miri/
eval.rs

1//! Main evaluator loop and setting up the initial stack frame.
2
3use std::ffi::{OsStr, OsString};
4use std::panic::{self, AssertUnwindSafe};
5use std::path::PathBuf;
6use std::rc::Rc;
7use std::task::Poll;
8use std::{iter, thread};
9
10use rustc_abi::ExternAbi;
11use rustc_data_structures::fx::{FxHashMap, FxHashSet};
12use rustc_hir::def::Namespace;
13use rustc_hir::def_id::DefId;
14use rustc_middle::ty::layout::{HasTyCtxt, HasTypingEnv, LayoutCx};
15use rustc_middle::ty::{self, Ty, TyCtxt};
16use rustc_session::config::EntryFnType;
17
18use crate::concurrency::GenmcCtx;
19use crate::concurrency::thread::TlsAllocAction;
20use crate::diagnostics::report_leaks;
21use crate::shims::{global_ctor, tls};
22use crate::*;
23
24#[derive(Copy, Clone, Debug)]
25pub enum MiriEntryFnType {
26    MiriStart,
27    Rustc(EntryFnType),
28}
29
30/// When the main thread would exit, we will yield to any other thread that is ready to execute.
31/// But we must only do that a finite number of times, or a background thread running `loop {}`
32/// will hang the program.
33const MAIN_THREAD_YIELDS_AT_SHUTDOWN: u32 = 256;
34
35#[derive(Copy, Clone, Debug, PartialEq)]
36pub enum AlignmentCheck {
37    /// Do not check alignment.
38    None,
39    /// Check alignment "symbolically", i.e., using only the requested alignment for an allocation and not its real base address.
40    Symbolic,
41    /// Check alignment on the actual physical integer address.
42    Int,
43}
44
45#[derive(Copy, Clone, Debug, PartialEq)]
46pub enum RejectOpWith {
47    /// Isolated op is rejected with an abort of the machine.
48    Abort,
49
50    /// If not Abort, miri returns an error for an isolated op.
51    /// Following options determine if user should be warned about such error.
52    /// Do not print warning about rejected isolated op.
53    NoWarning,
54
55    /// Print a warning about rejected isolated op, with backtrace.
56    Warning,
57
58    /// Print a warning about rejected isolated op, without backtrace.
59    WarningWithoutBacktrace,
60}
61
62#[derive(Copy, Clone, Debug, PartialEq)]
63pub enum IsolatedOp {
64    /// Reject an op requiring communication with the host. By
65    /// default, miri rejects the op with an abort. If not, it returns
66    /// an error code, and prints a warning about it. Warning levels
67    /// are controlled by `RejectOpWith` enum.
68    Reject(RejectOpWith),
69
70    /// Execute op requiring communication with the host, i.e. disable isolation.
71    Allow,
72}
73
74#[derive(Debug, Copy, Clone, PartialEq, Eq)]
75pub enum BacktraceStyle {
76    /// Prints a terser backtrace which ideally only contains relevant information.
77    Short,
78    /// Prints a backtrace with all possible information.
79    Full,
80    /// Prints only the frame that the error occurs in.
81    Off,
82}
83
84#[derive(Debug, Copy, Clone, PartialEq, Eq)]
85pub enum ValidationMode {
86    /// Do not perform any kind of validation.
87    No,
88    /// Validate the interior of the value, but not things behind references.
89    Shallow,
90    /// Fully recursively validate references.
91    Deep,
92}
93
94/// Configuration needed to spawn a Miri instance.
95#[derive(Clone)]
96pub struct MiriConfig {
97    /// The host environment snapshot to use as basis for what is provided to the interpreted program.
98    /// (This is still subject to isolation as well as `forwarded_env_vars`.)
99    pub env: Vec<(OsString, OsString)>,
100    /// Determine if validity checking is enabled.
101    pub validation: ValidationMode,
102    /// Determines if Stacked Borrows or Tree Borrows is enabled.
103    pub borrow_tracker: Option<BorrowTrackerMethod>,
104    /// Controls alignment checking.
