android_kernel_motorola_sm6225/drivers/char/riscom8.c

1758 lines
43 KiB
C
Raw Normal View History

/*
* linux/drivers/char/riscom.c -- RISCom/8 multiport serial driver.
*
* Copyright (C) 1994-1996 Dmitry Gorodchanin (pgmdsg@ibi.com)
*
* This code is loosely based on the Linux serial driver, written by
* Linus Torvalds, Theodore T'so and others. The RISCom/8 card
* programming info was obtained from various drivers for other OSes
* (FreeBSD, ISC, etc), but no source code from those drivers were
* directly included in this driver.
*
*
* This program is free software; you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation; either version 2 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program; if not, write to the Free Software
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
*
* Revision 1.1
*
* ChangeLog:
* Arnaldo Carvalho de Melo <acme@conectiva.com.br> - 27-Jun-2001
* - get rid of check_region and several cleanups
*/
#include <linux/module.h>
#include <asm/io.h>
#include <linux/kernel.h>
#include <linux/sched.h>
#include <linux/ioport.h>
#include <linux/interrupt.h>
#include <linux/errno.h>
#include <linux/tty.h>
#include <linux/mm.h>
#include <linux/serial.h>
#include <linux/fcntl.h>
#include <linux/major.h>
#include <linux/init.h>
#include <linux/delay.h>
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
#include <linux/tty_flip.h>
#include <asm/uaccess.h>
#include "riscom8.h"
#include "riscom8_reg.h"
/* Am I paranoid or not ? ;-) */
#define RISCOM_PARANOIA_CHECK
/*
* Crazy InteliCom/8 boards sometimes has swapped CTS & DSR signals.
* You can slightly speed up things by #undefing the following option,
* if you are REALLY sure that your board is correct one.
*/
#define RISCOM_BRAIN_DAMAGED_CTS
/*
* The following defines are mostly for testing purposes. But if you need
* some nice reporting in your syslog, you can define them also.
*/
#undef RC_REPORT_FIFO
#undef RC_REPORT_OVERRUN
#define RISCOM_LEGAL_FLAGS \
(ASYNC_HUP_NOTIFY | ASYNC_SAK | ASYNC_SPLIT_TERMIOS | \
ASYNC_SPD_HI | ASYNC_SPEED_VHI | ASYNC_SESSION_LOCKOUT | \
ASYNC_PGRP_LOCKOUT | ASYNC_CALLOUT_NOHUP)
#define RS_EVENT_WRITE_WAKEUP 0
static struct riscom_board * IRQ_to_board[16];
static struct tty_driver *riscom_driver;
static struct riscom_board rc_board[RC_NBOARD] = {
{
.base = RC_IOBASE1,
},
{
.base = RC_IOBASE2,
},
{
.base = RC_IOBASE3,
},
{
.base = RC_IOBASE4,
},
};
static struct riscom_port rc_port[RC_NBOARD * RC_NPORT];
/* RISCom/8 I/O ports addresses (without address translation) */
static unsigned short rc_ioport[] = {
#if 1
0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x09, 0x0a, 0x0b, 0x0c,
#else
0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x09, 0x0a, 0x0b, 0x0c, 0x10,
0x11, 0x12, 0x18, 0x28, 0x31, 0x32, 0x39, 0x3a, 0x40, 0x41, 0x61, 0x62,
0x63, 0x64, 0x6b, 0x70, 0x71, 0x78, 0x7a, 0x7b, 0x7f, 0x100, 0x101
#endif
};
#define RC_NIOPORT ARRAY_SIZE(rc_ioport)
static inline int rc_paranoia_check(struct riscom_port const * port,
char *name, const char *routine)
{
#ifdef RISCOM_PARANOIA_CHECK
static const char badmagic[] = KERN_INFO
"rc: Warning: bad riscom port magic number for device %s in %s\n";
static const char badinfo[] = KERN_INFO
"rc: Warning: null riscom port for device %s in %s\n";
if (!port) {
printk(badinfo, name, routine);
return 1;
}
if (port->magic != RISCOM8_MAGIC) {
printk(badmagic, name, routine);
return 1;
}
#endif
return 0;
}
/*
*
* Service functions for RISCom/8 driver.
*
*/
/* Get board number from pointer */
static inline int board_No (struct riscom_board const * bp)
{
return bp - rc_board;
}
/* Get port number from pointer */
static inline int port_No (struct riscom_port const * port)
{
return RC_PORT(port - rc_port);
}
/* Get pointer to board from pointer to port */
static inline struct riscom_board * port_Board(struct riscom_port const * port)
{
return &rc_board[RC_BOARD(port - rc_port)];
}
/* Input Byte from CL CD180 register */
static inline unsigned char rc_in(struct riscom_board const * bp, unsigned short reg)
{
return inb(bp->base + RC_TO_ISA(reg));
}
/* Output Byte to CL CD180 register */
static inline void rc_out(struct riscom_board const * bp, unsigned short reg,
unsigned char val)
{
outb(val, bp->base + RC_TO_ISA(reg));
}
/* Wait for Channel Command Register ready */
static inline void rc_wait_CCR(struct riscom_board const * bp)
{
unsigned long delay;
/* FIXME: need something more descriptive then 100000 :) */
for (delay = 100000; delay; delay--)
if (!rc_in(bp, CD180_CCR))
return;
printk(KERN_INFO "rc%d: Timeout waiting for CCR.\n", board_No(bp));
}
/*
* RISCom/8 probe functions.
