most of
* the PTEs in the io_vma will be unmapped and only the bottom
* page will be mapped. But the code allows for weird cards
* that might want IO ports > 4K.
*/
sp->io_base = p3_ioremap(paddrbase + pstart, psize, pgprot_val(prot));
/*
* Change the mapping used by inb() outb() etc
*/
hd64465_port_map(io->start,
io->stop - io->start + 1,
(unsigned long)sp->io_base + io->start, 0);
} else {
hd64465_port_unmap(sio->start, sio->stop - sio->start + 1);
p3_iounmap(sp->io_base);
}
*sio = *io;
return 0;
}
/*============================================================*/
static int hs_set_mem_map(struct pcmcia_socket *s, struct pccard_mem_map *mem)
{
hs_socket_t *sp = container_of(s, struct hs_socket_t, socket);
struct pccard_mem_map *smem;
int map = mem->map;
unsigned long paddr;
#if 0
DPRINTK("hs_set_mem_map(sock=%d, map=%d, flags=0x%x, card_start=0x%08x)\n",
sock, map, mem->flags, mem->card_start);
#endif
if (map >= MAX_WIN)
return -EINVAL;
smem = &sp->mem_maps[map];
paddr = sp->mem_base; /* base of Attribute mapping */
if (!(mem->flags & MAP_ATTRIB))
paddr += HD64465_PCC_WINDOW; /* base of Common mapping */
paddr += mem->card_start;
/* Because we specified SS_CAP_STATIC_MAP, we are obliged
* at this time to report the system address corresponding
* to the card address requested. This is how Socket Services
* queries our fixed mapping. I wish this fact had been
* documented - Greg Banks.
*/
mem->static_start = paddr;
*smem = *mem;
return 0;
}
/* TODO: do we need to use the MMU to access Common memory ??? */
/*============================================================*/
/*
* This function is registered with the HD64465 glue code to do a
* secondary demux step on the PCMCIA interrupts. It handles
* mapping the IREQ request from the card to a standard Linux
* IRQ, as requested by SocketServices.
*/
static int hs_irq_demux(int irq, void *dev)
{
hs_socket_t *sp = dev;
u_int cscr;
DPRINTK("hs_irq_demux(irq=%d)\n", irq);
if (sp->state.io_irq &&
(cscr = hs_in(sp, CSCR)) & HD64465_PCCCSCR_PIREQ) {
cscr &= ~HD64465_PCCCSCR_PIREQ;
hs_out(sp, cscr, CSCR);
return sp->state.io_irq;
}
return irq;
}
/*============================================================*/
/*
* Interrupt handling routine.
*/
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 hs_interrupt(int irq, void *dev)
{
hs_socket_t *sp = dev;
u_int events = 0;
u_int cscr;
cscr = hs_in(sp, CSCR);
DPRINTK("hs_interrupt, cscr=%04x\n", cscr);
/* check for bus-related changes to be reported to Socket Services */
if (cscr & HD64465_PCCCSCR_PCDC) {
/* double-check for a 16-bit card, as we don't support CardBus */
if ((hs_in(sp, ISR) & HD64465_PCCISR_PCD_MASK) != 0) {
printk(KERN_NOTICE MODNAME
": socket %d, card not a supported card type or not inserted correctly\n",
sp->number);
/* Don't do the rest unless a card is present */
cscr &= ~(HD64465_PCCCSCR_PCDC|