105    pub check_alignment: AlignmentCheck,
106    /// Action for an op requiring communication with the host.
107    pub isolated_op: IsolatedOp,
108    /// Determines if memory leaks should be ignored.
109    pub ignore_leaks: bool,
110    /// Environment variables that should always be forwarded from the host.
111    pub forwarded_env_vars: Vec<String>,
112    /// Additional environment variables that should be set in the interpreted program.
113    pub set_env_vars: FxHashMap<String, String>,
114    /// Command-line arguments passed to the interpreted program.
115    pub args: Vec<String>,
116    /// The seed to use when non-determinism or randomness are required (e.g. ptr-to-int cast, `getrandom()`).
117    pub seed: Option<u64>,
118    /// The stacked borrows pointer ids to report about.
119    pub tracked_pointer_tags: FxHashSet<BorTag>,
120    /// The allocation ids to report about.
121    pub tracked_alloc_ids: FxHashSet<AllocId>,
122    /// For the tracked alloc ids, also report read/write accesses.
123    pub track_alloc_accesses: bool,
124    /// Determine if data race detection should be enabled.
125    pub data_race_detector: bool,
126    /// Determine if weak memory emulation should be enabled. Requires data race detection to be enabled.
127    pub weak_memory_emulation: bool,
128    /// Determine if we are running in GenMC mode. In this mode, Miri will explore multiple concurrent executions of the given program.
129    pub genmc_mode: bool,
130    /// Track when an outdated (weak memory) load happens.
131    pub track_outdated_loads: bool,
132    /// Rate of spurious failures for compare_exchange_weak atomic operations,
133    /// between 0.0 and 1.0, defaulting to 0.8 (80% chance of failure).
134    pub cmpxchg_weak_failure_rate: f64,
135    /// If `Some`, enable the `measureme` profiler, writing results to a file
136    /// with the specified prefix.
137    pub measureme_out: Option<String>,
138    /// Which style to use for printing backtraces.
139    pub backtrace_style: BacktraceStyle,
140    /// Which provenance to use for int2ptr casts.
141    pub provenance_mode: ProvenanceMode,
142    /// Whether to ignore any output by the program. This is helpful when debugging miri
143    /// as its messages don't get intermingled with the program messages.
144    pub mute_stdout_stderr: bool,
145    /// The probability of the active thread being preempted at the end of each basic block.
146    pub preemption_rate: f64,
147    /// Report the current instruction being executed every N basic blocks.
148    pub report_progress: Option<u32>,
149    /// Whether Stacked Borrows and Tree Borrows retagging should recurse into fields of datatypes.
150    pub retag_fields: RetagFields,
151    /// The location of the shared object files to load when calling external functions
152    pub native_lib: Vec<PathBuf>,
153    /// Whether to enable the new native lib tracing system.
154    pub native_lib_enable_tracing: bool,
155    /// Run a garbage collector for BorTags every N basic blocks.
156    pub gc_interval: u32,
157    /// The number of CPUs to be reported by miri.
158    pub num_cpus: u32,
159    /// Requires Miri to emulate pages of a certain size.
160    pub page_size: Option<u64>,
161    /// Whether to collect a backtrace when each allocation is created, just in case it leaks.
162    pub collect_leak_backtraces: bool,
163    /// Probability for address reuse.
164    pub address_reuse_rate: f64,
165    /// Probability for address reuse across threads.
166    pub address_reuse_cross_thread_rate: f64,
167    /// Round Robin scheduling with no preemption.
168    pub fixed_scheduling: bool,
169    /// Always prefer the intrinsic fallback body over the native Miri implementation.
170    pub force_intrinsic_fallback: bool,
171    /// Whether floating-point operations can behave non-deterministically.
