Add parser for hostname and hosts

src/libos/Cargo.lock

@@ -1,5 +1,7 @@
 # This file is automatically @generated by Cargo.
 # It is not intended for manual editing.
+version = 3
+
 [[package]]
 name = "Occlum"
 version = "0.27.0"
@@ -22,6 +24,7 @@ dependencies = [
  "rcore-fs-ramfs",
  "rcore-fs-sefs",
  "rcore-fs-unionfs",
+ "regex",
  "resolv-conf",
  "ringbuf",
  "scroll",
@@ -518,6 +521,23 @@ dependencies = [
  "rand_core 0.3.1",
 ]
 
+[[package]]
+name = "regex"
+version = "1.3.1"
+source = "git+https://github.com/mesalock-linux/regex-sgx#76aef86f9836532d17764523d0fa23bb7d2e31cf"
+dependencies = [
+ "regex-syntax",
+ "sgx_tstd",
+]
+
+[[package]]
+name = "regex-syntax"
+version = "0.6.12"
+source = "git+https://github.com/mesalock-linux/regex-sgx#76aef86f9836532d17764523d0fa23bb7d2e31cf"
+dependencies = [
+ "sgx_tstd",
+]
+
 [[package]]
 name = "resolv-conf"
 version = "0.7.0"

src/libos/Cargo.toml

@@ -29,6 +29,7 @@ memoffset = "0.6.1"
 scroll = { version = "0.10.2", default-features = false }
 itertools = { version = "0.10.0", default-features = false, features = ["use_alloc"]  }
 ctor = "0.1"
+regex = { git = "https://github.com/mesalock-linux/regex-sgx", default-features = false, features = ["std", "unicode", "mesalock_sgx"] }
 
 [patch.'https://github.com/apache/teaclave-sgx-sdk.git']
 sgx_tstd = { path = "../../deps/rust-sgx-sdk/sgx_tstd" }

src/libos/include/edl/occlum_edl_types.h

@@ -29,11 +29,11 @@ typedef struct itimerspec{
     struct _timespec it_value;
 } itimerspec_t;
 
-// todo: more detailed description
+// The host_file_buffer struct includes /etc/resolv.conf, /etc/hosts and /etc/hostname buffer
 struct host_file_buffer {
-    const char* resolv_conf_ptr;
-    const char* hosts_ptr;
-    const char* hostname_ptr;
+    const char* resolv_conf_buf;
+    const char* hosts_buf;
+    const char* hostname_buf;
 };
 
 #define FD_SETSIZE 1024

src/libos/src/entry.rs

@@ -98,7 +98,7 @@ pub extern "C" fn occlum_ecall_init(
     });
 
     // Parse host file
-    let resolv_conf_ptr = unsafe { (*file_buffer).resolv_conf_ptr };
+    let resolv_conf_ptr = unsafe { (*file_buffer).resolv_conf_buf };
     match parse_host_file(HostFile::RESOLV_CONF, resolv_conf_ptr) {
         Err(e) => {
             error!("failed to parse /etc/resolv.conf: {}", e.backtrace());
@@ -111,12 +111,13 @@ pub extern "C" fn occlum_ecall_init(
         }
     }
 
-    let hostname_ptr = unsafe { (*file_buffer).hostname_ptr };
+    let hostname_ptr = unsafe { (*file_buffer).hostname_buf };
     match parse_host_file(HostFile::HOSTNAME, hostname_ptr) {
         Err(e) => {
             error!("failed to parse /etc/hostname: {}", e.backtrace());
         }
         Ok(hostname_str) => {
+            misc::init_nodename(&hostname_str);
             *HOSTNAME_STR.write().unwrap() = Some(hostname_str);
             if let Err(e) = write_host_file(HostFile::HOSTNAME) {
                 error!("failed to write /etc/hostname: {}", e.backtrace());
@@ -124,7 +125,7 @@ pub extern "C" fn occlum_ecall_init(
         }
     }
 
