Skip to main content

memf_linux/
envvars.rs

1//! Linux process environment variable walker.
2//!
3//! Reads environment variables from `mm_struct.env_start`..`env_end`
4//! for each process. The environment region contains null-separated
5//! `KEY=VALUE\0` strings. Requires that the memory pages are accessible
6//! through the ObjectReader's VAS (typically the process's own CR3).
7
8use memf_core::object_reader::ObjectReader;
9use memf_format::PhysicalMemoryProvider;
10
11use crate::{EnvVarInfo, Error, Result};
12
13/// Maximum environment region size to read (256 KiB safety limit).
14const MAX_ENV_SIZE: u64 = 256 * 1024;
15
16/// Walk environment variables for all processes in the task list.
17pub fn walk_envvars<P: PhysicalMemoryProvider>(
18    reader: &ObjectReader<P>,
19) -> Result<Vec<EnvVarInfo>> {
20    let init_task_addr = reader
21        .symbols()
22        .symbol_address("init_task")
23        .ok_or_else(|| Error::MissingKernelSymbol {
24            name: "init_task".into(),
25        })?;
26
27    let tasks_offset = reader
28        .symbols()
29        .field_offset("task_struct", "tasks")
30        .ok_or_else(|| Error::MissingField {
31            struct_name: "task_struct".into(),
32            field_name: "tasks".into(),
33        })?;
34
35    let head_vaddr = init_task_addr + tasks_offset;
36    let task_addrs = reader.walk_list(head_vaddr, "task_struct", "tasks")?;
37
38    let mut all_vars = Vec::new();
39
40    collect_process_envvars(reader, init_task_addr, &mut all_vars);
41
42    for &task_addr in &task_addrs {
43        collect_process_envvars(reader, task_addr, &mut all_vars);
44    }
45
46    Ok(all_vars)
47}
48
49fn collect_process_envvars<P: PhysicalMemoryProvider>(
50    reader: &ObjectReader<P>,
51    task_addr: u64,
52    out: &mut Vec<EnvVarInfo>,
53) {
54    let mm_ptr: u64 = match reader.read_field(task_addr, "task_struct", "mm") {
55        Ok(v) => v,
56        Err(_) => return,
57    };
58    if mm_ptr == 0 {
59        return;
60    }
61
62    if let Ok(vars) = walk_process_envvars(reader, task_addr) {
63        out.extend(vars);
64    }
65}
66
67/// Walk environment variables for a single process.
68pub fn walk_process_envvars<P: PhysicalMemoryProvider>(
69    reader: &ObjectReader<P>,
70    task_addr: u64,
71) -> Result<Vec<EnvVarInfo>> {
72    let pid: u32 = reader.read_field(task_addr, "task_struct", "pid")?;
73    let comm = reader.read_field_string(task_addr, "task_struct", "comm", 16)?;
74    let mm_ptr: u64 = reader.read_field(task_addr, "task_struct", "mm")?;
75
76    if mm_ptr == 0 {
77        return Err(Error::WalkFailed {
78            walker: "walk_process_envvars",
79            reason: format!("task {comm} (PID {pid}) has NULL mm (kernel thread)"),
80        });
81    }
82
83    let env_start: u64 = reader.read_field(mm_ptr, "mm_struct", "env_start")?;
84    let env_end: u64 = reader.read_field(mm_ptr, "mm_struct", "env_end")?;
85
86    if env_start == 0 || env_end <= env_start {
87        return Ok(Vec::new());
88    }
89
90    let size = (env_end - env_start).min(MAX_ENV_SIZE);
91    let data = reader.read_bytes(env_start, size as usize)?;
92
93    Ok(parse_env_region(&data, u64::from(pid), &comm))
94}
95
96fn parse_env_region(data: &[u8], pid: u64, comm: &str) -> Vec<EnvVarInfo> {
97    let mut vars = Vec::new();
98
99    for chunk in data.split(|&b| b == 0) {
100        if chunk.is_empty() {
101            continue;
102        }
103        let s = String::from_utf8_lossy(chunk);
104        if let Some(eq_pos) = s.find('=') {
105            let key = s[..eq_pos].to_string();
106            let value = s[eq_pos + 1..].to_string();
107            vars.push(EnvVarInfo {
108                pid,
109                comm: comm.to_string(),
110                key,
111                value,
112            });
113        }
114    }
115
