CSAPP-Architecture lab
这个实验有三个任务，partA,partB,partC。
partA
这部分任务需要在archlab/sim/misc 目录下完成
重要代码：
./yas *.ys
./yis *.yo
gcc -c examples.c
objdump -d examples.c > examples.d
需要查看examples.c反汇编的X86代码，然后写出相应的Y86版本代码：
5 Disassembly of section .text:
6
7 0000000000000000 :
8 0: 55 push %rbp
9 1: 48 89 e5 mov %rsp,%rbp
10 4: 48 89 7d e8 mov %rdi,-0x18(%rbp)
11 8: 48 c7 45 f8 00 00 00 movq $0x0,-0x8(%rbp)
12 f: 00
13 10: eb 17 jmp 29
14 12: 48 8b 45 e8 mov -0x18(%rbp),%rax
15 16: 48 8b 00 mov (%rax),%rax
16 19: 48 01 45 f8 add %rax,-0x8(%rbp)
17 1d: 48 8b 45 e8 mov -0x18(%rbp),%rax
18 21: 48 8b 40 08 mov 0x8(%rax),%rax
19 25: 48 89 45 e8 mov %rax,-0x18(%rbp)
20 29: 48 83 7d e8 00 cmpq $0x0,-0x18(%rbp)
21 2e: 75 e2 jne 12
22 30: 48 8b 45 f8 mov -0x8(%rbp),%rax
23 34: 5d pop %rbp
24 35: c3 retq
26 0000000000000036 :
27 36: 55 push %rbp
28 37: 48 89 e5 mov %rsp,%rbp
29 3a: 48 83 ec 20 sub $0x20,%rsp
30 3e: 48 89 7d e8 mov %rdi,-0x18(%rbp)
31 42: 48 83 7d e8 00 cmpq $0x0,-0x18(%rbp)
32 47: 75 07 jne 50
33 49: b8 00 00 00 00 mov $0x0,%eax
34 4e: eb 2a jmp 7a
35 50: 48 8b 45 e8 mov -0x18(%rbp),%rax
36 54: 48 8b 00 mov (%rax),%rax
37 57: 48 89 45 f0 mov %rax,-0x10(%rbp)
38 5b: 48 8b 45 e8 mov -0x18(%rbp),%rax
39 5f: 48 8b 40 08 mov 0x8(%rax),%rax
40 63: 48 89 c7 mov %rax,%rdi
41 66: e8 00 00 00 00 callq 6b
42 6b: 48 89 45 f8 mov %rax,-0x8(%rbp)
43 6f: 48 8b 55 f0 mov -0x10(%rbp),%rdx
44 73: 48 8b 45 f8 mov -0x8(%rbp),%rax
45 77: 48 01 d0 add %rdx,%rax
46 7a: c9 leaveq
47 7b: c3 retq
49 000000000000007c :
50 7c: 55 push %rbp
51 7d: 48 89 e5 mov %rsp,%rbp
52 80: 48 89 7d e8 mov %rdi,-0x18(%rbp)
53 84: 48 89 75 e0 mov %rsi,-0x20(%rbp)
54 88: 48 89 55 d8 mov %rdx,-0x28(%rbp)
55 8c: 48 c7 45 f0 00 00 00 movq $0x0,-0x10(%rbp)
56 93: 00
57 94: eb 33 jmp c9
58 96: 48 8b 45 e8 mov -0x18(%rbp),%rax
59 9a: 48 8d 50 08 lea 0x8(%rax),%rdx
60 9e: 48 89 55 e8 mov %rdx,-0x18(%rbp)
61 a2: 48 8b 00 mov (%rax),%rax
62 a5: 48 89 45 f8 mov %rax,-0x8(%rbp)
63 a9: 48 8b 45 e0 mov -0x20(%rbp),%rax
64 ad: 48 8d 50 08 lea 0x8(%rax),%rdx
65 b1: 48 89 55 e0 mov %rdx,-0x20(%rbp)
66 b5: 48 8b 55 f8 mov -0x8(%rbp),%rdx
67 b9: 48 89 10 mov %rdx,(%rax)
68 bc: 48 8b 45 f8 mov -0x8(%rbp),%rax
69 c0: 48 31 45 f0 xor %rax,-0x10(%rbp)
70 c4: 48 83 6d d8 01 subq $0x1,-0x28(%rbp)
71 c9: 48 83 7d d8 00 cmpq $0x0,-0x28(%rbp)
72 ce: 7f c6 jg 96
73 d0: 48 8b 45 f0 mov -0x10(%rbp),%rax
74 d4: 5d pop %rbp
75 d5: c3 retq
查看X86的汇编代码，然后照着写Y86的汇编代码。
sum.ys
1 .pos 0
2 irmovq stack,%rsp
3 call main
4 halt
5
6 .align 8
7 ele1:
8 .quad 0x00a
9 .quad ele2
10 ele2:
11 .quad 0x0b0
12 .quad ele3
13 ele3:
14 .quad 0xc00
15 .quad 0
16
17 main:
18 irmovq ele1,%rdi
19 call sum_list
20 ret
21 sum_list:
22 irmovq $8,%r8
23 rrmovq %rdi,%rbx
24 xorq %rax,%rax
25 andq %rbx,%rbx
26 je end
27 loop:
28 mrmovq (%rbx),%rcx
29 addq %rcx,%rax
30 addq %r8,%rbx
31 mrmovq (%rbx),%rbx
32 andq %rbx,%rbx
33 jne loop
34 end:
35 ret
36
37 .pos 0x200
38 stack:
运行：
./yas sum.ys
./yis sum.yo
结果：
Stopped in 30 steps at PC = 0x13. Status 'HLT', CC Z=1 S=0 O=0
Changes to registers:
%rax: 0x0000000000000000 0x0000000000000cba