-    let hosts_ptr = unsafe { (*file_buffer).hosts_ptr };
+    let hosts_ptr = unsafe { (*file_buffer).hosts_buf };
     match parse_host_file(HostFile::HOSTS, hosts_ptr) {
         Err(e) => {
             error!("failed to parse /etc/hosts: {}", e.backtrace());

src/libos/src/misc/mod.rs

@@ -10,4 +10,4 @@ mod uname;
 pub use self::random::{do_getrandom, get_random, RandFlags};
 pub use self::rlimit::{do_prlimit, resource_t, rlimit_t, ResourceLimits};
 pub use self::sysinfo::{do_sysinfo, sysinfo_t};
-pub use self::uname::{do_uname, utsname_t};
+pub use self::uname::{do_uname, init_nodename, utsname_t};

src/libos/src/misc/uname.rs

@@ -1,7 +1,5 @@
 use super::*;
-use crate::fs::{AccessMode, CreationFlags, FileMode, FsView};
 use std::ffi::{CStr, CString};
-use std::str;
 /// A sample of `struct utsname`
 /// ```
 ///   sysname = Linux
@@ -26,7 +24,7 @@ pub struct utsname_t {
 
 pub fn do_uname(name: &mut utsname_t) -> Result<()> {
     copy_from_cstr_to_u8_array(&SYSNAME, &mut name.sysname);
-    obtain_nodename(&mut name.nodename);
+    copy_from_cstr_to_u8_array(&NODENAME.read().unwrap(), &mut name.nodename);
     copy_from_cstr_to_u8_array(&RELEASE, &mut name.release);
     copy_from_cstr_to_u8_array(&VERSION, &mut name.version);
     copy_from_cstr_to_u8_array(&MACHINE, &mut name.machine);
@@ -36,7 +34,7 @@ pub fn do_uname(name: &mut utsname_t) -> Result<()> {
 
 lazy_static! {
     static ref SYSNAME: CString = CString::new("Occlum").unwrap();
-    static ref NODENAME: CString = CString::new("occlum-node").unwrap();
+    static ref NODENAME: RwLock<CString> = RwLock::new(CString::new("occlum-node").unwrap());
     static ref RELEASE: CString = CString::new("0.1").unwrap();
     static ref VERSION: CString = CString::new("0.1").unwrap();
     static ref MACHINE: CString = CString::new("x86-64").unwrap();
@@ -50,31 +48,8 @@ fn copy_from_cstr_to_u8_array(src: &CStr, dst: &mut [u8]) {
     dst[len] = 0;
 }
 
-fn obtain_nodename(dst: &mut [u8]) {
-    const HOSTNAME_PATH: &'static str = "/etc/hostname";
-
-    let fs_view = FsView::new();
-
-    let hostname_file = match fs_view.open_file(
-        HOSTNAME_PATH,
-        AccessMode::O_RDONLY as u32,
-        FileMode::from_bits(0o666).unwrap(),
-    ) {
-        Ok(file) => file,
-        Err(e) => {
-            // If failed to open hostname file, use "occlum-node" nodename.
-            error!("failed to open /etc/hostname: {}", e.backtrace());
-            copy_from_cstr_to_u8_array(&NODENAME, dst);
-            return;
-        }
-    };
-
-    let mut nodename: [u8; 65] = [0; 65];
-    hostname_file.read(&mut nodename);
-
-    // The \n need to be eliminated.
-    let nodename_string = str::from_utf8(&nodename).unwrap().replace("\n", "");
-    let len = nodename_string.len();
-    dst[..len].copy_from_slice(&nodename_string.into_bytes());
-    dst[len] = 0;
+pub fn init_nodename(nodename_str: &str) {
+    let nodename_cstr = CString::new(nodename_str).unwrap();
+    let mut nodename = NODENAME.write().unwrap();
+    *nodename = nodename_cstr;
 }

src/libos/src/util/host_file_util.rs

@@ -6,9 +6,9 @@ use std::str;
 
 #[repr(C)]
 pub struct host_file_buffer {
-    pub resolv_conf_ptr: *const c_char,
-    pub hosts_ptr: *const c_char,
-    pub hostname_ptr: *const c_char,
+    pub resolv_conf_buf: *const c_char,
+    pub hosts_buf: *const c_char,
+    pub hostname_buf: *const c_char,
 }
 
 pub enum HostFile {
@@ -57,15 +57,22 @@ pub fn parse_host_file(host_file: HostFile, host_file_ptr: *const c_char) -> Res
     let host_file_str = str::from_utf8(host_file_bytes)
         .map_err(|_| errno!(EINVAL, "host file contains non UTF-8 characters"))?;
 
