Category : Network Files
Archive   : TCP_SRC.ZIP
Filename : 3C503.ASM
version equ 4
include defs.asm
;/* PC/FTP Packet Driver source, conforming to version 1.05 of the spec,
;* for the 3-Com 3C503 interface card.
;* Robert C Clements, K1BC, 14 February, 1989
;* Portions (C) Copyright 1988, 1989 Robert C Clements
;*
; Copyright, 1988, 1989, Russell Nelson
; This program is free software; you can redistribute it and/or modify
; it under the terms of the GNU General Public License as published by
; the Free Software Foundation, version 1.
;
; This program is distributed in the hope that it will be useful,
; but WITHOUT ANY WARRANTY; without even the implied warranty of
; MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
; GNU General Public License for more details.
;
; You should have received a copy of the GNU General Public License
; along with this program; if not, write to the Free Software
; Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
;* Change history:
;* Updated to driver spec version 1.08 Feb. 17, 1989 by Russell Nelson.
;* Changes 27 Jul 89 by Bob Clements (/Rcc)
;* Added Thick versus Thin Ethernet switch 27 Jul 89 by Bob Clements (/Rcc)
;* Added call to memory_test.
;* Added rcv_mode logic. Started, but didn't finish, multicast logic.
;* Fixed get_address to return current, not PROM, address.
;* Minor races fixed.
comment /
From: "James A. Harvey"
Subject: Patches for 6.x packet drivers; lockup problem fixed!
Now for the best part, the lockup problem fix. I think this may be one that
I keep hearing about that for most people the machine locks up for a minute
on startup, but then continues. For me it was worse because it appears that
the "recovery" time is only short on heavily loaded networks. The lockup is
caused by the "first page for RX" being set improperly in etopen; I finally
figured it out by looking at code from other drivers that used the DS8390
chip. One must switch to the page 1 command registers first.
/
code segment word public
assume cs:code, ds:code
; Stuff specific to the 3-Com 3C503 Ethernet controller board
; WD version in C by Bob Clements, K1BC, May 1988 for the KA9Q TCP/IP package
; 3Com version based on WD8003E version in .ASM, also by Bob Clements, dated
; 19 August 1988. The WD and 3Com cards both use the National DS8390.
; Symbol prefix "EN" is for Ethernet, National chip
; Symbol prefix "E33" is for _E_thernet, _3_Com 50_3_
; Symbol prefix "E33G" is for registers in the Gate array ASIC.
; The E33 registers - For the ASIC on the 3C503 card:
; Offsets from the board's base address, which can be set by
; jumpers to be one of the following 8 values (hex):
; 350, 330, 310, 300, 2E0, 2A0, 280, 250
; Factory default address is 300H.
; The card occupies a block of 16 I/O addresses.
; It also occupies 16 addresses at base+400 through base+40F.
; These high-addressed registers are in the ASIC.
; Recall that the normal PC I/O decoding is only 10 bits. The 11'th
; bit (400H) can be used on the same card for additional registers.
; This offset requires word, not byte, arithmetic
; on the DX register for the setport macro. Current SETPORT is OK.
; The card can also be jumpered to have the shared memory disabled
; or enabled at one of four addresses: C8000, CC000, D8000 or DC000.
; This version of the driver REQUIRES the shared memory to be
; enabled somewhere.
; The card can be operated using direct I/O instructions or by
; using the PC's DMA channels instead of the shared memory, but
; I haven't included the code for those other two methods.
; They would be needed in a system where all four possible addresses
; for the shared memory are in use by other devices. /Rcc
; Blocks of I/O addresses:
E33GA equ 400h ; Registers in the gate array.
E33_SAPROM equ 000h ; Window on station addr prom (if
; E33G_CNTRL bits 3,2 = 0,1
; These appear at Base+0 through Base+0F when bits 3,2 of
; E33G_CNTRL are 0,0.
