d/freebsd-nq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity
8b2e5491db
freebsd-nq/sys/isa/sio.c

1768 lines
44 KiB
C
Raw Normal View History

Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*-
* Copyright (c) 1991 The Regents of the University of California.
* All rights reserved.
*
* Redistribution and use in source and binary forms, with or without
* modification, are permitted provided that the following conditions
* are met:
* 1. Redistributions of source code must retain the above copyright
* notice, this list of conditions and the following disclaimer.
* 2. Redistributions in binary form must reproduce the above copyright
* notice, this list of conditions and the following disclaimer in the
* documentation and/or other materials provided with the distribution.
* 3. All advertising materials mentioning features or use of this software
* must display the following acknowledgement:
* This product includes software developed by the University of
* California, Berkeley and its contributors.
* 4. Neither the name of the University nor the names of its contributors
* may be used to endorse or promote products derived from this software
* without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
* FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
Removed all patch kit headers, sccsid and rcsid strings, put $Id$ in, some minor cleanup. Added $Id$ to files that did not have any version info, etc
1993-10-16 13:48:52 +00:00
* from: @(#)com.c 7.5 (Berkeley) 5/16/91
Remove CALLOUTMASK, if non-bidir case.
1994-01-31 19:07:59 +00:00
* $Id: sio.c,v 1.24 1994/01/31 08:52:12 ache Exp $
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
#include "sio.h"
#if NSIO > 0
/*
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
* Serial driver, based on 386BSD-0.1 com driver.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* Mostly rewritten to use pseudo-DMA.
* Works for National Semiconductor NS8250-NS16550AF UARTs.
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
* COM driver, based on HP dca driver.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
#include "param.h"
#include "systm.h"
#include "ioctl.h"
#include "tty.h"
#include "proc.h"
#include "user.h"
#include "conf.h"
#include "file.h"
#include "uio.h"
#include "kernel.h"
#include "syslog.h"
#include "i386/isa/isa.h"
#include "i386/isa/isa_device.h"
#include "i386/isa/comreg.h"
#include "i386/isa/ic/ns16550.h"
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
#define FAKE_DCD(unit) ((unit) == comconsole)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#define LOTS_OF_EVENTS 64 /* helps separate urgent events from input */
#define RB_I_HIGH_WATER (RBSZ - 2 * RS_IBUFSIZE)
#define RB_I_LOW_WATER ((RBSZ - 2 * RS_IBUFSIZE) * 7 / 8)
#define RS_IBUFSIZE 256
#define TTY_BI TTY_FE /* XXX */
#define TTY_OE TTY_PE /* XXX */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifndef COM_BIDIR
#define UNIT(x) (minor(x)) /* XXX */
#else /* COM_BIDIR */
#define COM_UNITMASK 0x7f
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
#define COM_CALLOUTMASK 0x80 /* for both minor and dev */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#define UNIT(x) (minor(x) & COM_UNITMASK)
#define CALLOUT(x) (minor(x) & COM_CALLOUTMASK)
#endif /* COM_BIDIR */
#ifdef COM_MULTIPORT
/* checks in flags for multiport and which is multiport "master chip"
* for a given card
*/
#define COM_ISMULTIPORT(dev) ((dev)->id_flags & 0x01)
#define COM_MPMASTER(dev) (((dev)->id_flags >> 8) & 0x0ff)
#endif /* COM_MULTIPORT */
Patch from Brian Smith (modified a little by me) to allow kernel config file override to disable fifo on 16550s: I bought a board with two 16550's, but one of those ports has a mouse on it. The sio driver always enables the fifo, which is a bad thing for mice and X. The mouse is jerky and hard to use. The simple thing is be to treat one of the ports as a non-fifo'ed UART, and I use the flags option in my config file. So, my config file has: device sio0 at isa? port "IO_COM1" tty irq 4 flags 0x2 vector siointr device sio1 at isa? port "IO_COM2" tty irq 3 vector siointr (patch deleted)
1994-01-31 06:12:18 +00:00
#define COM_NOFIFO(dev) ((dev)->id_flags & 0x02)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
#ifndef FIFO_TRIGGER
/*
* This driver is fast enough to work with any value and for high values
* to be only slightly more efficient. Low values may be better because
* they give lower latency.
* TODO: always use low values for low speeds. Mouse movements are jerky
* if more than one packet arrives at once. The low speeds used for
* serial mice help avoid this, but not if (large) fifos are enabled.
*/
#define FIFO_TRIGGER FIFO_TRIGGER_14
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#define com_scr 7 /* scratch register for 16450-16550 (R/W) */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
#define setsofttty() (ipending |= 1 << 4) /* XXX */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* Input buffer watermarks.
* The external device is asked to stop sending when the buffer exactly reaches
* high water, or when the high level requests it.
* The high level is notified immediately (rather than at a later clock tick)
* when this watermark is reached.
* The buffer size is chosen so the watermark should almost never be reached.
* The low watermark is invisibly 0 since the buffer is always emptied all at
* once.
*/
#define RS_IHIGHWATER (3 * RS_IBUFSIZE / 4)
/*
* com state bits.
* (CS_BUSY | CS_TTGO) and (CS_BUSY | CS_TTGO | CS_ODEVREADY) must be higher
* than the other bits so that they can be tested as a group without masking
* off the low bits.
*
* The following com and tty flags correspond closely:
* TS_BUSY = CS_BUSY (maintained by comstart() and comflush())
* CS_TTGO = ~TS_TTSTOP (maintained by comstart() and siostop())
* CS_CTS_OFLOW = CCTS_OFLOW (maintained by comparam())
* CS_RTS_IFLOW = CRTS_IFLOW (maintained by comparam())
* TS_FLUSH is not used.
* Bug: I think TIOCSETA doesn't clear TS_TTSTOP when it clears IXON.
*/
#define CS_BUSY 0x80 /* output in progress */
#define CS_TTGO 0x40 /* output not stopped by XOFF */
#define CS_ODEVREADY 0x20 /* external device h/w ready (CTS) */
#define CS_CHECKMSR 1 /* check of MSR scheduled */
#define CS_CTS_OFLOW 2 /* use CTS output flow control */
#define CS_ODONE 4 /* output completed */
#define CS_RTS_IFLOW 8 /* use RTS input flow control */
static char *error_desc[] = {
#define CE_OVERRUN 0
"silo overflow",
#define CE_INTERRUPT_BUF_OVERFLOW 1
"interrupt-level buffer overflow",
#define CE_TTY_BUF_OVERFLOW 2
"tty-level buffer overflow",
};
#define CE_NTYPES 3
#define CE_RECORD(com, errnum) (++(com)->delta_error_counts[errnum])
/* types. XXX - should be elsewhere */
typedef u_int Port_t; /* hardware port */
typedef u_char bool_t; /* boolean */
/* com device structure */
struct com_s {
u_char state; /* miscellaneous flag bits */
u_char cfcr_image; /* copy of value written to CFCR */
bool_t hasfifo; /* nonzero for 16550 UARTs */
u_char mcr_image; /* copy of value written to MCR */
#ifdef COM_BIDIR
bool_t bidir; /* is this unit bidirectional? */
bool_t active; /* is the port active _at all_? */
bool_t active_in; /* is the incoming port in use? */
bool_t active_out; /* is the outgoing port in use? */
#endif /* COM_BIDIR */
#ifdef COM_MULTIPORT
bool_t multiport; /* is this unit part of a multiport device? */
#endif /* COM_MULTIPORT */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
int dtr_wait; /* time to hold DTR down on close (* 1/HZ) */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
u_int tx_fifo_size;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* The high level of the driver never reads status registers directly
* because there would be too many side effects to handle conveniently.
* Instead, it reads copies of the registers stored here by the
* interrupt handler.
