freebsd-nq/sys/dev/aic7xxx/aic7xxx_reg.h

777 lines
21 KiB
C
Raw Normal View History

/*
* Aic7xxx register and scratch ram definitions.
*
* Copyright (c) 1994, 1995, 1996 Justin T. Gibbs.
* 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 immediately at the beginning of the file, without modification,
* 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. The name of the author may not be used to endorse or promote products
* derived from this software without specific prior written permission.
*
* THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
* ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
* IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
* ARE DISCLAIMED. IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
* ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
* DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
* OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
* HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
* LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
* OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
* SUCH DAMAGE.
*
* $Id: aic7xxx_reg.h,v 1.8 1996/04/20 21:20:31 gibbs Exp $
*/
/*
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* This header is shared by the sequencer code and the kernel level driver.
*
* All page numbers refer to the Adaptec AIC-7770 Data Book availible from
* Adaptec's Technical Documents Department 1-800-934-2766
*/
/*
* SCSI Sequence Control (p. 3-11).
* Each bit, when set starts a specific SCSI sequence on the bus
*/
#define SCSISEQ 0x000
#define TEMODEO 0x80
#define ENSELO 0x40
#define ENSELI 0x20
#define ENRSELI 0x10
#define ENAUTOATNO 0x08
#define ENAUTOATNI 0x04
#define ENAUTOATNP 0x02
#define SCSIRSTO 0x01
/*
* SCSI Transfer Control 0 Register (pp. 3-13).
* Controls the SCSI module data path.
*/
#define SXFRCTL0 0x001
#define DFON 0x80
#define DFPEXP 0x40
#define ULTRAEN 0x20
#define CLRSTCNT 0x10
#define SPIOEN 0x08
#define SCAMEN 0x04
#define CLRCHN 0x02
/* UNUSED 0x01 */
/*
* SCSI Transfer Control 1 Register (pp. 3-14,15).
* Controls the SCSI module data path.
*/
#define SXFRCTL1 0x002
#define BITBUCKET 0x80
#define SWRAPEN 0x40
#define ENSPCHK 0x20
#define STIMESEL 0x18
#define ENSTIMER 0x04
#define ACTNEGEN 0x02
#define STPWEN 0x01 /* Powered Termination */
/*
* SCSI Control Signal Read Register (p. 3-15).
* Reads the actual state of the SCSI bus pins
*/
#define SCSISIGI 0x003
#define CDI 0x80
#define IOI 0x40
#define MSGI 0x20
#define ATNI 0x10
#define SELI 0x08
#define BSYI 0x04
#define REQI 0x02
#define ACKI 0x01
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* Possible phases in SCSISIGI
*/
#define PHASE_MASK 0xe0
#define P_DATAOUT 0x00
#define P_DATAIN 0x40
#define P_COMMAND 0x80
#define P_MESGOUT 0xa0
#define P_STATUS 0xc0
#define P_MESGIN 0xe0
/*
* SCSI Contol Signal Write Register (p. 3-16).
* Writing to this register modifies the control signals on the bus. Only
* those signals that are allowed in the current mode (Initiator/Target) are
* asserted.
*/
#define SCSISIGO 0x003
#define CDO 0x80
#define IOO 0x40
#define MSGO 0x20
#define ATNO 0x10
#define SELO 0x08
#define BSYO 0x04
#define REQO 0x02
#define ACKO 0x01
/*
* SCSI Rate Control (p. 3-17).
* Contents of this register determine the Synchronous SCSI data transfer
* rate and the maximum synchronous Req/Ack offset. An offset of 0 in the
* SOFS (3:0) bits disables synchronous data transfers. Any offset value
* greater than 0 enables synchronous transfers.
*/
#define SCSIRATE 0x004
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define WIDEXFER 0x80 /* Wide transfer control */
#define SXFR 0x70 /* Sync transfer rate */
#define SOFS 0x0f /* Sync offset */
/*
* SCSI ID (p. 3-18).
* Contains the ID of the board and the current target on the
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* selected channel.
*/
#define SCSIID 0x005
#define TID 0xf0 /* Target ID mask */
#define OID 0x0f /* Our ID mask */
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* SCSI Latched Data (p. 3-19).
* Read/Write latchs used to transfer data on the SCSI bus during
* Automatic or Manual PIO mode. SCSIDATH can be used for the
* upper byte of a 16bit wide asyncronouse data phase transfer.
