linux/drivers/serial/sunsab.c

1153 lines
29 KiB
C
Raw Normal View History

/* sunsab.c: ASYNC Driver for the SIEMENS SAB82532 DUSCC.
*
* Copyright (C) 1997 Eddie C. Dost (ecd@skynet.be)
* Copyright (C) 2002, 2006 David S. Miller (davem@davemloft.net)
*
* Rewrote buffer handling to use CIRC(Circular Buffer) macros.
* Maxim Krasnyanskiy <maxk@qualcomm.com>
*
* Fixed to use tty_get_baud_rate, and to allow for arbitrary baud
* rates to be programmed into the UART. Also eliminated a lot of
* duplicated code in the console setup.
* Theodore Ts'o <tytso@mit.edu>, 2001-Oct-12
*
* Ported to new 2.5.x UART layer.
* David S. Miller <davem@davemloft.net>
*/
#include <linux/module.h>
#include <linux/kernel.h>
#include <linux/errno.h>
#include <linux/tty.h>
#include <linux/tty_flip.h>
#include <linux/major.h>
#include <linux/string.h>
#include <linux/ptrace.h>
#include <linux/ioport.h>
#include <linux/circ_buf.h>
#include <linux/serial.h>
#include <linux/sysrq.h>
#include <linux/console.h>
#include <linux/spinlock.h>
#include <linux/slab.h>
#include <linux/delay.h>
#include <linux/init.h>
#include <linux/of_device.h>
#include <asm/io.h>
#include <asm/irq.h>
#include <asm/prom.h>
#if defined(CONFIG_SERIAL_SUNSAB_CONSOLE) && defined(CONFIG_MAGIC_SYSRQ)
#define SUPPORT_SYSRQ
#endif
#include <linux/serial_core.h>
#include "suncore.h"
#include "sunsab.h"
struct uart_sunsab_port {
struct uart_port port; /* Generic UART port */
union sab82532_async_regs __iomem *regs; /* Chip registers */
unsigned long irqflags; /* IRQ state flags */
int dsr; /* Current DSR state */
unsigned int cec_timeout; /* Chip poll timeout... */
unsigned int tec_timeout; /* likewise */
unsigned char interrupt_mask0;/* ISR0 masking */
unsigned char interrupt_mask1;/* ISR1 masking */
unsigned char pvr_dtr_bit; /* Which PVR bit is DTR */
unsigned char pvr_dsr_bit; /* Which PVR bit is DSR */
unsigned int gis_shift;
int type; /* SAB82532 version */
/* Setting configuration bits while the transmitter is active
* can cause garbage characters to get emitted by the chip.
* Therefore, we cache such writes here and do the real register
* write the next time the transmitter becomes idle.
*/
unsigned int cached_ebrg;
unsigned char cached_mode;
unsigned char cached_pvr;
unsigned char cached_dafo;
};
/*
* This assumes you have a 29.4912 MHz clock for your UART.
*/
#define SAB_BASE_BAUD ( 29491200 / 16 )
static char *sab82532_version[16] = {
"V1.0", "V2.0", "V3.2", "V(0x03)",
"V(0x04)", "V(0x05)", "V(0x06)", "V(0x07)",
"V(0x08)", "V(0x09)", "V(0x0a)", "V(0x0b)",
"V(0x0c)", "V(0x0d)", "V(0x0e)", "V(0x0f)"
};
#define SAB82532_MAX_TEC_TIMEOUT 200000 /* 1 character time (at 50 baud) */
#define SAB82532_MAX_CEC_TIMEOUT 50000 /* 2.5 TX CLKs (at 50 baud) */
#define SAB82532_RECV_FIFO_SIZE 32 /* Standard async fifo sizes */
#define SAB82532_XMIT_FIFO_SIZE 32
static __inline__ void sunsab_tec_wait(struct uart_sunsab_port *up)
{
int timeout = up->tec_timeout;
while ((readb(&up->regs->r.star) & SAB82532_STAR_TEC) && --timeout)
udelay(1);
}
static __inline__ void sunsab_cec_wait(struct uart_sunsab_port *up)
{
int timeout = up->cec_timeout;
while ((readb(&up->regs->r.star) & SAB82532_STAR_CEC) && --timeout)
udelay(1);
}
static struct tty_struct *
receive_chars(struct uart_sunsab_port *up,
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 13:55:46 +00:00
union sab82532_irq_status *stat)
{
struct tty_struct *tty = NULL;
unsigned char buf[32];
int saw_console_brk = 0;
int free_fifo = 0;
int count = 0;
int i;
if (up->port.state != NULL) /* Unopened serial console */
tty = up->port.state->port.tty;
/* Read number of BYTES (Character + Status) available. */
if (stat->sreg.isr0 & SAB82532_ISR0_RPF) {
count = SAB82532_RECV_FIFO_SIZE;
free_fifo++;
}
if (stat->sreg.isr0 & SAB82532_ISR0_TCD) {
count = readb(&up->regs->r.rbcl) & (SAB82532_RECV_FIFO_SIZE - 1);
free_fifo++;
}
/* Issue a FIFO read command in case we where idle. */
if (stat->sreg.isr0 & SAB82532_ISR0_TIME) {
sunsab_cec_wait(up);
writeb(SAB82532_CMDR_RFRD, &up->regs->w.cmdr);
return tty;
}
if (stat->sreg.isr0 & SAB82532_ISR0_RFO)
free_fifo++;
/* Read the FIFO. */
for (i = 0; i < count; i++)