*/
u_char last_modem_status; /* last MSR read by intr handler */
u_char prev_modem_status; /* last MSR handled by high level */
u_char *ibuf; /* start of input buffer */
u_char *ibufend; /* end of input buffer */
u_char *ihighwater; /* threshold in input buffer */
u_char *iptr; /* next free spot in input buffer */
u_char *obufend; /* end of output buffer */
int ocount; /* original count for current output */
u_char *optr; /* next char to output */
Port_t data_port; /* i/o ports */
Port_t int_id_port;
Port_t iobase;
Port_t modem_ctl_port;
Port_t line_status_port;
Port_t modem_status_port;
struct tty *tp; /* cross reference */
u_long bytes_in; /* statistics */
u_long bytes_out;
u_int delta_error_counts[CE_NTYPES];
u_int error_counts[CE_NTYPES];
/*
* Ping-pong input buffers. The extra factor of 2 in the sizes is
* to allow for an error byte for each input byte.
*/
#define CE_INPUT_OFFSET RS_IBUFSIZE
u_char ibuf1[2 * RS_IBUFSIZE];
u_char ibuf2[2 * RS_IBUFSIZE];
};
/*
* These functions in the com module ought to be declared (with a prototype)
* in a com-driver system header. The void ones may need to be int to match
* ancient devswitch declarations, but they don't actually return anything.
*/
#define Dev_t int /* promoted dev_t */
struct consdev;
int sioclose __P((Dev_t dev, int fflag, int devtype,
struct proc *p));
void siointr __P((int unit));
int sioioctl __P((Dev_t dev, int cmd, caddr_t data,
int fflag, struct proc *p));
int siocngetc __P((Dev_t dev));
void siocninit __P((struct consdev *cp));
void siocnprobe __P((struct consdev *cp));
void siocnputc __P((Dev_t dev, int c));
int sioopen __P((Dev_t dev, int oflags, int devtype,
struct proc *p));
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
void siopoll __P((void));
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
int sioread __P((Dev_t dev, struct uio *uio, int ioflag));
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
int sioselect __P((Dev_t dev, int rw, struct proc *p));
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
void siostop __P((struct tty *tp, int rw));
int siowrite __P((Dev_t dev, struct uio *uio, int ioflag));
void softsio1 __P((void));
static int sioattach __P((struct isa_device *dev));
static void comflush __P((struct com_s *com));
static void comhardclose __P((struct com_s *com));
static void cominit __P((int unit, int rate));
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
#ifdef COM_MULTIPORT
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
static void comintr1 __P((struct com_s *com));
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
#endif /* COM_MULTIPORT */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
static void commctl __P((struct com_s *com, int bits, int how));
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
static int comparam __P((struct tty *tp, struct termios *t));
static int sioprobe __P((struct isa_device *dev));
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
static void comstart __P((struct tty *tp));
static void comwakeup __P((caddr_t chan, int ticks));
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
static int tiocm_xxx2mcr __P((int tiocm_xxx));
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/* table and macro for fast conversion from a unit number to its com struct */
static struct com_s *p_com_addr[NSIO];
#define com_addr(unit) (p_com_addr[unit])
static struct com_s com_structs[NSIO];
struct isa_driver siodriver = {
sioprobe, sioattach, "sio"
};
#ifdef COMCONSOLE
static int comconsole = COMCONSOLE;
#else
static int comconsole = -1;
#endif
static bool_t comconsinit;
static speed_t comdefaultrate = TTYDEF_SPEED;
static u_int com_events; /* input chars + weighted output completions */
static int commajor;
struct tty sio_tty[NSIO];
extern struct tty *constty;
extern u_int ipending; /* XXX */
extern int tk_nin; /* XXX */
extern int tk_rawcc; /* XXX */
#ifdef KGDB
#include "machine/remote-sl.h"
extern int kgdb_dev;
extern int kgdb_rate;
extern int kgdb_debug_init;
#endif
static struct speedtab comspeedtab[] = {
0, 0,
50, COMBRD(50),
75, COMBRD(75),
110, COMBRD(110),
134, COMBRD(134),
150, COMBRD(150),
200, COMBRD(200),
300, COMBRD(300),
600, COMBRD(600),
1200, COMBRD(1200),
1800, COMBRD(1800),
2400, COMBRD(2400),
4800, COMBRD(4800),
9600, COMBRD(9600),
19200, COMBRD(19200),
38400, COMBRD(38400),
57600, COMBRD(57600),
115200, COMBRD(115200),
-1, -1
};
/* XXX - configure this list */
static Port_t likely_com_ports[] = { 0x3f8, 0x2f8, 0x3e8, 0x2e8, };
static int
sioprobe(dev)
struct isa_device *dev;
{
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
static bool_t already_init;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
Port_t *com_ptr;
Port_t iobase;
int result;
if (!already_init) {
/*
* Turn off MCR_IENABLE for all likely serial ports. An unused
* port with its MCR_IENABLE gate open will inhibit interrupts
* from any used port that shares the interrupt vector.
*/
for (com_ptr = likely_com_ports;
com_ptr < &likely_com_ports[sizeof likely_com_ports
/ sizeof likely_com_ports[0]];
++com_ptr)
outb(*com_ptr + com_mcr, 0);
already_init = TRUE;
}
iobase = dev->id_iobase;
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
result = IO_COMSIZE;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* We don't want to get actual interrupts, just masked ones.
* Interrupts from this line should already be masked in the ICU,
* but mask them in the processor as well in case there are some
* (misconfigured) shared interrupts.
*/
disable_intr();
/*
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
* Initialize the speed so that any junk in the THR or output fifo will
* be transmitted in a known time. (There may be lots of junk after a
* soft reboot, and output interrupts don't work right after a master
* reset, at least for 16550s. (The speed is undefined after MR, but
* MR empties the THR and the TSR so it's not clear why this matters)).
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* Enable output interrupts (only) and check the following:
* o the CFCR, IER and MCR in UART hold the values written to them
* (the values happen to be all distinct - this is good for
* avoiding false positive tests from bus echoes).
* o an output interrupt is generated and its vector is correct.
* o the interrupt goes away when the IIR in the UART is read.
*/
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(iobase + com_cfcr, CFCR_DLAB);
outb(iobase + com_dlbl, COMBRD(9600) & 0xff);
outb(iobase + com_dlbh, (u_int) COMBRD(9600) >> 8);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(iobase + com_cfcr, CFCR_8BITS); /* ensure IER is addressed */
outb(iobase + com_mcr, MCR_IENABLE); /* open gate early */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(iobase + com_ier, 0); /* ensure edge on next intr */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(iobase + com_ier, IER_ETXRDY); /* generate interrupt */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
DELAY((16 + 1) * 9600 / 10); /* enough to drain 16 bytes */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if ( inb(iobase + com_cfcr) != CFCR_8BITS
|| inb(iobase + com_ier) != IER_ETXRDY
|| inb(iobase + com_mcr) != MCR_IENABLE
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
#ifndef COM_MULTIPORT /* XXX - need to do more to enable interrupts */
|| !isa_irq_pending(dev)
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
|| (inb(iobase + com_iir) & IIR_IMASK) != IIR_TXRDY
|| isa_irq_pending(dev)
|| (inb(iobase + com_iir) & IIR_IMASK) != IIR_NOPEND)
result = 0;
/*
* Turn off all device interrupts and check that they go off properly.
* Leave MCR_IENABLE set. It gates the OUT2 output of the UART to
* the ICU input. Closing the gate would give a floating ICU input
* (unless there is another device driving at) and spurious interrupts.
* (On the system that this was first tested on, the input floats high
* and gives a (masked) interrupt as soon as the gate is closed.)
*/
outb(iobase + com_ier, 0);
outb(iobase + com_mcr, MCR_IENABLE); /* dummy to avoid bus echo */
if ( inb(iobase + com_ier) != 0
|| isa_irq_pending(dev)
|| (inb(iobase + com_iir) & IIR_IMASK) != IIR_NOPEND)
result = 0;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (result == 0)
outb(iobase + com_mcr, 0);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
enable_intr();
return (result);
}
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
static int
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
sioattach(isdp)
struct isa_device *isdp;
{
struct com_s *com;
static bool_t comwakeup_started = FALSE;
Port_t iobase;
int s;
int unit;
iobase = isdp->id_iobase;
unit = isdp->id_unit;
if (unit == comconsole)
DELAY(1000); /* XXX */
s = spltty();
/*
* sioprobe() has initialized the device registers as follows:
* o cfcr = CFCR_8BITS.