EN_OFF equ 0h
ENDCFG_BM8 equ 48h
include 8390.inc
; Registers in the 3-Com custom Gate Array
E33G_STARTPG equ E33GA+00h ; Start page, must match EN0_STARTPG
E33G_STOPPG equ E33GA+01h ; Stop page, must match EN0_STOPPG
E33G_NBURST equ E33GA+02h ; Size of DMA burst before relinquishing bus
E33G_IOBASE equ E33GA+03h ; Bit coded: where I/O regs are jumpered.
; (Which you have to know already to read it)
E33G_ROMBASE equ E33GA+04h ; Bit coded: Where/whether EEPROM&DPRAM exist
E33G_GACFR equ E33GA+05h ; Config/setup bits for the ASIC GA
E33G_CNTRL equ E33GA+06h ; Board's main control register
E33G_STATUS equ E33GA+07h ; Status on completions.
E33G_IDCFR equ E33GA+08h ; Interrupt/DMA config register
; (Which IRQ to assert, DMA chan to use)
E33G_DMAAH equ E33GA+09h ; High byte of DMA address reg
E33G_DMAAL equ E33GA+0ah ; Low byte of DMA address reg
E33G_VP2 equ E33GA+0bh ; Vector pointer - for clearing RAM select
E33G_VP1 equ E33GA+0ch ; on a system reset, to re-enable EPROM.
E33G_VP0 equ E33GA+0dh ; 3Com says set this to Ctrl-Alt-Del handler
E33G_FIFOH equ E33GA+0eh ; FIFO for programmed I/O data moves ...
E33G_FIFOL equ E33GA+0fh ; .. low byte of above.
; Bits in E33G_CNTRL register:
ECNTRL_RESET equ 001h ; Software reset of the ASIC and 8390
ECNTRL_THIN equ 002h ; Onboard thin-net xcvr enable
ECNTRL_SAPROM equ 004h ; Map the station address prom
ECNTRL_DBLBFR equ 020h ; FIFO configuration bit
ECNTRL_OUTPUT equ 040h ; PC-to-3C503 direction if 1
ECNTRL_START equ 080h ; Start the DMA logic
; Bits in E33G_STATUS register:
ESTAT_DPRDY equ 080h ; Data port (of FIFO) ready
ESTAT_UFLW equ 040h ; Tried to read FIFO when it was empty
ESTAT_OFLW equ 020h ; Tried to write FIFO when it was full
ESTAT_DTC equ 010h ; Terminal Count from PC bus DMA logic
ESTAT_DIP equ 008h ; DMA In Progress
; Bits in E33G_GACFR register:
EGACFR_NORM equ 049h ; Enable 8K shared mem, no DMA TC int
EGACFR_IRQOFF equ 0c9h ; Above, and disable 8390 IRQ line
; Shared memory management parameters
SM_TSTART_PG equ 020h ; First page of TX buffer
SM_RSTART_PG equ 026h ; Starting page of RX ring
SM_RSTOP_PG equ 040h ; Last page +1 of RX ring
; End of 3C503 parameter definitions
pause_ macro
jmp $+2
endm
longpause macro
push cx
mov cx,0
loop $
pop cx
endm
ram_enable macro
setport E33G_GACFR ; Make sure gate array is set up and
mov al, EGACFR_NORM ; the RAM is enabled (not EPROM)
out dx, al ; ..
endm
reset_8390 macro
loadport ; First, pulse the board reset
setport E33G_CNTRL
mov al,thin_bit ; Thick or thin cable bit
or al,ECNTRL_RESET
out dx,al ; Turn on board reset bit
mov al,thin_bit ; Thick or thin cable bit
out dx,al ; Turn off board reset bit
call do_reset
loadport
endm
terminate_board macro
endm
; The following three values may be overridden from the command line.
; If they are omitted from the command line, these defaults are used.