buf[i] = readb(&up->regs->r.rfifo[i]);
/* Issue Receive Message Complete command. */
if (free_fifo) {
sunsab_cec_wait(up);
writeb(SAB82532_CMDR_RMC, &up->regs->w.cmdr);
}
/* Count may be zero for BRK, so we check for it here */
if ((stat->sreg.isr1 & SAB82532_ISR1_BRK) &&
(up->port.line == up->port.cons->index))
saw_console_brk = 1;
for (i = 0; i < count; i++) {
[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 04:54:13 +00:00
unsigned char ch = buf[i], flag;
if (tty == NULL) {
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 13:55:46 +00:00
uart_handle_sysrq_char(&up->port, ch);
continue;
}
[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 04:54:13 +00:00
flag = TTY_NORMAL;
up->port.icount.rx++;
if (unlikely(stat->sreg.isr0 & (SAB82532_ISR0_PERR |
SAB82532_ISR0_FERR |
SAB82532_ISR0_RFO)) ||
unlikely(stat->sreg.isr1 & SAB82532_ISR1_BRK)) {
/*
* For statistics only
*/
if (stat->sreg.isr1 & SAB82532_ISR1_BRK) {
stat->sreg.isr0 &= ~(SAB82532_ISR0_PERR |
SAB82532_ISR0_FERR);
up->port.icount.brk++;
/*
* We do the SysRQ and SAK checking
* here because otherwise the break
* may get masked by ignore_status_mask
* or read_status_mask.
*/
if (uart_handle_break(&up->port))
continue;
} else if (stat->sreg.isr0 & SAB82532_ISR0_PERR)
up->port.icount.parity++;
else if (stat->sreg.isr0 & SAB82532_ISR0_FERR)
up->port.icount.frame++;
if (stat->sreg.isr0 & SAB82532_ISR0_RFO)
up->port.icount.overrun++;
/*
* Mask off conditions which should be ingored.
*/
stat->sreg.isr0 &= (up->port.read_status_mask & 0xff);
stat->sreg.isr1 &= ((up->port.read_status_mask >> 8) & 0xff);
if (stat->sreg.isr1 & SAB82532_ISR1_BRK) {
[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 04:54:13 +00:00
flag = TTY_BREAK;
} else if (stat->sreg.isr0 & SAB82532_ISR0_PERR)
[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 04:54:13 +00:00
flag = TTY_PARITY;
else if (stat->sreg.isr0 & SAB82532_ISR0_FERR)
[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 04:54:13 +00:00
flag = TTY_FRAME;
}
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 13:55:46 +00:00
if (uart_handle_sysrq_char(&up->port, ch))
continue;
if ((stat->sreg.isr0 & (up->port.ignore_status_mask & 0xff)) == 0 &&
[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 04:54:13 +00:00
(stat->sreg.isr1 & ((up->port.ignore_status_mask >> 8) & 0xff)) == 0)
tty_insert_flip_char(tty, ch, flag);
if (stat->sreg.isr0 & SAB82532_ISR0_RFO)
tty_insert_flip_char(tty, 0, TTY_OVERRUN);
}
if (saw_console_brk)
sun_do_break();
return tty;
}
static void sunsab_stop_tx(struct uart_port *);
static void sunsab_tx_idle(struct uart_sunsab_port *);
static void transmit_chars(struct uart_sunsab_port *up,
union sab82532_irq_status *stat)
{
struct circ_buf *xmit = &up->port.state->xmit;
int i;
if (stat->sreg.isr1 & SAB82532_ISR1_ALLS) {
up->interrupt_mask1 |= SAB82532_IMR1_ALLS;
writeb(up->interrupt_mask1, &up->regs->w.imr1);
set_bit(SAB82532_ALLS, &up->irqflags);
}
#if 0 /* bde@nwlink.com says this check causes problems */
if (!(stat->sreg.isr1 & SAB82532_ISR1_XPR))
return;
#endif
if (!(readb(&up->regs->r.star) & SAB82532_STAR_XFW))
return;
set_bit(SAB82532_XPR, &up->irqflags);
sunsab_tx_idle(up);
if (uart_circ_empty(xmit) || uart_tx_stopped(&up->port)) {
up->interrupt_mask1 |= SAB82532_IMR1_XPR;
writeb(up->interrupt_mask1, &up->regs->w.imr1);
return;
}
up->interrupt_mask1 &= ~(SAB82532_IMR1_ALLS|SAB82532_IMR1_XPR);
writeb(up->interrupt_mask1, &up->regs->w.imr1);
clear_bit(SAB82532_ALLS, &up->irqflags);
/* Stuff 32 bytes into Transmit FIFO. */
clear_bit(SAB82532_XPR, &up->irqflags);
for (i = 0; i < up->port.fifosize; i++) {
writeb(xmit->buf[xmit->tail],
&up->regs->w.xfifo[i]);
xmit->tail = (xmit->tail + 1) & (UART_XMIT_SIZE - 1);
up->port.icount.tx++;
if (uart_circ_empty(xmit))
break;
}
/* Issue a Transmit Frame command. */
sunsab_cec_wait(up);
writeb(SAB82532_CMDR_XF, &up->regs->w.cmdr);
if (uart_circ_chars_pending(xmit) < WAKEUP_CHARS)
uart_write_wakeup(&up->port);
if (uart_circ_empty(xmit))
sunsab_stop_tx(&up->port);
}
static void check_status(struct uart_sunsab_port *up,
union sab82532_irq_status *stat)
{
if (stat->sreg.isr0 & SAB82532_ISR0_CDSC)
uart_handle_dcd_change(&up->port,