* It is most important that CFCR_DLAB is off, so that the
* data port is not hidden when we enable interrupts.
* o ier = 0.
* Interrupts are only enabled when the line is open.
* o mcr = MCR_IENABLE.
* Keeping MCR_DTR and MCR_RTS off might stop the external
* device from sending before we are ready.
*/
com = &com_structs[unit];
com->cfcr_image = CFCR_8BITS;
com->mcr_image = MCR_IENABLE;
Change default dtrwait value to 2 seconds. Change C-style into static initializer per Bruce request.
1993-12-16 04:38:27 +00:00
com->dtr_wait = 200;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
com->tx_fifo_size = 1;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com->iptr = com->ibuf = com->ibuf1;
com->ibufend = com->ibuf1 + RS_IBUFSIZE;
com->ihighwater = com->ibuf1 + RS_IHIGHWATER;
com->iobase = iobase;
com->data_port = iobase + com_data;
com->int_id_port = iobase + com_iir;
com->modem_ctl_port = iobase + com_mcr;
com->line_status_port = iobase + com_lsr;
com->modem_status_port = iobase + com_msr;
com->tp = &sio_tty[unit];
/* attempt to determine UART type */
Fix attach message, make separate line for it (old version assumed that line is probe line)
1993-11-17 23:38:23 +00:00
printf("sio%d: type", unit);
Change to sio for recognising the last serial port on a multiport card correctly. Updated syscons to 0.2e, added features for upcoming XFree86 2.0, fixed crashbug.
1993-09-28 00:01:18 +00:00
#ifdef COM_MULTIPORT
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (!COM_ISMULTIPORT(isdp))
Change to sio for recognising the last serial port on a multiport card correctly. Updated syscons to 0.2e, added features for upcoming XFree86 2.0, fixed crashbug.
1993-09-28 00:01:18 +00:00
#endif
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
{
u_char scr;
u_char scr1;
u_char scr2;
scr = inb(iobase + com_scr);
outb(iobase + com_scr, 0xa5);
scr1 = inb(iobase + com_scr);
outb(iobase + com_scr, 0x5a);
scr2 = inb(iobase + com_scr);
outb(iobase + com_scr, scr);
if (scr1 != 0xa5 || scr2 != 0x5a) {
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
printf(" 8250");
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
goto determined_type;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
outb(iobase + com_fifo, FIFO_ENABLE | FIFO_TRIGGER_14);
DELAY(100);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
switch (inb(com->int_id_port) & IIR_FIFO_MASK) {
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
case FIFO_TRIGGER_1:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
printf(" 16450");
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
break;
case FIFO_TRIGGER_4:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
printf(" 16450?");
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
break;
case FIFO_TRIGGER_8:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
printf(" 16550?");
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
break;
case FIFO_TRIGGER_14:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
printf(" 16550A");
Patch from Brian Smith (modified a little by me) to allow kernel config file override to disable fifo on 16550s: I bought a board with two 16550's, but one of those ports has a mouse on it. The sio driver always enables the fifo, which is a bad thing for mice and X. The mouse is jerky and hard to use. The simple thing is be to treat one of the ports as a non-fifo'ed UART, and I use the flags option in my config file. So, my config file has: device sio0 at isa? port "IO_COM1" tty irq 4 flags 0x2 vector siointr device sio1 at isa? port "IO_COM2" tty irq 3 vector siointr (patch deleted)
1994-01-31 06:12:18 +00:00
if (COM_NOFIFO(isdp)) {
printf(" fifo software disabled");
} else {
com->hasfifo = TRUE;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
com->tx_fifo_size = 16;
Patch from Brian Smith (modified a little by me) to allow kernel config file override to disable fifo on 16550s: I bought a board with two 16550's, but one of those ports has a mouse on it. The sio driver always enables the fifo, which is a bad thing for mice and X. The mouse is jerky and hard to use. The simple thing is be to treat one of the ports as a non-fifo'ed UART, and I use the flags option in my config file. So, my config file has: device sio0 at isa? port "IO_COM1" tty irq 4 flags 0x2 vector siointr device sio1 at isa? port "IO_COM2" tty irq 3 vector siointr (patch deleted)
1994-01-31 06:12:18 +00:00
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
break;
}
outb(iobase + com_fifo, 0);
determined_type: ;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef COM_MULTIPORT
if (COM_ISMULTIPORT(isdp)) {
struct isa_device *masterdev;
com->multiport = TRUE;
printf(" (multiport)");
/* set the master's common-interrupt-enable reg.,
* as appropriate. YYY See your manual
*/
/* enable only common interrupt for port */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(com->modem_ctl_port, com->mcr_image = 0);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
masterdev = find_isadev(isa_devtab_tty, &siodriver,
COM_MPMASTER(isdp));
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(masterdev->id_iobase + com_scr, 0x80);
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com->multiport = FALSE;
#endif /* COM_MULTIPORT */
Fix attach message, make separate line for it (old version assumed that line is probe line)
1993-11-17 23:38:23 +00:00
printf("\n");
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef KGDB
if (kgdb_dev == makedev(commajor, unit)) {
if (comconsole == unit)
kgdb_dev = -1; /* can't debug over console port */
else {
cominit(unit, kgdb_rate);
if (kgdb_debug_init) {
/*
* Print prefix of device name,
* let kgdb_connect print the rest.
*/
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
printf("sio%d: ", unit);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
kgdb_connect(1);
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
printf("sio%d: kgdb enabled\n", unit);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
}
#endif
/*
* Need to reset baud rate, etc. of next print so reset comconsinit.
*/
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (unit == comconsole)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
comconsinit = FALSE;
com_addr(unit) = com;
splx(s);
if (!comwakeup_started) {
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
comwakeup((caddr_t) NULL, 0);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
comwakeup_started = TRUE;
}
return (1);
}
/* ARGSUSED */
int
sioopen(dev, flag, mode, p)
dev_t dev;
int flag;
int mode;
struct proc *p;
{
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
#ifdef COM_BIDIR
bool_t callout;
#endif /* COM_BIDIR */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
struct com_s *com;
int error = 0;
Port_t iobase;
int s;
struct tty *tp;
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
int unit;
unit = UNIT(dev);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if ((u_int) unit >= NSIO || (com = com_addr(unit)) == NULL)
return (ENXIO);
#ifdef COM_BIDIR
/* if it's a callout device, and bidir not possible on that dev, die */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
callout = CALLOUT(dev);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (callout && !(com->bidir))
return (ENXIO);
#endif /* COM_BIDIR */
tp = com->tp;
s = spltty();
#ifdef COM_BIDIR
bidir_open_top:
/* if it's bidirectional, we've gotta deal with it... */
if (com->bidir) {
if (callout) {
if (com->active_in) {
/* it's busy. die */
splx(s);
return (EBUSY);
} else {
/* it's ours. lock it down, and set it up */
com->active_out = TRUE;
}
} else {
if (com->active_out) {
/* it's busy, outgoing. wait, if possible */
if (flag & O_NONBLOCK) {
/* can't wait; bail */
splx(s);
return (EBUSY);
} else {
/* wait for it... */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
error = tsleep((caddr_t)&com->active_out,
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
TTIPRI|PCATCH,
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
"siooth",
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
0);
/* if there was an error, take off. */
if (error != 0) {
splx(s);
return (error);
}
/* else take it from the top */
goto bidir_open_top;
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
} else if (com->prev_modem_status & MSR_DCD
|| FAKE_DCD(unit)) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/* there's a carrier on the line; we win */
com->active_in = TRUE;
} else {
/* there is no carrier on the line */
if (flag & O_NONBLOCK) {
/* can't wait; let it open */
com->active_in = TRUE;
} else {
/* put DTR & RTS up */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
/* XXX - bring up RTS earlier? */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, MCR_DTR | MCR_RTS, DMSET);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(com->iobase + com_ier, IER_EMSC);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/* wait for it... */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
error = tsleep((caddr_t)&com->active_in,
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
TTIPRI|PCATCH,
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
"siodcd",
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
0);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/* if not active, turn intrs and DTR off */
if (!com->active) {
outb(com->iobase + com_ier, 0);
commctl(com, MCR_DTR, DMBIC);
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/* if there was an error, take off. */
if (error != 0) {
splx(s);
return (error);
}
/* else take it from the top */
goto bidir_open_top;
}
}
}
}
com->active = TRUE;
#endif /* COM_BIDIR */
tp->t_oproc = comstart;
tp->t_param = comparam;
tp->t_dev = dev;
if (!(tp->t_state & TS_ISOPEN)) {
tp->t_state |= TS_WOPEN;
ttychars(tp);
if (tp->t_ispeed == 0) {
/*
* We no longer use the flags from <sys/ttydefaults.h>
* since those are only relevant for logins. It's
* important to have echo off initially so that the
* line doesn't start blathering before the echo flag
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
* can be turned off.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
tp->t_iflag = 0;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
tp->t_oflag = 0;
set output flags to ttydefault if COMCONSOLE is defined.