+    // Parse and inspect host file
     match host_file {
         HostFile::HOSTS => {
-            // TODO: Parsing hosts
-        }
-        HostFile::HOSTNAME => {
-            // TODO: Parsing hostname
+            if let Err(_) = hosts_parser_util::parse_hosts_buffer(host_file_bytes) {
+                return_errno!(EINVAL, "malformated host /etc/hosts");
+            }
         }
+        HostFile::HOSTNAME => match hosts_parser_util::parse_hostname_buffer(host_file_bytes) {
+            Err(_) => {
+                return_errno!(EINVAL, "malformated host /etc/hostname");
+            }
+            Ok(hostname_str) => {
+                return Ok(hostname_str);
+            }
+        },
         HostFile::RESOLV_CONF => {
-            // Parse and inspect host file
             if let Err(_) = resolv_conf::Config::parse(host_file_bytes) {
                 return_errno!(EINVAL, "malformated host /etc/resolv.conf");
             }

src/libos/src/util/hosts_parser_util.rs

@@ -0,0 +1,103 @@
+use super::*;
+use regex::Regex;
+use std::net::IpAddr;
+use std::path::Path;
+use std::str;
+use std::str::FromStr;
+
+lazy_static! {
+    // The hostname regex is compliant with RFC1123
+    static ref HOSTNAME_RE: Regex = Regex::new(r"^(([a-zA-Z0-9]|[a-zA-Z0-9][a-zA-Z0-9\-]*[a-zA-Z0-9])\.)*([A-Za-z0-9]|[A-Za-z0-9][A-Za-z0-9\-]*[A-Za-z0-9])$").unwrap();
+}
+
+#[derive(Debug, Default, Clone)]
+pub struct HostEntry {
+    ip: String,
+    hostname: Vec<String>,
+}
+
+#[derive(Debug, Default, Clone)]
+pub struct Hosts {
+    pub entries: Vec<HostEntry>,
+}
+
+impl FromStr for HostEntry {
+    type Err = error::Error;
+    fn from_str(line: &str) -> Result<Self> {
+        let slice: Vec<String> = line.split_whitespace().map(|s| s.to_string()).collect();
+
+        // check IP:
+        let ip = match slice.first() {
+            Some(ip) => ip,
+            None => {
+                return_errno!(EINVAL, "malformated ip in hosts file");
+            }
+        };
+
+        let _ip_addr: IpAddr = match ip.parse() {
+            Ok(ip) => ip,
+            Err(_) => {
+                return_errno!(EINVAL, "malformated ip in hosts file");
+            }
+        };
+
+        let mut hostname: Vec<String> = Vec::new();
+        for i in slice[1..].to_vec() {
+            if !HOSTNAME_RE.is_match(&i) {
+                return_errno!(EINVAL, "malformated hostname in hosts file");
+            }
+            hostname.push(i.to_owned());
+        }
+
+        if hostname.is_empty() {
+            return_errno!(EINVAL, "malformated hostname in hosts file");
+        }
+        Ok(HostEntry {
+            ip: ip.to_string(),
+            hostname,
+        })
+    }
+}
+
+pub fn parse_hosts_buffer(bytes: &[u8]) -> Result<Hosts> {
+    let mut hosts: Hosts = Default::default();
+    for (_, line) in bytes.split(|&x| x == b'\n').enumerate() {
+        let line = str::from_utf8(line).unwrap();
+        let line = line.trim_start();
+        match line.chars().next() {
+            // comment
+            Some('#') => continue,
+            // empty line
+            None => continue,
+            // valid line
+            Some(_) => {}
+        }
+        hosts.entries.push(line.parse()?);
+    }
+    Ok(hosts)
+}
+
+pub fn parse_hostname_buffer(bytes: &[u8]) -> Result<String> {
+    let mut hostname: Vec<String> = Vec::new();
+    for (_, line) in bytes.split(|&x| x == b'\n').enumerate() {
+        let line = str::from_utf8(line).unwrap();
+        let line = line.trim_start();
+        match line.chars().next() {
+            // comment
+            Some('#') => continue,
+            // empty line
+            None => continue,
+            // valid line
+            Some(_) => {}
+        }
+        if (!HOSTNAME_RE.is_match(&line)) | (line.len() > 64) {
+            return_errno!(EINVAL, "malformated hostname in /etc/hostname file");
+        }
+        hostname.push(line.to_owned());
+    }
+
+    if hostname.len() != 1 {
+        return_errno!(EINVAL, "malformated hostname in /etc/hostname file");
+    }
+    Ok(hostname[0].clone())
+}

src/libos/src/util/mod.rs

@@ -2,6 +2,7 @@ use super::*;
 
 pub mod dirty;
 pub mod host_file_util;
+pub mod hosts_parser_util;
 pub mod log;
 pub mod mem_util;
 pub mod mpx_util;

src/pal/include/occlum_pal_api.h

@@ -93,11 +93,10 @@ struct occlum_pal_create_process_args {
     int *pid;
 };
 
