Merge pull request #249 from cakebaker/pmap_device_format

pmap: implement `--device`

Author: sylvestre

src/uu/pmap/src/maps_format_parser.rs

@@ -3,17 +3,59 @@
 // For the full copyright and license information, please view the LICENSE
 // file that was distributed with this source code.
 
+use std::fmt;
 use std::io::{Error, ErrorKind};
 
-// Represents a parsed single line from /proc/<PID>/maps.
+// Represents a parsed single line from /proc/<PID>/maps for the default and device formats. It
+// omits the inode information because it's not used by those formats.
 #[derive(Debug, PartialEq)]
 pub struct MapLine {
     pub address: String,
     pub size_in_kb: u64,
-    pub perms: String,
+    pub perms: Perms,
+    pub offset: String,
+    pub device: String,
     pub mapping: String,
 }
 
+// Represents a set of permissions from the "perms" column of /proc/<PID>/maps.
+#[derive(Clone, Copy, Debug, PartialEq)]
+pub struct Perms {
+    pub readable: bool,
+    pub writable: bool,
+    pub executable: bool,
+    pub shared: bool,
+}
+
+impl From<&str> for Perms {
+    fn from(s: &str) -> Self {
+        let mut chars = s.chars();
+
+        Self {
+            readable: chars.next() == Some('r'),
+            writable: chars.next() == Some('w'),
+            executable: chars.next() == Some('x'),
+            shared: chars.next() == Some('s'),
+        }
+    }
+}
+
+// Please note: While `Perms` has four boolean fields, it's string representation has five
+// characters because pmap's default and device formats use five characters for the perms,
+// with the last character always being '-'.
+impl fmt::Display for Perms {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        write!(
+            f,
+            "{}{}{}{}-",
+            if self.readable { 'r' } else { '-' },
+            if self.writable { 'w' } else { '-' },
+            if self.executable { 'x' } else { '-' },
+            if self.shared { 's' } else { '-' },
+        )
+    }
+}
+
 // Parses a single line from /proc/<PID>/maps. See
 // https://www.kernel.org/doc/html/latest/filesystems/proc.html for details about the expected
 // format.
@@ -30,16 +72,29 @@ pub fn parse_map_line(line: &str) -> Result<MapLine, Error> {
     let (perms, rest) = rest
         .split_once(' ')
         .ok_or_else(|| Error::from(ErrorKind::InvalidData))?;
-    let perms = parse_perms(perms);
+    let perms = Perms::from(perms);
+
+    let (offset, rest) = rest
+        .split_once(' ')
+        .ok_or_else(|| Error::from(ErrorKind::InvalidData))?;
+    let offset = format!("{offset:0>16}");
+
+    let (device, rest) = rest
+        .split_once(' ')
+        .ok_or_else(|| Error::from(ErrorKind::InvalidData))?;
+    let device = parse_device(device)?;
 
-    let mapping: String = rest.splitn(4, ' ').skip(3).collect();
+    // skip the "inode" column
+    let mapping: String = rest.splitn(2, ' ').skip(1).collect();
     let mapping = mapping.trim_ascii_start();
     let mapping = parse_mapping(mapping);
 
     Ok(MapLine {
         address,
         size_in_kb,
         perms,
+        offset,
+        device,
         mapping,
     })
 }
@@ -58,13 +113,12 @@ fn parse_address(memory_range: &str) -> Result<(String, u64), Error> {
     Ok((format!("{start:0>16}"), size_in_kb))
 }
 
-// Turns a 4-char perms string from /proc/<PID>/maps into a 5-char perms string. The first three
-// chars are left untouched.
-fn parse_perms(perms: &str) -> String {
-    let perms = perms.replace("p", "-");
-
-    // the fifth char seems to be always '-' in the original pmap
-    format!("{perms}-")
+// Pads the device info from /proc/<PID>/maps with zeros and turns AB:CD into 0AB:000CD.
+fn parse_device(device: &str) -> Result<String, Error> {
+    let (major, minor) = device
+        .split_once(':')
+        .ok_or_else(|| Error::from(ErrorKind::InvalidData))?;
+    Ok(format!("{major:0>3}:{minor:0>5}"))
 }
 
 fn parse_mapping(mapping: &str) -> String {
@@ -86,44 +140,60 @@ fn parse_mapping(mapping: &str) -> String {
 mod test {
     use super::*;
 
-    fn create_map_line(address: &str, size_in_kb: u64, perms: &str, mapping: &str) -> MapLine {
+    fn create_map_line(
+        address: &str,
+        size_in_kb: u64,
+        perms: Perms,
+        offset: &str,
+        device: &str,
+        mapping: &str,
+    ) -> MapLine {
         MapLine {
             address: address.to_string(),
             size_in_kb,
-            perms: perms.to_string(),
+            perms,
+            offset: offset.to_string(),
+            device: device.to_string(),
             mapping: mapping.to_string(),
         }
     }
 