116    vars
117}
118
119#[cfg(test)]
120mod tests {
121    use super::*;
122    use memf_core::test_builders::{flags, PageTableBuilder, SyntheticPhysMem};
123    use memf_core::vas::{TranslationMode, VirtualAddressSpace};
124    use memf_symbols::isf::IsfResolver;
125    use memf_symbols::test_builders::IsfBuilder;
126
127    fn make_test_reader(
128        data: &[u8],
129        vaddr: u64,
130        paddr: u64,
131        extra_mappings: &[(u64, u64, &[u8])],
132    ) -> ObjectReader<SyntheticPhysMem> {
133        let isf = IsfBuilder::new()
134            .add_struct("task_struct", 128)
135            .add_field("task_struct", "pid", 0, "int")
136            .add_field("task_struct", "state", 4, "long")
137            .add_field("task_struct", "tasks", 16, "list_head")
138            .add_field("task_struct", "comm", 32, "char")
139            .add_field("task_struct", "mm", 48, "pointer")
140            .add_struct("list_head", 16)
141            .add_field("list_head", "next", 0, "pointer")
142            .add_field("list_head", "prev", 8, "pointer")
143            .add_struct("mm_struct", 128)
144            .add_field("mm_struct", "pgd", 0, "pointer")
145            .add_field("mm_struct", "env_start", 64, "unsigned long")
146            .add_field("mm_struct", "env_end", 72, "unsigned long")
147            .add_symbol("init_task", vaddr)
148            .build_json();
149
150        let resolver = IsfResolver::from_value(&isf).unwrap();
151        let mut builder = PageTableBuilder::new()
152            .map_4k(vaddr, paddr, flags::WRITABLE)
153            .write_phys(paddr, data);
154
155        for &(ev, ep, edata) in extra_mappings {
156            builder = builder
157                .map_4k(ev, ep, flags::WRITABLE)
158                .write_phys(ep, edata);
159        }
160
161        let (cr3, mem) = builder.build();
162        let vas = VirtualAddressSpace::new(mem, cr3, TranslationMode::X86_64FourLevel);
163        ObjectReader::new(vas, Box::new(resolver))
164    }
165
166    #[test]
167    fn walk_single_process_envvars() {
168        let vaddr: u64 = 0xFFFF_8000_0010_0000;
169        let paddr: u64 = 0x0080_0000;
170        let mut data = vec![0u8; 4096];
171
172        data[0..4].copy_from_slice(&1u32.to_le_bytes());
173        let tasks_addr = vaddr + 16;
174        data[16..24].copy_from_slice(&tasks_addr.to_le_bytes());
175        data[24..32].copy_from_slice(&tasks_addr.to_le_bytes());
176        data[32..36].copy_from_slice(b"bash");
177        let mm_addr = vaddr + 0x200;
178        data[48..56].copy_from_slice(&mm_addr.to_le_bytes());
179
180        let env_vaddr: u64 = 0xFFFF_8000_0020_0000;
181        let env_paddr: u64 = 0x0090_0000;
182        data[0x200..0x208].copy_from_slice(&0x1000u64.to_le_bytes());
183        data[0x240..0x248].copy_from_slice(&env_vaddr.to_le_bytes());
184        let env_data = b"HOME=/root\0PATH=/usr/bin:/bin\0SHELL=/bin/bash\0";
185        let env_end = env_vaddr + env_data.len() as u64;
186        data[0x248..0x250].copy_from_slice(&env_end.to_le_bytes());
187
188        let reader = make_test_reader(
189            &data,
190            vaddr,
191            paddr,
192            &[(env_vaddr, env_paddr, env_data.as_slice())],
193        );
194        let vars = walk_envvars(&reader).unwrap();
195
196        assert_eq!(vars.len(), 3);
197        assert_eq!(vars[0].pid, 1);
198        assert_eq!(vars[0].comm, "bash");
199        assert_eq!(vars[0].key, "HOME");
200        assert_eq!(vars[0].value, "/root");
201        assert_eq!(vars[1].key, "PATH");
202        assert_eq!(vars[1].value, "/usr/bin:/bin");
203        assert_eq!(vars[2].key, "SHELL");
204        assert_eq!(vars[2].value, "/bin/bash");
205    }
206
207    #[test]
208    fn walk_envvars_skips_kernel_threads() {
209        let vaddr: u64 = 0xFFFF_8000_0010_0000;