-// todo: more detailed description
 struct host_file_buffer {
-    const char *resolv_conf_ptr;
-    const char *hosts_ptr;
-    const char *hostname_ptr;
+    const char *resolv_conf_buf;
+    const char *hosts_buf;
+    const char *hostname_buf;
 };
 
 /*
@@ -165,6 +164,8 @@ int occlum_pal_kill(int pid, int sig);
  */
 int occlum_pal_destroy(void);
 
+void free_host_file_buffer(struct host_file_buffer file_buffer);
+
 #ifdef __cplusplus
 }
 #endif

src/pal/src/pal_api.c

@@ -108,9 +108,9 @@ int occlum_pal_init(const struct occlum_pal_attr *attr) {
 
     int ecall_ret = 0;
     struct host_file_buffer file_buffer = {
-        .hostname_ptr = pal_load_file("/etc/hostname"),
-        .hosts_ptr = pal_load_file("/etc/hosts"),
-        .resolv_conf_ptr = pal_load_file("/etc/resolv.conf"),
+        .hostname_buf = pal_load_file_to_string("/etc/hostname"),
+        .hosts_buf = pal_load_file_to_string("/etc/hosts"),
+        .resolv_conf_buf = pal_load_file_to_string("/etc/resolv.conf"),
     };
 
     const struct host_file_buffer *file_buffer_ptr = &file_buffer;
@@ -118,12 +118,7 @@ int occlum_pal_init(const struct occlum_pal_attr *attr) {
     sgx_status_t ecall_status = occlum_ecall_init(eid, &ecall_ret, attr->log_level,
                                 resolved_path, file_buffer_ptr);
 
-    free((void *)file_buffer.hostname_ptr);
-    file_buffer.hostname_ptr = NULL;
-    free((void *)file_buffer.hosts_ptr);
-    file_buffer.hosts_ptr = NULL;
-    free((void *)file_buffer.resolv_conf_ptr);
-    file_buffer.resolv_conf_ptr = NULL;
+    free_host_file_buffer(file_buffer);
 
     if (ecall_status != SGX_SUCCESS) {
         const char *sgx_err = pal_get_sgx_error_msg(ecall_status);
@@ -285,6 +280,17 @@ int occlum_pal_destroy(void) {
     return ret;
 }
 
+void free_host_file_buffer(struct host_file_buffer file_buffer) {
+    free((void *)file_buffer.hostname_buf);
+    file_buffer.hostname_buf = NULL;
+
+    free((void *)file_buffer.hosts_buf);
+    file_buffer.hosts_buf = NULL;
+
+    free((void *)file_buffer.resolv_conf_buf);
+    file_buffer.resolv_conf_buf = NULL;
+}
+
 int pal_get_version(void) __attribute__((weak, alias ("occlum_pal_get_version")));
 
 int pal_init(const struct occlum_pal_attr *attr)\

src/pal/src/pal_load_file.c

@@ -2,7 +2,7 @@
 #include <stdlib.h>
 #include "pal_log.h"
 
-char *pal_load_file(const char *filename) {
+char *pal_load_file_to_string(const char *filename) {
     FILE *fp = fopen(filename, "rb");
 
     if (fp == NULL) {
@@ -18,7 +18,7 @@ char *pal_load_file(const char *filename) {
         return NULL;
     }
     fread(file_buffer, 1, fsize, fp);
-    file_buffer[fsize] = 0;
+    file_buffer[fsize] = '\0';
     fclose(fp);
     return file_buffer;
 }

src/pal/src/pal_load_file.h

@@ -1,6 +1,6 @@
 #ifndef __PAL_LOAD_FILE_H__
 #define __PAL_LOAD_FILE_H__
 
-char *pal_load_file(const char *filename);
+char *pal_load_file_to_string(const char *filename);
 
 #endif /* __PAL_LOAD_FILE_H__ */