172    pub float_nondet: bool,
173}
174
175impl Default for MiriConfig {
176    fn default() -> MiriConfig {
177        MiriConfig {
178            env: vec![],
179            validation: ValidationMode::Shallow,
180            borrow_tracker: Some(BorrowTrackerMethod::StackedBorrows),
181            check_alignment: AlignmentCheck::Int,
182            isolated_op: IsolatedOp::Reject(RejectOpWith::Abort),
183            ignore_leaks: false,
184            forwarded_env_vars: vec![],
185            set_env_vars: FxHashMap::default(),
186            args: vec![],
187            seed: None,
188            tracked_pointer_tags: FxHashSet::default(),
189            tracked_alloc_ids: FxHashSet::default(),
190            track_alloc_accesses: false,
191            data_race_detector: true,
192            weak_memory_emulation: true,
193            genmc_mode: false,
194            track_outdated_loads: false,
195            cmpxchg_weak_failure_rate: 0.8, // 80%
196            measureme_out: None,
197            backtrace_style: BacktraceStyle::Short,
198            provenance_mode: ProvenanceMode::Default,
199            mute_stdout_stderr: false,
200            preemption_rate: 0.01, // 1%
201            report_progress: None,
202            retag_fields: RetagFields::Yes,
203            native_lib: vec![],
204            native_lib_enable_tracing: false,
205            gc_interval: 10_000,
206            num_cpus: 1,
207            page_size: None,
208            collect_leak_backtraces: true,
209            address_reuse_rate: 0.5,
210            address_reuse_cross_thread_rate: 0.1,
211            fixed_scheduling: false,
212            force_intrinsic_fallback: false,
213            float_nondet: true,
214        }
215    }
216}
217
218/// The state of the main thread. Implementation detail of `on_main_stack_empty`.
219#[derive(Debug)]
220enum MainThreadState<'tcx> {
221    GlobalCtors {
222        ctor_state: global_ctor::GlobalCtorState<'tcx>,
223        /// The main function to call.
224        entry_id: DefId,
225        entry_type: MiriEntryFnType,
226        /// Arguments passed to `main`.
227        argc: ImmTy<'tcx>,
228        argv: ImmTy<'tcx>,
229    },
230    Running,
231    TlsDtors(tls::TlsDtorsState<'tcx>),
232    Yield {
233        remaining: u32,
234    },
235    Done,
236}
237
238impl<'tcx> MainThreadState<'tcx> {
239    fn on_main_stack_empty(
240        &mut self,
241        this: &mut MiriInterpCx<'tcx>,
242    ) -> InterpResult<'tcx, Poll<()>> {
243        use MainThreadState::*;
244        match self {
245            GlobalCtors { ctor_state, entry_id, entry_type, argc, argv } => {
246                match ctor_state.on_stack_empty(this)? {
247                    Poll::Pending => {} // just keep going
248                    Poll::Ready(()) => {
249                        call_main(this, *entry_id, *entry_type, argc.clone(), argv.clone())?;
250                        *self = Running;
251                    }
252                }
253            }
254            Running => {
255                *self = TlsDtors(Default::default());
256            }
257            TlsDtors(state) =>
258                match state.on_stack_empty(this)? {
259                    Poll::Pending => {} // just keep going
260                    Poll::Ready(()) => {
261                        if this.machine.data_race.as_genmc_ref().is_some() {
262                            // In GenMC mode, we don't yield at the end of the main thread.
263                            // Instead, the `GenmcCtx` will ensure that unfinished threads get a chance to run at this point.
264                            *self = Done;
265                        } else {
266                            // Give background threads a chance to finish by yielding the main thread a
267                            // couple of times -- but only if we would also preempt threads randomly.
268                            if this.machine.preemption_rate > 0.0 {
269                                // There is a non-zero chance they will yield back to us often enough to
270                                // make Miri terminate eventually.
271                                *self = Yield { remaining: MAIN_THREAD_YIELDS_AT_SHUTDOWN };
272                            } else {
273                                // The other threads did not get preempted, so no need to yield back to
274                                // them.
275                                *self = Done;
276                            }
277                        }
278                    }
279                },
280            Yield { remaining } =>
281                match remaining.checked_sub(1) {
282                    None => *self = Done,
283                    Some(new_remaining) => {
284                        *remaining = new_remaining;
285                        this.yield_active_thread();
286                    }
287                },
288            Done => {
289                // Figure out exit code.
