diff --git a/fat32/fat32.s b/fat32/fat32.s
index 4f8aec12..9a62ebbb 100644
--- a/fat32/fat32.s
+++ b/fat32/fat32.s
@@ -8,6 +8,7 @@
 .include "lib.inc"
 .include "sdcard.inc"
 .include "text_input.inc"
+.include "65c816.inc"
 
 .import sector_buffer, sector_buffer_end, sector_lba, sdcard_set_fast_mode
 
@@ -3189,9 +3190,12 @@ x16_banked_copy:
 	inx
 	lda #$9f
 	sta fat32_ptr+1
+	set_carry_if_65c816
+	bcs @816_9f_page
 @nowrap:
 	cpy tmp_done
 	bne @loop
+@end_banked_read:
 	; restore temporary zero page
 	stx bank_save
 	pla
@@ -3199,6 +3203,42 @@ x16_banked_copy:
 	pla
 	sta krn_ptr1
 	jmp fat32_read_cont1
+@816_9f_page:
+	; early exit
+	cpy tmp_done
+	beq @end_banked_read
+	; in order to avoid an indexed write into I/O space
+	; on the 65C816, which could have side effects, we
+	; resort to an alternate method here which avoids
+	; this condition, and is at least two cycles shorter
+	; in the loop construct, not counting the setup
+
+	; save old ptr
+	lda fat32_ptr+1
+	pha
+	lda fat32_ptr
+	pha
+
+	stz fat32_ptr
+	lda #$a0
+	sta fat32_ptr+1
+@wrapped_loop:
+	lda (fat32_bufptr),y
+	stx ram_bank
+	sta (fat32_ptr)
+	stz ram_bank
+	iny
+	; we will always have less than 256 bytes to copy
+	; before loop end here, so we only ever need to increment
+	; the low byte of the destination ptr
+	inc fat32_ptr
+	cpy tmp_done
+	bne @wrapped_loop
+	pla
+	sta fat32_ptr
+	pla
+	sta fat32_ptr+1
+	bra @end_banked_read
 
 x16_stream_copy:
 	; move Y (bytecnt) into X for countdown
@@ -3478,9 +3518,12 @@ fat32_write:
 	inx
 	lda #$9f
 	sta fat32_ptr+1
+	set_carry_if_65c816
+	bcs @816_9f_page
 @nowrap:
 	cpy tmp_done
 	bne @loop
+@end_banked_read:
 	; restore temporary zero page
 	stx bank_save
 	pla
@@ -3488,6 +3531,42 @@ fat32_write:
 	pla
 	sta krn_ptr1
 	jmp @4c
+@816_9f_page:
+	; early exit
+	cpy tmp_done
+	beq @end_banked_read
+	; in order to avoid an indexed read from I/O space
+	; on the 65C816, which could have side effects, we
+	; resort to an alternate method here which avoids
+	; this condition, and is at least a cycle shorter
+	; in the loop construct, not counting the setup
+
+	; save old ptr
+	lda fat32_ptr+1
+	pha
+	lda fat32_ptr
+	pha
+
+	stz fat32_ptr
+	lda #$a0
+	sta fat32_ptr+1
+@wrapped_loop:
+	stx ram_bank
+	lda (fat32_ptr)
+	stz ram_bank
+	sta (fat32_bufptr),y
+	iny
+	; we will always have less than 256 bytes to copy
+	; before loop end here, so we only ever need to increment
+	; the low byte of the source ptr
+	inc fat32_ptr
+	cpy tmp_done
+	bne @wrapped_loop
+	pla
+	sta fat32_ptr
+	pla
+	sta fat32_ptr+1
+	bra @end_banked_read
 
 
 ;-----------------------------------------------------------------------------