1993-12-13 13:20:50 +00:00
#ifdef COMCONSOLE
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (unit == comconsole)
set output flags to ttydefault if COMCONSOLE is defined.
1993-12-13 13:20:50 +00:00
tp->t_oflag = TTYDEF_OFLAG;
#endif
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
tp->t_cflag = CREAD | CS8 | HUPCL;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
tp->t_lflag = 0;
tp->t_ispeed = tp->t_ospeed = comdefaultrate;
}
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/*
* XXX the full state after a first open() needs to be
* programmable and separate for callin and callout.
*/
1) Fix error with port recognition, speed initialization code added to probe. 2) Force CLOCAL=on for outgoing ports and CLOCAL=off for incoming ports into open in bidirectional case. 3) Add DELAY after writing to com_ier for fifo drain into probe.
1993-12-09 17:24:19 +00:00
#ifdef COM_BIDIR
Force HUPCL in bidirectional port open. This don't allow it hangs forever.
1993-12-11 00:36:14 +00:00
if (com->bidir) {
1) Fix error with port recognition, speed initialization code added to probe. 2) Force CLOCAL=on for outgoing ports and CLOCAL=off for incoming ports into open in bidirectional case. 3) Add DELAY after writing to com_ier for fifo drain into probe.
1993-12-09 17:24:19 +00:00
if (callout)
tp->t_cflag |= CLOCAL;
else
tp->t_cflag &= ~CLOCAL;
Force HUPCL in bidirectional port open. This don't allow it hangs forever.
1993-12-11 00:36:14 +00:00
}
1) Fix error with port recognition, speed initialization code added to probe. 2) Force CLOCAL=on for outgoing ports and CLOCAL=off for incoming ports into open in bidirectional case. 3) Add DELAY after writing to com_ier for fifo drain into probe.
1993-12-09 17:24:19 +00:00
#endif
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, MCR_DTR | MCR_RTS, DMSET);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
error = comparam(tp, &tp->t_termios);
if (error != 0)
goto out;
ttsetwater(tp);
iobase = com->iobase;
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (com->hasfifo) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/* (re)enable and drain FIFO */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(iobase + com_fifo, FIFO_ENABLE | FIFO_TRIGGER
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
| FIFO_RCV_RST | FIFO_XMT_RST);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
DELAY(100);
}
disable_intr();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
(void) inb(com->line_status_port);
(void) inb(com->data_port);
com->last_modem_status =
com->prev_modem_status = inb(com->modem_status_port);
outb(iobase + com_ier, IER_ERXRDY | IER_ETXRDY | IER_ERLS
| IER_EMSC);
enable_intr();
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (com->prev_modem_status & MSR_DCD || FAKE_DCD(unit))
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
tp->t_state |= TS_CARR_ON;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else if (tp->t_state & TS_XCLUDE && p->p_ucred->cr_uid != 0) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
splx(s);
return (EBUSY);
}
while (!(flag & O_NONBLOCK) && !(tp->t_cflag & CLOCAL)
#ifdef COM_BIDIR
/* We went through a lot of trouble to open it,
* but it's certain we have a carrier now, so
* don't spend any time on it now.
*/
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
&& !(com->bidir)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#endif /* COM_BIDIR */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
&& !(tp->t_state & TS_CARR_ON)) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
tp->t_state |= TS_WOPEN;
error = ttysleep(tp, (caddr_t)&tp->t_raw, TTIPRI | PCATCH,
ttopen, 0);
if (error != 0)
break;
}
out:
splx(s);
if (error == 0)
Make the LINT kernel compile with -W -Wreturn-type -Wcomment -Werror, and add same (sans -Werror) to Makefile for future compilations.
1993-11-25 01:38:01 +00:00
error = (*linesw[tp->t_line].l_open)(dev, tp, 0);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef COM_BIDIR
/* wakeup sleepers */
wakeup((caddr_t) &com->active_in);
#endif /* COM_BIDIR */
/*
* XXX - the next step was once not done, so interrupts, DTR and RTS
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
* remained hot if the process was killed while it was sleeping
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* waiting for carrier. Now there is the opposite problem. If several
* processes are sleeping waiting for carrier on the same line and one
* is killed, interrupts are turned off so the other processes will
* never see the carrier rise.
*/
if (error != 0 && !(tp->t_state & TS_ISOPEN))
comhardclose(com);
tp->t_state &= ~TS_WOPEN;
return (error);
}
/*ARGSUSED*/
int
sioclose(dev, flag, mode, p)
dev_t dev;
int flag;
int mode;
struct proc *p;
{
struct com_s *com;
struct tty *tp;
com = com_addr(UNIT(dev));
tp = com->tp;
(*linesw[tp->t_line].l_close)(tp, flag);
comhardclose(com);
ttyclose(tp);
return (0);
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
static void
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
comhardclose(com)
struct com_s *com;
{
Port_t iobase;
int s;
struct tty *tp;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
int unit;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
s = spltty();
iobase = com->iobase;
outb(iobase + com_cfcr, com->cfcr_image &= ~CFCR_SBREAK);
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
unit = com - &com_structs[0];
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef KGDB
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/* do not disable interrupts or hang up if debugging */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (kgdb_dev != makedev(commajor, unit))
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#endif
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
{
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(iobase + com_ier, 0);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
tp = com->tp;
if (tp->t_cflag & HUPCL || tp->t_state & TS_WOPEN
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
|| !(com->prev_modem_status & MSR_DCD) && !FAKE_DCD(unit)
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
|| !(tp->t_state & TS_ISOPEN)) {
commctl(com, MCR_RTS, DMSET);
if (com->dtr_wait != 0)
/*
* Uninterruptible sleep since we want to
* wait a fixed time.
* XXX - delay in open() (if necessary),
* not here (always).
*/
tsleep((caddr_t)&com->dtr_wait, TTIPRI,
"sioclose", com->dtr_wait);
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
#ifdef COM_BIDIR
com->active = com->active_in = com->active_out = FALSE;
/* wakeup sleepers who are waiting for out to finish */
wakeup((caddr_t) &com->active_out);
#endif /* COM_BIDIR */
splx(s);
}
int
sioread(dev, uio, flag)
dev_t dev;
struct uio *uio;
int flag;
{
struct tty *tp = com_addr(UNIT(dev))->tp;
return ((*linesw[tp->t_line].l_read)(tp, uio, flag));
}
int
siowrite(dev, uio, flag)
dev_t dev;
struct uio *uio;
int flag;
{
int unit = UNIT(dev);
struct tty *tp = com_addr(unit)->tp;
/*
* (XXX) We disallow virtual consoles if the physical console is
* a serial port. This is in case there is a display attached that
* is not the console. In that situation we don't need/want the X
* server taking over the console.