290                let ret_place = this.machine.main_fn_ret_place.clone().unwrap();
291                let exit_code = this.read_target_isize(&ret_place)?;
292                // Rust uses `isize` but the underlying type of an exit code is `i32`.
293                // Do a saturating cast.
294                let exit_code = i32::try_from(exit_code).unwrap_or(if exit_code >= 0 {
295                    i32::MAX
296                } else {
297                    i32::MIN
298                });
299                // Deal with our thread-local memory. We do *not* want to actually free it, instead we consider TLS
300                // to be like a global `static`, so that all memory reached by it is considered to "not leak".
301                this.terminate_active_thread(TlsAllocAction::Leak)?;
302
303                // Stop interpreter loop.
304                throw_machine_stop!(TerminationInfo::Exit { code: exit_code, leak_check: true });
305            }
306        }
307        interp_ok(Poll::Pending)
308    }
309}
310
311/// Returns a freshly created `InterpCx`.
312/// Public because this is also used by `priroda`.
313pub fn create_ecx<'tcx>(
314    tcx: TyCtxt<'tcx>,
315    entry_id: DefId,
316    entry_type: MiriEntryFnType,
317    config: &MiriConfig,
318    genmc_ctx: Option<Rc<GenmcCtx>>,
319) -> InterpResult<'tcx, InterpCx<'tcx, MiriMachine<'tcx>>> {
320    let typing_env = ty::TypingEnv::fully_monomorphized();
321    let layout_cx = LayoutCx::new(tcx, typing_env);
322    let mut ecx = InterpCx::new(
323        tcx,
324        rustc_span::DUMMY_SP,
325        typing_env,
326        MiriMachine::new(config, layout_cx, genmc_ctx),
327    );
328
329    // Make sure we have MIR. We check MIR for some stable monomorphic function in libcore.
330    let sentinel =
331        helpers::try_resolve_path(tcx, &["core", "ascii", "escape_default"], Namespace::ValueNS);
332    if !matches!(sentinel, Some(s) if tcx.is_mir_available(s.def.def_id())) {
333        tcx.dcx().fatal(
334            "the current sysroot was built without `-Zalways-encode-mir`, or libcore seems missing. \
335            Use `cargo miri setup` to prepare a sysroot that is suitable for Miri."
336        );
337    }
338
339    // Compute argc and argv from `config.args`.
340    let argc =
341        ImmTy::from_int(i64::try_from(config.args.len()).unwrap(), ecx.machine.layouts.isize);
342    let argv = {
343        // Put each argument in memory, collect pointers.
344        let mut argvs = Vec::<Immediate<Provenance>>::with_capacity(config.args.len());
345        for arg in config.args.iter() {
346            // Make space for `0` terminator.
347            let size = u64::try_from(arg.len()).unwrap().strict_add(1);
348            let arg_type = Ty::new_array(tcx, tcx.types.u8, size);
349            let arg_place =
350                ecx.allocate(ecx.layout_of(arg_type)?, MiriMemoryKind::Machine.into())?;
351            ecx.write_os_str_to_c_str(OsStr::new(arg), arg_place.ptr(), size)?;
352            ecx.mark_immutable(&arg_place);
353            argvs.push(arg_place.to_ref(&ecx));
354        }
355        // Make an array with all these pointers, in the Miri memory.
356        let u8_ptr_type = Ty::new_imm_ptr(tcx, tcx.types.u8);
357        let u8_ptr_ptr_type = Ty::new_imm_ptr(tcx, u8_ptr_type);
358        let argvs_layout =
359            ecx.layout_of(Ty::new_array(tcx, u8_ptr_type, u64::try_from(argvs.len()).unwrap()))?;
360        let argvs_place = ecx.allocate(argvs_layout, MiriMemoryKind::Machine.into())?;
361        for (arg, idx) in argvs.into_iter().zip(0..) {
362            let place = ecx.project_index(&argvs_place, idx)?;
363            ecx.write_immediate(arg, &place)?;
364        }
365        ecx.mark_immutable(&argvs_place);
366        // Store `argc` and `argv` for macOS `_NSGetArg{c,v}`, and for the GC to see them.
