Implement kernel command line with init configuring

Cargo.lock

@@ -531,6 +531,7 @@ dependencies = [
  "log",
  "multiboot2",
  "pod",
+ "regex",
  "spin 0.9.8",
  "trapframe",
  "volatile",
@@ -979,6 +980,31 @@ dependencies = [
  "bitflags 1.3.2",
 ]
 
+[[package]]
+name = "regex"
+version = "1.9.1"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b2eae68fc220f7cf2532e4494aded17545fce192d59cd996e0fe7887f4ceb575"
+dependencies = [
+ "regex-automata",
+ "regex-syntax",
+]
+
+[[package]]
+name = "regex-automata"
+version = "0.3.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "39354c10dd07468c2e73926b23bb9c2caca74c5501e38a35da70406f1d923310"
+dependencies = [
+ "regex-syntax",
+]
+
+[[package]]
+name = "regex-syntax"
+version = "0.7.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "e5ea92a5b6195c6ef2a0295ea818b312502c6fc94dde986c5553242e18fd4ce2"
+
 [[package]]
 name = "ringbuf"
 version = "0.3.2"

build/grub/conf/grub.cfg

@@ -2,7 +2,7 @@ set timeout_style=menu
 set timeout=1
 
 menuentry 'jinux' {
-    multiboot2 /boot/jinux
+    multiboot2 /boot/jinux SHELL="/bin/sh" LOGNAME="root" HOME="/" USER="root" PATH="/bin" init=/usr/bin/busybox -- sh -l
     module2 --nounzip /boot/ramdisk.cpio.gz
     boot
 }

framework/jinux-frame/Cargo.toml

@@ -18,12 +18,13 @@ log = "0.4"
 lazy_static = { version = "1.0", features = ["spin_no_std"] }
 trapframe = { git = "https://github.com/sdww0/trapframe-rs", rev = "e886763" }
 inherit-methods-macro = { git = "https://github.com/jinzhao-dev/inherit-methods-macro", rev = "98f7e3e" }
-multiboot2 = "0.16.0"
+regex = { version = "1.9.1", default-features = false, features = ["unicode"] }
 
 [target.x86_64-custom.dependencies]
 x86_64 = "0.14.2"
 x86 = "0.52.0"
 acpi = "4.1.1"
 aml = "0.16.3"
+multiboot2 = { version = "0.16.0", features = [] }
 
 [features]

framework/jinux-frame/src/arch/x86/boot/multiboot2.rs

@@ -1,9 +1,13 @@
-use alloc::{string::{ToString, String}, vec::Vec};
+use alloc::{
+    string::{String, ToString},
+    vec::Vec,
+};
 use multiboot2::{BootInformation, BootInformationHeader, MemoryAreaType};
 
 use crate::boot::{
+    kcmdline::KCmdlineArg,
     memory_region::{MemoryRegion, MemoryRegionType},
-    kcmdline::KCmdlineArg, BootloaderAcpiArg, BootloaderFramebufferArg,
+    BootloaderAcpiArg, BootloaderFramebufferArg,
 };
 use core::{arch::global_asm, mem::swap};
 use spin::Once;
@@ -35,9 +39,9 @@ fn init_kernel_commandline(kernel_cmdline: &'static Once<KCmdlineArg>) {
             .get()
             .unwrap()
             .command_line_tag()
-            .expect("Kernel commandline not found from the Multiboot2 header!")
+            .expect("Kernel command-line not found from the Multiboot2 header!")
             .cmdline()
-            .expect("UTF-8 error: failed to parse kernel commandline!")
+            .expect("UTF-8 error: failed to parse kernel command-line!")
             .into()
     });
 }

framework/jinux-frame/src/boot/kcmdline.rs

@@ -1,23 +1,189 @@
-//! The module to parse kernel commandline arguments.
+//! The module to parse kernel command-line arguments.
+//!
+//! The format of the Jinux command line string conforms
+//! to the Linux kernel command line rules:
+//!
+//! https://www.kernel.org/doc/html/v6.4/admin-guide/kernel-parameters.html
 //!
 
-use alloc::string::String;
+use alloc::{
+    collections::BTreeMap,
+    ffi::CString,
+    string::{String, ToString},
+    vec,
+    vec::Vec,
+};
+use log::debug;
+use regex::Regex;
 
-/// The struct to store parsed kernel commandline arguments.
+#[derive(PartialEq, Debug)]
+struct InitprocArg {
+    // Since environment arguments can precede the init path argument, we
+    // have no choice but to wrap the path in `Option` and check it later.
+    path: Option<String>,
+    argv: Vec<CString>,
+    envp: Vec<CString>,
+}
+
+/// The struct to store the parsed kernel command-line arguments.
+#[derive(Debug)]
 pub struct KCmdlineArg {
-    initproc: Option<String>,
+    initproc: Option<InitprocArg>,
+    module_args: BTreeMap<String, Vec<CString>>,
 }
 