*/
if (constty && unit == comconsole)
constty = NULL;
return ((*linesw[tp->t_line].l_write)(tp, uio, flag));
}
void
siointr(unit)
int unit;
{
struct com_s *com;
#ifndef COM_MULTIPORT
u_char line_status;
u_char modem_status;
u_char *ioptr;
u_char recv_data;
com = com_addr(unit);
#else /* COM_MULTIPORT */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
bool_t possibly_more_intrs;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
/*
* Loop until there is no activity on any port. This is necessary
* to get an interrupt edge more than to avoid another interrupt.
* If the IRQ signal is just an OR of the IRQ signals from several
* devices, then the edge from one may be lost because another is
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
* on.
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
*/
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
do {
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
possibly_more_intrs = FALSE;
for (unit = 0; unit < NSIO; ++unit) {
com = com_addr(unit);
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (com != NULL
&& (inb(com->int_id_port) & IIR_IMASK)
!= IIR_NOPEND) {
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/*
* XXX call comintr1() instead of here from
* comwakeup(). The interrupt edge problem
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
* only exists for real interrupts.
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
*/
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
comintr1(com);
possibly_more_intrs = TRUE;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
}
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
} while (possibly_more_intrs);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
return;
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
static void
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
comintr1(com)
struct com_s *com;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
u_char line_status;
u_char modem_status;
u_char *ioptr;
u_char recv_data;
#endif /* COM_MULTIPORT */
while (TRUE) {
line_status = inb(com->line_status_port);
/* input event? (check first to help avoid overruns) */
while (line_status & LSR_RCV_MASK) {
/* break/unnattached error bits or real input? */
if (!(line_status & LSR_RXRDY))
recv_data = 0;
else
recv_data = inb(com->data_port);
++com->bytes_in;
#ifdef KGDB
/* trap into kgdb? (XXX - needs testing and optim) */
if (recv_data == FRAME_END
&& !(com->tp->t_state & TS_ISOPEN)
&& kgdb_dev == makedev(commajor, unit)) {
kgdb_connect(0);
continue;
}
#endif /* KGDB */
ioptr = com->iptr;
if (ioptr >= com->ibufend)
CE_RECORD(com, CE_INTERRUPT_BUF_OVERFLOW);
else {
++com_events;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
#if 0 /* for testing input latency vs efficiency */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (com->iptr - com->ibuf == 8)
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
setsofttty();
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
ioptr[0] = recv_data;
ioptr[CE_INPUT_OFFSET] = line_status;
com->iptr = ++ioptr;
if (ioptr == com->ihighwater
&& com->state & CS_RTS_IFLOW)
outb(com->modem_ctl_port,
com->mcr_image &= ~MCR_RTS);
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
/* XXX - move this out of isr */
if (line_status & LSR_OE)
CE_RECORD(com, CE_OVERRUN);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/*
* "& 0x7F" is to avoid the gcc-1.40 generating a slow
* jump from the top of the loop to here
*/
line_status = inb(com->line_status_port) & 0x7F;
}
/* modem status change? (always check before doing output) */
modem_status = inb(com->modem_status_port);
if (modem_status != com->last_modem_status) {
/*
* Schedule high level to handle DCD changes. Note
* that we don't use the delta bits anywhere. Some
* UARTs mess them up, and it's easy to remember the
* previous bits and calculate the delta.
*/
com->last_modem_status = modem_status;
if (!(com->state & CS_CHECKMSR)) {
com_events += LOTS_OF_EVENTS;
com->state |= CS_CHECKMSR;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
setsofttty();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
/* handle CTS change immediately for crisp flow ctl */
if (com->state & CS_CTS_OFLOW) {
if (modem_status & MSR_CTS)
com->state |= CS_ODEVREADY;
else
com->state &= ~CS_ODEVREADY;
}
}
/* output queued and everything ready? */
if (line_status & LSR_TXRDY
&& com->state >= (CS_ODEVREADY | CS_BUSY | CS_TTGO)) {
ioptr = com->optr;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (com->tx_fifo_size > 1) {
u_int ocount;
ocount = com->obufend - ioptr;
if (ocount > com->tx_fifo_size)
ocount = com->tx_fifo_size;
com->bytes_out += ocount;
do
outb(com->data_port, *ioptr++);
while (--ocount != 0);
} else {
outb(com->data_port, *ioptr++);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
++com->bytes_out;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
}
com->optr = ioptr;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (ioptr >= com->obufend) {
/* output just completed */
com_events += LOTS_OF_EVENTS;
com->state ^= (CS_ODONE | CS_BUSY);
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
setsofttty(); /* handle at high level ASAP */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
}
/* finished? */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
#ifndef COM_MULTIPORT
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if ((inb(com->int_id_port) & IIR_IMASK) == IIR_NOPEND)
#endif /* COM_MULTIPORT */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
return;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
}
From guido@gvr.win.tue.nl Wed Sep 8 13:34:16 1993 That is because TIOCMGET was broken. Yes...this is known for some time and no, we (Bruce and me) never posted it. Why? Simply because we choose to post fixes when we fixed most of the bugs. Anyway..now that the slip problems are coming, here is a fix for correct TIOCMGET behaviour. -Guido Note: this should be tested first (Rich?). Tested by rgrimes
1993-09-10 16:59:16 +00:00
static int
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
tiocm_xxx2mcr(tiocm_xxx)
int tiocm_xxx;
From guido@gvr.win.tue.nl Wed Sep 8 13:34:16 1993 That is because TIOCMGET was broken. Yes...this is known for some time and no, we (Bruce and me) never posted it. Why? Simply because we choose to post fixes when we fixed most of the bugs. Anyway..now that the slip problems are coming, here is a fix for correct TIOCMGET behaviour. -Guido Note: this should be tested first (Rich?). Tested by rgrimes
1993-09-10 16:59:16 +00:00
{
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
int mcr;
mcr = 0;
if (tiocm_xxx & TIOCM_DTR)
mcr |= MCR_DTR;
if (tiocm_xxx & TIOCM_RTS)
mcr |= MCR_RTS;
return (mcr);
From guido@gvr.win.tue.nl Wed Sep 8 13:34:16 1993 That is because TIOCMGET was broken. Yes...this is known for some time and no, we (Bruce and me) never posted it. Why? Simply because we choose to post fixes when we fixed most of the bugs. Anyway..now that the slip problems are coming, here is a fix for correct TIOCMGET behaviour. -Guido Note: this should be tested first (Rich?). Tested by rgrimes
1993-09-10 16:59:16 +00:00
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
int
sioioctl(dev, cmd, data, flag, p)
dev_t dev;
int cmd;
caddr_t data;
int flag;
struct proc *p;
{
struct com_s *com;
int error;
Port_t iobase;
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
int mcr;
int msr;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
int s;
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
int tiocm_xxx;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
struct tty *tp;
com = com_addr(UNIT(dev));
tp = com->tp;
error = (*linesw[tp->t_line].l_ioctl)(tp, cmd, data, flag);
if (error >= 0)
return (error);
error = ttioctl(tp, cmd, data, flag);
Temporary plug CLOCAL security hole, while sticky bits yet not implemented. Only for bidirectional case.
1994-01-02 10:17:29 +00:00
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
#ifdef COM_BIDIR
/* XXX: plug security hole while sticky bits not yet implemented */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (com->bidir && com->active_in && p->p_ucred->cr_uid != 0)
Temporary plug CLOCAL security hole, while sticky bits yet not implemented. Only for bidirectional case.
1994-01-02 10:17:29 +00:00
tp->t_cflag &= ~CLOCAL;
#endif
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (error >= 0)
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
return (error);
Temporary plug CLOCAL security hole, while sticky bits yet not implemented. Only for bidirectional case.