!(readb(&up->regs->r.vstr) & SAB82532_VSTR_CD));
if (stat->sreg.isr1 & SAB82532_ISR1_CSC)
uart_handle_cts_change(&up->port,
(readb(&up->regs->r.star) & SAB82532_STAR_CTS));
if ((readb(&up->regs->r.pvr) & up->pvr_dsr_bit) ^ up->dsr) {
up->dsr = (readb(&up->regs->r.pvr) & up->pvr_dsr_bit) ? 0 : 1;
up->port.icount.dsr++;
}
wake_up_interruptible(&up->port.state->port.delta_msr_wait);
}
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 13:55:46 +00:00
static irqreturn_t sunsab_interrupt(int irq, void *dev_id)
{
struct uart_sunsab_port *up = dev_id;
struct tty_struct *tty;
union sab82532_irq_status status;
unsigned long flags;
unsigned char gis;
spin_lock_irqsave(&up->port.lock, flags);
status.stat = 0;
gis = readb(&up->regs->r.gis) >> up->gis_shift;
if (gis & 1)
status.sreg.isr0 = readb(&up->regs->r.isr0);
if (gis & 2)
status.sreg.isr1 = readb(&up->regs->r.isr1);
tty = NULL;
if (status.stat) {
if ((status.sreg.isr0 & (SAB82532_ISR0_TCD | SAB82532_ISR0_TIME |
SAB82532_ISR0_RFO | SAB82532_ISR0_RPF)) ||
(status.sreg.isr1 & SAB82532_ISR1_BRK))
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 13:55:46 +00:00
tty = receive_chars(up, &status);
if ((status.sreg.isr0 & SAB82532_ISR0_CDSC) ||
(status.sreg.isr1 & SAB82532_ISR1_CSC))
check_status(up, &status);
if (status.sreg.isr1 & (SAB82532_ISR1_ALLS | SAB82532_ISR1_XPR))
transmit_chars(up, &status);
}
spin_unlock_irqrestore(&up->port.lock, flags);
if (tty)
tty_flip_buffer_push(tty);
return IRQ_HANDLED;
}
/* port->lock is not held. */
static unsigned int sunsab_tx_empty(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
int ret;
/* Do not need a lock for a state test like this. */
if (test_bit(SAB82532_ALLS, &up->irqflags))
ret = TIOCSER_TEMT;
else
ret = 0;
return ret;
}
/* port->lock held by caller. */
static void sunsab_set_mctrl(struct uart_port *port, unsigned int mctrl)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
if (mctrl & TIOCM_RTS) {
up->cached_mode &= ~SAB82532_MODE_FRTS;
up->cached_mode |= SAB82532_MODE_RTS;
} else {
up->cached_mode |= (SAB82532_MODE_FRTS |
SAB82532_MODE_RTS);
}
if (mctrl & TIOCM_DTR) {
up->cached_pvr &= ~(up->pvr_dtr_bit);
} else {
up->cached_pvr |= up->pvr_dtr_bit;
}
set_bit(SAB82532_REGS_PENDING, &up->irqflags);
if (test_bit(SAB82532_XPR, &up->irqflags))
sunsab_tx_idle(up);
}
/* port->lock is held by caller and interrupts are disabled. */
static unsigned int sunsab_get_mctrl(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
unsigned char val;
unsigned int result;
result = 0;
val = readb(&up->regs->r.pvr);
result |= (val & up->pvr_dsr_bit) ? 0 : TIOCM_DSR;
val = readb(&up->regs->r.vstr);
result |= (val & SAB82532_VSTR_CD) ? 0 : TIOCM_CAR;
val = readb(&up->regs->r.star);
result |= (val & SAB82532_STAR_CTS) ? TIOCM_CTS : 0;
return result;
}
/* port->lock held by caller. */
static void sunsab_stop_tx(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
up->interrupt_mask1 |= SAB82532_IMR1_XPR;
writeb(up->interrupt_mask1, &up->regs->w.imr1);
}
/* port->lock held by caller. */
static void sunsab_tx_idle(struct uart_sunsab_port *up)
{
if (test_bit(SAB82532_REGS_PENDING, &up->irqflags)) {
u8 tmp;
clear_bit(SAB82532_REGS_PENDING, &up->irqflags);
writeb(up->cached_mode, &up->regs->rw.mode);
writeb(up->cached_pvr, &up->regs->rw.pvr);
writeb(up->cached_dafo, &up->regs->w.dafo);
writeb(up->cached_ebrg & 0xff, &up->regs->w.bgr);
tmp = readb(&up->regs->rw.ccr2);
tmp &= ~0xc0;
tmp |= (up->cached_ebrg >> 2) & 0xc0;
writeb(tmp, &up->regs->rw.ccr2);
}
}
/* port->lock held by caller. */
static void sunsab_start_tx(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
struct circ_buf *xmit = &up->port.state->xmit;
int i;
up->interrupt_mask1 &= ~(SAB82532_IMR1_ALLS|SAB82532_IMR1_XPR);
writeb(up->interrupt_mask1, &up->regs->w.imr1);
if (!test_bit(SAB82532_XPR, &up->irqflags))
return;
clear_bit(SAB82532_ALLS, &up->irqflags);
clear_bit(SAB82532_XPR, &up->irqflags);
for (i = 0; i < up->port.fifosize; i++) {
writeb(xmit->buf[xmit->tail],
&up->regs->w.xfifo[i]);
xmit->tail = (xmit->tail + 1) & (UART_XMIT_SIZE - 1);
up->port.icount.tx++;
if (uart_circ_empty(xmit))
break;
}
/* Issue a Transmit Frame command. */
