ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Copyright (C) 1994 Poul-Henning Kamp
|
|
|
|
*
|
|
|
|
* All rights reserved.
|
|
|
|
*
|
|
|
|
* This file contains some material which are covered by the message after
|
|
|
|
* this message.
|
|
|
|
*
|
|
|
|
* The rest of this file is covered by the following clause:
|
|
|
|
* ----------------------------------------------------------------------------
|
|
|
|
* "THE BEER-WARE LICENSE" (Revision 42):
|
|
|
|
* <phk@login.dkuug.dk> wrote this file. As long as you retain this notice you
|
|
|
|
* can do whatever you want with this stuff. If we meet some day, and you think
|
|
|
|
* this stuff is worth it, you can buy me a beer in return. Poul-Henning Kamp
|
|
|
|
* ----------------------------------------------------------------------------
|
|
|
|
*
|
1994-10-23 21:28:03 +00:00
|
|
|
* $Id: pas.c,v 1.5 1994/10/18 03:53:12 phk Exp $
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
*
|
|
|
|
* This is a driver for the one particular kind of the "ProAudioSpectrum"
|
|
|
|
* card from MediaVision. To find out if your card is supported, you can
|
|
|
|
* either try out the driver, or you can look for a chip a little less than
|
|
|
|
* 1" square in one end of the card, with writing on it that say ...5380...
|
|
|
|
*
|
|
|
|
* Up to four of these cards can be in the same computer. If you have
|
|
|
|
* multiple cards, you need to set the "card-id" jumpers correspondingly.
|
|
|
|
*
|
|
|
|
* The driver uses no interrupts, so don't expect record-breaking performance.
|
|
|
|
*
|
|
|
|
* Poul-Henning Kamp <phk@freefall.cdrom.com>
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
1994-08-19 00:01:08 +00:00
|
|
|
* Copyright (C) 1993 Allen K. Briggs, Chris P. Caputo,
|
|
|
|
* Michael L. Finch, Bradley A. Grantham, and
|
|
|
|
* Lawrence A. Kesteloot
|
|
|
|
* 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 Alice Group.
|
|
|
|
* 4. The names of the Alice Group or any of its members may not be used
|
|
|
|
* to endorse or promote products derived from this software without
|
|
|
|
* specific prior written permission.
|
|
|
|
*
|
|
|
|
* THIS SOFTWARE IS PROVIDED BY THE ALICE GROUP ``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 ALICE GROUP 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.
|
|
|
|
*
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
* Was: Id: pas.c,v 1.8.2.1 1994/07/11 00:02:32 cgd Exp
|
1994-08-19 00:01:08 +00:00
|
|
|
*
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
* Modified for use with the pc532 by Phil Nelson, Feb 94.
|
1994-08-19 00:01:08 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
|
|
|
|
#include "pas.h"
|
|
|
|
#if NPAS > 0
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
1994-10-23 21:28:03 +00:00
|
|
|
#include <sys/param.h>
|
|
|
|
#include <sys/systm.h>
|
|
|
|
#include <sys/buf.h>
|
|
|
|
#include <sys/devconf.h>
|
|
|
|
|
|
|
|
#include <i386/isa/isa_device.h>
|
|
|
|
#include <i386/isa/ic/ncr_5380.h>
|
1994-08-19 00:01:08 +00:00
|
|
|
#include <scsi/scsi_all.h>
|
|
|
|
#include <scsi/scsiconf.h>
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Define this macro to disable the PSEUDO-DMA transfers.
|
|
|
|
*/
|
|
|
|
#undef NO_PAS16_PSEUDO_DMA
|
|
|
|
|
|
|
|
/*
|
|
|
|
* This parameter determines how many byte we "insb" between each check for
|
|
|
|
* a change of phase.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#ifndef PAS16_PSEUDO_DMA_SIZE
|
|
|
|
# define PAS16_PSEUDO_DMA_SIZE 512
|
|
|
|
#endif
|
|
|
|
|
|
|
|
/*
|
|
|
|
* How many times we attempt to select a device.
|
|
|
|
*/
|
|
|
|
#ifndef ATTEMPT_SELECTION
|
|
|
|
# define ATTEMPT_SELECTION 5
|
|
|
|
#endif /* ATTEMPT_SELECTION */
|
|
|
|
|
|
|
|
/*
|
|
|
|
* How many microseconds between each attempt.