1994-01-02 10:17:29 +00:00
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
iobase = com->iobase;
s = spltty();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
switch (cmd) {
case TIOCSBRK:
outb(iobase + com_cfcr, com->cfcr_image |= CFCR_SBREAK);
break;
case TIOCCBRK:
outb(iobase + com_cfcr, com->cfcr_image &= ~CFCR_SBREAK);
break;
case TIOCSDTR:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, MCR_DTR, DMBIS);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
case TIOCCDTR:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, MCR_DTR, DMBIC);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
case TIOCMSET:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, tiocm_xxx2mcr(*(int *)data), DMSET);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
case TIOCMBIS:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, tiocm_xxx2mcr(*(int *)data), DMBIS);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
case TIOCMBIC:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
commctl(com, tiocm_xxx2mcr(*(int *)data), DMBIC);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
case TIOCMGET:
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
tiocm_xxx = TIOCM_LE; /* XXX - always enabled while open */
mcr = com->mcr_image;
if (mcr & MCR_DTR)
tiocm_xxx |= TIOCM_DTR;
if (mcr & MCR_RTS)
tiocm_xxx |= TIOCM_RTS;
msr = com->prev_modem_status;
if (msr & MSR_CTS)
tiocm_xxx |= TIOCM_CTS;
if (msr & MSR_DCD)
tiocm_xxx |= TIOCM_CD;
if (msr & MSR_DSR)
tiocm_xxx |= TIOCM_DSR;
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/*
* XXX - MSR_RI is naturally volatile, and we make MSR_TERI
* more volatile by reading the modem status a lot. Perhaps
* we should latch both bits until the status is read here.
*/
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (msr & (MSR_RI | MSR_TERI))
tiocm_xxx |= TIOCM_RI;
*(int *)data = tiocm_xxx;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
#ifdef COM_BIDIR
case TIOCMSBIDIR:
/* must be root to set bidir. capability */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
error = suser(p->p_ucred, &p->p_acflag);
if (error != 0) {
splx(s);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
return(EPERM);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/* if it's the console, can't do it (XXX why?) */
if (UNIT(dev) == comconsole) {
splx(s);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
return(ENOTTY);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
#if 0
/* XXX - can't do the next, for obvious reasons...
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* but there are problems to be looked at...
*/
/* if the port is active, don't do it */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (com->active) {
splx(s);
return(EBUSY);
}
#endif
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com->bidir = *(int *)data;
break;
case TIOCMGBIDIR:
*(int *)data = com->bidir;
break;
#endif /* COM_BIDIR */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
case TIOCMSDTRWAIT:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
/* must be root since the wait applies to following logins */
error = suser(p->p_ucred, &p->p_acflag);
if (error != 0) {
splx(s);
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
return(EPERM);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
/* if it's the console, can't do it (XXX why?) */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (UNIT(dev) == comconsole) {
splx(s);
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
return(ENOTTY);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
}
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
com->dtr_wait = *(int *)data;
break;
case TIOCMGDTRWAIT:
*(int *)data = com->dtr_wait;
break;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
default:
splx(s);
return (ENOTTY);
}
splx(s);
return (0);
}
/* cancel pending output */
static void
comflush(com)
struct com_s *com;
{
struct ringb *rbp;
disable_intr();
if (com->state & CS_ODONE)
com_events -= LOTS_OF_EVENTS;
com->state &= ~(CS_ODONE | CS_BUSY);
enable_intr();
rbp = &com->tp->t_out;
rbp->rb_hd += com->ocount;
rbp->rb_hd = RB_ROLLOVER(rbp, rbp->rb_hd);
com->ocount = 0;
com->tp->t_state &= ~TS_BUSY;
}
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
void
siopoll()
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
static bool_t awake = FALSE;
int s;
int unit;
if (com_events == 0)
return;
disable_intr();
if (awake) {
enable_intr();
return;
}
awake = TRUE;
enable_intr();
s = spltty();
repeat:
for (unit = 0; unit < NSIO; ++unit) {
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
u_char *buf;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
struct com_s *com;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
u_char *ibuf;
int incc;
struct tty *tp;
com = com_addr(unit);
if (com == NULL)
continue;
tp = com->tp;
/* switch the role of the low-level input buffers */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
if (com->iptr == (ibuf = com->ibuf)) {
buf = NULL; /* not used, but compiler can't tell */
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
incc = 0;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
buf = ibuf;
disable_intr();
incc = com->iptr - buf;
com_events -= incc;
if (ibuf == com->ibuf1)
ibuf = com->ibuf2;
else
ibuf = com->ibuf1;
com->ibufend = ibuf + RS_IBUFSIZE;
com->ihighwater = ibuf + RS_IHIGHWATER;
com->iptr = ibuf;
/*
* There is now room for another low-level buffer full
* of input, so enable RTS if it is now disabled and
* there is room in the high-level buffer.
*/
if (!(com->mcr_image & MCR_RTS)
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
&& !(tp->t_state & TS_RTS_IFLOW))
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(com->modem_ctl_port,
com->mcr_image |= MCR_RTS);
enable_intr();
com->ibuf = ibuf;
}
if (com->state & CS_CHECKMSR) {
u_char delta_modem_status;
disable_intr();
delta_modem_status = com->last_modem_status
^ com->prev_modem_status;
com->prev_modem_status = com->last_modem_status;
com_events -= LOTS_OF_EVENTS;
com->state &= ~CS_CHECKMSR;
enable_intr();
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (delta_modem_status & MSR_DCD && !FAKE_DCD(unit)) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (com->prev_modem_status & MSR_DCD) {
(*linesw[tp->t_line].l_modem)(tp, 1);
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
#ifdef COM_BIDIR
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
wakeup((caddr_t) &com->active_in);
#endif /* COM_BIDIR */
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
(*linesw[tp->t_line].l_modem)(tp, 0);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
}
/* XXX */
if (TRUE) {
u_int delta;
int errnum;
u_long total;
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
for (errnum = 0; errnum < CE_NTYPES; ++errnum) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
disable_intr();
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
delta = com->delta_error_counts[errnum];
com->delta_error_counts[errnum] = 0;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
enable_intr();
if (delta != 0) {
total =
com->error_counts[errnum] += delta;
log(LOG_WARNING,
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
"sio%d: %u more %s%s (total %lu)\n",
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
unit, delta, error_desc[errnum],
delta == 1 ? "" : "s", total);
}
}
}
if (com->state & CS_ODONE) {
comflush(com);
/* XXX - why isn't the table used for t_line == 0? */
if (tp->t_line != 0)
(*linesw[tp->t_line].l_start)(tp);
else
comstart(tp);
}
if (incc <= 0 || !(tp->t_state & TS_ISOPEN))
continue;
if (com->state & CS_RTS_IFLOW
&& RB_LEN(&tp->t_raw) + incc >= RB_I_HIGH_WATER
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
&& !(tp->t_state & TS_RTS_IFLOW)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* XXX - need RTS flow control for all line disciplines.
* Only have it in standard one now.
*/
&& linesw[tp->t_line].l_rint == ttyinput) {
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
tp->t_state |= TS_RTS_IFLOW;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
ttstart(tp);
}
/*
* Avoid the grotesquely inefficient lineswitch routine
* (ttyinput) in "raw" mode. It usually takes about 450
* instructions (that's without canonical processing or echo!).
* slinput is reasonably fast (usually 40 instructions plus
* call overhead).