*/
#define SCSIDATL 0x006
#define SCSIDATH 0x007
/*
* SCSI Transfer Count (pp. 3-19,20)
* These registers count down the number of bytes transfered
* across the SCSI bus. The counter is decremented only once
* the data has been safely transfered. SDONE in SSTAT0 is
* set when STCNT goes to 0
*/
#define STCNT 0x008
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define STCNT0 0x008
#define STCNT1 0x009
#define STCNT2 0x00a
/*
* Clear SCSI Interrupt 0 (p. 3-20)
* Writing a 1 to a bit clears the associated SCSI Interrupt in SSTAT0.
*/
#define CLRSINT0 0x00b
#define CLRSELDO 0x40
#define CLRSELDI 0x20
#define CLRSELINGO 0x10
#define CLRSWRAP 0x08
/* UNUSED 0x04 */
#define CLRSPIORDY 0x02
/* UNUSED 0x01 */
/*
* SCSI Status 0 (p. 3-21)
* Contains one set of SCSI Interrupt codes
* These are most likely of interest to the sequencer
*/
#define SSTAT0 0x00b
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define TARGET 0x80 /* Board acting as target */
#define SELDO 0x40 /* Selection Done */
#define SELDI 0x20 /* Board has been selected */
#define SELINGO 0x10 /* Selection In Progress */
#define SWRAP 0x08 /* 24bit counter wrap */
#define SDONE 0x04 /* STCNT = 0x000000 */
#define SPIORDY 0x02 /* SCSI PIO Ready */
#define DMADONE 0x01 /* DMA transfer completed */
/*
* Clear SCSI Interrupt 1 (p. 3-23)
* Writing a 1 to a bit clears the associated SCSI Interrupt in SSTAT1.
*/
#define CLRSINT1 0x00c
#define CLRSELTIMEO 0x80
#define CLRATNO 0x40
#define CLRSCSIRSTI 0x20
/* UNUSED 0x10 */
#define CLRBUSFREE 0x08
#define CLRSCSIPERR 0x04
#define CLRPHASECHG 0x02
#define CLRREQINIT 0x01
/*
* SCSI Status 1 (p. 3-24)
*/
#define SSTAT1 0x00c
#define SELTO 0x80
#define ATNTARG 0x40
#define SCSIRSTI 0x20
#define PHASEMIS 0x10
#define BUSFREE 0x08
#define SCSIPERR 0x04
#define PHASECHG 0x02
#define REQINIT 0x01
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* SCSI Interrupt Mode 1 (pp. 3-28,29)
* Setting any bit will enable the corresponding function
* in SIMODE1 to interrupt via the IRQ pin.
*/
#define SIMODE1 0x011
#define ENSELTIMO 0x80
#define ENATNTARG 0x40
#define ENSCSIRST 0x20
#define ENPHASEMIS 0x10
#define ENBUSFREE 0x08
#define ENSCSIPERR 0x04
#define ENPHASECHG 0x02
#define ENREQINIT 0x01
/*
* SCSI Data Bus (High) (p. 3-29)
* This register reads data on the SCSI Data bus directly.
*/
#define SCSIBUSL 0x012
#define SCSIBUSH 0x013
/*
* SCSI/Host Address (p. 3-30)
* These registers hold the host address for the byte about to be
* transfered on the SCSI bus. They are counted up in the same
* manner as STCNT is counted down. SHADDR should always be used
* to determine the address of the last byte transfered since HADDR
* can be squewed by write ahead.
*/
#define SHADDR 0x014
#define SHADDR0 0x014
#define SHADDR1 0x015
#define SHADDR2 0x016
#define SHADDR3 0x017
/*
* Selection/Reselection ID (p. 3-31)
* Upper four bits are the device id. The ONEBIT is set when the re/selecting
* device did not set its own ID.
*/
#define SELID 0x019
#define SELID_MASK 0xf0
#define ONEBIT 0x08
/* UNUSED 0x07 */
/*
* SCSI Block Control (p. 3-32)
* Controls Bus type and channel selection. In a twin channel configuration
* addresses 0x00-0x1e are gated to the appropriate channel based on this
* register. SELWIDE allows for the coexistence of 8bit and 16bit devices
* on a wide bus.
