# Envrionment Setting

My Computer information (which is required):

OS: Linux x86_64 (6.1.53-1-MANJARO)
CPU: 13th Gen Intel i5-13500H (16) @ 4.700GHz
GPU: Intel Raptor Lake-P [Iris Xe Graphics]
Memory: 16GB
Editor: VS Code
C Compiler: gcc(version 13.2.1) clang(version 16.0.6)
C++ Compiler: g++(version 13.2.1) clang++(version 16.0.6)
C/C++ build tools: cmake(version 3.27.6) xmake(version 2.8.3) make(version 4.4.1)
Rust: rustc 1.73.0-nightly
Python: Python 3.11.5
Java: openjdk 21 2023-09-19
Go: version go1.21.1 linux/amd64

# How to build

Run command below in your shell:

git clone https://gitee.com/ipads-lab/chcore-lab-v2.git
cd chcore-lab-v2
make build

and then it will pull a image from docker hub, which is called chcore_builder.

If error like “failed to add the host (veth28449b0) <=> sandbox (veth45f4af9) pair interfaces: operation not supported.” occurs, try to restart your computer or docker

Now, you’ve already build a file named “kernel.img” in directory “build/”.

# How to run

Simply run :

make qemu

to run Chcore

Note that there is no output, which is normal. It won’t give any output until you finish lab1

# How to debug

You need install gdb-multiarch to debug. For example, in my manjaro, run :

yay -S gdb-multiarch

If you’re in Ubuntu, just run :

sudo apt-get install gdb-multiarch

After installing, using two bash(or anything else such as tmux), one run :

make qemu-gdb

to activate qemu’s gdb mode, the other run :

make gdb

to connect the remote target(i.e. the qemu’s gdb mode).

If you see this, then it’s success.

# How to edit code

I strongly recommand VS Code, just cd into this directory and type code ., then you can use it to edit the code.

# Machine Boot

The lab document is in docs/, follow it to finish the lab.

# Pre

Modify Makefile like this :

qemu: build
	$(V)$(_QEMU) $(QEMU_OPTS)

qemu-gdb: build
	$(V)$(_QEMU) -S -gdb tcp::$(QEMU_GDB_PORT) $(QEMU_OPTS)

That would avoid run code but do not update it

# Q1

阅读 _start 函数的开头，尝试说明 ChCore 是如何让其中一个核首先进入初始化流程，并让其他核暂停执行的。

Read the cpu id from mpidr_el1 register to the x8 register
Get the low 8-bit value of the register x8
Check if low 8-bit is 0, if true, then jump to function primary, else, make a infnity loop to build a fence to stop others cores at this instruction
function primary will enter a initial procedure, including set the exception level to EL1 and stack size, transfer control to the kernel

for more: register mpidr_el1 store the information of core, in multi-processor system, every processor has a unique value of mpidr_el1.

# M2

练习题 2：在 arm64_elX_to_el1 函数的 LAB 1 TODO 1 处填写一行汇编代码，获取 CPU 当前异常级别。

mrs x9, CurrentEL

# M3

Look at label .Lno_gic_sr, that is how to jump to el1 from el2.

So we write code like this :

	adr x9, .Ltarget
	msr elr_el3, x9
	mov x9, SPSR_ELX_DAIF | SPSR_ELX_EL1H
	msr spsr_el3, x9

to ensure our implementation is right, we use gdb below :

we will see that it jump back _start from function arm64_elX_to_el1

# Q4

结合此前 ICS 课的知识，并参考 kernel.img 的反汇编（通过 aarch64-linux-gnu-objdump -S 可获得），说明为什么要在进入 C 函数之前设置启动栈。如果不设置，会发生什么？

We can see the assembly below :

   80010:	94001ffc 	bl	88000 <arm64_elX_to_el1>
   80014:	580000a0 	ldr	x0, 80028 <primary+0x18>
   80018:	91400400 	add	x0, x0, #0x1, lsl #12
   8001c:	9100001f 	mov	sp, x0
   80020:	940020e2 	bl	883a8 <init_c>

note that =boot_cpu_stack is *0x80028(i.e. 0x00088980), and INIT_STACK_SIZE is 0x1000(i.e. 0x1, lsl #12)

Because init_c is a function which require a stack frame. That’s why we should initial the stack pointer and allocate stack space.

more details in rCore

# Q5

在实验 1 中，其实不调用 clear_bss 也不影响内核的执行，请思考不清理 .bss 段在之后的何种情况下会导致内核无法工作。

Anser from GPT:

当不清零 .bss 段时，未显式初始化的全局变量和静态变量的初始值将是未定义的。这可能导致以下问题和错误：
未定义的行为：访问未初始化的变量可能导致未定义的行为。这包括读取随机的内存值、产生不可预测的结果以及可能导致程序崩溃或错误的行为。
依赖于零值的假设：在程序设计中，有时会假设未初始化的全局变量或静态变量的初始值为零。如果这些变量的初始值不是零，那么这些假设就会被破坏，可能导致程序逻辑错误或不一致。
安全问题：如果未初始化的变量被认为是零值，而实际上它们存储了其他敏感信息，可能会引发安全问题。例如，如果未初始化的变量被用于存储密码、密钥或其他敏感数据，那么不清零可能导致这些敏感信息泄露。
跨平台兼容性问题：不同的编译器和操作系统对未初始化变量的处理可能不同。在某些平台上，编译器或操作系统可能为未初始化变量提供默认的零值，而在其他平台上可能不会。因此，不清零 .bss 段可能导致在不同的平台上出现不一致的行为，影响跨平台的兼容性。

# M6

void uart_send_string(char *str)
{
        /* LAB 1 TODO 3 BEGIN */

        while (*str != '\0') {
                early_uart_send((unsigned int)(*str));
                str++;
        }

        /* LAB 1 TODO 3 END */
}

# M7

I wrote this in first

orr x8, x8, #0x82f

which enable all flags.

However, we can see that below here:

	bic     x8, x8, #SCTLR_EL1_A
	bic     x8, x8, #SCTLR_EL1_SA0
	bic     x8, x8, #SCTLR_EL1_SA
	orr     x8, x8, #SCTLR_EL1_nAA

it reminds us that we should just enable one flag here, i.e. flag M, so the correct code is :

orr x8, x8, #SCTLR_EL1_M

# How to Grade

Remember to install expect， in Manjaro you can :

yay -S expect

then type :

make grade

you will see something like this:

Chcore lab0&1 Machine Boot

Set the envrionment and boot the machine