|
|
|
|
*/
|
|
|
|
#ifndef SELECTION_DELAY
|
|
|
|
# define SELECTION_DELAY 10000
|
|
|
|
#endif /* SELECTION_DELAY */
|
|
|
|
|
|
|
|
#ifndef PAS16_TIMEOUT
|
|
|
|
# define PAS16_TIMEOUT 1000000
|
|
|
|
#endif /* PAS16_TIMEOUT */
|
1994-08-19 00:01:08 +00:00
|
|
|
/* What we need to debug the driver */
|
|
|
|
|
|
|
|
#ifdef PAS_DEBUG
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
# if PAS_DEBUG == 0
|
|
|
|
# undef PAS_DEBUG
|
|
|
|
# else
|
|
|
|
static int pas_show_scsi_cmd(struct scsi_xfer *xs);
|
|
|
|
# ifndef PAS_DEBUG_REQUEST_SENSE
|
|
|
|
# define PAS_DEBUG_REQUEST_SENSE
|
|
|
|
# endif
|
|
|
|
# endif
|
1994-08-19 00:01:08 +00:00
|
|
|
#endif /* PAS_DEBUG */
|
|
|
|
|
|
|
|
#define SCSI_PHASE_DATA_OUT 0x0
|
|
|
|
#define SCSI_PHASE_DATA_IN 0x1
|
|
|
|
#define SCSI_PHASE_CMD 0x2
|
|
|
|
#define SCSI_PHASE_STATUS 0x3
|
|
|
|
#define SCSI_PHASE_UNSPEC1 0x4
|
|
|
|
#define SCSI_PHASE_UNSPEC2 0x5
|
|
|
|
#define SCSI_PHASE_MESSAGE_OUT 0x6
|
|
|
|
#define SCSI_PHASE_MESSAGE_IN 0x7
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
#define SCSI_CUR_PHASE(x) ((x>>2)&0x7)
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
#define SCSI_RET_SUCCESS 0
|
|
|
|
#define SCSI_RET_RETRY 1
|
|
|
|
#define SCSI_RET_DEVICE_DOWN 2
|
|
|
|
#define SCSI_RET_COMMAND_FAIL 3
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
/* Our per device (card) structure */
|
|
|
|
static
|
|
|
|
struct pas_softc {
|
|
|
|
struct scsi_link sc_link;
|
|
|
|
int mv_unit;
|
|
|
|
u_short iobase;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
#define xor_0000 iobase
|
|
|
|
u_short xor_0001;
|
|
|
|
u_short xor_0002;
|
|
|
|
u_short xor_0003;
|
|
|
|
u_short xor_2000;
|
|
|
|
u_short xor_2001;
|
|
|
|
u_short xor_2002;
|
|
|
|
u_short xor_2003;
|
|
|
|
u_short xor_4000;
|
|
|
|
u_short xor_4001;
|
|
|
|
u_short xor_4003;
|
1994-08-19 00:01:08 +00:00
|
|
|
} s_pas[NPAS];
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Register Access-macros use these
|
|
|
|
*/
|
|
|
|
#define r_csdr(ptr) (ptr->xor_0000)
|
|
|
|
#define w_odr(ptr) (ptr->xor_0000)
|
|
|
|
#define r_icr(ptr) (ptr->xor_0001)
|
|
|
|
#define w_icr(ptr) (ptr->xor_0001)
|
|
|
|
#define rw_mr(ptr) (ptr->xor_0002)
|
|
|
|
#define r_tcr(ptr) (ptr->xor_0003)
|
|
|
|
#define w_tcr(ptr) (ptr->xor_0003)
|
|
|
|
#define r_cscr(ptr) (ptr->xor_2000)
|
|
|
|
#define w_ser(ptr) (ptr->xor_2000)
|
|
|
|
#define r_bsr(ptr) (ptr->xor_2001)
|
|
|
|
#define w_sdsr(ptr) (ptr->xor_2001)
|
|
|
|
#define r_idr(ptr) (ptr->xor_2002)
|
|
|
|
#define w_sdtr(ptr) (ptr->xor_2002)
|
|
|
|
#define r_rpir(ptr) (ptr->xor_2003)
|
|
|
|
#define w_sdir(ptr) (ptr->xor_2003)
|
|
|
|
#define pas_data(ptr) (ptr->xor_4000)
|
|
|
|
#define pas_stat(ptr) (ptr->xor_4001)
|
|
|
|
#define R_PAS_STAT_DREQ 0x80
|
|
|
|
#define pas_irq(ptr) (ptr->xor_4003)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* The actual macros used to access the chip
|
|
|
|
*/
|
1994-08-19 00:01:08 +00:00
|
|
|
#define P_PAS struct pas_softc *
|
|
|
|
#define R_PAS(ptr,foo) inb(foo(ptr))
|
|
|
|
#define W_PAS(ptr,foo,val) outb(foo(ptr),val)
|
|
|
|
#define M_PAS(ptr,foo,opr,arg) outb(foo(ptr),R_PAS(ptr,foo) opr arg)
|
|
|
|
|
|
|
|
#define WAIT_FOR_NOT_REQ(ptr) { \
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
int scsi_timeout = PAS16_TIMEOUT; \
|
|
|
|
while ( (R_PAS(ptr,r_cscr) & R_CSCR_REQ) && \
|
|
|
|
(R_PAS(ptr,r_cscr) & R_CSCR_REQ) && \
|
|
|
|
(R_PAS(ptr,r_cscr) & R_CSCR_REQ) && \
|
1994-08-19 00:01:08 +00:00
|
|
|
(--scsi_timeout) ); \
|
|
|
|
if (!scsi_timeout) { \
|
|
|
|
printf("scsi timeout--WAIT_FOR_NOT_REQ-- pas.c:%d.\n", __LINE__); \
|
|
|
|
goto scsi_timeout_error; \
|
|
|
|
} \
|
|
|
|
}
|
|
|
|
#define WAIT_FOR_REQ(ptr) { \
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
int scsi_timeout = PAS16_TIMEOUT; \
|
|
|
|
while ( ((R_PAS(ptr,r_cscr) & R_CSCR_REQ) == 0) && \
|
|
|
|
((R_PAS(ptr,r_cscr) & R_CSCR_REQ) == 0) && \
|
|
|
|
((R_PAS(ptr,r_cscr) & R_CSCR_REQ) == 0) && \
|
1994-08-19 00:01:08 +00:00
|
|
|
(--scsi_timeout) ); \
|
|
|
|
if (!scsi_timeout) { \
|
|
|
|
printf("scsi timeout--WAIT_FOR_REQ-- pas.c:%d.\n", __LINE__); \
|
|
|
|
goto scsi_timeout_error; \
|
|
|
|
} \
|
|
|
|
}
|
|
|
|
|
|
|
|
static u_int32 pas_adapter_info(int adapter_number);
|
|
|
|
static void pas_minphys(struct buf *bp);
|
|
|
|
static int32 pas_scsi_cmd(struct scsi_xfer *xs);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