+    #[test]
+    fn test_perms_to_string() {
+        assert_eq!("-----", Perms::from("---p").to_string());
+        assert_eq!("---s-", Perms::from("---s").to_string());
+        assert_eq!("rwx--", Perms::from("rwxp").to_string());
+    }
+
     #[test]
     fn test_parse_map_line() {
         let data = [
             (
-                create_map_line("000062442eb9e000", 16, "r----", "konsole"),
+                create_map_line("000062442eb9e000", 16, Perms::from("r--p"), "0000000000000000", "008:00008", "konsole"),
                 "62442eb9e000-62442eba2000 r--p 00000000 08:08 10813151                   /usr/bin/konsole"
             ),
             (
-                create_map_line("000071af50000000", 132, "rw---", "  [ anon ]"),
+                create_map_line("000071af50000000", 132, Perms::from("rw-p"), "0000000000000000", "000:00000", "  [ anon ]"),
                 "71af50000000-71af50021000 rw-p 00000000 00:00 0 "
             ),
             (
-                create_map_line("00007ffc3f8df000", 132, "rw---", "  [ stack ]"),
+                create_map_line("00007ffc3f8df000", 132, Perms::from("rw-p"), "0000000000000000", "000:00000", "  [ stack ]"),
                 "7ffc3f8df000-7ffc3f900000 rw-p 00000000 00:00 0                          [stack]"
             ),
             (
-                create_map_line("000071af8c9e6000", 16, "rw-s-", "  [ anon ]"),
+                create_map_line("000071af8c9e6000", 16, Perms::from("rw-s"), "0000000105830000", "000:00010", "  [ anon ]"),
                 "71af8c9e6000-71af8c9ea000 rw-s 105830000 00:10 1075                      anon_inode:i915.gem"
             ),
             (
-                create_map_line("000071af6cf0c000", 3560, "rw-s-", "memfd:wayland-shm (deleted)"),
+                create_map_line("000071af6cf0c000", 3560, Perms::from("rw-s"), "0000000000000000", "000:00001", "memfd:wayland-shm (deleted)"),
                 "71af6cf0c000-71af6d286000 rw-s 00000000 00:01 256481                     /memfd:wayland-shm (deleted)"
             ),
             (
-                create_map_line("ffffffffff600000", 4, "--x--", "  [ anon ]"),
+                create_map_line("ffffffffff600000", 4, Perms::from("--xp"), "0000000000000000", "000:00000", "  [ anon ]"),
                 "ffffffffff600000-ffffffffff601000 --xp 00000000 00:00 0                  [vsyscall]"
             ),
             (
-                create_map_line("00005e8187da8000", 24, "r----", "hello   world"),
+                create_map_line("00005e8187da8000", 24, Perms::from("r--p"), "0000000000000000", "008:00008", "hello   world"),
                 "5e8187da8000-5e8187dae000 r--p 00000000 08:08 9524160                    /usr/bin/hello   world"
             ),
         ];
@@ -161,10 +231,14 @@ mod test {
     }
 
     #[test]
-    fn test_parse_perms() {
-        assert_eq!("-----", parse_perms("---p"));
-        assert_eq!("---s-", parse_perms("---s"));
-        assert_eq!("rwx--", parse_perms("rwxp"));
+    fn test_parse_device() {
+        assert_eq!("012:00034", parse_device("12:34").unwrap());
+        assert_eq!("000:00000", parse_device("00:00").unwrap());
+    }
+
+    #[test]
+    fn test_parse_device_without_colon() {
+        assert!(parse_device("1234").is_err());
     }
 
     #[test]

src/uu/pmap/src/pmap.rs

@@ -4,7 +4,7 @@
 // file that was distributed with this source code.
 
 use clap::{crate_version, Arg, ArgAction, Command};
-use maps_format_parser::parse_map_line;
+use maps_format_parser::{parse_map_line, MapLine};
 use std::env;
 use std::fs;
 use std::io::Error;
@@ -49,11 +49,12 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
             }
         }
 
-        match parse_maps(pid) {
-            Ok(total) => println!(" total {total:>16}K"),
-            Err(_) => {
-                set_exit_code(1);
-            }
+        if matches.get_flag(options::DEVICE) {
+            output_device_format(pid).map_err(|_| set_exit_code(1)).ok();
+        } else {
+            output_default_format(pid)
+                .map_err(|_| set_exit_code(1))
+                .ok();
         }
     }
 
@@ -74,21 +75,70 @@ fn parse_cmdline(pid: &str) -> Result<String, Error> {
     Ok(cmdline.into())
 }
 