*/
static inline int rc_request_io_range(struct riscom_board * const bp)
{
int i;
for (i = 0; i < RC_NIOPORT; i++)
if (!request_region(RC_TO_ISA(rc_ioport[i]) + bp->base, 1,
"RISCom/8")) {
goto out_release;
}
return 0;
out_release:
printk(KERN_INFO "rc%d: Skipping probe at 0x%03x. IO address in use.\n",
board_No(bp), bp->base);
while(--i >= 0)
release_region(RC_TO_ISA(rc_ioport[i]) + bp->base, 1);
return 1;
}
static inline void rc_release_io_range(struct riscom_board * const bp)
{
int i;
for (i = 0; i < RC_NIOPORT; i++)
release_region(RC_TO_ISA(rc_ioport[i]) + bp->base, 1);
}
/* Reset and setup CD180 chip */
static void __init rc_init_CD180(struct riscom_board const * bp)
{
unsigned long flags;
save_flags(flags); cli();
rc_out(bp, RC_CTOUT, 0); /* Clear timeout */
rc_wait_CCR(bp); /* Wait for CCR ready */
rc_out(bp, CD180_CCR, CCR_HARDRESET); /* Reset CD180 chip */
sti();
msleep(50); /* Delay 0.05 sec */
cli();
rc_out(bp, CD180_GIVR, RC_ID); /* Set ID for this chip */
rc_out(bp, CD180_GICR, 0); /* Clear all bits */
rc_out(bp, CD180_PILR1, RC_ACK_MINT); /* Prio for modem intr */
rc_out(bp, CD180_PILR2, RC_ACK_TINT); /* Prio for transmitter intr */
rc_out(bp, CD180_PILR3, RC_ACK_RINT); /* Prio for receiver intr */
/* Setting up prescaler. We need 4 ticks per 1 ms */
rc_out(bp, CD180_PPRH, (RC_OSCFREQ/(1000000/RISCOM_TPS)) >> 8);
rc_out(bp, CD180_PPRL, (RC_OSCFREQ/(1000000/RISCOM_TPS)) & 0xff);
restore_flags(flags);
}
/* Main probing routine, also sets irq. */
static int __init rc_probe(struct riscom_board *bp)
{
unsigned char val1, val2;
int irqs = 0;
int retries;
bp->irq = 0;
if (rc_request_io_range(bp))
return 1;
/* Are the I/O ports here ? */
rc_out(bp, CD180_PPRL, 0x5a);
outb(0xff, 0x80);
val1 = rc_in(bp, CD180_PPRL);
rc_out(bp, CD180_PPRL, 0xa5);
outb(0x00, 0x80);
val2 = rc_in(bp, CD180_PPRL);
if ((val1 != 0x5a) || (val2 != 0xa5)) {
printk(KERN_ERR "rc%d: RISCom/8 Board at 0x%03x not found.\n",
board_No(bp), bp->base);
goto out_release;
}
/* It's time to find IRQ for this board */
for (retries = 0; retries < 5 && irqs <= 0; retries++) {
irqs = probe_irq_on();
rc_init_CD180(bp); /* Reset CD180 chip */
rc_out(bp, CD180_CAR, 2); /* Select port 2 */
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_TXEN); /* Enable transmitter */
rc_out(bp, CD180_IER, IER_TXRDY); /* Enable tx empty intr */
msleep(50);
irqs = probe_irq_off(irqs);
val1 = rc_in(bp, RC_BSR); /* Get Board Status reg */
val2 = rc_in(bp, RC_ACK_TINT); /* ACK interrupt */
rc_init_CD180(bp); /* Reset CD180 again */
if ((val1 & RC_BSR_TINT) || (val2 != (RC_ID | GIVR_IT_TX))) {
printk(KERN_ERR "rc%d: RISCom/8 Board at 0x%03x not "
"found.\n", board_No(bp), bp->base);
goto out_release;
}
}
if (irqs <= 0) {
printk(KERN_ERR "rc%d: Can't find IRQ for RISCom/8 board "
"at 0x%03x.\n", board_No(bp), bp->base);
goto out_release;
}
bp->irq = irqs;
bp->flags |= RC_BOARD_PRESENT;
printk(KERN_INFO "rc%d: RISCom/8 Rev. %c board detected at "
"0x%03x, IRQ %d.\n",
board_No(bp),
(rc_in(bp, CD180_GFRCR) & 0x0f) + 'A', /* Board revision */
bp->base, bp->irq);
return 0;
out_release:
rc_release_io_range(bp);
return 1;
}
/*
*
* Interrupt processing routines.
*
*/
static inline void rc_mark_event(struct riscom_port * port, int event)
{
set_bit(event, &port->event);
schedule_work(&port->tqueue);
}
static inline struct riscom_port * rc_get_port(struct riscom_board const * bp,
unsigned char const * what)
{
unsigned char channel;
struct riscom_port * port;
channel = rc_in(bp, CD180_GICR) >> GICR_CHAN_OFF;
if (channel < CD180_NCH) {
port = &rc_port[board_No(bp) * RC_NPORT + channel];
if (port->flags & ASYNC_INITIALIZED) {
return port;
}
}
printk(KERN_ERR "rc%d: %s interrupt from invalid port %d\n",
board_No(bp), what, channel);
return NULL;
}
static inline void rc_receive_exc(struct riscom_board const * bp)
{
struct riscom_port *port;
struct tty_struct *tty;
unsigned char status;
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
unsigned char ch, flag;
if (!(port = rc_get_port(bp, "Receive")))
return;
tty = port->tty;
#ifdef RC_REPORT_OVERRUN
status = rc_in(bp, CD180_RCSR);
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
if (status & RCSR_OE)
port->overrun++;
status &= port->mark_mask;
#else
status = rc_in(bp, CD180_RCSR) & port->mark_mask;
#endif
ch = rc_in(bp, CD180_RDR);
if (!status) {
return;
}
if (status & RCSR_TOUT) {
printk(KERN_WARNING "rc%d: port %d: Receiver timeout. "
"Hardware problems ?\n",
board_No(bp), port_No(port));
return;
} else if (status & RCSR_BREAK) {
printk(KERN_INFO "rc%d: port %d: Handling break...\n",
board_No(bp), port_No(port));
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
flag = TTY_BREAK;
if (port->flags & ASYNC_SAK)
do_SAK(tty);
} else if (status & RCSR_PE)
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
flag = TTY_PARITY;
else if (status & RCSR_FE)
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
flag = TTY_FRAME;
else if (status & RCSR_OE)
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
flag = TTY_OVERRUN;
else
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
flag = TTY_NORMAL;
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
tty_insert_flip_char(tty, ch, flag);
tty_flip_buffer_push(tty);
}
static inline void rc_receive(struct riscom_board const * bp)
{
struct riscom_port *port;
struct tty_struct *tty;
unsigned char count;
if (!(port = rc_get_port(bp, "Receive")))
return;
tty = port->tty;
count = rc_in(bp, CD180_RDCR);
#ifdef RC_REPORT_FIFO
port->hits[count > 8 ? 9 : count]++;
#endif
while (count--) {
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
if (tty_buffer_request_room(tty, 1) == 0) {
printk(KERN_WARNING "rc%d: port %d: Working around "
"flip buffer overflow.\n",
board_No(bp), port_No(port));
break;
}
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
tty_insert_flip_char(tty, rc_in(bp, CD180_RDR), TTY_NORMAL);
}
[PATCH] TTY layer buffering revamp The API and code have been through various bits of initial review by serial driver people but they definitely need to live somewhere for a while so the unconverted drivers can get knocked into shape, existing drivers that have been updated can be better tuned and bugs whacked out. This replaces the tty flip buffers with kmalloc objects in rings. In the normal situation for an IRQ driven serial port at typical speeds the behaviour is pretty much the same, two buffers end up allocated and the kernel cycles between them as before. When there are delays or at high speed we now behave far better as the buffer pool can grow a bit rather than lose characters. This also means that we can operate at higher speeds reliably. For drivers that receive characters in blocks (DMA based, USB and especially virtualisation) the layer allows a lot of driver specific code that works around the tty layer with private secondary queues to be removed. The IBM folks need this sort of layer, the smart serial port people do, the virtualisers do (because a virtualised tty typically operates at infinite speed rather than emulating 9600 baud). Finally many drivers had invalid and unsafe attempts to avoid buffer overflows by directly invoking tty methods extracted out of the innards of work queue structs. These are no longer needed and all go away. That fixes various random hangs with serial ports on overflow. The other change in here is to optimise the receive_room path that is used by some callers. It turns out that only one ldisc uses receive room except asa constant and it updates it far far less than the value is read. We thus