1
0
mirror of https://github.com/sgmarz/osblog.git synced 2024-11-28 04:13:31 +04:00
osblog/risc_v/src/test.rs

148 lines
5.2 KiB
Rust
Raw Normal View History

2020-04-22 04:30:09 +04:00
// test.rs
2020-04-26 06:25:32 +04:00
use crate::{cpu::{build_satp,
memcpy,
satp_fence_asid,
CpuMode,
SatpMode,
TrapFrame},
kmem::{kfree, kmalloc},
page::{map, zalloc, EntryBits, Table, PAGE_SIZE},
process::{Process,
2020-04-26 06:25:32 +04:00
ProcessData,
ProcessState,
NEXT_PID,
PROCESS_LIST,
PROCESS_STARTING_ADDR,
STACK_ADDR,
2020-04-26 06:25:32 +04:00
STACK_PAGES},
syscall::syscall_fs_read};
pub fn test_block() {
2020-04-26 06:25:32 +04:00
// The bytes to read would usually come from the inode, but we are in an
// interrupt context right now, so we cannot pause. Usually, this would be done
// by an exec system call.
2020-04-26 00:06:29 +04:00
let bytes_to_read = 1024 * 50;
let buffer = kmalloc(bytes_to_read);
2020-04-26 06:25:32 +04:00
// Read the file from the disk.
let bytes_read = syscall_fs_read(8, 8, buffer, bytes_to_read as u32, 0);
2020-04-26 06:25:32 +04:00
// After compiling our program, I manually looked and saw it was 12,288
// bytes. So, to make sure we got the right one, I do a manual check
// here.
if bytes_read != 12288 {
println!(
"Unable to load program at inode 8, which should be \
12,288 bytes, got {}",
bytes_read
);
}
else {
// Let's get this program running!
2020-04-26 06:25:32 +04:00
// Everything is "page" based since we're going to map pages to
// user space. So, we need to know how many program pages we
// need. Each page is 4096 bytes.
let program_pages = (bytes_read / PAGE_SIZE) + 1;
let my_pid = unsafe { NEXT_PID + 1 };
unsafe {
NEXT_PID += 1;
}
2020-04-26 06:25:32 +04:00
let mut my_proc =
Process { frame: zalloc(1) as *mut TrapFrame,
stack: zalloc(STACK_PAGES),
pid: my_pid,
root: zalloc(1) as *mut Table,
state: ProcessState::Running,
data: ProcessData::zero(),
2020-04-26 06:25:32 +04:00
sleep_until: 0,
program: zalloc(program_pages), };
// Map the program in the MMU.
let ptr = my_proc.program;
unsafe {
memcpy(ptr, buffer, bytes_read);
}
let table = unsafe { my_proc.root.as_mut().unwrap() };
2020-04-26 06:25:32 +04:00
// This will map all of the program pages. Notice that in linker.lds in userspace
// we set the entry point address to 0x2000_0000. This is the same address as
// PROCESS_STARTING_ADDR, and they must match.
for i in 0..program_pages {
2020-04-26 06:25:32 +04:00
let vaddr = PROCESS_STARTING_ADDR + i * PAGE_SIZE;
let paddr = ptr as usize + i * PAGE_SIZE;
// We don't have an ELF loader yet, so we're loading raw binaries into memory. Since
// it is a flat binary, all .data, .rodata, and .bss sections get wrapped into
// the .text section. Normally, we don't want the .text section to be writeable,
// however because of this "flattening", we don't have a choice.
// Notice that USER shows up here. Since we're running in user mode, this bit MUST
// BE SET! Otherwise, we'll get a page fault from the beginning.
map(
table,
vaddr,
paddr,
EntryBits::UserReadWriteExecute.val(),
0,
);
}
// Map the stack
let ptr = my_proc.stack as *mut u8;
for i in 0..STACK_PAGES {
2020-04-26 06:25:32 +04:00
let vaddr = STACK_ADDR + i * PAGE_SIZE;
let paddr = ptr as usize + i * PAGE_SIZE;
// We create the stack. We don't load a stack from the disk. This is why I don't
// need to make the stack executable.
map(
table,
vaddr,
paddr,
EntryBits::UserReadWrite.val(),
0,
);
}
// Set everything up in the trap frame
unsafe {
2020-04-26 06:25:32 +04:00
// The program counter is a virtual memory address and is loaded into mepc
// when we execute mret.
(*my_proc.frame).pc = PROCESS_STARTING_ADDR;
2020-04-26 06:25:32 +04:00
// Stack pointer. The stack starts at the bottom and works its way up, so we have to
// set the stack pointer to the bottom.
(*my_proc.frame).regs[2] =
2020-04-26 06:25:32 +04:00
STACK_ADDR as usize + STACK_PAGES * PAGE_SIZE;
// USER MODE! This is how we set what'll go into mstatus when we run the process.
(*my_proc.frame).mode = CpuMode::User as usize;
(*my_proc.frame).pid = my_proc.pid as usize;
2020-04-26 06:25:32 +04:00
// The SATP register is used for the MMU, so we need to
// map our table into that register. The switch_to_user
// function will load .satp into the actual register
// when the time comes.
(*my_proc.frame).satp =
build_satp(
SatpMode::Sv39,
my_proc.pid as usize,
my_proc.root as usize,
);
}
2020-04-26 06:25:32 +04:00
// We don't reuse PIDs, so this really shouldn't matter.
satp_fence_asid(my_pid as usize);
// I took a different tact here than in process.rs. In there I created the process
// while holding onto the process list. It doesn't really matter since this is synchronous,
// but it might get dicey
if let Some(mut pl) = unsafe { PROCESS_LIST.take() } {
2020-04-26 06:25:32 +04:00
println!(
"Added user process to the scheduler...get \
ready for take-off!"
);
// As soon as we push this process on the list, it'll be schedule-able.
pl.push_back(my_proc);
unsafe {
PROCESS_LIST.replace(pl);
2020-04-26 00:06:29 +04:00
}
}
else {
println!("Unable to spawn process.");
2020-04-26 06:25:32 +04:00
// Since my_proc couldn't enter the process list, it
// will be dropped and all of the associated allocations
// will be deallocated.
}
2020-04-26 00:06:29 +04:00
}
println!();
kfree(buffer);
}