HD64465_PCCCSCR_PRC|
HD64465_PCCCSCR_PBW|
HD64465_PCCCSCR_PBD|
HD64465_PCCCSCR_PSC);
} else {
cscr &= ~HD64465_PCCCSCR_PCDC;
events |= SS_DETECT; /* card insertion or removal */
}
}
if (cscr & HD64465_PCCCSCR_PRC) {
cscr &= ~HD64465_PCCCSCR_PRC;
events |= SS_READY; /* ready signal changed */
}
if (cscr & HD64465_PCCCSCR_PBW) {
cscr &= ~HD64465_PCCCSCR_PSC;
events |= SS_BATWARN; /* battery warning */
}
if (cscr & HD64465_PCCCSCR_PBD) {
cscr &= ~HD64465_PCCCSCR_PSC;
events |= SS_BATDEAD; /* battery dead */
}
if (cscr & HD64465_PCCCSCR_PSC) {
cscr &= ~HD64465_PCCCSCR_PSC;
events |= SS_STSCHG; /* STSCHG (status changed) signal */
}
if (cscr & HD64465_PCCCSCR_PIREQ) {
cscr &= ~HD64465_PCCCSCR_PIREQ;
/* This should have been dealt with during irq demux */
printk(KERN_NOTICE MODNAME ": unexpected IREQ from card\n");
}
hs_out(sp, cscr, CSCR);
if (events)
pcmcia_parse_events(&sp->socket, events);
return IRQ_HANDLED;
}
/*============================================================*/
static struct pccard_operations hs_operations = {
.init = hs_init,
.get_status = hs_get_status,
.set_socket = hs_set_socket,
.set_io_map = hs_set_io_map,
.set_mem_map = hs_set_mem_map,
};
static int hs_init_socket(hs_socket_t *sp, int irq, unsigned long mem_base,
unsigned int ctrl_base)
{
unsigned short v;
int i, err;
memset(sp, 0, sizeof(*sp));
sp->irq = irq;
sp->mem_base = mem_base;
sp->mem_length = 4*HD64465_PCC_WINDOW; /* 16MB */
sp->ctrl_base = ctrl_base;
for (i=0 ; i<MAX_IO_WIN ; i++)
sp->io_maps[i].map = i;
for (i=0 ; i<MAX_WIN ; i++)
sp->mem_maps[i].map = i;
hd64465_register_irq_demux(sp->irq, hs_irq_demux, sp);
if ((err = request_irq(sp->irq, hs_interrupt, IRQF_DISABLED, MODNAME, sp)) < 0)
return err;
if (request_mem_region(sp->mem_base, sp->mem_length, MODNAME) == 0) {
sp->mem_base = 0;
return -ENOMEM;
}
/* According to section 3.2 of the PCMCIA standard, low-voltage
* capable cards must implement cold insertion, i.e. Vpp and
* Vcc set to 0 before card is inserted.
*/
/*hs_set_voltages(sp, 0, 0);*/
/* hi-Z the outputs to the card and set 16MB map mode */
v = hs_in(sp, GCR);
v &= ~HD64465_PCCGCR_PCCT; /* memory-only card */
hs_out(sp, v, GCR);
v = hs_in(sp, GCR);
v |= HD64465_PCCGCR_PDRV; /* enable outputs to card */
hs_out(sp, v, GCR);
v = hs_in(sp, GCR);
v |= HD64465_PCCGCR_PMMOD; /* 16MB mapping mode */
hs_out(sp, v, GCR);
v = hs_in(sp, GCR);
/* lowest 16MB of Common */
v &= ~(HD64465_PCCGCR_PPA25|HD64465_PCCGCR_PPA24);
hs_out(sp, v, GCR);
hs_reset_socket(sp, 1);
printk(KERN_INFO "HD64465 PCMCIA bridge socket %d at 0x%08lx irq %d\n",
i, sp->mem_base, sp->irq);
return 0;
}
static void hs_exit_socket(hs_socket_t *sp)
{
unsigned short cscier, gcr;
unsigned long flags;
local_irq_save(flags);
/* turn off interrupts in hardware */
cscier = hs_in(sp, CSCIER);
cscier = (cscier & IER_MASK) | IER_OFF;
hs_out(sp, cscier, CSCIER);