static int pas_reset(int);
|
1994-08-19 00:01:08 +00:00
|
|
|
static int pas_send_cmd(struct scsi_xfer *xs);
|
|
|
|
static int scsi_req(P_PAS,int,int,u_char *,int,
|
|
|
|
u_char *, int , int *, int *);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
static int sci_data_out(P_PAS, int, int, u_char *);
|
|
|
|
static int sci_data_in(P_PAS, int, int, u_char *);
|
|
|
|
static int select_target(P_PAS, int);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
extern int pasprobe(struct isa_device *dev);
|
|
|
|
extern int pasattach(struct isa_device *dev);
|
|
|
|
struct isa_driver pasdriver = { pasprobe, pasattach, "pas" };
|
|
|
|
|
|
|
|
|
|
|
|
static
|
|
|
|
struct scsi_adapter pas_adapter = {
|
|
|
|
pas_scsi_cmd, /* scsi_cmd() */
|
|
|
|
pas_minphys, /* scsi_minphys() */
|
|
|
|
0, /* open_target_lu() */
|
|
|
|
0, /* close_target_lu() */
|
|
|
|
pas_adapter_info, /* adapter_info() */
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
"pas", /* name */
|
1994-08-19 00:01:08 +00:00
|
|
|
{ 0, 0 } /* spare[2] */
|
|
|
|
};
|
|
|
|
|
|
|
|
static
|
|
|
|
struct scsi_device pas_dev = {
|
|
|
|
NULL, /* Use default error handler. */
|
|
|
|
NULL, /* have a queue, served by this (?) */
|
|
|
|
NULL, /* have no async handler. */
|
|
|
|
NULL, /* Use default "done" routine. */
|
|
|
|
"pas",
|
|
|
|
0,
|
|
|
|
{ 0, 0 }
|
|
|
|
};
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
static char *mv_type[] = {
|
|
|
|
0,0,0,0,0,0,"PAS+",0,0,0,0,0,"PAS16D",0,"CDPC","PAS16"};
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
int
|
|
|
|
pasprobe(struct isa_device *dev)
|
|
|
|
{
|
|
|
|
int port = dev->id_iobase;
|
|
|
|
int base = port - 0x1c00;
|
|
|
|
int unit = dev->id_unit;
|
|
|
|
int i, j;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if(base & 3) {
|
|
|
|
printf("pas%d cannot operate on address %x.", unit,port);
|
|
|
|
printf(" Address must be a multiple of four.\n");
|
|
|
|
return 0;
|
|
|
|
}
|
1994-08-19 00:01:08 +00:00
|
|
|
/* Tell the PAS16 we want to talk to, where to listen */
|
|
|
|
outb(0x9a01,0xbc + unit);
|
|
|
|
outb(0x9a01,base >> 2);
|
|
|
|
|
1994-09-13 06:44:39 +00:00
|
|
|
#ifdef PAS_DEBUG
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
printf("%x: 0x803=%x 0xec03=%x 0xff88=%x\n",
|
|
|
|
base,inb(base^0x803),inb(base^0xec03),inb(base^0xfc00));
|
1994-09-13 06:44:39 +00:00
|
|
|
#endif
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
/* Killer one */
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
i = inb(base ^ 0x803);
|
1994-08-19 00:01:08 +00:00
|
|
|
if(i == 0xff) return 0;
|
|
|
|
|
|
|
|
/* killer two */
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if((0x03 & inb(base^0xec03)) != 0x03) return 0;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/* killer three */
|
1994-08-19 00:01:08 +00:00
|
|
|
for(i=0;i<4;i++) {
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if(inb(port) != 0xff || inb(port^0x2000) != 0xff)
|
|
|
|
goto ok;
|
1994-08-19 00:01:08 +00:00
|
|
|
}
|
|
|
|
return 0;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
ok:
|
|
|
|
/* killer four */
|
|
|
|
i = inb(base ^ 0x803);
|
|
|
|
outb(base^0x803,i ^ 0xe0);
|
|
|
|
j = inb(base^0x803);
|
|
|
|
outb(base^0x803,1);
|
|
|
|
if(i != j) return 0;
|
|
|
|
|
1994-09-13 06:44:39 +00:00
|
|
|
return 4;
|
1994-08-19 00:01:08 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static u_int32
|
|
|
|
pas_adapter_info(int adapter_number)
|
|
|
|
{
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
1994-10-23 21:28:03 +00:00
|
|
|
static struct kern_devconf kdc_pas[NPAS] = { {
|
|
|
|
0, 0, 0, /* filled in by dev_attach */
|
|
|
|
"pas", 0, { MDDT_ISA, 0, "bio" },
|
|
|
|
isa_generic_externalize, 0, 0, ISA_EXTERNALLEN,
|
|
|
|
&kdc_isa0, /* parent */
|
|
|
|
0, /* parentdata */
|
|
|
|
DC_BUSY, /* host adapters are always busy */
|
|
|
|
"Media Vision ProAudioSpectrum SCSI host adapter"
|
|
|
|
} };
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
pas_registerdev(struct isa_device *id)
|
|
|
|
{
|
|
|
|
if(id->id_unit)
|
|
|
|
kdc_pas[id->id_unit] = kdc_pas[0];
|
|
|
|
kdc_pas[id->id_unit].kdc_unit = id->id_unit;
|
|
|
|
kdc_pas[id->id_unit].kdc_parentdata = id;
|
|
|
|
dev_attach(&kdc_pas[id->id_unit]);
|
|
|
|
}
|
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
int
|
|
|
|
pasattach(struct isa_device *dev)
|
|
|
|
{
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