*/
if (!(tp->t_iflag & (ICRNL | IGNCR | IMAXBEL | INLCR | ISTRIP
| IXOFF | IXON))
&& !(tp->t_lflag & (ECHO | ECHONL | ICANON | IEXTEN | ISIG
| PENDIN))
&& !(tp->t_state & (TS_CNTTB | TS_LNCH))
&& linesw[tp->t_line].l_rint == ttyinput) {
tk_nin += incc;
tk_rawcc += incc;
tp->t_rawcc += incc;
com->delta_error_counts[CE_TTY_BUF_OVERFLOW]
+= incc - rb_write(&tp->t_raw, (char *) buf,
incc);
ttwakeup(tp);
if (tp->t_state & TS_TTSTOP
&& (tp->t_iflag & IXANY
|| tp->t_cc[VSTART] == tp->t_cc[VSTOP])) {
tp->t_state &= ~TS_TTSTOP;
tp->t_lflag &= ~FLUSHO;
ttstart(tp);
}
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
do {
u_char line_status;
int recv_data;
line_status = (u_char) buf[CE_INPUT_OFFSET];
recv_data = (u_char) *buf++;
if (line_status
& (LSR_BI | LSR_FE | LSR_OE | LSR_PE)) {
if (line_status & LSR_BI)
recv_data |= TTY_BI;
if (line_status & LSR_FE)
recv_data |= TTY_FE;
if (line_status & LSR_OE)
recv_data |= TTY_OE;
if (line_status & LSR_PE)
recv_data |= TTY_PE;
}
(*linesw[tp->t_line].l_rint)(recv_data, tp);
} while (--incc > 0);
}
if (com_events == 0)
break;
}
if (com_events >= LOTS_OF_EVENTS)
goto repeat;
splx(s);
awake = FALSE;
}
static int
comparam(tp, t)
struct tty *tp;
struct termios *t;
{
u_int cfcr;
int cflag;
struct com_s *com;
int divisor;
int error;
Port_t iobase;
int s;
int unit;
/* check requested parameters */
divisor = ttspeedtab(t->c_ospeed, comspeedtab);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (t->c_ispeed == 0)
t->c_ispeed = t->c_ospeed;
if (divisor < 0 || t->c_ispeed != t->c_ospeed)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
return (EINVAL);
/* parameters are OK, convert them to the com struct and the device */
unit = UNIT(tp->t_dev);
com = com_addr(unit);
iobase = com->iobase;
s = spltty();
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (divisor == 0)
commctl(com, MCR_DTR, DMBIC); /* hang up line */
else
commctl(com, MCR_DTR, DMBIS);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
cflag = t->c_cflag;
switch (cflag & CSIZE) {
case CS5:
cfcr = CFCR_5BITS;
break;
case CS6:
cfcr = CFCR_6BITS;
break;
case CS7:
cfcr = CFCR_7BITS;
break;
default:
cfcr = CFCR_8BITS;
break;
}
if (cflag & PARENB) {
cfcr |= CFCR_PENAB;
if (!(cflag & PARODD))
cfcr |= CFCR_PEVEN;
}
if (cflag & CSTOPB)
cfcr |= CFCR_STOPB;
/*
* Some UARTs lock up if the divisor latch registers are selected
* while the UART is doing output (they refuse to transmit anything
* more until given a hard reset). Fix this by stopping filling
* the device buffers and waiting for them to drain. Reading the
* line status port outside of siointr() might lose some receiver
* error bits, but that is acceptable here.
*/
disable_intr();
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
retry:
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com->state &= ~CS_TTGO;
enable_intr();
while ((inb(com->line_status_port) & (LSR_TSRE | LSR_TXRDY))
!= (LSR_TSRE | LSR_TXRDY)) {
error = ttysleep(tp, (caddr_t)&tp->t_raw, TTIPRI | PCATCH,
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
"sioparam", 1);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (error != 0 && error != EAGAIN) {
if (!(tp->t_state & TS_TTSTOP)) {
disable_intr();
com->state |= CS_TTGO;
enable_intr();
}
splx(s);
return (error);
}
}
disable_intr(); /* very important while com_data is hidden */
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
/*
* XXX - clearing CS_TTGO is not sufficient to stop further output,
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
* because siopoll() calls comstart() which usually sets it again
* because TS_TTSTOP is clear. Setting TS_TTSTOP would not be
* sufficient, for similar reasons.
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
*/
if ((inb(com->line_status_port) & (LSR_TSRE | LSR_TXRDY))
!= (LSR_TSRE | LSR_TXRDY))
goto retry;
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
if (divisor != 0) {
outb(iobase + com_cfcr, cfcr | CFCR_DLAB);
outb(iobase + com_dlbl, divisor & 0xFF);
outb(iobase + com_dlbh, (u_int) divisor >> 8);
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(iobase + com_cfcr, com->cfcr_image = cfcr);
if (!(tp->t_state & TS_TTSTOP))
com->state |= CS_TTGO;
if (cflag & CRTS_IFLOW)
com->state |= CS_RTS_IFLOW; /* XXX - secondary changes? */
else
com->state &= ~CS_RTS_IFLOW;
/*
* Set up state to handle output flow control.
* XXX - worth handling MDMBUF (DCD) flow control at the lowest level?
* Now has 16+ msec latency, while CTS flow has 50- usec latency.
*/
com->state &= ~CS_CTS_OFLOW;
com->state |= CS_ODEVREADY;
if (cflag & CCTS_OFLOW) {
com->state |= CS_CTS_OFLOW;
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (!(com->last_modem_status & MSR_CTS))
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com->state &= ~CS_ODEVREADY;
}
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
/*
* Recover from fiddling with CS_TTGO. We used to call siointr()
* unconditionally, but that defeated the careful discarding of
* stale input in sioopen().
*
* XXX sioopen() is not careful waiting for carrier for the callout
* case.
*/
if (com->state >= (CS_BUSY | CS_TTGO))
siointr(unit);
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
enable_intr();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
splx(s);
return (0);
}
Make the LINT kernel compile with -W -Wreturn-type -Wcomment -Werror, and add same (sans -Werror) to Makefile for future compilations.
1993-11-25 01:38:01 +00:00
static void
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
comstart(tp)
struct tty *tp;
{
struct com_s *com;
int s;
int unit;
unit = UNIT(tp->t_dev);
com = com_addr(unit);
s = spltty();
disable_intr();
if (tp->t_state & TS_TTSTOP)
com->state &= ~CS_TTGO;
else
com->state |= CS_TTGO;
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (tp->t_state & TS_RTS_IFLOW) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (com->mcr_image & MCR_RTS && com->state & CS_RTS_IFLOW)
outb(com->modem_ctl_port, com->mcr_image &= ~MCR_RTS);
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else {
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
if (!(com->mcr_image & MCR_RTS) && com->iptr < com->ihighwater)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(com->modem_ctl_port, com->mcr_image |= MCR_RTS);
}
enable_intr();
if (tp->t_state & (TS_TIMEOUT | TS_TTSTOP))
goto out;
if (RB_LEN(&tp->t_out) <= tp->t_lowat) {
if (tp->t_state & TS_ASLEEP) {
tp->t_state &= ~TS_ASLEEP;
wakeup((caddr_t)&tp->t_out);
}
if (tp->t_wsel) {
selwakeup(tp->t_wsel, tp->t_state & TS_WCOLL);
tp->t_wsel = 0;
tp->t_state &= ~TS_WCOLL;
}
}
if (com->ocount != 0) {
disable_intr();
siointr(unit);
enable_intr();
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
} else if (RB_LEN(&tp->t_out) != 0) {
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
tp->t_state |= TS_BUSY;
com->ocount = RB_CONTIGGET(&tp->t_out);
disable_intr();
com->obufend = (com->optr = (u_char *) tp->t_out.rb_hd)
+ com->ocount;
com->state |= CS_BUSY;
siointr(unit); /* fake interrupt to start output */
enable_intr();
}
out:
splx(s);
}
void
siostop(tp, rw)
struct tty *tp;
int rw;
{
struct com_s *com;
com = com_addr(UNIT(tp->t_dev));
if (rw & FWRITE)
comflush(com);
disable_intr();
Add missing read flush code to siostop
1993-11-08 19:36:32 +00:00
if (rw & FREAD) {
com_events -= (com->iptr - com->ibuf);
com->iptr = com->ibuf;
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (tp->t_state & TS_TTSTOP)
com->state &= ~CS_TTGO;
else
com->state |= CS_TTGO;
enable_intr();
}
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
int
sioselect(dev, rw, p)
dev_t dev;
int rw;
struct proc *p;
{
Remove CALLOUTMASK, if non-bidir case.