make it a variable not a function call. I expect the code to contain bugs due to the size alone but I'll be watching and squashing them and feeding out new patches as it goes. Because the buffers now dynamically expand you should only run out of buffering when the kernel runs out of memory for real. That means a lot of the horrible hacks high performance drivers used to do just aren't needed any more. Description: tty_insert_flip_char is an old API and continues to work as before, as does tty_flip_buffer_push() [this is why many drivers dont need modification]. It does now also return the number of chars inserted There are also tty_buffer_request_room(tty, len) which asks for a buffer block of the length requested and returns the space found. This improves efficiency with hardware that knows how much to transfer. and tty_insert_flip_string_flags(tty, str, flags, len) to insert a string of characters and flags For a smart interface the usual code is len = tty_request_buffer_room(tty, amount_hardware_says); tty_insert_flip_string(tty, buffer_from_card, len); More description! At the moment tty buffers are attached directly to the tty. This is causing a lot of the problems related to tty layer locking, also problems at high speed and also with bursty data (such as occurs in virtualised environments) I'm working on ripping out the flip buffers and replacing them with a pool of dynamically allocated buffers. This allows both for old style "byte I/O" devices and also helps virtualisation and smart devices where large blocks of data suddenely materialise and need storing. So far so good. Lots of drivers reference tty->flip.*. Several of them also call directly and unsafely into function pointers it provides. This will all break. Most drivers can use tty_insert_flip_char which can be kept as an API but others need more. At the moment I've added the following interfaces, if people think more will be needed now is a good time to say int tty_buffer_request_room(tty, size) Try and ensure at least size bytes are available, returns actual room (may be zero). At the moment it just uses the flipbuf space but that will change. Repeated calls without characters being added are not cumulative. (ie if you call it with 1, 1, 1, and then 4 you'll have four characters of space. The other functions will also try and grow buffers in future but this will be a more efficient way when you know block sizes. int tty_insert_flip_char(tty, ch, flag) As before insert a character if there is room. Now returns 1 for success, 0 for failure. int tty_insert_flip_string(tty, str, len) Insert a block of non error characters. Returns the number inserted. int tty_prepare_flip_string(tty, strptr, len) Adjust the buffer to allow len characters to be added. Returns a buffer pointer in strptr and the length available. This allows for hardware that needs to use functions like insl or mencpy_fromio. Signed-off-by: Alan Cox <alan@redhat.com> Cc: Paul Fulghum <paulkf@microgate.com> Signed-off-by: Hirokazu Takata <takata@linux-m32r.org> Signed-off-by: Serge Hallyn <serue@us.ibm.com> Signed-off-by: Jeff Dike <jdike@addtoit.com> Signed-off-by: John Hawkes <hawkes@sgi.com> Signed-off-by: Martin Schwidefsky <schwidefsky@de.ibm.com> Signed-off-by: Adrian Bunk <bunk@stusta.de> Signed-off-by: Andrew Morton <akpm@osdl.org> Signed-off-by: Linus Torvalds <torvalds@osdl.org>
2006-01-10 05:54:13 +01:00
tty_flip_buffer_push(tty);
}
static inline void rc_transmit(struct riscom_board const * bp)
{
struct riscom_port *port;
struct tty_struct *tty;
unsigned char count;
if (!(port = rc_get_port(bp, "Transmit")))
return;
tty = port->tty;
if (port->IER & IER_TXEMPTY) {
/* FIFO drained */
rc_out(bp, CD180_CAR, port_No(port));
port->IER &= ~IER_TXEMPTY;
rc_out(bp, CD180_IER, port->IER);
return;
}
if ((port->xmit_cnt <= 0 && !port->break_length)
|| tty->stopped || tty->hw_stopped) {
rc_out(bp, CD180_CAR, port_No(port));
port->IER &= ~IER_TXRDY;
rc_out(bp, CD180_IER, port->IER);
return;
}
if (port->break_length) {
if (port->break_length > 0) {
if (port->COR2 & COR2_ETC) {
rc_out(bp, CD180_TDR, CD180_C_ESC);
rc_out(bp, CD180_TDR, CD180_C_SBRK);
port->COR2 &= ~COR2_ETC;
}
count = min_t(int, port->break_length, 0xff);
rc_out(bp, CD180_TDR, CD180_C_ESC);
rc_out(bp, CD180_TDR, CD180_C_DELAY);
rc_out(bp, CD180_TDR, count);
if (!(port->break_length -= count))
port->break_length--;
} else {
rc_out(bp, CD180_TDR, CD180_C_ESC);
rc_out(bp, CD180_TDR, CD180_C_EBRK);
rc_out(bp, CD180_COR2, port->COR2);
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_CORCHG2);
port->break_length = 0;
}
return;
}
count = CD180_NFIFO;
do {
rc_out(bp, CD180_TDR, port->xmit_buf[port->xmit_tail++]);
port->xmit_tail = port->xmit_tail & (SERIAL_XMIT_SIZE-1);
if (--port->xmit_cnt <= 0)
break;
} while (--count > 0);
if (port->xmit_cnt <= 0) {
rc_out(bp, CD180_CAR, port_No(port));
port->IER &= ~IER_TXRDY;
rc_out(bp, CD180_IER, port->IER);
}
if (port->xmit_cnt <= port->wakeup_chars)
rc_mark_event(port, RS_EVENT_WRITE_WAKEUP);
}
static inline void rc_check_modem(struct riscom_board const * bp)
{
struct riscom_port *port;
struct tty_struct *tty;
unsigned char mcr;
if (!(port = rc_get_port(bp, "Modem")))
return;
tty = port->tty;
mcr = rc_in(bp, CD180_MCR);
if (mcr & MCR_CDCHG) {
if (rc_in(bp, CD180_MSVR) & MSVR_CD)
wake_up_interruptible(&port->open_wait);
else
schedule_work(&port->tqueue_hangup);
}
#ifdef RISCOM_BRAIN_DAMAGED_CTS
if (mcr & MCR_CTSCHG) {
if (rc_in(bp, CD180_MSVR) & MSVR_CTS) {
tty->hw_stopped = 0;
port->IER |= IER_TXRDY;
if (port->xmit_cnt <= port->wakeup_chars)
rc_mark_event(port, RS_EVENT_WRITE_WAKEUP);
} else {
tty->hw_stopped = 1;
port->IER &= ~IER_TXRDY;
}
rc_out(bp, CD180_IER, port->IER);
}
if (mcr & MCR_DSRCHG) {
if (rc_in(bp, CD180_MSVR) & MSVR_DSR) {
tty->hw_stopped = 0;
port->IER |= IER_TXRDY;
if (port->xmit_cnt <= port->wakeup_chars)
rc_mark_event(port, RS_EVENT_WRITE_WAKEUP);
} else {
tty->hw_stopped = 1;
port->IER &= ~IER_TXRDY;
}
rc_out(bp, CD180_IER, port->IER);
}
#endif /* RISCOM_BRAIN_DAMAGED_CTS */
/* Clear change bits */
rc_out(bp, CD180_MCR, 0);
}
/* The main interrupt processing 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 15:55:46 +02:00
static irqreturn_t rc_interrupt(int irq, void * dev_id)
{
unsigned char status;
unsigned char ack;
struct riscom_board *bp;
unsigned long loop = 0;
int handled = 0;
bp = IRQ_to_board[irq];
if (!(bp->flags & RC_BOARD_ACTIVE))
return IRQ_NONE;
while ((++loop < 16) && ((status = ~(rc_in(bp, RC_BSR))) &
(RC_BSR_TOUT | RC_BSR_TINT |
RC_BSR_MINT | RC_BSR_RINT))) {
handled = 1;
if (status & RC_BSR_TOUT)
printk(KERN_WARNING "rc%d: Got timeout. Hardware "
"error?\n", board_No(bp));
else if (status & RC_BSR_RINT) {
ack = rc_in(bp, RC_ACK_RINT);
if (ack == (RC_ID | GIVR_IT_RCV))
rc_receive(bp);
else if (ack == (RC_ID | GIVR_IT_REXC))
rc_receive_exc(bp);
else
printk(KERN_WARNING "rc%d: Bad receive ack "
"0x%02x.\n",
board_No(bp), ack);
} else if (status & RC_BSR_TINT) {
ack = rc_in(bp, RC_ACK_TINT);
if (ack == (RC_ID | GIVR_IT_TX))
rc_transmit(bp);
else
printk(KERN_WARNING "rc%d: Bad transmit ack "
"0x%02x.\n",
board_No(bp), ack);
} else /* if (status & RC_BSR_MINT) */ {
ack = rc_in(bp, RC_ACK_MINT);
if (ack == (RC_ID | GIVR_IT_MODEM))
rc_check_modem(bp);
else
printk(KERN_WARNING "rc%d: Bad modem ack "
"0x%02x.\n",
board_No(bp), ack);
}
rc_out(bp, CD180_EOIR, 0); /* Mark end of interrupt */
rc_out(bp, RC_CTOUT, 0); /* Clear timeout flag */
}
return IRQ_RETVAL(handled);
}
/*
* Routines for open & close processing.