/* hi-Z the outputs to the card */
gcr = hs_in(sp, GCR);
gcr &= HD64465_PCCGCR_PDRV;
hs_out(sp, gcr, GCR);
/* power the card down */
hs_set_voltages(sp, 0, 0);
if (sp->mem_base != 0)
release_mem_region(sp->mem_base, sp->mem_length);
if (sp->irq != 0) {
free_irq(sp->irq, hs_interrupt);
hd64465_unregister_irq_demux(sp->irq);
}
local_irq_restore(flags);
}
static struct device_driver hd64465_driver = {
.name = "hd64465-pcmcia",
.bus = &platform_bus_type,
.suspend = pcmcia_socket_dev_suspend,
.resume = pcmcia_socket_dev_resume,
};
static struct platform_device hd64465_device = {
.name = "hd64465-pcmcia",
.id = 0,
};
static int __init init_hs(void)
{
int i;
unsigned short v;
/* hd64465_io_debug = 1; */
if (driver_register(&hd64465_driver))
return -EINVAL;
/* Wake both sockets out of STANDBY mode */
/* TODO: wait 15ms */
v = inw(HD64465_REG_SMSCR);
v &= ~(HD64465_SMSCR_PC0ST|HD64465_SMSCR_PC1ST);
outw(v, HD64465_REG_SMSCR);
/* keep power controller out of shutdown mode */
v = inb(HD64465_REG_PCC0SCR);
v |= HD64465_PCCSCR_SHDN;
outb(v, HD64465_REG_PCC0SCR);
/* use serial (TPS2206) power controller */
v = inb(HD64465_REG_PCC0CSCR);
v |= HD64465_PCCCSCR_PSWSEL;
outb(v, HD64465_REG_PCC0CSCR);
/*
* Setup hs_sockets[] structures and request system resources.
* TODO: on memory allocation failure, power down the socket
* before quitting.
*/
for (i=0; i<HS_MAX_SOCKETS; i++) {
hs_set_voltages(&hs_sockets[i], 0, 0);
hs_sockets[i].socket.features |= SS_CAP_PCCARD | SS_CAP_STATIC_MAP; /* mappings are fixed in host memory */
hs_sockets[i].socket.resource_ops = &pccard_static_ops;
hs_sockets[i].socket.irq_mask = 0xffde;/*0xffff*/ /* IRQs mapped in s/w so can do any, really */
hs_sockets[i].socket.map_size = HD64465_PCC_WINDOW; /* 16MB fixed window size */
hs_sockets[i].socket.owner = THIS_MODULE;
hs_sockets[i].socket.ss_entry = &hs_operations;
}
i = hs_init_socket(&hs_sockets[0],
HD64465_IRQ_PCMCIA0,
HD64465_PCC0_BASE,
HD64465_REG_PCC0ISR);
if (i < 0) {
unregister_driver(&hd64465_driver);
return i;
}
i = hs_init_socket(&hs_sockets[1],
HD64465_IRQ_PCMCIA1,
HD64465_PCC1_BASE,
HD64465_REG_PCC1ISR);
if (i < 0) {
unregister_driver(&hd64465_driver);
return i;
}
/* hd64465_io_debug = 0; */
platform_device_register(&hd64465_device);
for (i=0; i<HS_MAX_SOCKETS; i++) {
unsigned int ret;
hs_sockets[i].socket.dev.parent = &hd64465_device.dev;
hs_sockets[i].number = i;
ret = pcmcia_register_socket(&hs_sockets[i].socket);
if (ret && i)
pcmcia_unregister_socket(&hs_sockets[0].socket);
}
return 0;
}
static void __exit exit_hs(void)
{
int i;
for (i=0 ; i<HS_MAX_SOCKETS ; i++) {
pcmcia_unregister_socket(&hs_sockets[i].socket);
hs_exit_socket(&hs_sockets[i]);
}
platform_device_unregister(&hd64465_device);
unregister_driver(&hd64465_driver);
}
module_init(init_hs);
module_exit(exit_hs);
/*============================================================*/
/*END*/