367        {
368            let argc_place =
369                ecx.allocate(ecx.machine.layouts.isize, MiriMemoryKind::Machine.into())?;
370            ecx.write_immediate(*argc, &argc_place)?;
371            ecx.mark_immutable(&argc_place);
372            ecx.machine.argc = Some(argc_place.ptr());
373
374            let argv_place =
375                ecx.allocate(ecx.layout_of(u8_ptr_ptr_type)?, MiriMemoryKind::Machine.into())?;
376            ecx.write_pointer(argvs_place.ptr(), &argv_place)?;
377            ecx.mark_immutable(&argv_place);
378            ecx.machine.argv = Some(argv_place.ptr());
379        }
380        // Store command line as UTF-16 for Windows `GetCommandLineW`.
381        if tcx.sess.target.os == "windows" {
382            // Construct a command string with all the arguments.
383            let cmd_utf16: Vec<u16> = args_to_utf16_command_string(config.args.iter());
384
385            let cmd_type =
386                Ty::new_array(tcx, tcx.types.u16, u64::try_from(cmd_utf16.len()).unwrap());
387            let cmd_place =
388                ecx.allocate(ecx.layout_of(cmd_type)?, MiriMemoryKind::Machine.into())?;
389            ecx.machine.cmd_line = Some(cmd_place.ptr());
390            // Store the UTF-16 string. We just allocated so we know the bounds are fine.
391            for (&c, idx) in cmd_utf16.iter().zip(0..) {
392                let place = ecx.project_index(&cmd_place, idx)?;
393                ecx.write_scalar(Scalar::from_u16(c), &place)?;
394            }
395            ecx.mark_immutable(&cmd_place);
396        }
397        let imm = argvs_place.to_ref(&ecx);
398        let layout = ecx.layout_of(u8_ptr_ptr_type)?;
399        ImmTy::from_immediate(imm, layout)
400    };
401
402    // Some parts of initialization require a full `InterpCx`.
403    MiriMachine::late_init(&mut ecx, config, {
404        let mut main_thread_state = MainThreadState::GlobalCtors {
405            entry_id,
406            entry_type,
407            argc,
408            argv,
409            ctor_state: global_ctor::GlobalCtorState::default(),
410        };
411
412        // Cannot capture anything GC-relevant here.
413        // `argc` and `argv` *are* GC_relevant, but they also get stored in `machine.argc` and
414        // `machine.argv` so we are good.
415        Box::new(move |m| main_thread_state.on_main_stack_empty(m))
416    })?;
417
418    interp_ok(ecx)
419}
420
421// Call the entry function.
422fn call_main<'tcx>(
423    ecx: &mut MiriInterpCx<'tcx>,
424    entry_id: DefId,
425    entry_type: MiriEntryFnType,
426    argc: ImmTy<'tcx>,
427    argv: ImmTy<'tcx>,
428) -> InterpResult<'tcx, ()> {
429    let tcx = ecx.tcx();
430
431    // Setup first stack frame.
432    let entry_instance = ty::Instance::mono(tcx, entry_id);
433
434    // Return place (in static memory so that it does not count as leak).
435    let ret_place = ecx.allocate(ecx.machine.layouts.isize, MiriMemoryKind::Machine.into())?;
436    ecx.machine.main_fn_ret_place = Some(ret_place.clone());
437
438    // Call start function.
439    match entry_type {
440        MiriEntryFnType::Rustc(EntryFnType::Main { .. }) => {
441            let start_id = tcx.lang_items().start_fn().unwrap_or_else(|| {
442                tcx.dcx().fatal("could not find start lang item");
443            });
444            let main_ret_ty = tcx.fn_sig(entry_id).no_bound_vars().unwrap().output();
445            let main_ret_ty = main_ret_ty.no_bound_vars().unwrap();
446            let start_instance = ty::Instance::try_resolve(
447                tcx,
448                ecx.typing_env(),
449                start_id,
450                tcx.mk_args(&[ty::GenericArg::from(main_ret_ty)]),
451            )
452            .unwrap()
453            .unwrap();
454
455            let main_ptr = ecx.fn_ptr(FnVal::Instance(entry_instance));
456
457            // Always using DEFAULT is okay since we don't support signals in Miri anyway.