*/
#define SBLKCTL 0x01f
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define DIAGLEDEN 0x80 /* Aic78X0 only */
#define DIAGLEDON 0x40 /* Aic78X0 only */
#define AUTOFLUSHDIS 0x20
/* UNUSED 0x10 */
#define SELBUS_MASK 0x0a
#define SELBUSB 0x08
/* UNUSED 0x04 */
#define SELWIDE 0x02
/* UNUSED 0x01 */
#define SELNARROW 0x00
/*
* Sequencer Control (p. 3-33)
* Error detection mode and speed configuration
*/
#define SEQCTL 0x060
#define PERRORDIS 0x80
#define PAUSEDIS 0x40
#define FAILDIS 0x20
#define FASTMODE 0x10
#define BRKADRINTEN 0x08
#define STEP 0x04
#define SEQRESET 0x02
#define LOADRAM 0x01
/*
* Sequencer RAM Data (p. 3-34)
* Single byte window into the Scratch Ram area starting at the address
* specified by SEQADDR0 and SEQADDR1. To write a full word, simply write
* four bytes in sucessesion. The SEQADDRs will increment after the most
* significant byte is written
*/
#define SEQRAM 0x061
/*
* Sequencer Address Registers (p. 3-35)
* Only the first bit of SEQADDR1 holds addressing information
*/
#define SEQADDR0 0x062
#define SEQADDR1 0x063
#define SEQADDR1_MASK 0x01
/*
* Accumulator
* We cheat by passing arguments in the Accumulator up to the kernel driver
*/
#define ACCUM 0x064
#define SINDEX 0x065
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define DINDEX 0x066
#define ALLZEROS 0x06a
#define NONE 0x06a
#define SINDIR 0x06c
#define DINDIR 0x06d
#define FUNCTION1 0x06e
/*
* Host Address (p. 3-48)
* This register contains the address of the byte about
* to be transfered across the host bus.
*/
#define HADDR 0x088
#define HADDR0 0x088
#define HADDR1 0x089
#define HADDR2 0x08a
#define HADDR3 0x08b
#define HCNT 0x08c
#define HCNT0 0x08c
#define HCNT1 0x08d
#define HCNT2 0x08e
/*
* SCB Pointer (p. 3-49)
* Gate one of the four SCBs into the SCBARRAY window.
*/
#define SCBPTR 0x090
/*
* Board Control (p. 3-43)
*/
#define BCTL 0x084
/* RSVD 0xf0 */
#define ACE 0x08 /* Support for external processors */
/* RSVD 0x06 */
#define ENABLE 0x01
/*
* On the aic78X0 chips, Board Control is replaced by the DSCommand
* register (p. 4-64)
*/
#define DSCOMMAND 0x084
#define CACHETHEN 0x80 /* Cache Threshold enable */
#define DPARCKEN 0x40 /* Data Parity Check Enable */
#define MPARCKEN 0x20 /* Memory Parity Check Enable */
#define EXTREQLCK 0x10 /* External Request Lock */
/*
* Bus On/Off Time (p. 3-44)
*/
#define BUSTIME 0x085
#define BOFF 0xf0
#define BON 0x0f
/*
* Bus Speed (p. 3-45)
*/
#define BUSSPD 0x086
#define DFTHRSH 0xc0
#define STBOFF 0x38
#define STBON 0x07
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define DFTHRSH_100 0xc0
/*
* Host Control (p. 3-47) R/W
* Overal host control of the device.
*/
#define HCNTRL 0x087
/* UNUSED 0x80 */
#define POWRDN 0x40
/* UNUSED 0x20 */
#define SWINT 0x10
#define IRQMS 0x08
#define PAUSE 0x04
#define INTEN 0x02
#define CHIPRST 0x01
#define CHIPRSTACK 0x01
/*
* Interrupt Status (p. 3-50)
* Status for system interrupts
*/
#define INTSTAT 0x091
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define SEQINT_MASK 0xf1 /* SEQINT Status Codes */
#define BAD_PHASE 0x01 /* unknown scsi bus phase */
#define SEND_REJECT 0x11 /* sending a message reject */
#define NO_IDENT 0x21 /* no IDENTIFY after reconnect*/
#define NO_MATCH 0x31 /* no cmd match for reconnect */
#define SDTR_MSG 0x41 /* SDTR message received */
#define WDTR_MSG 0x51 /* WDTR message received */
#define REJECT_MSG 0x61 /* Reject message received */
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define BAD_STATUS 0x71 /* Bad status from target */
#define RESIDUAL 0x81 /* Residual byte count != 0 */
#define ABORT_TAG 0x91 /* Sent an ABORT_TAG message */
#define AWAITING_MSG 0xa1 /*
* Kernel requested to specify
* a message to this target
* (command was null), so tell
* it that it can fill the
* message buffer.