%rcx: 0x0000000000000000 0x0000000000000c00
%rsp: 0x0000000000000000 0x0000000000000200
%rdi: 0x0000000000000000 0x0000000000000018
%r8: 0x0000000000000000 0x0000000000000008
Changes to memory:
0x01f0: 0x0000000000000000 0x000000000000005b
0x01f8: 0x0000000000000000 0x0000000000000013
rsum.ys
1 .pos 0
2 irmovq stack,%rsp
3 call main
4 halt
5
6 .align 8
7 ele1:
8 .quad 0x00a
9 .quad ele2
10 ele2:
11 .quad 0x0b0
12 .quad ele3
13 ele3:
14 .quad 0xc00
15 .quad 0
16
17 main:
18 irmovq ele1,%rdi
19 xorq %rax,%rax
20 xorq %rbx,%rbx
21 call rsum_list
22 ret
23 rsum_list:
24 andq %rdi,%rdi
25 je end
26 mrmovq (%rdi),%rcx
27 pushq %rcx
28 irmovq $8,%r8
29 addq %r8,%rdi
30 mrmovq (%rdi),%rdi
31 call rsum_list
32 popq %rcx
33 addq %rcx,%rax
34 ret
35 end:
36 ret
37
38 .pos 1000
39 stack:
运行结果：
Stopped in 44 steps at PC = 0x13. Status 'HLT', CC Z=0 S=0 O=0
Changes to registers:
%rax: 0x0000000000000000 0x0000000000000cba
%rcx: 0x0000000000000000 0x000000000000000a
%rsp: 0x0000000000000000 0x00000000000003e8
%r8: 0x0000000000000000 0x0000000000000008
Changes to memory:
0x03a8: 0x0000000000000000 0x0000000000000096
0x03b0: 0x0000000000000000 0x0000000000000c00
0x03b8: 0x0000000000000000 0x0000000000000096
0x03c0: 0x0000000000000000 0x00000000000000b0
0x03c8: 0x0000000000000000 0x0000000000000096
0x03d0: 0x0000000000000000 0x000000000000000a
0x03d8: 0x0000000000000000 0x000000000000005f
0x03e0: 0x0000000000000000 0x0000000000000013
copy_block.ys
1 .pos 0
2 irmovq stack,%rsp
3 call main
4 halt
5
6 .align 8
7 src:
8 .quad 0x00a
9 .quad 0x0b0
10 .quad 0xc00
11 dist:
12 .quad 0x111
13 .quad 0x222
14 .quad 0x333
15
16 main:
17 irmovq src,%rdi
18 irmovq dist,%rsi
19 irmovq $3,%rdx
20 call copy_block
21 ret
22 copy_block:
23 irmovq $8,%r8
24 irmovq $1,%r9
25 xorq %rax,%rax
26 andq %rdx,%rdx
27 je end
28 loop:
29 mrmovq (%rdi),%rbx
30 addq %r8,%rdi
31 rmmovq %rbx,(%rsi)
32 addq %r8,%rsi
33 xorq %rbx,%rax
34 subq %r9,%rdx
35 andq %rdx,%rdx
36 jne loop
37 end:
38 ret
39
40 .pos 1000
41 stack:
运行结果：
Stopped in 38 steps at PC = 0x13. Status 'HLT', CC Z=1 S=0 O=0
Changes to registers:
%rax: 0x0000000000000000 0x0000000000000cba
%rbx: 0x0000000000000000 0x0000000000000c00
%rsp: 0x0000000000000000 0x00000000000003e8
%rsi: 0x0000000000000000 0x0000000000000048
%rdi: 0x0000000000000000 0x0000000000000030
%r8: 0x0000000000000000 0x0000000000000008
%r9: 0x0000000000000000 0x0000000000000001
Changes to memory:
0x0030: 0x0000000000000111 0x000000000000000a
0x0038: 0x0000000000000222 0x00000000000000b0
0x0040: 0x0000000000000333 0x0000000000000c00
0x03d8: 0x0000000000000000 0x000000000000006f
0x03e0: 0x0000000000000000 0x0000000000000013
partB
需要实现一个iaddq指令，可参考下面这几个来实现
iaddq的流程：
210 icode:ifun <-- M1[PC] # 取指
211 rA:rB <-- M1[PC+1]
212 valC <-- M8[PC+2]
213 valP <-- PC+10
214
215 valB <-- R[rB] # 译码
216
217 valE <-- valB + valC # 执行
218
219 R[rB] <-- valE # 写回
220 PC <-- valP # 更新 PC
注意出现IIADDQ的地方
223 然后根据实现修改 hcl 代码：
224
225 ```
226 # 取指
227 # 指令是否有效？
228 bool instr_valid = icode in
230
231 # Does fetched instruction require a regid byte?