; The shared memory base is set by a jumper. We read it from the
; card and set up accordingly.
public int_no, io_addr, thin_not_thick
int_no db 2,0,0,0 ; Interrupt level
io_addr dw 0300h,0 ; I/O address for card (jumpers)
thin_not_thick dw 1,0 ; Non-zero means thin net
public mem_base
mem_base dw 00000h,0 ; Shared memory addr (jumpers)
; (Not changeable by software in 3C503) ; (0 if disabled by jumpers)
thin_bit db ECNTRL_THIN ; Default to thin cable
public driver_class, driver_type, driver_name, driver_function, parameter_list
driver_class db BLUEBOOK, IEEE8023, 0 ;from the packet spec
driver_type db 12 ;from the packet spec
driver_name db '3C503',0 ;name of the driver.
driver_function db 2
parameter_list label byte
db 1 ;major rev of packet driver
db 9 ;minor rev of packet driver
db 14 ;length of parameter list
db EADDR_LEN ;length of MAC-layer address
dw GIANT ;MTU, including MAC headers
dw MAX_MULTICAST * EADDR_LEN ;buffer size of multicast addrs
dw 0 ;(# of back-to-back MTU rcvs) - 1
dw 0 ;(# of successive xmits) - 1
int_num dw 0 ;Interrupt # to hook for post-EOI
;processing, 0 == none,
is_186 db 0
include movemem.asm
block_output:
;enter with cx = byte count, ds:si = buffer location, ax = buffer address
assume ds:nothing
cmp mem_base,0 ;memory or I/O?
je block_o ;I/O.
mov es,mem_base ; Set up ES:DI at the shared RAM
mov di,ax ; ..
loadport ; Set up for address of TX buffer.
ram_enable ; Make sure the RAM is actually there.
call movemem
clc
ret
block_o:
setport E33G_CNTRL
mov al,thin_bit
or al,0c0h ;start dma, write to board.
out dx,al
;;; we should really have another copy of this loop for 186 I/O instructions.
setport E33G_STATUS
block_o_0:
jcxz block_o_2 ;if there is none, exit.
in al,dx ;wait for the FIFO to be ready.
test al,ESTAT_DPRDY
je block_o_0
setport E33G_FIFOH ;now get ready to read data.
cmp cx,8 ;do we have eight more to do?
jb block_o_1 ;no, do them one by one.
lodsw ;yes, output eight all at once.
out dx,ax
lodsw
out dx,ax
lodsw
out dx,ax
lodsw
out dx,ax
sub cx,8 ;reduce the count by what we write.
setport E33G_STATUS ;go back to the status bit.
jmp block_o_0
block_o_1:
lodsb
out dx,ax
loop block_o_1
block_o_2:
loadport
setport E33G_CNTRL
mov al,thin_bit ;stop dma.
out dx,al
ret
block_input:
;enter with cx = byte count, es:di = buffer location, ax = board address.
cmp mem_base,0 ; memory or I/O
je block_i ; I/O
push ds
assume ds:nothing
mov ds,mem_base ; ds:si points at first byte to move
mov si,ax
add ax,cx ; Find the end of this frame.
cmp ah,byte ptr cs:sm_rstop_ptr ; Over the top of the ring?
jb rcopy_one_piece ; Go move it
rcopy_wrap:
; Copy in two pieces due to buffer wraparound.
mov ah,byte ptr cs:sm_rstop_ptr ; Compute length of first part
xor al,al
sub ax,si ; as all of the pages up to wrap point
sub cx,ax ; Move the rest in second part
push cx ; Save count of second part
mov cx,ax ; Count for first move
call rcopy_subr
mov si,SM_RSTART_PG*256 ; Offset to start of first receive page
pop cx ; Bytes left to move
rcopy_one_piece:
call rcopy_subr
pop ds
ret
rcopy_subr:
shr cx,1 ; convert byte count to word count
rep movsw
jnc rcv_wrap_even ; odd byte left over?
