Error when compiling simple SPARC ASM math code
Writing SPARC asm code to evaluate a hardcoded statement, but I'm getting an error I don't understand. I've searched all over, and while it seems to come up a lot in some bug reports out there, there's no real clues that I've found for programmers. Yes, it is homework, and yes, I'm not finished, and yes I have branch delays all over the place. I'll get to them on my own, but I need to know what the error is. This error's not telling me anything useful, and the books I have aren't any good for it either.
I'm really new at this, so any help would be greatly appreciated.
1 /*Justin Reeves*/
2 /*max{x^3-14x^2+56x-64} from [-2,8]*/
3 /*for (x = lwr, x <= upr, x++) */
4 define(lwr_b, -2) !lower bound
5 define(upr_b, 8) !upper bound
6 define(x_r, %l0) !x
7 define(sum_r, %l1) !sum, each pass of loop may update
8 define(max_r, %l2) !max, cmp to sum, store in max if larger
9
10 .global main
11 main:
12 save %sp, -64, %sp
13
14 ba loop_test
15 mov lwr_b, x_r /*init x_r = -2 */
16
17 loop_test:
18 cmp x_r, upr_b
19 ble sum_loop
20 nop
21 /*then x > upr_b and the max has been found*/
22 /*odd spot for it...but this is the end of the program*/
23
24
25 sum_loop: ! starting backwards to give us an intial nonzero constant sum
26 mov -64, sum_r /* sum = -64 */
27
28 mov x_r, %o0 /*56x*/
29 mov 56, %o0
30 clr %o2
31 call .mul /*AFAIK 56x should now be in %o0*/
32 nop
33 add sum_r, %o0, sum_r /* sum = 56x-64 */
34
35 mov x_r, %o0 /* 14x^2 */
36 mov x_r, %o1
37 mov -14, %o2
38 call .mul
39 nop
40 add sum_r, %o0, sum_r /* sum = -14x^2+56x-64 */
41
42 mov x_r, %o0
43 mov x_r, %o1
44 mov x_r, %o2
45 call .mul
46 nop
47 add sum_r, %o0, sum_r /*sum = x^3-14x^2+56x-64 */
48
49 add x_r, 1, 开发者_运维问答 x_r /* x++ */
50 cmp sum_r, max_r
51 bge collect_lrg /*branches if sum > max*/
52 nop
53
54 collect_lrg:
55 mov sum_r, max_r
56 ba loop_test
57
58 mov 1, %g1 /*exit request*/
59 ta 0 /*trap to system*/
then when I try to define the macros and compile I get:
cs32107@matrix:~$ m4 polynomialv2.m > polynomial.s
cs32107@matrix:~$ gcc -g polynomial.s -o polynomial ld: fatal: relocation error: R_SPARC_32: file /var/tmp//ccVOrnx2.o: symbol : offset 0xfb5d11dd is non-aligned ld: fatal: relocation error: R_SPARC_32: file /var/tmp//ccVOrnx2.o: symbol : offset 0xfb5d120f is non-aligned ld: fatal: relocation error: R_SPARC_32: file /var/tmp//ccVOrnx2.o: symbol : offset 0xfb5d1215 is non-aligned ld: fatal: relocation error: R_SPARC_32: file /var/tmp//ccVOrnx2.o: symbol : offset 0xfb5d1219 is non-aligned ld: fatal: relocation error: R_SPARC_32: file /var/tmp//ccVOrnx2.o: symbol : offset 0xfb5d121d is non-aligned ld: fatal: relocation error: R_SPARC_32: file /var/tmp//ccVOrnx2.o: symbol : offset 0xfb5d1266 is non-aligned collect2: ld returned 1 exit status cs32107@matrix:~$A bug in the assembler causes debugging code to add unaligned data access. Don't use -g but perhaps -gstabs if you need debugging information. There maybe an update for gas that fixes the problem too.
精彩评论