232 bool need_regids =
233 icode in { IRRMOVQ, IOPQ, IPUSHQ, IPOPQ, IIRMOVQ, IRMMOVQ, IMRMOVQ, IIADDQ };
234
235 # Does fetched instruction require a constant word?
236 bool need_valC =
237 icode in { IIRMOVQ, IRMMOVQ, IMRMOVQ, IJXX, ICALL, IIADDQ };
238
239 # 译码和写回，指定读入和写入
240 ## What register should be used as the B source?
241 word srcB = [
242 icode in { IOPQ, IRMMOVQ, IMRMOVQ ，IIADDQ } : rB;
243 icode in { IPUSHQ, IPOPQ, ICALL, IRET } : RRSP;
244 1 : RNONE; # Don't need register
245 ];
246
247 ## What register should be used as the E destination?
252 1 : RNONE; # Don't write any register
253 ];
254
260 icode in { ICALL, IPUSHQ } : -8;
261 icode in { IRET, IPOPQ } : 8;
264
265 ## Select input B to ALU
232 bool need_regids =
233 icode in { IRRMOVQ, IOPQ, IPUSHQ, IPOPQ, IIRMOVQ, IRMMOVQ, IMRMOVQ, IIADDQ };
234
235 # Does fetched instruction require a constant word?
236 bool need_valC =
237 icode in { IIRMOVQ, IRMMOVQ, IMRMOVQ, IJXX, ICALL, IIADDQ };
238
239 # 译码和写回，指定读入和写入
240 ## What register should be used as the B source?
241 word srcB = [
242 icode in { IOPQ, IRMMOVQ, IMRMOVQ ，IIADDQ } : rB;
243 icode in { IPUSHQ, IPOPQ, ICALL, IRET } : RRSP;
244 1 : RNONE; # Don't need register
245 ];
246
247 ## What register should be used as the E destination?
252 1 : RNONE; # Don't write any register
253 ];
254
260 icode in { ICALL, IPUSHQ } : -8;
261 icode in { IRET, IPOPQ } : 8;
264
265 ## Select input B to ALU
266 word aluB = [
267 icode in { IRMMOVQ, IMRMOVQ, IOPQ, ICALL, IPUSHQ, IRET, IPOPQ, IIADDQ } : valB;
268 icode in { IRRMOVQ, IIRMOVQ } : 0;
269 # Other instructions don't need ALU
270 ];
271
272 ## Should the condition codes be updated?
273 bool set_cc = icode in { IOPQ, IIADDQ };
接着测试：
完成后 `make` ，注意如果无法使用 gui 界面修改 `Makefile` ，注释掉相关内容，
使用官方文档所给的命令 `unix> ./ssim -t ../y86-code/asumi.yo`
在命令行下测试得到 `ISA Check Succeeds` ，
再用 `unix> (cd ../y86-code; make testssim)` 和
`unix> (cd ../ptest; make SIM=../seq/ssim)` 测试均成功。
接着我用的partA的sum.ys来测试
在程序结束前给%r8再加1，成功了
Stopped in 31 steps at PC = 0x13. Status 'HLT', CC Z=0 S=0 O=0
Changes to registers:
%rax: 0x0000000000000000 0x0000000000000cba
%rcx: 0x0000000000000000 0x0000000000000c00
%rsp: 0x0000000000000000 0x0000000000000200
%rdi: 0x0000000000000000 0x0000000000000018
%r8: 0x0000000000000000 0x0000000000000009
Changes to memory:
0x01f0: 0x0000000000000000 0x000000000000005b
0x01f8: 0x0000000000000000 0x0000000000000013
partC
修改ncopy.ys，目的是熟悉流水线操作，加快程序的执行效率。