int i,base=dev->id_iobase-0x1c00;
|
1994-08-19 00:01:08 +00:00
|
|
|
struct pas_softc *ppas;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
i = inb(base^0xEC03) & 0x0f;
|
|
|
|
printf("pas%d: Type = %d <%s>\n",dev->id_unit,i,mv_type[i]);
|
1994-08-19 00:01:08 +00:00
|
|
|
ppas = s_pas + dev->id_unit;
|
|
|
|
ppas->sc_link.adapter_unit = dev->id_unit;
|
|
|
|
ppas->sc_link.adapter_targ = 7;
|
|
|
|
ppas->sc_link.adapter = &pas_adapter;
|
|
|
|
ppas->sc_link.device = &pas_dev;
|
|
|
|
ppas->iobase=dev->id_iobase;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
ppas->xor_0001 = ppas->iobase^0x0001;
|
|
|
|
ppas->xor_0002 = ppas->iobase^0x0002;
|
|
|
|
ppas->xor_0003 = ppas->iobase^0x0003;
|
|
|
|
ppas->xor_2000 = ppas->iobase^0x2000;
|
|
|
|
ppas->xor_2001 = ppas->iobase^0x2001;
|
|
|
|
ppas->xor_2002 = ppas->iobase^0x2002;
|
|
|
|
ppas->xor_2003 = ppas->iobase^0x2003;
|
|
|
|
ppas->xor_4000 = ppas->iobase^0x4000;
|
|
|
|
ppas->xor_4001 = ppas->iobase^0x4001;
|
|
|
|
ppas->xor_4003 = ppas->iobase^0x4003;
|
|
|
|
|
|
|
|
/* Various magic */
|
|
|
|
outb(base^0x4000,0x30);
|
|
|
|
outb(base^0x4001,0x01);
|
|
|
|
outb(base^0xbc00,0x01);
|
|
|
|
outb(base^0x8003,0x4d);
|
|
|
|
|
1994-10-23 21:28:03 +00:00
|
|
|
pas_registerdev(dev);
|
1994-08-19 00:01:08 +00:00
|
|
|
scsi_attachdevs(&(ppas->sc_link));
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
pas_reset(dev->id_unit);
|
1994-08-19 00:01:08 +00:00
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
pas_minphys(struct buf *bp)
|
|
|
|
{
|
|
|
|
#define MIN_PHYS 65536 /*BARF!!!!*/
|
|
|
|
if (bp->b_bcount > MIN_PHYS) {
|
|
|
|
printf("Uh-oh... pas_minphys setting bp->b_bcount = %x.\n", MIN_PHYS);
|
|
|
|
bp->b_bcount = MIN_PHYS;
|
|
|
|
}
|
|
|
|
#undef MIN_PHYS
|
|
|
|
}
|
|
|
|
|
|
|
|
static int32
|
|
|
|
pas_scsi_cmd(struct scsi_xfer *xs)
|
|
|
|
{
|
|
|
|
int flags, s, r;
|
|
|
|
|
|
|
|
flags = xs->flags;
|
|
|
|
if (xs->bp) flags |= (SCSI_NOSLEEP);
|
|
|
|
if ( flags & ITSDONE ) {
|
|
|
|
printf("Already done?");
|
|
|
|
xs->flags &= ~ITSDONE;
|
|
|
|
}
|
|
|
|
if ( ! ( flags & INUSE ) ) {
|
|
|
|
printf("Not in use?");
|
|
|
|
xs->flags |= INUSE;
|
|
|
|
}
|
|
|
|
|
|
|
|
if ( flags & SCSI_RESET ) {
|
|
|
|
printf("flags & SCSIRESET.\n");
|
|
|
|
if ( ! ( flags & SCSI_NOSLEEP ) ) {
|
|
|
|
s = splbio();
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
pas_reset(xs->sc_link->adapter_unit);
|
1994-08-19 00:01:08 +00:00
|
|
|
splx(s);
|
|
|
|
return(SUCCESSFULLY_QUEUED);
|
|
|
|
} else {
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
pas_reset(xs->sc_link->adapter_unit);
|
1994-08-19 00:01:08 +00:00
|
|
|
return (COMPLETE);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
r = pas_send_cmd(xs);
|
|
|
|
xs->flags |= ITSDONE;
|
|
|
|
scsi_done(xs);
|
|
|
|
if (xs->flags&SCSI_NOMASK) {
|
|
|
|
return (r);
|
|
|
|
}
|
|
|
|
return SUCCESSFULLY_QUEUED;
|
|
|
|
}
|
|
|
|
|
|
|
|
#ifdef PAS_DEBUG
|
|
|
|
static int
|
|
|
|
pas_show_scsi_cmd(struct scsi_xfer *xs)
|
|
|
|
{
|
|
|
|
u_char *b = (u_char *) xs->cmd;
|
|
|
|
int i = 0;
|
|
|
|
|
|
|
|
if ( ! ( xs->flags & SCSI_RESET ) ) {
|
|
|
|
printf("pas(%d:%d:%d) ",
|
|
|
|
xs->sc_link->adapter_unit, xs->sc_link->target, xs->sc_link->lun);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
printf("%d@%x, %d@%x <",xs->cmdlen,(u_long)xs->cmd,
|
|
|
|
xs->datalen,(u_long)xs->data);
|
1994-08-19 00:01:08 +00:00
|
|
|
while (i < xs->cmdlen) {
|
|
|
|
if (i) printf(",");
|
|
|
|
printf("%x",b[i++]);
|
|
|
|
}
|
|
|
|
printf(">\n");
|
|
|
|
} else {
|
|
|
|
printf("pas(%d:%d:%d)-RESET-\n",
|
|
|
|
xs->sc_link->adapter_unit, xs->sc_link->target,
|
|
|
|
xs->sc_link->lun);
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
#endif /* PAS_DEBUG */
|
|
|
|
|
|
|
|
static int
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
pas_reset(int adapter)
|
1994-08-19 00:01:08 +00:00
|
|
|
{
|
|
|
|
struct pas_softc *ppas = s_pas + adapter;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Reset Hold Time is 25 uSec
|
|
|
|
*/
|
|
|
|
W_PAS(ppas,w_icr,W_ICR_ASSERT_RST);
|
|
|
|
DELAY(25);
|
|
|
|
W_PAS(ppas,w_icr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ppas,rw_mr,0);
|
|
|
|
W_PAS(ppas,w_ser,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
pas_send_cmd(struct scsi_xfer *xs)
|
|
|
|
{
|
|
|
|
P_PAS ptr = s_pas + xs->sc_link->adapter_unit;
|
|
|
|
int s,sent,ret;
|
|