 // Define get APIs.
 impl KCmdlineArg {
-    pub fn get_initproc(&self) -> Option<&str> {
-        self.initproc.as_deref()
+    /// Get the path of the initprocess.
+    pub fn get_initproc_path(&self) -> Option<&str> {
+        self.initproc
+            .as_ref()
+            .and_then(|i| i.path.as_ref())
+            .map(|s| s.as_str())
+    }
+    /// Get the argument vector(argv) of the initprocess.
+    pub fn get_initproc_argv(&self) -> Option<&Vec<CString>> {
+        self.initproc.as_ref().map(|i| &i.argv)
+    }
+    /// Get the environment vector(envp) of the initprocess.
+    pub fn get_initproc_envp(&self) -> Option<&Vec<CString>> {
+        self.initproc.as_ref().map(|i| &i.argv)
+    }
+    /// Get the argument vector of a kernel module.
+    pub fn get_module_args(&self, module: &str) -> Option<&Vec<CString>> {
+        self.module_args.get(module)
     }
 }
 
 // Define the way to parse a string to `KCmdlineArg`.
 impl From<&str> for KCmdlineArg {
     fn from(cmdline: &str) -> Self {
-        KCmdlineArg { initproc: None }
+        // What we construct.
+        let mut result = KCmdlineArg {
+            initproc: None,
+            module_args: BTreeMap::new(),
+        };
+
+        // Split the command line string by spaces but preserve
+        // ones that are protected by double quotes(`"`).
+        let re = Regex::new(r#"((\S*"[^"]*"\S*)+|\S+)"#).unwrap();
+        // Every thing after the "--" mark is the initproc arguments.
+        let mut kcmdline_end = false;
+
+        // The main parse loop. The processing steps are arranged (not very strictly)
+        // by the analysis over the Backus–Naur form syntax tree.
+        for arg in re.find_iter(cmdline).map(|m| m.as_str()) {
+            if arg == "" || arg == " " {
+                continue;
+            }
+            // Cmdline => KernelArg "--" InitArg
+            // KernelArg => Arg "\s+" KernelArg | %empty
+            // InitArg => Arg "\s+" InitArg | %empty
+            if kcmdline_end {
+                if let Some(&mut ref mut i) = result.initproc.as_mut() {
+                    i.argv.push(CString::new(arg).unwrap());
+                } else {
+                    panic!("Initproc arguments provided but no initproc path specified!");
+                }
+                continue;
+            }
+            if arg == "--" {
+                kcmdline_end = true;
+                continue;
+            }
+            // Arg => Entry | Entry "=" Value
+            let arg_pattern: Vec<_> = arg.split("=").collect();
+            let (entry, value) = match arg_pattern.len() {
+                1 => (arg_pattern[0], None),
+                2 => (arg_pattern[0], Some(arg_pattern[1])),
+                _ => {
+                    panic!("Unable to parse argument {}", arg);
+                }
+            };
+            // Entry => Module "." ModuleOptionName | KernelOptionName
+            let entry_pattern: Vec<_> = entry.split(".").collect();
+            let (node, option) = match entry_pattern.len() {
+                1 => (None, entry_pattern[0]),
+                2 => (Some(entry_pattern[0]), entry_pattern[1]),
+                _ => {
+                    panic!("Unable to parse entry {} in argument {}", entry, arg);
+                }
+            };
+            if let Some(modname) = node {
+                let modarg = if let Some(v) = value {
+                    CString::new(option.to_string() + "=" + v).unwrap()
+                } else {
+                    CString::new(option).unwrap()
+                };
+                result
+                    .module_args
+                    .entry(modname.to_string())
+                    .and_modify(|v| v.push(modarg.clone()))
+                    .or_insert(vec![modarg.clone()]);
+                continue;
+            }
+            // KernelOptionName => /*literal string alternatives*/ | /*init environment*/
+            if let Some(value) = value {
+                // The option has a value.
+                match option {
+                    "init" => {
+                        if let Some(&mut ref mut i) = result.initproc.as_mut() {
+                            if let Some(v) = &i.path {
+                                panic!("Initproc assigned twice in the command line!");
+                            }
+                            i.path = Some(value.to_string());
+                        } else {
+                            result.initproc = Some(InitprocArg {
+                                path: Some(value.to_string()),
+                                argv: Vec::new(),
+                                envp: Vec::new(),
+                            });
+                        }
+                    }
+                    _ => {
+                        // If the option is not recognized, it is passed to the initproc.
+                        // Pattern 'option=value' is treated as the init environment.
+                        let envp_entry = CString::new(option.to_string() + "=" + value).unwrap();
+                        if let Some(&mut ref mut i) = result.initproc.as_mut() {
+                            i.envp.push(envp_entry);
+                        } else {
+                            result.initproc = Some(InitprocArg {
+                                path: None,
+                                argv: Vec::new(),
+                                envp: vec![envp_entry],
+                            });
+                        }
+                    }
+                }
+            } else {
+                // There is no value, the entry is only a option.
+                match option {
+                    _ => {
+                        // If the option is not recognized, it is passed to the initproc.
+                        // Pattern 'option' without value is treated as the init argument.
+                        let argv_entry = CString::new(option.to_string()).unwrap();
+                        if let Some(&mut ref mut i) = result.initproc.as_mut() {
+                            i.argv.push(argv_entry);
+                        } else {
+                            result.initproc = Some(InitprocArg {
+                                path: None,
+                                argv: vec![argv_entry],
+                                envp: Vec::new(),
+                            });
+                        }
+                    }
+                }
+            }
+        }
+
+        debug!("{:?}", result);
+
+        if let Some(&ref i) = result.initproc.as_ref() {
+            if i.path == None {
+                panic!("Initproc arguments provided but no initproc! Maybe have bad option.");
+            }
+        }
+
+        result
     }
 }