sunsab_cec_wait(up);
writeb(SAB82532_CMDR_XF, &up->regs->w.cmdr);
}
/* port->lock is not held. */
static void sunsab_send_xchar(struct uart_port *port, char ch)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
unsigned long flags;
spin_lock_irqsave(&up->port.lock, flags);
sunsab_tec_wait(up);
writeb(ch, &up->regs->w.tic);
spin_unlock_irqrestore(&up->port.lock, flags);
}
/* port->lock held by caller. */
static void sunsab_stop_rx(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
up->interrupt_mask0 |= SAB82532_IMR0_TCD;
writeb(up->interrupt_mask1, &up->regs->w.imr0);
}
/* port->lock held by caller. */
static void sunsab_enable_ms(struct uart_port *port)
{
/* For now we always receive these interrupts. */
}
/* port->lock is not held. */
static void sunsab_break_ctl(struct uart_port *port, int break_state)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
unsigned long flags;
unsigned char val;
spin_lock_irqsave(&up->port.lock, flags);
val = up->cached_dafo;
if (break_state)
val |= SAB82532_DAFO_XBRK;
else
val &= ~SAB82532_DAFO_XBRK;
up->cached_dafo = val;
set_bit(SAB82532_REGS_PENDING, &up->irqflags);
if (test_bit(SAB82532_XPR, &up->irqflags))
sunsab_tx_idle(up);
spin_unlock_irqrestore(&up->port.lock, flags);
}
/* port->lock is not held. */
static int sunsab_startup(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
unsigned long flags;
unsigned char tmp;
int err = request_irq(up->port.irq, sunsab_interrupt,
IRQF_SHARED, "sab", up);
if (err)
return err;
spin_lock_irqsave(&up->port.lock, flags);
/*
* Wait for any commands or immediate characters
*/
sunsab_cec_wait(up);
sunsab_tec_wait(up);
/*
* Clear the FIFO buffers.
*/
writeb(SAB82532_CMDR_RRES, &up->regs->w.cmdr);
sunsab_cec_wait(up);
writeb(SAB82532_CMDR_XRES, &up->regs->w.cmdr);
/*
* Clear the interrupt registers.
*/
(void) readb(&up->regs->r.isr0);
(void) readb(&up->regs->r.isr1);
/*
* Now, initialize the UART
*/
writeb(0, &up->regs->w.ccr0); /* power-down */
writeb(SAB82532_CCR0_MCE | SAB82532_CCR0_SC_NRZ |
SAB82532_CCR0_SM_ASYNC, &up->regs->w.ccr0);
writeb(SAB82532_CCR1_ODS | SAB82532_CCR1_BCR | 7, &up->regs->w.ccr1);
writeb(SAB82532_CCR2_BDF | SAB82532_CCR2_SSEL |
SAB82532_CCR2_TOE, &up->regs->w.ccr2);
writeb(0, &up->regs->w.ccr3);
writeb(SAB82532_CCR4_MCK4 | SAB82532_CCR4_EBRG, &up->regs->w.ccr4);
up->cached_mode = (SAB82532_MODE_RTS | SAB82532_MODE_FCTS |
SAB82532_MODE_RAC);
writeb(up->cached_mode, &up->regs->w.mode);
writeb(SAB82532_RFC_DPS|SAB82532_RFC_RFTH_32, &up->regs->w.rfc);
tmp = readb(&up->regs->rw.ccr0);
tmp |= SAB82532_CCR0_PU; /* power-up */
writeb(tmp, &up->regs->rw.ccr0);
/*
* Finally, enable interrupts
*/
up->interrupt_mask0 = (SAB82532_IMR0_PERR | SAB82532_IMR0_FERR |
SAB82532_IMR0_PLLA);
writeb(up->interrupt_mask0, &up->regs->w.imr0);
up->interrupt_mask1 = (SAB82532_IMR1_BRKT | SAB82532_IMR1_ALLS |
SAB82532_IMR1_XOFF | SAB82532_IMR1_TIN |
SAB82532_IMR1_CSC | SAB82532_IMR1_XON |
SAB82532_IMR1_XPR);
writeb(up->interrupt_mask1, &up->regs->w.imr1);
set_bit(SAB82532_ALLS, &up->irqflags);
set_bit(SAB82532_XPR, &up->irqflags);
spin_unlock_irqrestore(&up->port.lock, flags);
return 0;
}
/* port->lock is not held. */
static void sunsab_shutdown(struct uart_port *port)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
unsigned long flags;
spin_lock_irqsave(&up->port.lock, flags);
/* Disable Interrupts */
up->interrupt_mask0 = 0xff;
writeb(up->interrupt_mask0, &up->regs->w.imr0);
up->interrupt_mask1 = 0xff;
writeb(up->interrupt_mask1, &up->regs->w.imr1);
/* Disable break condition */
up->cached_dafo = readb(&up->regs->rw.dafo);
up->cached_dafo &= ~SAB82532_DAFO_XBRK;
writeb(up->cached_dafo, &up->regs->rw.dafo);
/* Disable Receiver */
up->cached_mode &= ~SAB82532_MODE_RAC;
writeb(up->cached_mode, &up->regs->rw.mode);
/*
* XXX FIXME
*
* If the chip is powered down here the system hangs/crashes during
* reboot or shutdown. This needs to be investigated further,
* similar behaviour occurs in 2.4 when the driver is configured
* as a module only. One hint may be that data is sometimes
* transmitted at 9600 baud during shutdown (regardless of the
* speed the chip was configured for when the port was open).