*/
#define IMMEDDONE 0xb1 /*
* An immediate command has
* completed
*/
#define MSG_BUFFER_BUSY 0xc1 /*
* Sequencer wants to use the
* message buffer, but it
* already contains a message
*/
#define MSGIN_PHASEMIS 0xd1 /*
* Target changed phase on us
* when we were expecting
* another msgin byte.
*/
#define BRKADRINT 0x08
#define SCSIINT 0x04
#define CMDCMPLT 0x02
#define SEQINT 0x01
#define INT_PEND (BRKADRINT | SEQINT | SCSIINT | CMDCMPLT)
/*
* Hard Error (p. 3-53)
* Reporting of catastrophic errors. You usually cannot recover from
* these without a full board reset.
*/
#define ERROR 0x092
/* UNUSED 0xf0 */
#define PARERR 0x08
#define ILLOPCODE 0x04
#define ILLSADDR 0x02
#define ILLHADDR 0x01
/*
* Clear Interrupt Status (p. 3-52)
*/
#define CLRINT 0x092
#define CLRBRKADRINT 0x08
#define CLRSCSIINT 0x04
#define CLRCMDINT 0x02
#define CLRSEQINT 0x01
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define DFCNTRL 0x093
#define WIDEODD 0x40
#define SCSIEN 0x20
#define SDMAEN 0x10
#define SDMAENACK 0x10
#define HDMAEN 0x08
#define HDMAENACK 0x08
#define DIRECTION 0x04
#define FIFOFLUSH 0x02
#define FIFORESET 0x01
#define DFSTATUS 0x094
#define HDONE 0x08
#define FIFOEMP 0x01
#define DFDAT 0x099
/*
* SCB Auto Increment (p. 3-59)
* Byte offset into the SCB Array and an optional bit to allow auto
* incrementing of the address during download and upload operations
*/
#define SCBCNT 0x09a
#define SCBAUTO 0x80
#define SCBCNT_MASK 0x1f
/*
* Queue In FIFO (p. 3-60)
* Input queue for queued SCBs (commands that the seqencer has yet to start)
*/
#define QINFIFO 0x09b
/*
* Queue In Count (p. 3-60)
* Number of queued SCBs
*/
#define QINCNT 0x09c
/*
* Queue Out FIFO (p. 3-61)
* Queue of SCBs that have completed and await the host
*/
#define QOUTFIFO 0x09d
/*
* Queue Out Count (p. 3-61)
* Number of queued SCBs in the Out FIFO
*/
#define QOUTCNT 0x09e
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/*
* SCB Definition (p. 5-4)
* The two reserved bytes at SCBARRAY+1[23] are expected to be set to
* zero. Bit 3 in SCBARRAY+0 is used as an internal flag to indicate
* whether or not to DMA an SCB from host ram. This flag prevents the
* "re-fetching" of transactions that are requed because the target is
* busy with another command. We also use bits 6 & 7 to indicate whether
* or not to initiate SDTR or WDTR repectively when starting this command.
*/
#define SCBARRAY 0x0a0
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define SCB_CONTROL 0x0a0
#define NEEDWDTR 0x80
#define DISCENB 0x40
#define TAG_ENB 0x20
#define NEEDSDTR 0x10
#define DISCONNECTED 0x04
#define SCB_TAG_TYPE 0x03
#define SCB_TCL 0x0a1
#define SCB_TARGET_STATUS 0x0a2
#define SCB_SGCOUNT 0x0a3
#define SCB_SGPTR 0x0a4
#define SCB_SGPTR0 0x0a4
#define SCB_SGPTR1 0x0a5
#define SCB_SGPTR2 0x0a6
#define SCB_SGPTR3 0x0a7
#define SCB_RESID_SGCNT 0x0a8
#define SCB_RESID_DCNT 0x0a9
#define SCB_RESID_DCNT0 0x0a9
#define SCB_RESID_DCNT1 0x0aa
#define SCB_RESID_DCNT2 0x0ab
#define SCB_DATAPTR 0x0ac
#define SCB_DATAPTR0 0x0ac
#define SCB_DATAPTR1 0x0ad
#define SCB_DATAPTR2 0x0ae
#define SCB_DATAPTR3 0x0af
#define SCB_DATACNT 0x0b0
#define SCB_DATACNT0 0x0b0
#define SCB_DATACNT1 0x0b1
#define SCB_DATACNT2 0x0b2
/* UNUSED - QUAD PADDING 0x0b3 */
#define SCB_CMDPTR 0x0b4
#define SCB_CMDPTR0 0x0b4
#define SCB_CMDPTR1 0x0b5
#define SCB_CMDPTR2 0x0b6
#define SCB_CMDPTR3 0x0b7
#define SCB_CMDLEN 0x0b8
#define SCB_TAG 0x0b9
#define SCB_NEXT 0x0ba
#define SCB_PREV 0x0bb
#ifdef linux
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define SG_SIZEOF 0x0c /* sizeof(struct scatterlist) */
#else
#define SG_SIZEOF 0x08 /* sizeof(struct ahc_dma) */
#endif
/* --------------------- AHA-2840-only definitions -------------------- */
