This Technical Note presents short 65816 assembly language examples illustrating pitfalls and clever techniques.
Changes since November 1990: Added more explanations about the JSL table and corrected a comment.
The 65816 has a JSR ($aaaa,X) instruction for calling a selected subroutine from a table of addresses, but it has no JSL ($aaaa,X) instruction. If you need to dispatch to one of several routines that are not all in the same bank, you need an approach like the following. The idea is to perform a JSL to a routine which does a long jump by pushing a three-byte "RTL address" on the stack and then doing an RTL.
jsl LngJmp ;go jump to the routine ... LngJmp asl a ;take routine number in A and asl a ; multiply it by 4 tax ;put table index into X lda table+1,x ;get "middle" word of address pha ; and push it lda table,x ;get low word and dec a ; decrement it by one phb ;push a single throw-away byte sta 1,s ;store over low two of the 3 bytes rtl ;transfer control to the routine table dc.l routine1 ;table of 4-byte subroutine addresses dc.l routine2 dc.l routine3 ...
This code is correct because RTL pulls three bytes off the stack and increments the two low bytes without incrementing the high byte.
Note: This approach to a table-based JSL is more flexible than JML ($XXXX) because it does not require any fixed-location storage or bank zero space, other than the stack.
On the other hand, the following code is not correct. The approach here is to make a table of addresses minus one.
asl a W asl a R ;multiply index by 4 tax O ; and put it in X lda table+1,x N ;get the "middle" word pha G ; and push it lda table,x ! ;get the low word phb W ;push a single throw-away byte sta 1,s R ;store over low two bytes rtl O ;transfer control to the routine table dc.l routine1-1 N ;table of 4-byte addresses minus one dc.l routine2-1 G dc.l routine3-1 ! ...
This second sample code fragment fails if any of the routines in the table comes at the first byte of a bank. For example, if routine1 is at $060000, the address pushed is $05FFFF, and RTL transfers control to $050000, not $060000.
When your code gets called with the D register undefined, you must not use direct page addressing without setting D to a known good value. Preserving and restoring locations on the caller's direct page is not reliable, because D could be pointing at bytes below the stack pointer (which can be destroyed by interrupts) or even at the $C0xx soft switches (that would make your direct page accesses accidentally fiddle with hardware).
A common way to get temporary direct page space is to point D at part of your stack. This following code dereferences a handle stored in the A and X registers (if the handle is $E01234 and refers to a block of memory at $056789, then on entry A=$00E0 and X=$1234, and on exit A=$0005 and X=$6789).
phd ;save caller's direct-page register pha ;push high word of handle phx ;push low word of handle tsc ;get stack pointer in A tcd ;and put it in D lda [1] ;get low word of master pointer (no ",Y"!) tax ; and put it in X ldy #$0002 ;offset to high word of master pointer lda [1],y ;get high word ply ;remove low word of handle ply ; and high word pld ;restore the caller's direct-page register
Direct page addressing isn't the only way to address through pointers. Here's the same routine as before, but using the Data Bank register (B) instead of fiddling with D. (Note that handles do not have to be in bank $E0 or $E1, although they usually are.)
phb ;save caller's data bank register pha ;push high word of handle on stack plb ;sets B to the bank byte of the pointer lda |$0002,x ;load the high word of the master pointer pha ; and save it on the stack lda |$0000,x ;load the low word of the master pointer tax ;and return it in X pla ;restore the high word in A plb ;pull the handle's high word high byte off the ; stack plb ;restore the caller's data bank register
You don't have to switch into Native mode just to do an eight-bit operation with long addressing. Most 65816-specific instructions and addressing modes work in emulation mode in approximately the same way they work in eight-bit native mode. See the "Further Reference" for details.
This and all of the other Apple II Technical Notes have been converted to HTML by Aaron Heiss as a public service to the Apple II community, with permission by Apple Computer, Inc. Any and all trademarks, registered and otherwise, are properties of their owners.