*/
#if 0
/* Power Down */
tmp = readb(&up->regs->rw.ccr0);
tmp &= ~SAB82532_CCR0_PU;
writeb(tmp, &up->regs->rw.ccr0);
#endif
spin_unlock_irqrestore(&up->port.lock, flags);
free_irq(up->port.irq, up);
}
/*
* This is used to figure out the divisor speeds.
*
* The formula is: Baud = SAB_BASE_BAUD / ((N + 1) * (1 << M)),
*
* with 0 <= N < 64 and 0 <= M < 16
*/
static void calc_ebrg(int baud, int *n_ret, int *m_ret)
{
int n, m;
if (baud == 0) {
*n_ret = 0;
*m_ret = 0;
return;
}
/*
* We scale numbers by 10 so that we get better accuracy
* without having to use floating point. Here we increment m
* until n is within the valid range.
*/
n = (SAB_BASE_BAUD * 10) / baud;
m = 0;
while (n >= 640) {
n = n / 2;
m++;
}
n = (n+5) / 10;
/*
* We try very hard to avoid speeds with M == 0 since they may
* not work correctly for XTAL frequences above 10 MHz.
*/
if ((m == 0) && ((n & 1) == 0)) {
n = n / 2;
m++;
}
*n_ret = n - 1;
*m_ret = m;
}
/* Internal routine, port->lock is held and local interrupts are disabled. */
static void sunsab_convert_to_sab(struct uart_sunsab_port *up, unsigned int cflag,
unsigned int iflag, unsigned int baud,
unsigned int quot)
{
unsigned char dafo;
int bits, n, m;
/* Byte size and parity */
switch (cflag & CSIZE) {
case CS5: dafo = SAB82532_DAFO_CHL5; bits = 7; break;
case CS6: dafo = SAB82532_DAFO_CHL6; bits = 8; break;
case CS7: dafo = SAB82532_DAFO_CHL7; bits = 9; break;
case CS8: dafo = SAB82532_DAFO_CHL8; bits = 10; break;
/* Never happens, but GCC is too dumb to figure it out */
default: dafo = SAB82532_DAFO_CHL5; bits = 7; break;
}
if (cflag & CSTOPB) {
dafo |= SAB82532_DAFO_STOP;
bits++;
}
if (cflag & PARENB) {
dafo |= SAB82532_DAFO_PARE;
bits++;
}
if (cflag & PARODD) {
dafo |= SAB82532_DAFO_PAR_ODD;
} else {
dafo |= SAB82532_DAFO_PAR_EVEN;
}
up->cached_dafo = dafo;
calc_ebrg(baud, &n, &m);
up->cached_ebrg = n | (m << 6);
up->tec_timeout = (10 * 1000000) / baud;
up->cec_timeout = up->tec_timeout >> 2;
/* CTS flow control flags */
/* We encode read_status_mask and ignore_status_mask like so:
*
* ---------------------
* | ... | ISR1 | ISR0 |
* ---------------------
* .. 15 8 7 0
*/
up->port.read_status_mask = (SAB82532_ISR0_TCD | SAB82532_ISR0_TIME |
SAB82532_ISR0_RFO | SAB82532_ISR0_RPF |
SAB82532_ISR0_CDSC);
up->port.read_status_mask |= (SAB82532_ISR1_CSC |
SAB82532_ISR1_ALLS |
SAB82532_ISR1_XPR) << 8;
if (iflag & INPCK)
up->port.read_status_mask |= (SAB82532_ISR0_PERR |
SAB82532_ISR0_FERR);
if (iflag & (BRKINT | PARMRK))
up->port.read_status_mask |= (SAB82532_ISR1_BRK << 8);
/*
* Characteres to ignore
*/
up->port.ignore_status_mask = 0;
if (iflag & IGNPAR)
up->port.ignore_status_mask |= (SAB82532_ISR0_PERR |
SAB82532_ISR0_FERR);
if (iflag & IGNBRK) {
up->port.ignore_status_mask |= (SAB82532_ISR1_BRK << 8);
/*
* If we're ignoring parity and break indicators,
* ignore overruns too (for real raw support).
*/
if (iflag & IGNPAR)
up->port.ignore_status_mask |= SAB82532_ISR0_RFO;
}
/*
* ignore all characters if CREAD is not set
*/
if ((cflag & CREAD) == 0)
up->port.ignore_status_mask |= (SAB82532_ISR0_RPF |
SAB82532_ISR0_TCD);
uart_update_timeout(&up->port, cflag,
(up->port.uartclk / (16 * quot)));
/* Now schedule a register update when the chip's
* transmitter is idle.