|
|
int sense;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
u_char cmd[6];
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
#ifdef PAS_DEBUG
|
|
|
|
pas_show_scsi_cmd(xs);
|
|
|
|
#endif /* PAS_DEBUG */
|
|
|
|
s = splbio();
|
|
|
|
sense = scsi_req(ptr,
|
|
|
|
xs->sc_link->target, xs->sc_link->lun,
|
|
|
|
(u_char*)xs->cmd, xs->cmdlen, xs->data, xs->datalen,
|
|
|
|
&sent, &ret);
|
|
|
|
splx(s);
|
|
|
|
switch (sense) {
|
|
|
|
case 0x00:
|
|
|
|
xs->error = XS_NOERROR;
|
|
|
|
xs->resid = sent;
|
|
|
|
return (COMPLETE);
|
|
|
|
case 0x02: /* Check condition */
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
|
|
|
#ifdef PAS_DEBUG_REQUEST_SENSE
|
|
|
|
printf("pas%d, target%d: sent=%d,ret=%d,sense=%x check. cond.\n",
|
|
|
|
xs->sc_link->adapter_unit, xs->sc_link->target,sent,ret,sense);
|
|
|
|
printf("Cmd:<%x",((u_char*)(xs->cmd))[0]);
|
|
|
|
for(sent=1;sent < sizeof xs->cmdlen;sent++) {
|
|
|
|
printf(",%x",((u_char*)(xs->cmd))[sent]);
|
|
|
|
}
|
|
|
|
printf(">\n");
|
|
|
|
#endif /* PAS_DEBUG_REQUEST_SENSE */
|
|
|
|
cmd[0] = 3;
|
|
|
|
cmd[1] = xs->sc_link->lun<<5;
|
|
|
|
cmd[2] = 0;
|
|
|
|
cmd[3] = 0;
|
|
|
|
cmd[4] = sizeof xs->sense;
|
|
|
|
cmd[5] = 0;
|
|
|
|
|
|
|
|
DELAY(10);
|
1994-08-19 00:01:08 +00:00
|
|
|
s = splbio();
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
scsi_req(ptr, xs->sc_link->target, xs->sc_link->lun,
|
|
|
|
cmd, sizeof cmd,
|
|
|
|
(caddr_t) &(xs->sense), sizeof xs->sense,
|
|
|
|
&sent, &ret);
|
1994-08-19 00:01:08 +00:00
|
|
|
splx(s);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
#ifdef PAS_DEBUG_REQUEST_SENSE
|
|
|
|
printf("Sense result: (sent=%d, ret=%d)\n",sent,ret);
|
|
|
|
printf("<%x",((u_char*)(&xs->sense))[0]);
|
|
|
|
for(sent=1;sent < sizeof xs->sense;sent++) {
|
|
|
|
printf(",%x",((u_char*)(&xs->sense))[sent]);
|
|
|
|
}
|
|
|
|
printf(">\n");
|
|
|
|
#endif /* PAS_DEBUG_REQUEST_SENSE */
|
1994-08-19 00:01:08 +00:00
|
|
|
xs->error = XS_SENSE;
|
|
|
|
return HAD_ERROR;
|
|
|
|
case 0x08: /* Busy */
|
|
|
|
xs->error = XS_BUSY;
|
|
|
|
return HAD_ERROR;
|
|
|
|
default:
|
|
|
|
xs->error = XS_DRIVER_STUFFUP;
|
|
|
|
return HAD_ERROR;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Perform arbitration and selection. Figure 5-1, more or less.
|
|
|
|
*/
|
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
static int
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
select_target(P_PAS ptr, int tid)
|
1994-08-19 00:01:08 +00:00
|
|
|
{
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
int myid = ptr->sc_link.adapter_targ;
|
|
|
|
int ret = SCSI_RET_RETRY;
|
|
|
|
int tries = ATTEMPT_SELECTION;
|
|
|
|
int delay = SELECTION_DELAY;
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Convert the ID's to bit-maps.
|
|
|
|
*/
|
1994-08-19 00:01:08 +00:00
|
|
|
myid = 1 << myid;
|
|
|
|
tid = 1 << tid;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Not documented, but clearly needed for now.
|
|
|
|
* XXX is this missing somewhere else, and this is just a hack ?
|
|
|
|
*/
|
|
|
|
loop:
|
|
|
|
W_PAS(ptr,w_tcr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Write ID Bit to Output Register.
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_odr,myid);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Set "ARBITRATE" Bit.
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,rw_mr,RW_MR_ARBITRATE);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Check "Arbitration in progress" Bit.
|
|
|
|
*/
|
|
|
|
for (i = 0; i < 20; i++) /* 20usec circa */
|
|
|
|
if (R_PAS(ptr,r_icr) & R_ICR_ARBITRATION_IN_PROGRESS)
|
|
|
|
goto aip;
|
|
|
|
goto lost;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
aip:
|
|
|
|
/*
|
|
|
|
* Wait 2.2 usec Arbitration Delay.
|
|
|
|
*/
|
|
|
|
DELAY(2);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Check "Lost Arbitration" Bit.
|
|
|
|
*/
|
|
|
|
if (R_PAS(ptr,r_icr) & R_ICR_LOST_ARBITRATION)
|
1994-08-19 00:01:08 +00:00
|
|
|
goto lost;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
|
|
|
/*
|
|
|
|
* Higher priority ID present ?
|
|
|
|
*/
|
|
|
|
if ((R_PAS(ptr,r_csdr) & ~myid) > myid)
|
1994-08-19 00:01:08 +00:00
|
|
|
goto lost;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Check "Lost Arbitration" Bit.
|
|
|
|
*/
|
|
|
|
if (R_PAS(ptr,r_icr) & R_ICR_LOST_ARBITRATION)
|
|
|
|
goto lost;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Set "Assert SEL/".