*/
/* Called with disabled interrupts */
static inline int rc_setup_board(struct riscom_board * bp)
{
int error;
if (bp->flags & RC_BOARD_ACTIVE)
return 0;
error = request_irq(bp->irq, rc_interrupt, IRQF_DISABLED,
"RISCom/8", NULL);
if (error)
return error;
rc_out(bp, RC_CTOUT, 0); /* Just in case */
bp->DTR = ~0;
rc_out(bp, RC_DTR, bp->DTR); /* Drop DTR on all ports */
IRQ_to_board[bp->irq] = bp;
bp->flags |= RC_BOARD_ACTIVE;
return 0;
}
/* Called with disabled interrupts */
static inline void rc_shutdown_board(struct riscom_board *bp)
{
if (!(bp->flags & RC_BOARD_ACTIVE))
return;
bp->flags &= ~RC_BOARD_ACTIVE;
free_irq(bp->irq, NULL);
IRQ_to_board[bp->irq] = NULL;
bp->DTR = ~0;
rc_out(bp, RC_DTR, bp->DTR); /* Drop DTR on all ports */
}
/*
* Setting up port characteristics.
* Must be called with disabled interrupts
*/
static void rc_change_speed(struct riscom_board *bp, struct riscom_port *port)
{
struct tty_struct *tty;
unsigned long baud;
long tmp;
unsigned char cor1 = 0, cor3 = 0;
unsigned char mcor1 = 0, mcor2 = 0;
if (!(tty = port->tty) || !tty->termios)
return;
port->IER = 0;
port->COR2 = 0;
port->MSVR = MSVR_RTS;
baud = tty_get_baud_rate(tty);
/* Select port on the board */
rc_out(bp, CD180_CAR, port_No(port));
if (!baud) {
/* Drop DTR & exit */
bp->DTR |= (1u << port_No(port));
rc_out(bp, RC_DTR, bp->DTR);
return;
} else {
/* Set DTR on */
bp->DTR &= ~(1u << port_No(port));
rc_out(bp, RC_DTR, bp->DTR);
}
/*
* Now we must calculate some speed depended things
*/
/* Set baud rate for port */
tmp = (((RC_OSCFREQ + baud/2) / baud +
CD180_TPC/2) / CD180_TPC);
rc_out(bp, CD180_RBPRH, (tmp >> 8) & 0xff);
rc_out(bp, CD180_TBPRH, (tmp >> 8) & 0xff);
rc_out(bp, CD180_RBPRL, tmp & 0xff);
rc_out(bp, CD180_TBPRL, tmp & 0xff);
baud = (baud + 5) / 10; /* Estimated CPS */
/* Two timer ticks seems enough to wakeup something like SLIP driver */
tmp = ((baud + HZ/2) / HZ) * 2 - CD180_NFIFO;
port->wakeup_chars = (tmp < 0) ? 0 : ((tmp >= SERIAL_XMIT_SIZE) ?
SERIAL_XMIT_SIZE - 1 : tmp);
/* Receiver timeout will be transmission time for 1.5 chars */
tmp = (RISCOM_TPS + RISCOM_TPS/2 + baud/2) / baud;
tmp = (tmp > 0xff) ? 0xff : tmp;
rc_out(bp, CD180_RTPR, tmp);
switch (C_CSIZE(tty)) {
case CS5:
cor1 |= COR1_5BITS;
break;
case CS6:
cor1 |= COR1_6BITS;
break;
case CS7:
cor1 |= COR1_7BITS;
break;
case CS8:
cor1 |= COR1_8BITS;
break;
}
if (C_CSTOPB(tty))
cor1 |= COR1_2SB;
cor1 |= COR1_IGNORE;
if (C_PARENB(tty)) {
cor1 |= COR1_NORMPAR;
if (C_PARODD(tty))
cor1 |= COR1_ODDP;
if (I_INPCK(tty))
cor1 &= ~COR1_IGNORE;
}
/* Set marking of some errors */
port->mark_mask = RCSR_OE | RCSR_TOUT;
if (I_INPCK(tty))
port->mark_mask |= RCSR_FE | RCSR_PE;
if (I_BRKINT(tty) || I_PARMRK(tty))
port->mark_mask |= RCSR_BREAK;
if (I_IGNPAR(tty))
port->mark_mask &= ~(RCSR_FE | RCSR_PE);
if (I_IGNBRK(tty)) {
port->mark_mask &= ~RCSR_BREAK;
if (I_IGNPAR(tty))
/* Real raw mode. Ignore all */
port->mark_mask &= ~RCSR_OE;
}
/* Enable Hardware Flow Control */
if (C_CRTSCTS(tty)) {
#ifdef RISCOM_BRAIN_DAMAGED_CTS
port->IER |= IER_DSR | IER_CTS;
mcor1 |= MCOR1_DSRZD | MCOR1_CTSZD;
mcor2 |= MCOR2_DSROD | MCOR2_CTSOD;
tty->hw_stopped = !(rc_in(bp, CD180_MSVR) & (MSVR_CTS|MSVR_DSR));
#else
port->COR2 |= COR2_CTSAE;
#endif
}
/* Enable Software Flow Control. FIXME: I'm not sure about this */
/* Some people reported that it works, but I still doubt */
if (I_IXON(tty)) {
port->COR2 |= COR2_TXIBE;
cor3 |= (COR3_FCT | COR3_SCDE);
if (I_IXANY(tty))
port->COR2 |= COR2_IXM;
rc_out(bp, CD180_SCHR1, START_CHAR(tty));
rc_out(bp, CD180_SCHR2, STOP_CHAR(tty));
rc_out(bp, CD180_SCHR3, START_CHAR(tty));
rc_out(bp, CD180_SCHR4, STOP_CHAR(tty));
}
if (!C_CLOCAL(tty)) {
/* Enable CD check */
port->IER |= IER_CD;
mcor1 |= MCOR1_CDZD;
mcor2 |= MCOR2_CDOD;
}
if (C_CREAD(tty))