#define SEECTL_2840 0x0c0
/* UNUSED 0xf8 */
#define CS_2840 0x04
#define CK_2840 0x02
#define DO_2840 0x01
#define STATUS_2840 0x0c1
#define EEPROM_TF 0x80
#define BIOS_SEL 0x60
#define ADSEL 0x1e
#define DI_2840 0x01
/* --------------------- AIC-7870-only definitions -------------------- */
#define DSPCISTATUS 0x086
/*
* Serial EEPROM Control (p. 4-92 in 7870 Databook)
* Controls the reading and writing of an external serial 1-bit
* EEPROM Device. In order to access the serial EEPROM, you must
* first set the SEEMS bit that generates a request to the memory
* port for access to the serial EEPROM device. When the memory
* port is not busy servicing another request, it reconfigures
* to allow access to the serial EEPROM. When this happens, SEERDY
* gets set high to verify that the memory port access has been
* granted.
*
* After successful arbitration for the memory port, the SEECS bit of
* the SEECTL register is connected to the chip select. The SEECK,
* SEEDO, and SEEDI are connected to the clock, data out, and data in
* lines respectively. The SEERDY bit of SEECTL is useful in that it
* gives us an 800 nsec timer. After a write to the SEECTL register,
* the SEERDY goes high 800 nsec later. The one exception to this is
* when we first request access to the memory port. The SEERDY goes
* high to signify that access has been granted and, for this case, has
* no implied timing.
*
* See 93cx6.c for detailed information on the protocol necessary to
* read the serial EEPROM.
*/
#define SEECTL 0x01e
#define EXTARBACK 0x80
#define EXTARBREQ 0x40
#define SEEMS 0x20
#define SEERDY 0x10
#define SEECS 0x08
#define SEECK 0x04
#define SEEDO 0x02
#define SEEDI 0x01
/* ---------------------- Scratch RAM Offsets ------------------------- */
/* These offsets are either to values that are initialized by the board's
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* BIOS or are specified by the sequencer code.
*
* The host adapter card (at least the BIOS) uses 20-2f for SCSI
* device information, 32-33 and 5a-5f as well. As it turns out, the
* BIOS trashes 20-2f, writing the synchronous negotiation results
* on top of the BIOS values, so we re-use those for our per-target
* scratchspace (actually a value that can be copied directly into
* SCSIRATE). The kernel driver will enable synchronous negotiation
* for all targets that have a value other than 0 in the lower four
* bits of the target scratch space. This should work regardless of
* whether the bios has been installed.
*/
/*
* 1 byte per target starting at this address for configuration values
*/
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define TARG_SCRATCH 0x020
/*
* The sequencer will stick the frist byte of any rejected message here so
* we can see what is getting thrown away. Extended messages put the
* extended message type in REJBYTE_EXT.
*/
#define REJBYTE 0x030
#define REJBYTE_EXT 0x031
/*
* Bit vector of targets that have disconnection disabled.
*/
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define DISC_DSB 0x032
#define DISC_DSB_A 0x032
#define DISC_DSB_B 0x033
/*
* Length of pending message
*/
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define MSG_LEN 0x034
/* We reserve 8bytes to store outgoing messages */
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define MSG0 0x035
#define COMP_MSG0 0xcb /* 2's complement of MSG0 */
#define MSG1 0x036
#define MSG2 0x037
#define MSG3 0x038
#define MSG4 0x039
#define MSG5 0x03a
#define MSG6 0x03b
#define MSG7 0x03c
/*
* These are offsets into the card's scratch ram. Some of the values are
* specified in the AHA2742 technical reference manual and are initialized
* by the BIOS at boot time.