*/
up->cached_mode |= SAB82532_MODE_RAC;
set_bit(SAB82532_REGS_PENDING, &up->irqflags);
if (test_bit(SAB82532_XPR, &up->irqflags))
sunsab_tx_idle(up);
}
/* port->lock is not held. */
static void sunsab_set_termios(struct uart_port *port, struct ktermios *termios,
struct ktermios *old)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *) port;
unsigned long flags;
unsigned int baud = uart_get_baud_rate(port, termios, old, 0, 4000000);
unsigned int quot = uart_get_divisor(port, baud);
spin_lock_irqsave(&up->port.lock, flags);
sunsab_convert_to_sab(up, termios->c_cflag, termios->c_iflag, baud, quot);
spin_unlock_irqrestore(&up->port.lock, flags);
}
static const char *sunsab_type(struct uart_port *port)
{
struct uart_sunsab_port *up = (void *)port;
static char buf[36];
sprintf(buf, "SAB82532 %s", sab82532_version[up->type]);
return buf;
}
static void sunsab_release_port(struct uart_port *port)
{
}
static int sunsab_request_port(struct uart_port *port)
{
return 0;
}
static void sunsab_config_port(struct uart_port *port, int flags)
{
}
static int sunsab_verify_port(struct uart_port *port, struct serial_struct *ser)
{
return -EINVAL;
}
static struct uart_ops sunsab_pops = {
.tx_empty = sunsab_tx_empty,
.set_mctrl = sunsab_set_mctrl,
.get_mctrl = sunsab_get_mctrl,
.stop_tx = sunsab_stop_tx,
.start_tx = sunsab_start_tx,
.send_xchar = sunsab_send_xchar,
.stop_rx = sunsab_stop_rx,
.enable_ms = sunsab_enable_ms,
.break_ctl = sunsab_break_ctl,
.startup = sunsab_startup,
.shutdown = sunsab_shutdown,
.set_termios = sunsab_set_termios,
.type = sunsab_type,
.release_port = sunsab_release_port,
.request_port = sunsab_request_port,
.config_port = sunsab_config_port,
.verify_port = sunsab_verify_port,
};
static struct uart_driver sunsab_reg = {
.owner = THIS_MODULE,
.driver_name = "sunsab",
.dev_name = "ttyS",
.major = TTY_MAJOR,
};
static struct uart_sunsab_port *sunsab_ports;
#ifdef CONFIG_SERIAL_SUNSAB_CONSOLE
static void sunsab_console_putchar(struct uart_port *port, int c)
{
struct uart_sunsab_port *up = (struct uart_sunsab_port *)port;
sunsab_tec_wait(up);
writeb(c, &up->regs->w.tic);
}
static void sunsab_console_write(struct console *con, const char *s, unsigned n)
{
struct uart_sunsab_port *up = &sunsab_ports[con->index];
unsigned long flags;
int locked = 1;
local_irq_save(flags);
if (up->port.sysrq) {
locked = 0;
} else if (oops_in_progress) {
locked = spin_trylock(&up->port.lock);
} else
spin_lock(&up->port.lock);
uart_console_write(&up->port, s, n, sunsab_console_putchar);
sunsab_tec_wait(up);
if (locked)
spin_unlock(&up->port.lock);
local_irq_restore(flags);
}
static int sunsab_console_setup(struct console *con, char *options)
{
struct uart_sunsab_port *up = &sunsab_ports[con->index];
unsigned long flags;
unsigned int baud, quot;
/*
* The console framework calls us for each and every port
* registered. Defer the console setup until the requested
* port has been properly discovered. A bit of a hack,
* though...
*/
if (up->port.type != PORT_SUNSAB)
return -1;
printk("Console: ttyS%d (SAB82532)\n",
(sunsab_reg.minor - 64) + con->index);
sunserial_console_termios(con, up->port.dev->of_node);
switch (con->cflag & CBAUD) {
case B150: baud = 150; break;
case B300: baud = 300; break;
case B600: baud = 600; break;
case B1200: baud = 1200; break;
case B2400: baud = 2400; break;
case B4800: baud = 4800; break;
default: case B9600: baud = 9600; break;
case B19200: baud = 19200; break;
case B38400: baud = 38400; break;
case B57600: baud = 57600; break;
case B115200: baud = 115200; break;
case B230400: baud = 230400; break;
case B460800: baud = 460800; break;
};
/*
* Temporary fix.