lodsw ; yes, word fetch
stosb ; and byte store
rcv_wrap_even:
ret
block_i:
ret
include 8390.asm
public usage_msg
usage_msg db "usage: 3C503 [-n] [-d] [-w]
public copyright_msg
copyright_msg db "Packet driver for 3-Com 3C503, version "
db '0'+majver,".",'0'+version,".",'0'+dp8390_version,CR,LF
db "Portions Copyright 1989, Robert C. Clements, K1BC",CR,LF,'$'
cfg_err_msg db "3C503 Configuration failed. Check parameters.",CR,LF,'$'
mem_busted_msg db "Shared RAM on 3C503 card is defective or there is an address conflict.",CR,LF,'$'
int_no_name db "Interrupt number ",'$'
io_addr_name db "I/O port ",'$'
mem_base_name db "Memory address ",'$'
thin_msg db "Using the built-in transceiver (thinwire)",CR,LF,'$'
thick_msg db "Using the external transceiver (thickwire)",CR,LF,'$'
extrn set_recv_isr: near
;enter with si -> argument string, di -> word to store.
;if there is no number, don't change the number.
extrn get_number: near
;enter with dx -> name of word, di -> dword to print.
extrn print_number: near
public parse_args
parse_args:
;exit with nc if all went well, cy otherwise.
mov di,offset int_no ; May override interrupt channel
call get_number
mov di,offset io_addr ; May override I/O address
call get_number
mov di,offset thin_not_thick ; May override thick/thin cable flag
call get_number
mov ax,thin_not_thick ; Now make the right bit
cmp ax,0
je parse_thin1 ; If zero, leave bit off
mov al,ECNTRL_THIN ; Else the bit for the card
parse_thin1:
mov thin_bit,al ; Save for setting up the card
clc
ret
do_reset:
assume ds:code
loadport
cli ; Protect the E33G_CNTRL contents
setport E33G_CNTRL ; Switch control bits to enable SA PROM
mov al, thin_bit
or al, ECNTRL_SAPROM
out dx, al ; ..
setport E33_SAPROM ; Where the address prom is
cld ; Make sure string mode is right
push cs ; Point es:di at local copy space
pop es
mov di, offset curr_hw_addr
mov cx, EADDR_LEN ; Set count for loop
do_reset_1:
in al, dx ; Get a byte of address
stosb ; Feed it to caller
inc dx ; Next byte at next I/O port
loop do_reset_1 ; Loop over six bytes
loadport ; Re-establish I/O base after dx mods
setport E33G_CNTRL ; Switch control bits to turn off SA PROM
mov al, thin_bit
out dx, al ; Turn off SA PROM windowing
sti ; Ok for E33G_CNTRL to change now
call set_8390_eaddr
ret
init_card:
; Now get the board's physical address from on-board PROM into card_hw_addr
assume ds:code
loadport
cli ; Protect the E33G_CNTRL contents
setport E33G_CNTRL ; Switch control bits to enable SA PROM
mov al, thin_bit
or al, ECNTRL_SAPROM
out dx, al ; ..
setport E33_SAPROM ; Where the address prom is
cld ; Make sure string mode is right
push cs ; Point es:di at local copy space
pop es
mov di, offset curr_hw_addr
mov cx, EADDR_LEN ; Set count for loop
ini_addr_loop:
in al, dx ; Get a byte of address
stosb ; Feed it to caller
inc dx ; Next byte at next I/O port
loop ini_addr_loop ; Loop over six bytes
loadport ; Re-establish I/O base after dx mods
setport E33G_CNTRL ; Switch control bits to turn off SA PROM
mov al, thin_bit
out dx, al ; Turn off SA PROM windowing
sti ; Ok for E33G_CNTRL to change now
; Point the "Vector Pointer" registers off into the boonies so we
; don't get the shared RAM disabled on us while we're using it.