/* Enable receiver */
port->IER |= IER_RXD;
/* Set input FIFO size (1-8 bytes) */
cor3 |= RISCOM_RXFIFO;
/* Setting up CD180 channel registers */
rc_out(bp, CD180_COR1, cor1);
rc_out(bp, CD180_COR2, port->COR2);
rc_out(bp, CD180_COR3, cor3);
/* Make CD180 know about registers change */
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_CORCHG1 | CCR_CORCHG2 | CCR_CORCHG3);
/* Setting up modem option registers */
rc_out(bp, CD180_MCOR1, mcor1);
rc_out(bp, CD180_MCOR2, mcor2);
/* Enable CD180 transmitter & receiver */
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_TXEN | CCR_RXEN);
/* Enable interrupts */
rc_out(bp, CD180_IER, port->IER);
/* And finally set RTS on */
rc_out(bp, CD180_MSVR, port->MSVR);
}
/* Must be called with interrupts enabled */
static int rc_setup_port(struct riscom_board *bp, struct riscom_port *port)
{
unsigned long flags;
if (port->flags & ASYNC_INITIALIZED)
return 0;
if (!port->xmit_buf) {
/* We may sleep in get_zeroed_page() */
unsigned long tmp;
if (!(tmp = get_zeroed_page(GFP_KERNEL)))
return -ENOMEM;
if (port->xmit_buf) {
free_page(tmp);
return -ERESTARTSYS;
}
port->xmit_buf = (unsigned char *) tmp;
}
save_flags(flags); cli();
if (port->tty)
clear_bit(TTY_IO_ERROR, &port->tty->flags);
if (port->count == 1)
bp->count++;
port->xmit_cnt = port->xmit_head = port->xmit_tail = 0;
rc_change_speed(bp, port);
port->flags |= ASYNC_INITIALIZED;
restore_flags(flags);
return 0;
}
/* Must be called with interrupts disabled */
static void rc_shutdown_port(struct riscom_board *bp, struct riscom_port *port)
{
struct tty_struct *tty;
if (!(port->flags & ASYNC_INITIALIZED))
return;
#ifdef RC_REPORT_OVERRUN
printk(KERN_INFO "rc%d: port %d: Total %ld overruns were detected.\n",
board_No(bp), port_No(port), port->overrun);
#endif
#ifdef RC_REPORT_FIFO
{
int i;
printk(KERN_INFO "rc%d: port %d: FIFO hits [ ",
board_No(bp), port_No(port));
for (i = 0; i < 10; i++) {
printk("%ld ", port->hits[i]);
}
printk("].\n");
}
#endif
if (port->xmit_buf) {
free_page((unsigned long) port->xmit_buf);
port->xmit_buf = NULL;
}
if (!(tty = port->tty) || C_HUPCL(tty)) {
/* Drop DTR */
bp->DTR |= (1u << port_No(port));
rc_out(bp, RC_DTR, bp->DTR);
}
/* Select port */
rc_out(bp, CD180_CAR, port_No(port));
/* Reset port */
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_SOFTRESET);
/* Disable all interrupts from this port */
port->IER = 0;
rc_out(bp, CD180_IER, port->IER);
if (tty)
set_bit(TTY_IO_ERROR, &tty->flags);
port->flags &= ~ASYNC_INITIALIZED;
if (--bp->count < 0) {
printk(KERN_INFO "rc%d: rc_shutdown_port: "
"bad board count: %d\n",
board_No(bp), bp->count);
bp->count = 0;
}
/*
* If this is the last opened port on the board
* shutdown whole board
*/
if (!bp->count)
rc_shutdown_board(bp);
}
static int block_til_ready(struct tty_struct *tty, struct file * filp,
struct riscom_port *port)
{
DECLARE_WAITQUEUE(wait, current);
struct riscom_board *bp = port_Board(port);
int retval;
int do_clocal = 0;
int CD;
/*
* If the device is in the middle of being closed, then block
* until it's done, and then try again.
*/
if (tty_hung_up_p(filp) || port->flags & ASYNC_CLOSING) {
interruptible_sleep_on(&port->close_wait);
if (port->flags & ASYNC_HUP_NOTIFY)
return -EAGAIN;
else
return -ERESTARTSYS;
}
/*
* If non-blocking mode is set, or the port is not enabled,
* then make the check up front and then exit.
*/
if ((filp->f_flags & O_NONBLOCK) ||
(tty->flags & (1 << TTY_IO_ERROR))) {
port->flags |= ASYNC_NORMAL_ACTIVE;
return 0;
}
if (C_CLOCAL(tty))
do_clocal = 1;
/*
* Block waiting for the carrier detect and the line to become
* free (i.e., not in use by the callout). While we are in
* this loop, info->count is dropped by one, so that
* rs_close() knows when to free things. We restore it upon
* exit, either normal or abnormal.