210        let paddr: u64 = 0x0080_0000;
211        let mut data = vec![0u8; 4096];
212
213        data[0..4].copy_from_slice(&0u32.to_le_bytes());
214        let tasks_addr = vaddr + 16;
215        data[16..24].copy_from_slice(&tasks_addr.to_le_bytes());
216        data[24..32].copy_from_slice(&tasks_addr.to_le_bytes());
217        data[32..41].copy_from_slice(b"swapper/0");
218        data[48..56].copy_from_slice(&0u64.to_le_bytes());
219
220        let reader = make_test_reader(&data, vaddr, paddr, &[]);
221        let vars = walk_envvars(&reader).unwrap();
222        assert!(vars.is_empty());
223    }
224
225    #[test]
226    fn walk_process_envvars_null_mm_returns_error() {
227        let vaddr: u64 = 0xFFFF_8000_0010_0000;
228        let paddr: u64 = 0x0080_0000;
229        let mut data = vec![0u8; 4096];
230
231        data[48..56].copy_from_slice(&0u64.to_le_bytes());
232
233        let reader = make_test_reader(&data, vaddr, paddr, &[]);
234        let result = walk_process_envvars(&reader, vaddr);
235        assert!(result.is_err());
236    }
237
238    #[test]
239    fn parse_env_region_handles_malformed_entries() {
240        let data = b"GOOD=value\0MALFORMED\0ALSO_GOOD=ok\0";
241        let vars = parse_env_region(data, 1, "test");
242
243        assert_eq!(vars.len(), 2);
244        assert_eq!(vars[0].key, "GOOD");
245        assert_eq!(vars[0].value, "value");
246        assert_eq!(vars[1].key, "ALSO_GOOD");
247        assert_eq!(vars[1].value, "ok");
248    }
249
250    #[test]
251    fn parse_env_region_empty() {
252        let vars = parse_env_region(&[], 1, "test");
253        assert!(vars.is_empty());
254    }
255
256    #[test]
257    fn missing_init_task_symbol() {
258        let isf = IsfBuilder::new()
259            .add_struct("task_struct", 64)
260            .add_field("task_struct", "pid", 0, "int")
261            .add_field("task_struct", "tasks", 8, "list_head")
262            .add_struct("list_head", 16)
263            .add_field("list_head", "next", 0, "pointer")
264            .add_field("list_head", "prev", 8, "pointer")
265            .build_json();
266
267        let resolver = IsfResolver::from_value(&isf).unwrap();
268        let (cr3, mem) = PageTableBuilder::new().build();
269        let vas = VirtualAddressSpace::new(mem, cr3, TranslationMode::X86_64FourLevel);
270        let reader = ObjectReader::new(vas, Box::new(resolver));
271
272        let result = walk_envvars(&reader);
273        assert!(
274            matches!(result, Err(crate::Error::MissingKernelSymbol { ref name }) if name == "init_task"),
275            "expected MissingKernelSymbol {{name: \"init_task\"}}, got {result:?}"
276        );
277    }
278
279    #[test]
280    fn missing_tasks_field_returns_missing_field() {
281        let isf = IsfBuilder::new()
282            .add_struct("task_struct", 64)
283            .add_field("task_struct", "pid", 0, "int")
284            // tasks intentionally omitted
285            .add_struct("list_head", 16)
286            .add_field("list_head", "next", 0, "pointer")
287            .add_field("list_head", "prev", 8, "pointer")
288            .add_symbol("init_task", 0xFFFF_8000_0010_0000)
289            .build_json();
290        let resolver = IsfResolver::from_value(&isf).unwrap();
291        let (cr3, mem) = PageTableBuilder::new().build();
292        let vas = VirtualAddressSpace::new(mem, cr3, TranslationMode::X86_64FourLevel);
293        let reader: ObjectReader<SyntheticPhysMem> = ObjectReader::new(vas, Box::new(resolver));
294        let result = walk_envvars(&reader);
295        assert!(
296            matches!(result, Err(crate::Error::MissingField { ref struct_name, ref field_name }) if struct_name == "task_struct" && field_name == "tasks"),
297            "expected MissingField task_struct.tasks, got {result:?}"
298        );
299    }
300}