> a book i've read says to declare, then initialize, then other subroutines,
> then start, then main functions.

Some people do write that way. Personally I prefer my main code at
the start and routines after. I have trouble wading through subroutines
to get to what's calling them. Tidy code is a lot easier to debug

So, with that bias, my layout preferences and Dr Bartfai's comments
in mind -

( = indicates a space, I don't use tabs)

====list P=16F877
====include "c:\P16F877.inc

;----------------------------
;You can suppress error messages
;
=errorlevel -305, -302, -306

;----------------------------
;You need to set CONFIG, eg (all on one line)

=__CONFIG _CP_OFF & _XT_OSC & _WDT_OFF & _PWRTE_OFF & _BODEN_OFF & _LVP_OFF

;Declarations

d1====equ===20
d2====equ===21
d3====equ===22

;------------------------------
;You can also define variables in a block of memory

====cblock 0x20==========;this declares variable RAM address

====d1
====d2
====d3

====endc

;specifically define port pins eg

#define===led1====porta,0
#define===led2====porta,1
#define===led3====porta,2
#define===led4====porta,3

;eg   bsf    led1

;and make macros for bank selection (there is another way - Olin ?)

bank0=macro
=========bcf====status,rp0
=========bcf====status,rp1
=========bcf====status,irp
=========endm

bank1=macro
=========bsf====status,rp0
=========bcf====status,rp1
=========bcf====status,irp
=========endm
;------------------------------

====org=0x000
====goto=init

;-----------------------------
;If you're using interrupts

====org=0x004
====goto=isr
;----------------------------

Init=====movlw==b'00000000'======;all porta low
=======movwf==PORTA

=======bsf====STATUS,RP0=====;bank 1
=======bcf====STATUS,RP1

;or use the macro to change bank. Simply type

=======bank1

=======movlw==b'11110000'======;0,1,2,3 of porta are output
=======movwf==TRISA

=======bcf==== STATUS,RP0=====;bank 0

;or

=======bank0

Main===bcf=====PORTA,1========;be aware of read-modify-write
=======nop====================;problems that may occur when
=======bcf=====PORTA,2========;changing port pin data
=======nop
=======bcf=====PORTA,3
=======nop
=======bsf=====PORTA,0
=======call=====Delay
<snip>
=======goto====main

Delay===movlw 0x2C===========;4999993 cycles
<snip>
=======return

=end

=======================================================

> I get  a stack underflow error

Because you had org 0x000 set to INIT, not START. The first piece of
code encountered ended with a RETLW that had no CALL

--
http://www.piclist.com hint: The list server can filter out subtopics
(like ads or off topics) for you. See http://www.piclist.com/#topics