*/
spin_lock_init(&up->port.lock);
/*
* Initialize the hardware
*/
sunsab_startup(&up->port);
spin_lock_irqsave(&up->port.lock, flags);
/*
* Finally, enable interrupts
*/
up->interrupt_mask0 = SAB82532_IMR0_PERR | SAB82532_IMR0_FERR |
SAB82532_IMR0_PLLA | SAB82532_IMR0_CDSC;
writeb(up->interrupt_mask0, &up->regs->w.imr0);
up->interrupt_mask1 = SAB82532_IMR1_BRKT | SAB82532_IMR1_ALLS |
SAB82532_IMR1_XOFF | SAB82532_IMR1_TIN |
SAB82532_IMR1_CSC | SAB82532_IMR1_XON |
SAB82532_IMR1_XPR;
writeb(up->interrupt_mask1, &up->regs->w.imr1);
quot = uart_get_divisor(&up->port, baud);
sunsab_convert_to_sab(up, con->cflag, 0, baud, quot);
sunsab_set_mctrl(&up->port, TIOCM_DTR | TIOCM_RTS);
spin_unlock_irqrestore(&up->port.lock, flags);
return 0;
}
static struct console sunsab_console = {
.name = "ttyS",
.write = sunsab_console_write,
.device = uart_console_device,
.setup = sunsab_console_setup,
.flags = CON_PRINTBUFFER,
.index = -1,
.data = &sunsab_reg,
};
static inline struct console *SUNSAB_CONSOLE(void)
{
return &sunsab_console;
}
#else
#define SUNSAB_CONSOLE() (NULL)
#define sunsab_console_init() do { } while (0)
#endif
static int __devinit sunsab_init_one(struct uart_sunsab_port *up,
struct platform_device *op,
unsigned long offset,
int line)
{
up->port.line = line;
up->port.dev = &op->dev;
up->port.mapbase = op->resource[0].start + offset;
up->port.membase = of_ioremap(&op->resource[0], offset,
sizeof(union sab82532_async_regs),
"sab");
if (!up->port.membase)
return -ENOMEM;
up->regs = (union sab82532_async_regs __iomem *) up->port.membase;
up->port.irq = op->archdata.irqs[0];
up->port.fifosize = SAB82532_XMIT_FIFO_SIZE;
up->port.iotype = UPIO_MEM;
writeb(SAB82532_IPC_IC_ACT_LOW, &up->regs->w.ipc);
up->port.ops = &sunsab_pops;
up->port.type = PORT_SUNSAB;
up->port.uartclk = SAB_BASE_BAUD;
up->type = readb(&up->regs->r.vstr) & 0x0f;
writeb(~((1 << 1) | (1 << 2) | (1 << 4)), &up->regs->w.pcr);
writeb(0xff, &up->regs->w.pim);
if ((up->port.line & 0x1) == 0) {
up->pvr_dsr_bit = (1 << 0);
up->pvr_dtr_bit = (1 << 1);
up->gis_shift = 2;
} else {
up->pvr_dsr_bit = (1 << 3);
up->pvr_dtr_bit = (1 << 2);
up->gis_shift = 0;
}
up->cached_pvr = (1 << 1) | (1 << 2) | (1 << 4);
writeb(up->cached_pvr, &up->regs->w.pvr);
up->cached_mode = readb(&up->regs->rw.mode);
up->cached_mode |= SAB82532_MODE_FRTS;
writeb(up->cached_mode, &up->regs->rw.mode);
up->cached_mode |= SAB82532_MODE_RTS;
writeb(up->cached_mode, &up->regs->rw.mode);
up->tec_timeout = SAB82532_MAX_TEC_TIMEOUT;
up->cec_timeout = SAB82532_MAX_CEC_TIMEOUT;
return 0;
}
static int __devinit sab_probe(struct platform_device *op, const struct of_device_id *match)
{
static int inst;
struct uart_sunsab_port *up;
int err;
up = &sunsab_ports[inst * 2];
err = sunsab_init_one(&up[0], op,
0,
(inst * 2) + 0);
if (err)
goto out;
err = sunsab_init_one(&up[1], op,
sizeof(union sab82532_async_regs),
(inst * 2) + 1);
if (err)
goto out1;
sunserial_console_match(SUNSAB_CONSOLE(), op->dev.of_node,
&sunsab_reg, up[0].port.line,
false);
sunserial_console_match(SUNSAB_CONSOLE(), op->dev.of_node,
&sunsab_reg, up[1].port.line,
false);
err = uart_add_one_port(&sunsab_reg, &up[0].port);
if (err)
goto out2;
err = uart_add_one_port(&sunsab_reg, &up[1].port);
if (err)
goto out3;
dev_set_drvdata(&op->dev, &up[0]);
inst++;
return 0;
out3:
uart_remove_one_port(&sunsab_reg, &up[0].port);
out2:
of_iounmap(&op->resource[0],
up[1].port.membase,
sizeof(union sab82532_async_regs));
out1:
of_iounmap(&op->resource[0],
up[0].port.membase,
sizeof(union sab82532_async_regs));
out:
return err;
}
static int __devexit sab_remove(struct platform_device *op)
{
struct uart_sunsab_port *up = dev_get_drvdata(&op->dev);
uart_remove_one_port(&sunsab_reg, &up[1].port);
uart_remove_one_port(&sunsab_reg, &up[0].port);
of_iounmap(&op->resource[0],
up[1].port.membase,
sizeof(union sab82532_async_regs));
of_iounmap(&op->resource[0],
up[0].port.membase,
sizeof(union sab82532_async_regs));
dev_set_drvdata(&op->dev, NULL);
return 0;
}
static const struct of_device_id sab_match[] = {
{
.name = "se",
},
{
.name = "serial",
.compatible = "sab82532",
},
{},
};
MODULE_DEVICE_TABLE(of, sab_match);
static struct of_platform_driver sab_driver = {
.driver = {
.name = "sab",
.owner = THIS_MODULE,
.of_match_table = sab_match,
},
.probe = sab_probe,
.remove = __devexit_p(sab_remove),
};
static int __init sunsab_init(void)
{
struct device_node *dp;
int err;