458            // (This means we are effectively ignoring `-Zon-broken-pipe`.)
459            let sigpipe = rustc_session::config::sigpipe::DEFAULT;
460
461            ecx.call_function(
462                start_instance,
463                ExternAbi::Rust,
464                &[
465                    ImmTy::from_scalar(
466                        Scalar::from_pointer(main_ptr, ecx),
467                        // FIXME use a proper fn ptr type
468                        ecx.machine.layouts.const_raw_ptr,
469                    ),
470                    argc,
471                    argv,
472                    ImmTy::from_uint(sigpipe, ecx.machine.layouts.u8),
473                ],
474                Some(&ret_place),
475                ReturnContinuation::Stop { cleanup: true },
476            )?;
477        }
478        MiriEntryFnType::MiriStart => {
479            ecx.call_function(
480                entry_instance,
481                ExternAbi::Rust,
482                &[argc, argv],
483                Some(&ret_place),
484                ReturnContinuation::Stop { cleanup: true },
485            )?;
486        }
487    }
488
489    interp_ok(())
490}
491
492/// Evaluates the entry function specified by `entry_id`.
493/// Returns `Some(return_code)` if program execution completed.
494/// Returns `None` if an evaluation error occurred.
495pub fn eval_entry<'tcx>(
496    tcx: TyCtxt<'tcx>,
497    entry_id: DefId,
498    entry_type: MiriEntryFnType,
499    config: &MiriConfig,
500    genmc_ctx: Option<Rc<GenmcCtx>>,
501) -> Option<i32> {
502    // Copy setting before we move `config`.
503    let ignore_leaks = config.ignore_leaks;
504
505    if let Some(genmc_ctx) = &genmc_ctx {
506        genmc_ctx.handle_execution_start();
507    }
508
509    let mut ecx = match create_ecx(tcx, entry_id, entry_type, config, genmc_ctx).report_err() {
510        Ok(v) => v,
511        Err(err) => {
512            let (kind, backtrace) = err.into_parts();
513            backtrace.print_backtrace();
514            panic!("Miri initialization error: {kind:?}")
515        }
516    };
517
518    // Perform the main execution.
519    let res: thread::Result<InterpResult<'_, !>> =
520        panic::catch_unwind(AssertUnwindSafe(|| ecx.run_threads()));
521    let res = res.unwrap_or_else(|panic_payload| {
522        ecx.handle_ice();
523        panic::resume_unwind(panic_payload)
524    });
525    // `Ok` can never happen; the interpreter loop always exits with an "error"
526    // (but that "error" might be just "regular program termination").
527    let Err(err) = res.report_err();
528
529    // Show diagnostic, if any.
530    let (return_code, leak_check) = report_error(&ecx, err)?;
531
532    // We inform GenMC that the execution is complete.
533    if let Some(genmc_ctx) = ecx.machine.data_race.as_genmc_ref()
534        && let Err(error) = genmc_ctx.handle_execution_end(&ecx)
535    {
536        // FIXME(GenMC): Improve error reporting.
537        tcx.dcx().err(format!("GenMC returned an error: \"{error}\""));
538        return None;
539    }
540
541    // If we get here there was no fatal error.
542
543    // Possibly check for memory leaks.
544    if leak_check && !ignore_leaks {
545        // Check for thread leaks.
546        if !ecx.have_all_terminated() {
547            tcx.dcx().err("the main thread terminated without waiting for all remaining threads");
548            tcx.dcx().note("set `MIRIFLAGS=-Zmiri-ignore-leaks` to disable this check");
549            return None;
550        }
551        // Check for memory leaks.
552        info!("Additional static roots: {:?}", ecx.machine.static_roots);
553        let leaks = ecx.take_leaked_allocations(|ecx| &ecx.machine.static_roots);
554        if !leaks.is_empty() {
555            report_leaks(&ecx, leaks);
556            tcx.dcx().note("set `MIRIFLAGS=-Zmiri-ignore-leaks` to disable this check");
557            // Ignore the provided return code - let the reported error
558            // determine the return code.