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_icr,W_ICR_ASSERT_SEL);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Check "Lost Arbitration" Bit.
|
|
|
|
*/
|
|
|
|
if (R_PAS(ptr,r_icr) & R_ICR_LOST_ARBITRATION) {
|
1994-08-19 00:01:08 +00:00
|
|
|
goto nosel;
|
|
|
|
}
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Wait 1.2 usec minimum. (Bus Clear + Settle)
|
|
|
|
*/
|
|
|
|
DELAY(2);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Write Target and Initiator's ID bits to output Register.
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_odr,myid|tid);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Set "Assert BSY/" + "Assert Data Bus".
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_icr,
|
|
|
|
W_ICR_ASSERT_SEL | W_ICR_ASSERT_BSY | W_ICR_ASSERT_DATA_BUS);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Reset "ARBITRATE" Bit.
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,rw_mr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Reset "Select Enable" Register.
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_ser,0);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Reset "Assert BSY/".
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_icr,
|
|
|
|
W_ICR_ASSERT_SEL | W_ICR_ASSERT_DATA_BUS);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* "BSY/" Asserted within 250 msec ?
|
|
|
|
*/
|
|
|
|
for(i=0;i<250000;i+=100) {
|
|
|
|
if (R_PAS(ptr,r_cscr) & R_CSCR_BSY)
|
|
|
|
goto resp;
|
|
|
|
DELAY(100);
|
1994-08-19 00:01:08 +00:00
|
|
|
}
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
goto nodev;
|
|
|
|
resp:
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Reset "Assert SEL/" + "Assert Data Bus".
|
|
|
|
*/
|
|
|
|
W_PAS(ptr,w_icr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
return SCSI_RET_SUCCESS;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
nodev:
|
|
|
|
ret = SCSI_RET_DEVICE_DOWN;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
nosel:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_icr,0);
|
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
lost:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,rw_mr,0);
|
|
|
|
if(--tries) {
|
|
|
|
DELAY(delay);
|
|
|
|
goto loop;
|
|
|
|
}
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
/*
|
|
|
|
* Figure 5-2.
|
|
|
|
*/
|
1994-10-18 03:53:12 +00:00
|
|
|
|
1994-08-19 00:01:08 +00:00
|
|
|
static int
|
|
|
|
sci_data_out(P_PAS ptr, int phase, int count, u_char *data)
|
|
|
|
{
|
|
|
|
register unsigned char icmd;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
register int cnt=0,zonk;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if (SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
|
|
|
return cnt;
|
|
|
|
|
|
|
|
icmd = R_PAS(ptr,r_icr) & RW_ICR_MASK;
|
|
|
|
|
|
|
|
#ifndef NO_PAS16_PSEUDO_DMA
|
|
|
|
/*
|
|
|
|
* The PAS16 has special provisions for doing pseudo-dma.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Don't use pseudo-dma unless we need much data. This way we avoid all
|
|
|
|
* the variable length stuff entirely.
|
|
|
|
*
|
|
|
|
* XXX something screws up for big transfers :-(
|
|
|
|
*/
|
|
|
|
|
|
|
|
if(count >= 128 && count <= 8192) {
|
|
|
|
|
|
|
|
M_PAS(ptr,rw_mr,|,RW_MR_DMA_MODE);
|
|
|
|
|
|
|
|
W_PAS(ptr,w_icr,icmd | W_ICR_ASSERT_DATA_BUS);
|
|
|
|
|
|
|
|
W_PAS(ptr,w_sdsr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
while(count >= PAS16_PSEUDO_DMA_SIZE) {
|
|
|
|
zonk=0;
|
|
|
|
while(!(R_PAS(ptr,pas_stat) & R_PAS_STAT_DREQ))
|
|
|
|
if (++zonk > 1000 || SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
|
|
|
return cnt;
|
|
|
|
|
|
|
|
outsb(pas_data(ptr),data,PAS16_PSEUDO_DMA_SIZE);
|
|
|
|
|
|
|
|
data += PAS16_PSEUDO_DMA_SIZE;
|
|
|
|
cnt += PAS16_PSEUDO_DMA_SIZE;
|
|
|
|
count -= PAS16_PSEUDO_DMA_SIZE;
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
if(count) {
|
|
|
|
while(!(R_PAS(ptr,pas_stat) & R_PAS_STAT_DREQ))
|
|
|
|
if (SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
|
|
|
return cnt;
|
|
|
|
outsb(pas_data(ptr),data,count);
|
|
|
|
cnt += count;
|
|
|
|
data += count;
|
|
|
|
}
|
|
|
|
|
|
|
|
M_PAS(ptr,rw_mr,&,~RW_MR_DMA_MODE);
|
|
|
|
W_PAS(ptr,w_icr,icmd);
|
|
|
|
return cnt;
|
|
|
|
}
|
|
|
|
#endif /* NO_PAS16_PSEUDO_DMA */
|
1994-08-19 00:01:08 +00:00
|
|
|
loop:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if (SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
1994-08-19 00:01:08 +00:00
|
|
|
return cnt;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_odr, *data);
|
1994-08-19 00:01:08 +00:00
|
|
|
WAIT_FOR_REQ(ptr);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
icmd |= W_ICR_ASSERT_DATA_BUS;
|
|
|
|
W_PAS(ptr,w_icr,icmd);
|
|
|
|
icmd |= W_ICR_ASSERT_ACK;
|
|
|
|
W_PAS(ptr,w_icr,icmd);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
icmd &= ~(W_ICR_ASSERT_DATA_BUS|W_ICR_ASSERT_ACK);
|
1994-08-19 00:01:08 +00:00
|
|
|
WAIT_FOR_NOT_REQ(ptr);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_icr,icmd);
|
1994-08-19 00:01:08 +00:00
|
|
|
++cnt;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
data++;
|
1994-08-19 00:01:08 +00:00
|
|
|
if (--count > 0)
|
|
|
|
goto loop;
|
|
|
|
scsi_timeout_error:
|
|
|
|
return cnt;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
sci_data_in(P_PAS ptr, int phase, int count, u_char *data)
|
|
|
|
{
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
unsigned char icmd;
|
|
|
|
int cnt=0;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if (SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
|
|
|
return cnt;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
#ifndef NO_PAS16_PSEUDO_DMA
|
|
|
|
/*
|
|
|
|
* The PAS16 has special provisions for doing pseudo-dma.