*/
retval = 0;
add_wait_queue(&port->open_wait, &wait);
cli();
if (!tty_hung_up_p(filp))
port->count--;
sti();
port->blocked_open++;
while (1) {
cli();
rc_out(bp, CD180_CAR, port_No(port));
CD = rc_in(bp, CD180_MSVR) & MSVR_CD;
rc_out(bp, CD180_MSVR, MSVR_RTS);
bp->DTR &= ~(1u << port_No(port));
rc_out(bp, RC_DTR, bp->DTR);
sti();
set_current_state(TASK_INTERRUPTIBLE);
if (tty_hung_up_p(filp) ||
!(port->flags & ASYNC_INITIALIZED)) {
if (port->flags & ASYNC_HUP_NOTIFY)
retval = -EAGAIN;
else
retval = -ERESTARTSYS;
break;
}
if (!(port->flags & ASYNC_CLOSING) &&
(do_clocal || CD))
break;
if (signal_pending(current)) {
retval = -ERESTARTSYS;
break;
}
schedule();
}
__set_current_state(TASK_RUNNING);
remove_wait_queue(&port->open_wait, &wait);
if (!tty_hung_up_p(filp))
port->count++;
port->blocked_open--;
if (retval)
return retval;
port->flags |= ASYNC_NORMAL_ACTIVE;
return 0;
}
static int rc_open(struct tty_struct * tty, struct file * filp)
{
int board;
int error;
struct riscom_port * port;
struct riscom_board * bp;
board = RC_BOARD(tty->index);
if (board >= RC_NBOARD || !(rc_board[board].flags & RC_BOARD_PRESENT))
return -ENODEV;
bp = &rc_board[board];
port = rc_port + board * RC_NPORT + RC_PORT(tty->index);
if (rc_paranoia_check(port, tty->name, "rc_open"))
return -ENODEV;
if ((error = rc_setup_board(bp)))
return error;
port->count++;
tty->driver_data = port;
port->tty = tty;
if ((error = rc_setup_port(bp, port)))
return error;
if ((error = block_til_ready(tty, filp, port)))
return error;
return 0;
}
static void rc_close(struct tty_struct * tty, struct file * filp)
{
struct riscom_port *port = (struct riscom_port *) tty->driver_data;
struct riscom_board *bp;
unsigned long flags;
unsigned long timeout;
if (!port || rc_paranoia_check(port, tty->name, "close"))
return;
save_flags(flags); cli();
if (tty_hung_up_p(filp))
goto out;
bp = port_Board(port);
if ((tty->count == 1) && (port->count != 1)) {
printk(KERN_INFO "rc%d: rc_close: bad port count;"
" tty->count is 1, port count is %d\n",
board_No(bp), port->count);
port->count = 1;
}
if (--port->count < 0) {
printk(KERN_INFO "rc%d: rc_close: bad port count "
"for tty%d: %d\n",
board_No(bp), port_No(port), port->count);
port->count = 0;
}
if (port->count)
goto out;
port->flags |= ASYNC_CLOSING;
/*
* Now we wait for the transmit buffer to clear; and we notify
* the line discipline to only process XON/XOFF characters.
*/
tty->closing = 1;
if (port->closing_wait != ASYNC_CLOSING_WAIT_NONE)
tty_wait_until_sent(tty, port->closing_wait);
/*
* At this point we stop accepting input. To do this, we
* disable the receive line status interrupts, and tell the
* interrupt driver to stop checking the data ready bit in the
* line status register.
*/
port->IER &= ~IER_RXD;
if (port->flags & ASYNC_INITIALIZED) {
port->IER &= ~IER_TXRDY;
port->IER |= IER_TXEMPTY;
rc_out(bp, CD180_CAR, port_No(port));
rc_out(bp, CD180_IER, port->IER);
/*
* Before we drop DTR, make sure the UART transmitter
* has completely drained; this is especially
* important if there is a transmit FIFO!
*/
timeout = jiffies+HZ;
while(port->IER & IER_TXEMPTY) {
msleep_interruptible(jiffies_to_msecs(port->timeout));
if (time_after(jiffies, timeout))
break;
}
}
rc_shutdown_port(bp, port);
if (tty->driver->flush_buffer)
tty->driver->flush_buffer(tty);
tty_ldisc_flush(tty);
tty->closing = 0;
port->event = 0;
port->tty = NULL;
if (port->blocked_open) {
if (port->close_delay) {
msleep_interruptible(jiffies_to_msecs(port->close_delay));
}
wake_up_interruptible(&port->open_wait);
}
port->flags &= ~(ASYNC_NORMAL_ACTIVE|ASYNC_CLOSING);
wake_up_interruptible(&port->close_wait);
out: restore_flags(flags);
}
static int rc_write(struct tty_struct * tty,
const unsigned char *buf, int count)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board *bp;
int c, total = 0;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_write"))
return 0;
bp = port_Board(port);
if (!tty || !port->xmit_buf)
return 0;
save_flags(flags);
while (1) {
cli();
c = min_t(int, count, min(SERIAL_XMIT_SIZE - port->xmit_cnt - 1,
SERIAL_XMIT_SIZE - port->xmit_head));
if (c <= 0) {
restore_flags(flags);
break;
}
memcpy(port->xmit_buf + port->xmit_head, buf, c);
port->xmit_head = (port->xmit_head + c) & (SERIAL_XMIT_SIZE-1);
port->xmit_cnt += c;
restore_flags(flags);
buf += c;
count -= c;
total += c;
}
cli();
if (port->xmit_cnt && !tty->stopped && !tty->hw_stopped &&
!(port->IER & IER_TXRDY)) {
port->IER |= IER_TXRDY;
rc_out(bp, CD180_CAR, port_No(port));
rc_out(bp, CD180_IER, port->IER);
}
restore_flags(flags);
return total;
}
static void rc_put_char(struct tty_struct * tty, unsigned char ch)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_put_char"))
return;
if (!tty || !port->xmit_buf)
return;
save_flags(flags); cli();
if (port->xmit_cnt >= SERIAL_XMIT_SIZE - 1)
goto out;
port->xmit_buf[port->xmit_head++] = ch;
port->xmit_head &= SERIAL_XMIT_SIZE - 1;
port->xmit_cnt++;
out: restore_flags(flags);
}
static void rc_flush_chars(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_flush_chars"))
return;
if (port->xmit_cnt <= 0 || tty->stopped || tty->hw_stopped ||
!port->xmit_buf)
return;
save_flags(flags); cli();
port->IER |= IER_TXRDY;
rc_out(port_Board(port), CD180_CAR, port_No(port));
rc_out(port_Board(port), CD180_IER, port->IER);
restore_flags(flags);
}
static int rc_write_room(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
int ret;
if (rc_paranoia_check(port, tty->name, "rc_write_room"))
return 0;
ret = SERIAL_XMIT_SIZE - port->xmit_cnt - 1;
if (ret < 0)
ret = 0;
return ret;
}
static int rc_chars_in_buffer(struct tty_struct *tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
if (rc_paranoia_check(port, tty->name, "rc_chars_in_buffer"))
return 0;
return port->xmit_cnt;
}
static void rc_flush_buffer(struct tty_struct *tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_flush_buffer"))
return;
save_flags(flags); cli();
port->xmit_cnt = port->xmit_head = port->xmit_tail = 0;
restore_flags(flags);
tty_wakeup(tty);
}