1994-01-31 19:07:59 +00:00
#ifdef COM_BIDIR
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
return ttselect(dev & ~COM_CALLOUTMASK, rw, p);
Remove CALLOUTMASK, if non-bidir case.
1994-01-31 19:07:59 +00:00
#else
return ttselect(dev, rw, p);
#endif
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
}
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
static void
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
commctl(com, bits, how)
struct com_s *com;
int bits;
int how;
{
disable_intr();
switch (how) {
case DMSET:
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(com->modem_ctl_port,
com->mcr_image = bits | (com->mcr_image & MCR_IENABLE));
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
case DMBIS:
outb(com->modem_ctl_port, com->mcr_image |= bits);
break;
case DMBIC:
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
outb(com->modem_ctl_port, com->mcr_image &= ~bits);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
}
enable_intr();
}
Make the LINT kernel compile with -W -Wreturn-type -Wcomment -Werror, and add same (sans -Werror) to Makefile for future compilations.
1993-11-25 01:38:01 +00:00
static void
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
comwakeup(chan, ticks)
caddr_t chan;
int ticks;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
{
int unit;
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
timeout(comwakeup, (caddr_t) NULL, hz / 100);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
if (com_events != 0)
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
/* schedule siopoll() to run when the cpl allows */
setsofttty();
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/* recover from lost output interrupts */
for (unit = 0; unit < NSIO; ++unit) {
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
struct com_s *com;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com = com_addr(unit);
if (com != NULL && com->state >= (CS_BUSY | CS_TTGO)) {
disable_intr();
siointr(unit);
enable_intr();
}
}
Make the LINT kernel compile with -W -Wreturn-type -Wcomment -Werror, and add same (sans -Werror) to Makefile for future compilations.
1993-11-25 01:38:01 +00:00
return;
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
}
void
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
softsio1()
{
1) Set ipending back to 4, because (16 + 4) bit not in netmask or impmask. 2) Fixes from Bruce: o Changed name of schedsoftcom() to setsofttty() to match setsoftclock() o Bool_t isn't used. o tx_fifo_size is 1 for chips without fifos, 16 for 16550's, to help to output more efficiently for 16550's (LSR_TXRDY means that the fifo is empty, not that it has space for one char). o Changed name of softsio1() to siopoll() and merged compoll() into siopoll(). o The probe forgot to clear com_mcr after it failed. This is harmful for 4 single serial ports on 2 interrupts. It makes partial misconfigurations worse. o Don't bother initializing static variables that are 0 (bidir stuff). o Only initialize t_oflag to TTYDEF_OFLAG if unit == COMCONSOLE, not if COMCONSOLE is defined. o Don't call siointr() from comparam() if there is no output in progress. For the call from sioopen(), there's no output in progress, and siointr() often saw silo overflows for stale input because it was called before sioopen() discarded the input. o Let ttselect() do the work for select(), so that the fixes for ttselect() don't have to be duplicated in zillions of drivers.
1994-01-31 08:52:12 +00:00
siopoll();
New sio driver from Bruce.
1993-11-14 23:29:01 +00:00
}
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
* Following are all routines needed for SIO to act as console
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* XXX - not tested in this version
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
* XXX - i386/cons.c only knows about the com driver (NCOM and not NSIO)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
* XXX - check that the corresponding serial interrupts are never enabled
*/
#include "i386/i386/cons.h"
void
siocnprobe(cp)
struct consdev *cp;
{
int unit;
/* locate the major number */
for (commajor = 0; commajor < nchrdev; commajor++)
Make the LINT kernel compile with -W -Wreturn-type -Wcomment -Werror, and add same (sans -Werror) to Makefile for future compilations.
1993-11-25 01:38:01 +00:00
if (cdevsw[commajor].d_open == sioopen)
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
break;
/* XXX: ick */
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
unit = UNIT(CONUNIT);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
com_addr(unit) = &com_structs[unit];
com_addr(unit)->iobase = CONADDR;
/* make sure hardware exists? XXX */
/* initialize required fields */
cp->cn_dev = makedev(commajor, unit);
#ifdef COMCONSOLE
cp->cn_pri = CN_REMOTE; /* Force a serial port console */
#else
cp->cn_pri = CN_NORMAL;
#endif
}
void
siocninit(cp)
struct consdev *cp;
{
int unit;
unit = UNIT(cp->cn_dev);
cominit(unit, comdefaultrate);
comconsole = unit;
comconsinit = TRUE;
}
static void
cominit(unit, rate)
int unit;
int rate;
{
Port_t iobase;
int s;
iobase = com_addr(unit)->iobase;
s = splhigh();
outb(iobase + com_cfcr, CFCR_DLAB);
rate = ttspeedtab(comdefaultrate, comspeedtab);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(iobase + com_dlbl, rate & 0xFF);
outb(iobase + com_dlbh, rate >> 8);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
outb(iobase + com_cfcr, CFCR_8BITS);
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
outb(iobase + com_fifo,
FIFO_ENABLE | FIFO_TRIGGER | FIFO_RCV_RST | FIFO_XMT_RST);
DELAY(100);
(void) inb(iobase + com_lsr);
(void) inb(iobase + com_data);
(void) inb(iobase + com_msr);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
/*
* XXX - fishy to enable interrupts and then poll.
* It shouldn't be necessary to ready the iir.
*/
outb(iobase + com_ier, IER_ERXRDY | IER_ETXRDY | IER_ERLS | IER_EMSC);
(void) inb(iobase + com_iir);
splx(s);
}
int
siocngetc(dev)
dev_t dev;
{
int c;
Port_t iobase;
int s;
iobase = com_addr(UNIT(dev))->iobase;
s = splhigh();
while (!(inb(iobase + com_lsr) & LSR_RXRDY))
;
c = inb(iobase + com_data);
(void) inb(iobase + com_iir);
splx(s);
return (c);
}
void
siocnputc(dev, c)
dev_t dev;
int c;
{
Port_t iobase;
int s;
int timo;
iobase = com_addr(UNIT(dev))->iobase;
s = splhigh();
#ifdef KGDB
if (dev != kgdb_dev)
#endif
if (!comconsinit) {
From ache: Choose older MULTIPORT version, because lastest bde version not worked. Don't force HUPCL for bidirectional case. From bde: Use bit (1 << (16 + 4)) in schedsoftcom() to avoid clash with non-serial h/w on IRQ4. Allow FIFO_TRIGGER in config. Clear com->mcr_image when clearing mcr for init of 4port. The usual value MCR_IENABLE should have broken 4ports unless something happened to clear it later. Turn off interrupts as well as DTR after an error waiting for carrier (bidir dialin case). Drain fifo more carefully. Don't hang up if debugging. Rearrange siointr() -> siointr1() for multiport case for speed, lower latency and clarity. Use suser() to check perms. Provide missing splx() after failed perms checks. c_ispeed == 0 means c_ispeed = c_ospeed (POSIX). Set parameters (except speed) for c_ospeed == 0 as well as hanging up. Better initialization for console (fifo stuff...). Misc. cleanups. Fix dead hang if modem power is off.
1994-01-11 18:31:45 +00:00
cominit(UNIT(dev), comdefaultrate);
Initial import, 0.1 + pk 0.2.4-B1
1993-06-12 14:58:17 +00:00
comconsinit = TRUE;
}
/* wait for any pending transmission to finish */
timo = 50000;
while (!(inb(iobase + com_lsr) & LSR_TXRDY) && --timo)
;
outb(iobase + com_data, c);
/* wait for this transmission to complete */
timo = 1500000;
while (!(inb(iobase + com_lsr) & LSR_TXRDY) && --timo)
;
/* clear any interrupts generated by this transmission */
(void) inb(iobase + com_iir);
splx(s);
}
#endif /* NSIO > 0 */
Reference in New Issue Copy Permalink