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Don't use pseudo-dma unless we need much data. This way we avoid all
|
|
|
|
* the variable length stuff entirely.
|
|
|
|
*/
|
|
|
|
if(count >= 128) {
|
|
|
|
|
|
|
|
M_PAS(ptr,rw_mr,|,RW_MR_DMA_MODE);
|
|
|
|
|
|
|
|
W_PAS(ptr,w_sdir,0);
|
|
|
|
|
|
|
|
while(count > PAS16_PSEUDO_DMA_SIZE) {
|
|
|
|
while(!(R_PAS(ptr,pas_stat) & R_PAS_STAT_DREQ)) ;
|
|
|
|
|
|
|
|
insb(pas_data(ptr),data,PAS16_PSEUDO_DMA_SIZE);
|
|
|
|
|
|
|
|
data += PAS16_PSEUDO_DMA_SIZE;
|
|
|
|
cnt += PAS16_PSEUDO_DMA_SIZE;
|
|
|
|
count -= PAS16_PSEUDO_DMA_SIZE;
|
|
|
|
|
|
|
|
if (SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
|
|
|
return cnt;
|
|
|
|
}
|
|
|
|
|
|
|
|
while(!(R_PAS(ptr,pas_stat) & R_PAS_STAT_DREQ)) ;
|
|
|
|
|
|
|
|
insb(pas_data(ptr),data,count-1);
|
|
|
|
cnt += count-1;
|
|
|
|
data += count-1;
|
|
|
|
|
|
|
|
while(!(R_PAS(ptr,pas_stat) & R_PAS_STAT_DREQ)) ;
|
|
|
|
while((R_PAS(ptr,r_cscr) & R_CSCR_REQ));
|
|
|
|
*data = R_PAS(ptr,r_idr);
|
|
|
|
cnt++;
|
|
|
|
M_PAS(ptr,rw_mr,&,~RW_MR_DMA_MODE);
|
|
|
|
return cnt;
|
|
|
|
}
|
|
|
|
#endif /* NO_PAS16_PSEUDO_DMA */
|
|
|
|
|
|
|
|
icmd = R_PAS(ptr,r_icr) & RW_ICR_MASK;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
loop:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if (SCSI_CUR_PHASE(R_PAS(ptr,r_cscr)) != phase)
|
1994-08-19 00:01:08 +00:00
|
|
|
return cnt;
|
|
|
|
|
|
|
|
WAIT_FOR_REQ(ptr);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
*data++ = R_PAS(ptr,r_csdr);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
icmd |= W_ICR_ASSERT_ACK;
|
|
|
|
W_PAS(ptr,w_icr,icmd);
|
|
|
|
icmd &= ~W_ICR_ASSERT_ACK;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
WAIT_FOR_NOT_REQ(ptr);
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_icr,icmd);
|
1994-08-19 00:01:08 +00:00
|
|
|
++cnt;
|
|
|
|
if (--count > 0)
|
|
|
|
goto loop;
|
|
|
|
|
|
|
|
scsi_timeout_error:
|
|
|
|
return cnt;
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
static int
|
|
|
|
cmd_xfer(P_PAS ptr, int maxlen, u_char *data, u_char *status, u_char *msg)
|
|
|
|
{
|
|
|
|
int xfer=0, phase;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_icr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
while (1) {
|
|
|
|
|
|
|
|
WAIT_FOR_REQ(ptr);
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
phase = SCSI_CUR_PHASE(R_PAS(ptr,r_cscr));
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
switch (phase) {
|
|
|
|
case SCSI_PHASE_CMD:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_CMD);
|
1994-08-19 00:01:08 +00:00
|
|
|
xfer += sci_data_out(ptr, SCSI_PHASE_CMD, maxlen, data);
|
|
|
|
return xfer;
|
|
|
|
case SCSI_PHASE_DATA_IN:
|
|
|
|
printf("Data in phase in cmd_xfer?\n");
|
|
|
|
return 0;
|
|
|
|
case SCSI_PHASE_DATA_OUT:
|
|
|
|
printf("Data out phase in cmd_xfer?\n");
|
|
|
|
return 0;
|
|
|
|
case SCSI_PHASE_STATUS:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_STATUS);
|
1994-08-19 00:01:08 +00:00
|
|
|
printf("status in cmd_xfer.\n");
|
|
|
|
sci_data_in(ptr, SCSI_PHASE_STATUS, 1, status);
|
|
|
|
break;
|
|
|
|
case SCSI_PHASE_MESSAGE_IN:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_MESSAGE_IN);
|
1994-08-19 00:01:08 +00:00
|
|
|
printf("msgin in cmd_xfer.\n");
|
|
|
|
sci_data_in(ptr, SCSI_PHASE_MESSAGE_IN, 1, msg);
|
|
|
|
break;
|
|
|
|
case SCSI_PHASE_MESSAGE_OUT:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_MESSAGE_OUT);
|
1994-08-19 00:01:08 +00:00
|
|
|
sci_data_out(ptr, SCSI_PHASE_MESSAGE_OUT, 1, msg);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
printf("Unexpected phase 0x%x in cmd_xfer()\n", phase);
|
|
|
|
scsi_timeout_error:
|
|
|
|
return xfer;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
data_xfer(P_PAS ptr, int maxlen, u_char *data, u_char *status, u_char *msg)
|
|
|
|
{
|
|
|
|
int retlen = 0, xfer, phase;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_icr,0);
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
*status = 0;
|
|
|
|
|
|
|
|
while (1) {
|
|
|
|
|
|
|
|
WAIT_FOR_REQ(ptr);
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
