This page is for a prior offering of CS 3330. It is not up-to-date.
Changelog:
mrmovq 2, %rax
instruction.In this homework you’ll implement the rest of the SEQ.
You should create a file named seqhw.hcl
and work in that. You should start with the results from seqlab; if you don’t trust your own seqlab solution, an example solution will be available (this will only be posted after the lab is due; note: do not download this if you haven’t yet submitted seqlab!)
By the end of this homework, you will have a fully functioning Y86-64 simulator.
Stat
The Stat
should be
STAT_INS
if the icode
is not one of the ones the book discusses.STAT_HLT
if the icode
is halt
.STAT_AOK
otherwise.jXX
Add the conditional aspect to jXX
. Update the PC to the immediate value if (and only if) the conditions are met (you should already have something like wire conditionsMet:1
from implementing cmovXX
in lab).
If jXX
is correctly implemented, the following (which is found in y86/jxx.yo) should run for 19 steps, visiting hex addresses 0, a, 14, 27, 29, 1d, 14, 27, 29, 1d, 14, 27, 29 1d, 14, 27, 29, 1d, and 26, then halting at address 26:
irmovq $3, %rax
irmovq $-1, %rbx
a:
jmp b
c:
jge a
halt
b:
addq %rbx, %rax
jmp c
You can test this by running the output of the simulator through the grep
tool to select out just a subset of lines:
linux> ./hclrs seqhw.hcl y86/jxx.yo | grep 'pc ='
pc = 0x0; loaded [30 f0 03 00 00 00 00 00 00 00 : irmovq $0x3, %rax]
pc = 0xa; loaded [30 f3 ff ff ff ff ff ff ff ff : irmovq $0xffffffffffff, %rbx]
pc = 0x14; loaded [70 27 00 00 00 00 00 00 00 : jmp 0x27]
pc = 0x27; loaded [60 30 : addq %rbx, %rax]
pc = 0x29; loaded [70 1d 00 00 00 00 00 00 00 : jmp 0x1d]
pc = 0x1d; loaded [75 14 00 00 00 00 00 00 00 : jge 0x14]
pc = 0x14; loaded [70 27 00 00 00 00 00 00 00 : jmp 0x27]
pc = 0x27; loaded [60 30 : addq %rbx, %rax]
pc = 0x29; loaded [70 1d 00 00 00 00 00 00 00 : jmp 0x1d]
pc = 0x1d; loaded [75 14 00 00 00 00 00 00 00 : jge 0x14]
pc = 0x14; loaded [70 27 00 00 00 00 00 00 00 : jmp 0x27]
pc = 0x27; loaded [60 30 : addq %rbx, %rax]
pc = 0x29; loaded [70 1d 00 00 00 00 00 00 00 : jmp 0x1d]
pc = 0x1d; loaded [75 14 00 00 00 00 00 00 00 : jge 0x14]
pc = 0x14; loaded [70 27 00 00 00 00 00 00 00 : jmp 0x27]
pc = 0x27; loaded [60 30 : addq %rbx, %rax]
pc = 0x29; loaded [70 1d 00 00 00 00 00 00 00 : jmp 0x1d]
pc = 0x1d; loaded [75 14 00 00 00 00 00 00 00 : jge 0x14]
pc = 0x26; loaded [00 : halt]
mrmovq
Memory is accessed by setting mem_addr
to the memory address in question and either
mem_readbit
to 0
, mem_writebit
to 1
, and mem_input
to the value to write to memory, which will cause the memory system to write a 4-byte value to memory; ormem_readbit
to 1
and mem_writebit
to 0
, which will cause the memory system to read a 4-byte value from memory into mem_output
.You will also need to compute the memory address as reg_outputB
+ valC
(the book suggests you do this in the ALU, meaning the same mux you used for OPq
’s adding and subtracting).
If both memory moves are implemented correctly, the following (y86/rrmr.yo
) should result in %rdx
containing 0x20000 and address 0xa2 containing byte 0x02.
mrmovq 2, %rax
rmmovq %rax, 160(%rax)
mrmovq 158(%rax), %rdx
(The first instruction in the .yo is encoded as mrmovq 2(REG_NONE), %rax
where REG_NONE
is the register with number 15 — the one that makes the register file not write. When you try to read from the register, you should always get 0. Adding 2 to 0 should give the address 2, so this instruction should result in you reading memory from address 2.)
pushq
Decode: read rA
and %rsp
Execute: add −8 to %rsp
Memory: write reg_outputA
to the address computed by that subtraction
Writeback: write the result of the subtraction back into %rsp
The following code (y86/push.yo
)
irmovq $3, %rax
irmovq $256, %rsp
pushq %rax
should leave a 0x03 in address 0xf8 and an 0xf8 in %rsp
popq
Decode: read %rsp
Execute: add +8 to %rsp
Memory: read from the pre-added %rsp
address
Writeback: write both (1) the result of the addition back into %rsp
and (2) the results of the read into rA
The following code (y86/pop.yo
)
irmovq $4, %rsp
popq %rax
should leave a 0xc in %rsp
and a 0xfb0000000000000 in %rax
call
call
is like pushq
and jmp
in general form
Decode: read %rsp
Execute: add −8 to %rsp
Memory: write the next instruction address (valP
) to the address computed by that subtraction
Writeback: write the result of the subtraction back into %rsp
PC Update: the new PC (p_pc
) should be valC
, not valP
.
The following code (y86/call.yo
)
irmovq $256, %rsp
call a
addq %rsp, %rsp
a:
halt
should leave 0xf8 in %rsp
and a 0x13 in address 0xf8
ret
ret
is like popq
and jmp
in general form
Decode: read %rsp
Execute: add +8 to %rsp
Memory: read from the pre-added %rsp
address
Writeback: write the result of the addition back into %rsp
PC Update: the new PC (p_pc
) should be the value read from memory (mem_output
), not valP
.
The following code (y86/ret.yo
)
irmovq $256, %rsp
irmovq a, %rbx
rmmovq %rbx, (%rsp)
ret
halt
a:
irmovq $258, %rax
halt
should run 6 cycles, leave %rax
as 0x102, and %rsp
as 0x108
You can run the command make test-seqhw
to test your processor over almost all the .yo
files in the y86/
folder, comparing the output to supplied outputs in testdata/seq-reference
and testdata/seq-traces
. If your processor disagrees, you may find the traces in testdata/seq-traces
helpful for debugging. It is okay if you disagree on poptest.yo
. (poptest.yo
has been removed from the list of tests in testdata/seqhw-tests.txt
in the most recent version of hclrs.tar
)
In addition, your code should behave the same as tools/yis
when run on anything in the y86/
folder except
asumi.yo
and ins.yo
(which use instructions not in the y86 basic set). Note that you should still agree with our reference files (that make test-seqhw
uses) for these tests.pushquestion.yo
(which is ambiguous and may or may not work the same as yis
; we don’t care either way), poptest.yo
(which is similarly ambiguous)prog10.yo
(which gives an address error, a limitation we are not implementing in our simulator).Submit a file named seqhw.hcl
on the submission page.