static int rc_tiocmget(struct tty_struct *tty, struct file *file)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board * bp;
unsigned char status;
unsigned int result;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, __FUNCTION__))
return -ENODEV;
bp = port_Board(port);
save_flags(flags); cli();
rc_out(bp, CD180_CAR, port_No(port));
status = rc_in(bp, CD180_MSVR);
result = rc_in(bp, RC_RI) & (1u << port_No(port)) ? 0 : TIOCM_RNG;
restore_flags(flags);
result |= ((status & MSVR_RTS) ? TIOCM_RTS : 0)
| ((status & MSVR_DTR) ? TIOCM_DTR : 0)
| ((status & MSVR_CD) ? TIOCM_CAR : 0)
| ((status & MSVR_DSR) ? TIOCM_DSR : 0)
| ((status & MSVR_CTS) ? TIOCM_CTS : 0);
return result;
}
static int rc_tiocmset(struct tty_struct *tty, struct file *file,
unsigned int set, unsigned int clear)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
unsigned long flags;
struct riscom_board *bp;
if (rc_paranoia_check(port, tty->name, __FUNCTION__))
return -ENODEV;
bp = port_Board(port);
save_flags(flags); cli();
if (set & TIOCM_RTS)
port->MSVR |= MSVR_RTS;
if (set & TIOCM_DTR)
bp->DTR &= ~(1u << port_No(port));
if (clear & TIOCM_RTS)
port->MSVR &= ~MSVR_RTS;
if (clear & TIOCM_DTR)
bp->DTR |= (1u << port_No(port));
rc_out(bp, CD180_CAR, port_No(port));
rc_out(bp, CD180_MSVR, port->MSVR);
rc_out(bp, RC_DTR, bp->DTR);
restore_flags(flags);
return 0;
}
static inline void rc_send_break(struct riscom_port * port, unsigned long length)
{
struct riscom_board *bp = port_Board(port);
unsigned long flags;
save_flags(flags); cli();
port->break_length = RISCOM_TPS / HZ * length;
port->COR2 |= COR2_ETC;
port->IER |= IER_TXRDY;
rc_out(bp, CD180_CAR, port_No(port));
rc_out(bp, CD180_COR2, port->COR2);
rc_out(bp, CD180_IER, port->IER);
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_CORCHG2);
rc_wait_CCR(bp);
restore_flags(flags);
}
static inline int rc_set_serial_info(struct riscom_port * port,
struct serial_struct __user * newinfo)
{
struct serial_struct tmp;
struct riscom_board *bp = port_Board(port);
int change_speed;
unsigned long flags;
if (copy_from_user(&tmp, newinfo, sizeof(tmp)))
return -EFAULT;
#if 0
if ((tmp.irq != bp->irq) ||
(tmp.port != bp->base) ||
(tmp.type != PORT_CIRRUS) ||
(tmp.baud_base != (RC_OSCFREQ + CD180_TPC/2) / CD180_TPC) ||
(tmp.custom_divisor != 0) ||
(tmp.xmit_fifo_size != CD180_NFIFO) ||
(tmp.flags & ~RISCOM_LEGAL_FLAGS))
return -EINVAL;
#endif
change_speed = ((port->flags & ASYNC_SPD_MASK) !=
(tmp.flags & ASYNC_SPD_MASK));
if (!capable(CAP_SYS_ADMIN)) {
if ((tmp.close_delay != port->close_delay) ||
(tmp.closing_wait != port->closing_wait) ||
((tmp.flags & ~ASYNC_USR_MASK) !=
(port->flags & ~ASYNC_USR_MASK)))
return -EPERM;
port->flags = ((port->flags & ~ASYNC_USR_MASK) |
(tmp.flags & ASYNC_USR_MASK));
} else {
port->flags = ((port->flags & ~ASYNC_FLAGS) |
(tmp.flags & ASYNC_FLAGS));
port->close_delay = tmp.close_delay;
port->closing_wait = tmp.closing_wait;
}
if (change_speed) {
save_flags(flags); cli();
rc_change_speed(bp, port);
restore_flags(flags);
}
return 0;
}
static inline int rc_get_serial_info(struct riscom_port * port,
struct serial_struct __user *retinfo)
{
struct serial_struct tmp;
struct riscom_board *bp = port_Board(port);
memset(&tmp, 0, sizeof(tmp));
tmp.type = PORT_CIRRUS;
tmp.line = port - rc_port;
tmp.port = bp->base;
tmp.irq = bp->irq;
tmp.flags = port->flags;
tmp.baud_base = (RC_OSCFREQ + CD180_TPC/2) / CD180_TPC;
tmp.close_delay = port->close_delay * HZ/100;
tmp.closing_wait = port->closing_wait * HZ/100;
tmp.xmit_fifo_size = CD180_NFIFO;
return copy_to_user(retinfo, &tmp, sizeof(tmp)) ? -EFAULT : 0;
}
static int rc_ioctl(struct tty_struct * tty, struct file * filp,
unsigned int cmd, unsigned long arg)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
void __user *argp = (void __user *)arg;
int retval;
if (rc_paranoia_check(port, tty->name, "rc_ioctl"))
return -ENODEV;
switch (cmd) {
case TCSBRK: /* SVID version: non-zero arg --> no break */
retval = tty_check_change(tty);
if (retval)
return retval;
tty_wait_until_sent(tty, 0);
if (!arg)
rc_send_break(port, HZ/4); /* 1/4 second */
break;
case TCSBRKP: /* support for POSIX tcsendbreak() */
retval = tty_check_change(tty);
if (retval)
return retval;
tty_wait_until_sent(tty, 0);
rc_send_break(port, arg ? arg*(HZ/10) : HZ/4);
break;
case TIOCGSOFTCAR:
return put_user(C_CLOCAL(tty) ? 1 : 0, (unsigned __user *)argp);
case TIOCSSOFTCAR:
if (get_user(arg,(unsigned __user *) argp))
return -EFAULT;
tty->termios->c_cflag =
((tty->termios->c_cflag & ~CLOCAL) |
(arg ? CLOCAL : 0));
break;
case TIOCGSERIAL:
return rc_get_serial_info(port, argp);
case TIOCSSERIAL:
return rc_set_serial_info(port, argp);
default:
return -ENOIOCTLCMD;
}
return 0;
}
static void rc_throttle(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board *bp;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_throttle"))
return;
bp = port_Board(port);
save_flags(flags); cli();
port->MSVR &= ~MSVR_RTS;
rc_out(bp, CD180_CAR, port_No(port));
if (I_IXOFF(tty)) {
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_SSCH2);
rc_wait_CCR(bp);
}
rc_out(bp, CD180_MSVR, port->MSVR);
restore_flags(flags);
}
static void rc_unthrottle(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board *bp;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_unthrottle"))
return;
bp = port_Board(port);
save_flags(flags); cli();
port->MSVR |= MSVR_RTS;
rc_out(bp, CD180_CAR, port_No(port));
if (I_IXOFF(tty)) {
rc_wait_CCR(bp);
rc_out(bp, CD180_CCR, CCR_SSCH1);
rc_wait_CCR(bp);
}
rc_out(bp, CD180_MSVR, port->MSVR);
restore_flags(flags);
}
static void rc_stop(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board *bp;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_stop"))
return;
bp = port_Board(port);
save_flags(flags); cli();
port->IER &= ~IER_TXRDY;
rc_out(bp, CD180_CAR, port_No(port));
rc_out(bp, CD180_IER, port->IER);
restore_flags(flags);
}
static void rc_start(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board *bp;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_start"))
