As the avr is an 8-bit processor, it is important to handle data as 8-bit when
possible, even though the C standards force 8-bit data to be promoted to 16-bit
ints in many situations. Common cases are when doing logic operations on chars
- when legally valid (i.e., the results are the same as for 16-bit operations),
the generated code should use 8-bit operations. In particular, logic
operations using constants such as 0 and 0xff should be optimised.
Optimisation works well in many cases, but fails when the expressions get
slightly more complex:
extern uint8_t data[64];
uint8_t bar(uint8_t x, uint8_t y) {
return data[y ^ (x & 0x0f)];
}
uint8_t bar2(uint8_t x, uint8_t y) {
return data[(y ^ x) & 0x0f];
}
40 bar:
41 /* prologue: frame size=0 */
42 /* prologue end (size=0) */
43 0000 E82F mov r30,r24 ; x, x
44 0002 F0E0 ldi r31,lo8(0) ; x,
45 0004 EF70 andi r30,lo8(15) ; x,
46 0006 F070 andi r31,hi8(15) ; x,
47 0008 70E0 ldi r23,lo8(0) ; y,
48 000a E627 eor r30,r22 ; x, y
49 000c F727 eor r31,r23 ; x, y
50 000e E050 subi r30,lo8(-(data)) ; x,
51 0010 F040 sbci r31,hi8(-(data)) ; x,
52 0012 8081 ld r24,Z ; tmp51, data
53 0014 90E0 ldi r25,lo8(0) ; <result>,
54 /* epilogue: frame size=0 */
55 0016 0895 ret
56 /* epilogue end (size=1) */
57 /* function bar size 12 (11) */
61 bar2:
62 /* prologue: frame size=0 */
63 /* prologue end (size=0) */
64 0018 E62F mov r30,r22 ; y, y
65 001a E827 eor r30,r24 ; y, x
66 001c F0E0 ldi r31,lo8(0) ; ,
67 001e EF70 andi r30,lo8(15) ; tmp46,
68 0020 F070 andi r31,hi8(15) ; tmp46,
69 0022 E050 subi r30,lo8(-(data)) ; tmp46,
70 0024 F040 sbci r31,hi8(-(data)) ; tmp46,
71 0026 8081 ld r24,Z ; tmp50, data
72 0028 90E0 ldi r25,lo8(0) ; <result>,
73 /* epilogue: frame size=0 */
74 002a 0895 ret
75 /* epilogue end (size=1) */
76 /* function bar2 size 10 (9) */
The first function "bar" has several clear improvements - it does all the logic
operations as 16-bit. In the second case, the "eor" is nicely handled as
8-bit, but the "& 0x0f" is done as 16-bit - there is an unnecessary "andi r31,
hi8(15)" instruction.
--
Summary: Missed optimisation on avr - optimisation of 8-bit logic
sometimes fails
Product: gcc
Version: 4.2.2
Status: UNCONFIRMED
Severity: enhancement
Priority: P3
Component: c
AssignedTo: unassigned at gcc dot gnu dot org
ReportedBy: david at westcontrol dot com
GCC target triplet: avrgcc
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=34791