; Ideally a warm boot should reset this too, to get to ROM on this card,
; but I don't know a guaranteed way to determine that value.
setport E33G_VP2
mov al, 0ffh ; Point this at the ROM restart location
out dx, al ; of ffff0h.
setport E33G_VP1
out dx, al
xor al, al
setport E33G_VP0
out dx, al
;Make sure shared memory is jumpered on. Find its address.
setport E33G_ROMBASE ; Point at rom/ram cfg reg
xor bx,bx
in al, dx ; Read it
test al,0f0h ; Any bits on?
je memcfg_3 ; no - using I/O.
memcfg_1:
mov bx, 0c600h ; Build mem segment here
test al,0c0h ; DC00 or D800?
je memcfg_2 ; No
add bx, 01000h ; Yes, make Dx00
memcfg_2:
test al,0a0h ; DC00 or CC00?
je memcfg_3
add bx, 00400h ; Yes, make xC00
memcfg_3:
mov mem_base,bx ; Remember segment addr of memory
or bx,bx
je mem_works ; don't test the memory if we use I/O.
; Set up Gate Array's Config Reg to enable and size the RAM.
setport E33G_GACFR ; Make sure gate array is set up and
mov al, EGACFR_IRQOFF ; the RAM is enabled (not EPROM)
out dx, al ; ..
; Check the card's memory
mov ax, mem_base ; Set segment of the shared memory
add ax, 16*SM_TSTART_PG ; which starts 2000h up from "base"
mov cx, 2000h ; Length of RAM to test
call memory_test ; Check it out
jz mem_works ; Go if it's OK
jmp mem_busted ; Go report failure if it's bad
mem_works:
; Set up control of shared memory, buffer ring, etc.
loadport
setport E33G_STARTPG ; Set ASIC copy of rx's first buffer page
mov al, SM_RSTART_PG
out dx, al
setport E33G_STOPPG ; and ASIC copy of rx's last buffer page + 1
mov al,SM_RSTOP_PG
; mov al, byte ptr sm_rstop_ptr
out dx, al
; Set up interrupt/DMA control register in ASIC.
; For now, we won't use the DMA, so B0-B3 are zero.
xor ah, ah ; Get the interrupt level from arg line
mov al, int_no ; ..
cmp al, 9 ; If converted to 9, make back into 2
jne get_irq1 ; Not 9
mov al, 2 ; Card thinks it's IRQ2
get_irq1: ; Now should have level in range 2-5
sub ax, 2 ; Make 0-3 for tables
cmp ax, 5-2 ; In range?
jna get_irq2
jmp cfg_error ; If not, can't configure.
get_irq2:
xor cx, cx ; Make the bit for the ASIC
mov cl, al ; Shift count
mov al, 10h ; Bit for irq2
shl al, cl ; Shift over as needed.
setport E33G_IDCFR ; Point at ASIC reg for IRQ level
out dx, al ; Set the bit
setport E33G_NBURST ; Set burst size to 8
mov al, 8
out dx, al ; ..
setport E33G_DMAAH ; Set up transmit bfr in DMA addr
mov al, SM_TSTART_PG
out dx, al
xor ax, ax
setport E33G_DMAAL
out dx, al
ret
mem_busted:
mov dx, offset mem_busted_msg
jmp error
public print_parameters
print_parameters:
mov di, offset int_no ; May override interrupt channel
mov dx, offset int_no_name ; Message for it
call print_number
mov di, offset io_addr ; May override I/O address
mov dx, offset io_addr_name ; Message for it
call print_number
mov dx,offset thin_msg
cmp thin_not_thick,0 ; May override thick/thin cable flag
jne print_parameters_1
mov dx,offset thick_msg
print_parameters_1:
mov ah,9
int 21h
ret
include memtest.asm
code ends
end
Very nice! Thank you for this wonderful archive. I wonder why I found it only now. Long live the BBS file archives!
This is so awesome! 😀 I’d be cool if you could download an entire archive of this at once, though.
But one thing that puzzles me is the “mtswslnkmcjklsdlsbdmMICROSOFT” string. There is an article about it here. It is definitely worth a read: http://www.os2museum.com/wp/mtswslnk/