-fn parse_maps(pid: &str) -> Result<u64, Error> {
+fn process_maps<F>(pid: &str, mut process_line: F) -> Result<(), Error>
+where
+    F: FnMut(&MapLine),
+{
     let path = format!("/proc/{pid}/maps");
     let contents = fs::read_to_string(path)?;
-    let mut total = 0;
 
     for line in contents.lines() {
         let map_line = parse_map_line(line)?;
+        process_line(&map_line);
+    }
+
+    Ok(())
+}
+
+fn output_default_format(pid: &str) -> Result<(), Error> {
+    let mut total = 0;
+
+    process_maps(pid, |map_line| {
         println!(
             "{} {:>6}K {} {}",
             map_line.address, map_line.size_in_kb, map_line.perms, map_line.mapping
         );
         total += map_line.size_in_kb;
-    }
+    })?;
+
+    println!(" total {total:>16}K");
+
+    Ok(())
+}
+
+fn output_device_format(pid: &str) -> Result<(), Error> {
+    let mut total_mapped = 0;
+    let mut total_writeable_private = 0;
+    let mut total_shared = 0;
+
+    println!("Address           Kbytes Mode  Offset           Device    Mapping");
+
+    process_maps(pid, |map_line| {
+        println!(
+            "{} {:>7} {} {} {} {}",
+            map_line.address,
+            map_line.size_in_kb,
+            map_line.perms,
+            map_line.offset,
+            map_line.device,
+            map_line.mapping
+        );
+        total_mapped += map_line.size_in_kb;
 
-    Ok(total)
+        if map_line.perms.writable && !map_line.perms.shared {
+            total_writeable_private += map_line.size_in_kb;
+        }
+
+        if map_line.perms.shared {
+            total_shared += map_line.size_in_kb;
+        }
+    })?;
+
+    println!(
+        "mapped: {total_mapped}K    writeable/private: {total_writeable_private}K    shared: {total_shared}K"
+    );
+
+    Ok(())
 }
 
 pub fn uu_app() -> Command {

tests/by-util/test_pmap.rs

@@ -73,6 +73,22 @@ fn test_non_existing_pid() {
         .no_output();
 }
 
+#[test]
+#[cfg(target_os = "linux")]
+fn test_device() {
+    let pid = process::id();
+
+    for arg in ["-d", "--device"] {
+        let result = new_ucmd!()
+            .arg(arg)
+            .arg(pid.to_string())
+            .succeeds()
+            .stdout_move_str();
+
+        assert_device_format(pid, &result);
+    }
+}
+
 #[test]
 fn test_invalid_arg() {
     new_ucmd!().arg("--definitely-invalid").fails().code_is(1);
@@ -94,9 +110,46 @@ fn assert_format(pid: u32, s: &str) {
 
     let rest = rest.trim_end();
     let (memory_map, last_line) = rest.rsplit_once('\n').unwrap();
-    let re = Regex::new("(?m)^[0-9a-f]{16} +[1-9][0-9]*K (-|r)(-|w)(-|x)(-|s)- (  $$[ (anon|stack) $$]|[a-zA-Z0-9._-]+)$").unwrap();
+    let re = Regex::new(r"(?m)^[0-9a-f]{16} +[1-9][0-9]*K (-|r)(-|w)(-|x)(-|s)- (  \[ (anon|stack) \]|[a-zA-Z0-9._-]+)$").unwrap();
     assert!(re.is_match(memory_map));
 
     let re = Regex::new("^ total +[1-9][0-9]*K$").unwrap();
     assert!(re.is_match(last_line));
 }
+
+// Ensure `s` has the following device format (--device):
+//
+// 1234:   /some/path
+// Address           Kbytes Mode  Offset           Device    Mapping
+// 000073eb5f4c7000       8 rw--- 0000000000036000 008:00008 ld-linux-x86-64.so.2
+// 00007ffd588fc000     132 rw--- 0000000000000000 000:00000   [ stack ]
+// ffffffffff600000       4 --x-- 0000000000000000 000:00000   [ anon ]
+// ...
+// mapped: 3060K    writeable/private: 348K    shared: 0K
+#[cfg(target_os = "linux")]
+fn assert_device_format(pid: u32, s: &str) {
+    let lines: Vec<_> = s.lines().collect();
+    let line_count = lines.len();
+
+    let re = Regex::new(&format!("^{pid}:   .+[^ ]$")).unwrap();
+    assert!(re.is_match(lines[0]));
+
+    let expected_header = "Address           Kbytes Mode  Offset           Device    Mapping";
+    assert_eq!(expected_header, lines[1]);
+
+    let re = Regex::new(
+        r"^[0-9a-f]{16} +[1-9][0-9]* (-|r)(-|w)(-|x)(-|s)- [0-9a-f]{16} [0-9a-f]{3}:[0-9a-f]{5} (  \[ (anon|stack) \]|[a-zA-Z0-9._-]+)$",
+    )
+    .unwrap();
+
+    for line in lines.iter().take(line_count - 1).skip(2) {
+        assert!(re.is_match(line), "failing line: {line}");
+    }
+
+    let re = Regex::new(r"^mapped: \d+K\s{4}writeable/private: \d+K\s{4}shared: \d+K$").unwrap();
+    assert!(
+        re.is_match(lines[line_count - 1]),
+        "failing line: {}",
+        lines[line_count - 1]
+    );
+}