559            return None;
560        }
561    }
562    Some(return_code)
563}
564
565/// Turns an array of arguments into a Windows command line string.
566///
567/// The string will be UTF-16 encoded and NUL terminated.
568///
569/// Panics if the zeroth argument contains the `"` character because doublequotes
570/// in `argv[0]` cannot be encoded using the standard command line parsing rules.
571///
572/// Further reading:
573/// * [Parsing C++ command-line arguments](https://docs.microsoft.com/en-us/cpp/cpp/main-function-command-line-args?view=msvc-160#parsing-c-command-line-arguments)
574/// * [The C/C++ Parameter Parsing Rules](https://daviddeley.com/autohotkey/parameters/parameters.htm#WINCRULES)
575fn args_to_utf16_command_string<I, T>(mut args: I) -> Vec<u16>
576where
577    I: Iterator<Item = T>,
578    T: AsRef<str>,
579{
580    // Parse argv[0]. Slashes aren't escaped. Literal double quotes are not allowed.
581    let mut cmd = {
582        let arg0 = if let Some(arg0) = args.next() {
583            arg0
584        } else {
585            return vec![0];
586        };
587        let arg0 = arg0.as_ref();
588        if arg0.contains('"') {
589            panic!("argv[0] cannot contain a doublequote (\") character");
590        } else {
591            // Always surround argv[0] with quotes.
592            let mut s = String::new();
593            s.push('"');
594            s.push_str(arg0);
595            s.push('"');
596            s
597        }
598    };
599
600    // Build the other arguments.
601    for arg in args {
602        let arg = arg.as_ref();
603        cmd.push(' ');
604        if arg.is_empty() {
605            cmd.push_str("\"\"");
606        } else if !arg.bytes().any(|c| matches!(c, b'"' | b'\t' | b' ')) {
607            // No quote, tab, or space -- no escaping required.
608            cmd.push_str(arg);
609        } else {
610            // Spaces and tabs are escaped by surrounding them in quotes.
611            // Quotes are themselves escaped by using backslashes when in a
612            // quoted block.
613            // Backslashes only need to be escaped when one or more are directly
614            // followed by a quote. Otherwise they are taken literally.
615
616            cmd.push('"');
617            let mut chars = arg.chars().peekable();
618            loop {
619                let mut nslashes = 0;
620                while let Some(&'\\') = chars.peek() {
621                    chars.next();
622                    nslashes += 1;
623                }
624
625                match chars.next() {
626                    Some('"') => {
627                        cmd.extend(iter::repeat_n('\\', nslashes * 2 + 1));
628                        cmd.push('"');
629                    }
630                    Some(c) => {
631                        cmd.extend(iter::repeat_n('\\', nslashes));
632                        cmd.push(c);
633                    }
634                    None => {
635                        cmd.extend(iter::repeat_n('\\', nslashes * 2));
636                        break;
637                    }
638                }
639            }
640            cmd.push('"');
641        }
642    }
643
644    if cmd.contains('\0') {
645        panic!("interior null in command line arguments");
646    }
647    cmd.encode_utf16().chain(iter::once(0)).collect()
648}
649
650#[cfg(test)]
651mod tests {
652    use super::*;
653    #[test]
654    #[should_panic(expected = "argv[0] cannot contain a doublequote (\") character")]
655    fn windows_argv0_panic_on_quote() {
656        args_to_utf16_command_string(["\""].iter());
657    }
658    #[test]
659    fn windows_argv0_no_escape() {
660        // Ensure that a trailing backslash in argv[0] is not escaped.
661        let cmd = String::from_utf16_lossy(&args_to_utf16_command_string(
662            [r"C:\Program Files\", "arg1", "arg 2", "arg \" 3"].iter(),
663        ));
664        assert_eq!(cmd.trim_end_matches('\0'), r#""C:\Program Files\" arg1 "arg 2" "arg \" 3""#);
665    }
666}
miri/eval.rs

miri/
eval.rs