phase = SCSI_CUR_PHASE(R_PAS(ptr,r_cscr));
|
1994-08-19 00:01:08 +00:00
|
|
|
switch (phase) {
|
|
|
|
case SCSI_PHASE_CMD:
|
|
|
|
printf("Command phase in data_xfer().\n");
|
|
|
|
return retlen;
|
|
|
|
case SCSI_PHASE_DATA_IN:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_DATA_IN);
|
|
|
|
xfer = sci_data_in (ptr, SCSI_PHASE_DATA_IN, maxlen, data);
|
1994-08-19 00:01:08 +00:00
|
|
|
retlen += xfer;
|
|
|
|
maxlen -= xfer;
|
|
|
|
break;
|
|
|
|
case SCSI_PHASE_DATA_OUT:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_DATA_OUT);
|
|
|
|
xfer = sci_data_out (ptr, SCSI_PHASE_DATA_OUT, maxlen, data);
|
1994-08-19 00:01:08 +00:00
|
|
|
retlen += xfer;
|
|
|
|
maxlen -= xfer;
|
|
|
|
break;
|
|
|
|
case SCSI_PHASE_STATUS:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_STATUS);
|
1994-08-19 00:01:08 +00:00
|
|
|
sci_data_in(ptr, SCSI_PHASE_STATUS, 1, status);
|
|
|
|
break;
|
|
|
|
case SCSI_PHASE_MESSAGE_IN:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_MESSAGE_IN);
|
1994-08-19 00:01:08 +00:00
|
|
|
sci_data_in(ptr, SCSI_PHASE_MESSAGE_IN, 1, msg);
|
|
|
|
if (*msg == 0) {
|
|
|
|
return retlen;
|
|
|
|
} else {
|
|
|
|
printf( "message 0x%x in data_xfer.\n", *msg);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case SCSI_PHASE_MESSAGE_OUT:
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
W_PAS(ptr,w_tcr,SCSI_PHASE_MESSAGE_OUT);
|
1994-08-19 00:01:08 +00:00
|
|
|
sci_data_out(ptr, SCSI_PHASE_MESSAGE_OUT, 1, msg);
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
printf( "Unexpected phase 0x%x in data_xfer().\n",
|
|
|
|
phase);
|
|
|
|
scsi_timeout_error:
|
|
|
|
return retlen;
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int
|
|
|
|
scsi_req(P_PAS ptr, int target, int lun, u_char *cmd, int cmdlen,
|
|
|
|
u_char *databuf, int datalen, int *sent, int *ret)
|
|
|
|
{
|
|
|
|
/* Returns 0 on success, -1 on internal error, or the status byte */
|
|
|
|
int cmd_bytes_sent, r;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
u_char stat, msg, c;
|
1994-08-19 00:01:08 +00:00
|
|
|
|
|
|
|
*sent = 0;
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if ( ( r = select_target(ptr, target) ) != SCSI_RET_SUCCESS) {
|
|
|
|
*ret = r;
|
1994-08-19 00:01:08 +00:00
|
|
|
switch (r) {
|
|
|
|
case SCSI_RET_RETRY:
|
|
|
|
return 0x08;
|
|
|
|
default:
|
|
|
|
printf("select_target(target %d, lun %d) failed(%d).\n",
|
|
|
|
target, lun, r);
|
|
|
|
case SCSI_RET_DEVICE_DOWN:
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
c = 0x80 | lun;
|
|
|
|
|
ProAudioSpectrum 16 {Zilog|Ncr} 5380 based SCSI-driver.
This puppy is in good shape now.
It is a fully blown SCSI-driver, but it isn't a high performance one. It is
implemented entirely with polled I/O, and is intended to drive CD-ROM's, not
disks and tapes. It will run disks and tapes if asked to, but it isn't a
very good idea to do so. Transfer-rates max out at 600-700 kbyte/sec.
There is one problem: when write-requests get over 8192 bytes, the pseudo-DMA
stalls. This is only a problem if you dd(1) to a raw-device of some kind,
for mounting a disk it is ok. I have circumvented this by disabling the
pseudo-DMA in those cases.
It's very unlikely that I will spend more time on improving the performance
of this driver, it can do what I want it to now: install from a CD-ROM, and I
don't see any benefit in actually adding interrupts to the driver, considering
that performance never will be better than 700 kbyte/sec anyway.
You can install it under 1.1.5 too, by adding the lines to files.i386, your
config-file and copying pas.c and ic/ncr_5380.h over.
I will fix any bugs I can get a handle on.
Poul-Henning
1994-09-11 20:28:50 +00:00
|
|
|
if ((cmd_bytes_sent = cmd_xfer(ptr, cmdlen, cmd, &stat, &c)) != cmdlen) {
|
1994-08-19 00:01:08 +00:00
|
|
|
*ret = SCSI_RET_COMMAND_FAIL;
|
|
|
|
printf("Data underrun sending CCB (%d bytes of %d, sent).\n",
|
|
|
|
cmd_bytes_sent, cmdlen);
|
|
|
|
return -1;
|
|
|
|
}
|
|
|
|
|
|
|
|
*sent=data_xfer(ptr, datalen, databuf, &stat, &msg);
|
|
|
|
|
|
|
|
*ret = 0;
|
|
|
|
return stat;
|
|
|
|
}
|
|
|
|
|
|
|
|
#endif /* NPAS */
|