[SERIAL] sparc: Infrastructure to fix section mismatch bugs. This patch against 2.6.23 sparc-2.6.git contains a number of minor cleanups of the sparc serial drivers. Initially I fixed this build warning: WARNING: vmlinux.o(.text+0x107a2c): Section mismatch: reference to .init.text:add_preferred_console (between 'sunserial_console_match' and 'sunserial_console_termios') which is done by declaring sunserial_console_match() as __init. This resulted in build warnings on sunserial_current_minor. To resolve these the variable was changed so it is no longer global, and to hide operations on it inside 2 new functions. These functions handle the UART minor handling code that is common to all sparc serial drivers. These changes allowed to clean up the uart counters in all the sparc serial drivers, and the administration of minor device numbers. Lastly, sunserial_console_termios() does not need to be exported since it is only called from non-modular code. Sadly, the following build warning still exists: WARNING: vmlinux.o(__ksymtab+0x2910): Section mismatch: reference to .init.text:sunserial_console_match (between '__ksymtab_sunserial_console_match' and '__ksymtab_sunserial_unregister_minors') This could be resolved by not exporting sunserial_console_match(), but this is not possible at the moment because it is being called from modular code. On the other hand, this is a bogus warning since it comes from a ksymtab section. Signed-off-by: Martin Habets <errandir_news@mph.eclipse.co.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-11 11:37:04 +00:00
int num_channels = 0;
for_each_node_by_name(dp, "se")
num_channels += 2;
for_each_node_by_name(dp, "serial") {
if (of_device_is_compatible(dp, "sab82532"))
num_channels += 2;
}
if (num_channels) {
sunsab_ports = kzalloc(sizeof(struct uart_sunsab_port) *
num_channels, GFP_KERNEL);
if (!sunsab_ports)
return -ENOMEM;
[SERIAL] sparc: Infrastructure to fix section mismatch bugs. This patch against 2.6.23 sparc-2.6.git contains a number of minor cleanups of the sparc serial drivers. Initially I fixed this build warning: WARNING: vmlinux.o(.text+0x107a2c): Section mismatch: reference to .init.text:add_preferred_console (between 'sunserial_console_match' and 'sunserial_console_termios') which is done by declaring sunserial_console_match() as __init. This resulted in build warnings on sunserial_current_minor. To resolve these the variable was changed so it is no longer global, and to hide operations on it inside 2 new functions. These functions handle the UART minor handling code that is common to all sparc serial drivers. These changes allowed to clean up the uart counters in all the sparc serial drivers, and the administration of minor device numbers. Lastly, sunserial_console_termios() does not need to be exported since it is only called from non-modular code. Sadly, the following build warning still exists: WARNING: vmlinux.o(__ksymtab+0x2910): Section mismatch: reference to .init.text:sunserial_console_match (between '__ksymtab_sunserial_console_match' and '__ksymtab_sunserial_unregister_minors') This could be resolved by not exporting sunserial_console_match(), but this is not possible at the moment because it is being called from modular code. On the other hand, this is a bogus warning since it comes from a ksymtab section. Signed-off-by: Martin Habets <errandir_news@mph.eclipse.co.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-11 11:37:04 +00:00
err = sunserial_register_minors(&sunsab_reg, num_channels);
if (err) {
kfree(sunsab_ports);
sunsab_ports = NULL;
return err;
}
}
return of_register_platform_driver(&sab_driver);
}
static void __exit sunsab_exit(void)
{
of_unregister_platform_driver(&sab_driver);
[SERIAL] sparc: Infrastructure to fix section mismatch bugs. This patch against 2.6.23 sparc-2.6.git contains a number of minor cleanups of the sparc serial drivers. Initially I fixed this build warning: WARNING: vmlinux.o(.text+0x107a2c): Section mismatch: reference to .init.text:add_preferred_console (between 'sunserial_console_match' and 'sunserial_console_termios') which is done by declaring sunserial_console_match() as __init. This resulted in build warnings on sunserial_current_minor. To resolve these the variable was changed so it is no longer global, and to hide operations on it inside 2 new functions. These functions handle the UART minor handling code that is common to all sparc serial drivers. These changes allowed to clean up the uart counters in all the sparc serial drivers, and the administration of minor device numbers. Lastly, sunserial_console_termios() does not need to be exported since it is only called from non-modular code. Sadly, the following build warning still exists: WARNING: vmlinux.o(__ksymtab+0x2910): Section mismatch: reference to .init.text:sunserial_console_match (between '__ksymtab_sunserial_console_match' and '__ksymtab_sunserial_unregister_minors') This could be resolved by not exporting sunserial_console_match(), but this is not possible at the moment because it is being called from modular code. On the other hand, this is a bogus warning since it comes from a ksymtab section. Signed-off-by: Martin Habets <errandir_news@mph.eclipse.co.uk> Signed-off-by: David S. Miller <davem@davemloft.net>
2007-12-11 11:37:04 +00:00
if (sunsab_reg.nr) {
sunserial_unregister_minors(&sunsab_reg, sunsab_reg.nr);
}
kfree(sunsab_ports);
sunsab_ports = NULL;
}
module_init(sunsab_init);
module_exit(sunsab_exit);
MODULE_AUTHOR("Eddie C. Dost and David S. Miller");
MODULE_DESCRIPTION("Sun SAB82532 serial port driver");
MODULE_LICENSE("GPL");