*/
#define LASTPHASE 0x03d
#define ARG_1 0x03e
#define MAXOFFSET 0x01
#define RETURN_1 0x03f
#define SEND_WDTR 0x80
#define SEND_SDTR 0x60
#define SEND_SENSE 0x40
#define SEND_REJ 0x20
#define SCB_PAGEDIN 0x10
#define SIGSTATE 0x040
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define DMAPARAMS 0x041 /* Parameters for DMA Logic */
#define SG_COUNT 0x042
#define SG_NEXT 0x043 /* working value of SG pointer */
#define SG_NEXT0 0x043
#define SG_NEXT1 0x044
#define SG_NEXT2 0x045
#define SG_NEXT3 0x046
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define SCBCOUNT 0x047 /*
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* Number of SCBs supported by
* this card.
*/
#define COMP_SCBCOUNT 0x048 /*
* Two's compliment of SCBCOUNT
*/
#define QCNTMASK 0x049 /*
* Mask of bits to test against
* when looking at the Queue Count
* registers. Works around a bug
* on aic7850 chips.
*/
#define FLAGS 0x04a
#define SINGLE_BUS 0x00
#define TWIN_BUS 0x01
#define WIDE_BUS 0x02
#define PAGESCBS 0x04
#define DPHASE 0x10
#define SELECTED 0x20
#define IDENTIFY_SEEN 0x40
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define RESELECTED 0x80
#define SAVED_TCL 0x04b /*
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* Temporary storage for the
* target/channel/lun of a
* reconnecting target
*/
#define ACTIVE_A 0x04c
#define ACTIVE_B 0x04d
#define WAITING_SCBH 0x04e /*
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
* head of list of SCBs awaiting
* selection
*/
#define DISCONNECTED_SCBH 0x04f /*
* head of list of SCBs that are
* disconnected. Used for SCB
* paging.
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
*/
#define SCB_LIST_NULL 0xff
#define SAVED_LINKPTR 0x050
#define SAVED_SCBPTR 0x051
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define SCSICONF 0x05a
#define RESET_SCSI 0x40
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define HOSTCONF 0x05d
#define HA_274_BIOSCTRL 0x05f
#define BIOSMODE 0x30
#define BIOSDISABLED 0x30
#define CHANNEL_B_PRIMARY 0x08
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
/* Message codes */
#define MSG_EXTENDED 0x01
#define MSG_SDTR 0x01
#define MSG_WDTR 0x03
#define MSG_SDPTRS 0x02
#define MSG_RDPTRS 0x03
#define MSG_DISCONNECT 0x04
#define MSG_INITIATOR_DET_ERROR 0x05
#define MSG_ABORT 0x06
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define MSG_REJECT 0x07
#define MSG_NOP 0x08
#define MSG_MSG_PARITY_ERROR 0x09
#define MSG_BUS_DEVICE_RESET 0x0c
#define MSG_ABORT_TAG 0x0d
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define MSG_SIMPLE_TAG 0x20
#define MSG_IDENTIFY 0x80
/* WDTR Message values */
#define BUS_8_BIT 0x00
#define BUS_16_BIT 0x01
#define BUS_32_BIT 0x02
The long awaited stability patch set for the aic7xxx driver: 1) Use cpp to preprocess the sequencer code. 2) Convert all "magic numbers" to #defines shared by the sequencer and kernel driver via the aic7xxx_reg.h file. (The assembler still needs to be re-written in lex/yacc to allow ~|& type constructions). 3) Raise ATN on parity errors for "in" phases and send an initiator detected error or message-in parity error message as appropriate. 4) Turn off the reselection hardware from the time or a (re)connection to busfree. It seems that some fast targets were able to reconnect before the sequencer was able to see busfree. 5) The message buffer is considered "in-use" when there is a positive length count. The ACTIVE_MSG flag was unnecesary. 6) Properly set SCB_NEXT_WAITING to SCB_LIST_HEAD in scbs being added to the waiting scb list. This is a change in how the list code works to facilitate some planned work in the reset code. 7) The fields in the SCB have be re-arranged to be quad-word aligned. 8) The inb code has been rewritten to catch phasemisses and be more efficient. 9) Go back to "snooping the bus" to determine if the incomming identify message will be followed by a simple queue message. Its much faster than doing a search through the SCBs. 10) Implement better tag range checking for incomming tags. 11) Make sdtr_to_rate more accurate (use 25 instead of 24 in calculations - must have been asleep that night). 12) Rearrange some routines to reduce code complexity and size. 13) Update comments and formatting. 14) Fixed bugs I've forgotten about?? Reviewed by: David Greenman <davidg@FreeBSD.org>
1996-01-03 06:25:32 +00:00
#define MAX_OFFSET_8BIT 0x0f
#define MAX_OFFSET_16BIT 0x08