return;
bp = port_Board(port);
save_flags(flags); cli();
if (port->xmit_cnt && port->xmit_buf && !(port->IER & IER_TXRDY)) {
port->IER |= IER_TXRDY;
rc_out(bp, CD180_CAR, port_No(port));
rc_out(bp, CD180_IER, port->IER);
}
restore_flags(flags);
}
/*
* This routine is called from the work queue when the interrupt
* routine has signalled that a hangup has occurred. The path of
* hangup processing is:
*
* serial interrupt routine -> (workqueue) ->
* do_rc_hangup() -> tty->hangup() -> rc_hangup()
*
*/
static void do_rc_hangup(struct work_struct *ugly_api)
{
struct riscom_port *port = container_of(ugly_api, struct riscom_port, tqueue_hangup);
struct tty_struct *tty;
tty = port->tty;
if (tty)
tty_hangup(tty); /* FIXME: module removal race still here */
}
static void rc_hangup(struct tty_struct * tty)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
struct riscom_board *bp;
if (rc_paranoia_check(port, tty->name, "rc_hangup"))
return;
bp = port_Board(port);
rc_shutdown_port(bp, port);
port->event = 0;
port->count = 0;
port->flags &= ~ASYNC_NORMAL_ACTIVE;
port->tty = NULL;
wake_up_interruptible(&port->open_wait);
}
static void rc_set_termios(struct tty_struct * tty, struct ktermios * old_termios)
{
struct riscom_port *port = (struct riscom_port *)tty->driver_data;
unsigned long flags;
if (rc_paranoia_check(port, tty->name, "rc_set_termios"))
return;
if (tty->termios->c_cflag == old_termios->c_cflag &&
tty->termios->c_iflag == old_termios->c_iflag)
return;
save_flags(flags); cli();
rc_change_speed(port_Board(port), port);
restore_flags(flags);
if ((old_termios->c_cflag & CRTSCTS) &&
!(tty->termios->c_cflag & CRTSCTS)) {
tty->hw_stopped = 0;
rc_start(tty);
}
}
static void do_softint(struct work_struct *ugly_api)
{
struct riscom_port *port = container_of(ugly_api, struct riscom_port, tqueue);
struct tty_struct *tty;
if(!(tty = port->tty))
return;
if (test_and_clear_bit(RS_EVENT_WRITE_WAKEUP, &port->event))
tty_wakeup(tty);
}
static const struct tty_operations riscom_ops = {
.open = rc_open,
.close = rc_close,
.write = rc_write,
.put_char = rc_put_char,
.flush_chars = rc_flush_chars,
.write_room = rc_write_room,
.chars_in_buffer = rc_chars_in_buffer,
.flush_buffer = rc_flush_buffer,
.ioctl = rc_ioctl,
.throttle = rc_throttle,
.unthrottle = rc_unthrottle,
.set_termios = rc_set_termios,
.stop = rc_stop,
.start = rc_start,
.hangup = rc_hangup,
.tiocmget = rc_tiocmget,
.tiocmset = rc_tiocmset,
};
static inline int rc_init_drivers(void)
{
int error;
int i;
riscom_driver = alloc_tty_driver(RC_NBOARD * RC_NPORT);
if (!riscom_driver)
return -ENOMEM;
memset(IRQ_to_board, 0, sizeof(IRQ_to_board));
riscom_driver->owner = THIS_MODULE;
riscom_driver->name = "ttyL";
riscom_driver->major = RISCOM8_NORMAL_MAJOR;
riscom_driver->type = TTY_DRIVER_TYPE_SERIAL;
riscom_driver->subtype = SERIAL_TYPE_NORMAL;
riscom_driver->init_termios = tty_std_termios;
riscom_driver->init_termios.c_cflag =
B9600 | CS8 | CREAD | HUPCL | CLOCAL;
riscom_driver->init_termios.c_ispeed = 9600;
riscom_driver->init_termios.c_ospeed = 9600;
riscom_driver->flags = TTY_DRIVER_REAL_RAW;
tty_set_operations(riscom_driver, &riscom_ops);
if ((error = tty_register_driver(riscom_driver))) {
put_tty_driver(riscom_driver);
printk(KERN_ERR "rc: Couldn't register RISCom/8 driver, "
"error = %d\n",
error);
return 1;
}
memset(rc_port, 0, sizeof(rc_port));
for (i = 0; i < RC_NPORT * RC_NBOARD; i++) {
rc_port[i].magic = RISCOM8_MAGIC;
INIT_WORK(&rc_port[i].tqueue, do_softint);
INIT_WORK(&rc_port[i].tqueue_hangup, do_rc_hangup);
rc_port[i].close_delay = 50 * HZ/100;
rc_port[i].closing_wait = 3000 * HZ/100;
init_waitqueue_head(&rc_port[i].open_wait);
init_waitqueue_head(&rc_port[i].close_wait);
}
return 0;
}
static void rc_release_drivers(void)
{
unsigned long flags;
save_flags(flags);
cli();
tty_unregister_driver(riscom_driver);
put_tty_driver(riscom_driver);
restore_flags(flags);
}
#ifndef MODULE
/*
* Called at boot time.
*
* You can specify IO base for up to RC_NBOARD cards,
* using line "riscom8=0xiobase1,0xiobase2,.." at LILO prompt.
* Note that there will be no probing at default
* addresses in this case.
*
*/
static int __init riscom8_setup(char *str)
{
int ints[RC_NBOARD];
int i;
str = get_options(str, ARRAY_SIZE(ints), ints);
for (i = 0; i < RC_NBOARD; i++) {
if (i < ints[0])
rc_board[i].base = ints[i+1];
else
rc_board[i].base = 0;
}
return 1;
}
__setup("riscom8=", riscom8_setup);
#endif
static char banner[] __initdata =
KERN_INFO "rc: SDL RISCom/8 card driver v1.1, (c) D.Gorodchanin "
"1994-1996.\n";
static char no_boards_msg[] __initdata =
KERN_INFO "rc: No RISCom/8 boards detected.\n";
/*
* This routine must be called by kernel at boot time
*/
static int __init riscom8_init(void)
{
int i;
int found = 0;
printk(banner);
if (rc_init_drivers())
return -EIO;
for (i = 0; i < RC_NBOARD; i++)
if (rc_board[i].base && !rc_probe(&rc_board[i]))
found++;
if (!found) {
rc_release_drivers();
printk(no_boards_msg);
return -EIO;
}
return 0;
}
#ifdef MODULE
static int iobase;
static int iobase1;
static int iobase2;
static int iobase3;
module_param(iobase, int, 0);
module_param(iobase1, int, 0);
module_param(iobase2, int, 0);
module_param(iobase3, int, 0);
MODULE_LICENSE("GPL");
#endif /* MODULE */
/*
* You can setup up to 4 boards (current value of RC_NBOARD)
* by specifying "iobase=0xXXX iobase1=0xXXX ..." as insmod parameter.
*
*/
static int __init riscom8_init_module (void)
{
#ifdef MODULE
int i;
if (iobase || iobase1 || iobase2 || iobase3) {
for(i = 0; i < RC_NBOARD; i++)
rc_board[0].base = 0;
}
if (iobase)
rc_board[0].base = iobase;
if (iobase1)
rc_board[1].base = iobase1;
if (iobase2)
rc_board[2].base = iobase2;
if (iobase3)
rc_board[3].base = iobase3;
#endif /* MODULE */
return riscom8_init();
}
static void __exit riscom8_exit_module (void)
{
int i;
rc_release_drivers();
for (i = 0; i < RC_NBOARD; i++)
if (rc_board[i].flags & RC_BOARD_PRESENT)
rc_release_io_range(&rc_board[i]);
}
module_init(riscom8_init_module);
module_exit(riscom8_exit_module);