Skip to main content

rucc_driver/
lib.rs

1//! The driver: command line parsing, the phase graph, job scheduling and the linker
2//! invocation.
3//!
4//! Design: `spec/04-driver-and-cli.md`. Layer rank 13, see `spec/18-package-layout.md`.
5//!
6//! This is the only crate that is allowed to know the process exists. It reads the command
7//! line, touches the file system, spawns the linker and writes to the terminal, and it hands
8//! everything below it a [`Session`]. The binary crate is a `main` that calls
9//! [`run`] and nothing else, so that the whole driver is reachable from a test.
10//!
11//! # Status
12//!
13//! `--help`, `--version` and `--print-config` are real, which is the `M0` exit criterion in
14//! `spec/17-milestones.md`. The phase graph is real and `-###` prints it, and the scheduler
15//! that will run it is real and tested.
16//!
17//! Two phases run. `-E` reads the file, runs phase 4 over it and writes the result, to `-o` or
18//! to standard output. `--emit=tast` carries on through phase 7, the parse and the checking,
19//! and writes the typed tree. The flags those two read are real with them, which is `-D`, `-U`,
20//! `-I`, `-I-`, `-iquote`, `-isystem`, `-idirafter`, `-iprefix`, `-iwithprefix`,
21//! `-iwithprefixbefore`, `-include`, `-imacros`, `--sysroot=`, `-isysroot`, `-P`, `-std=`,
22//! `-fgnuc-version=`, `-ansi`, `-ffreestanding`, `-fno-builtin`, `-fno-builtin-<name>`,
23//! `-fgnu89-inline`, `-pedantic` and `-Werror`.
24//! The phases after them still say they are not implemented.
25//!
26//! This crate is tier 3 in `spec/18-package-layout.md` section 18.5: its Rust API is
27//! explicitly unstable and will change without a major version bump.
28
29#![doc(html_root_url = "https://docs.rs/rucc-driver/0.10.16")]
30
31pub mod compile;
32pub mod deps;
33pub mod library;
34pub mod link;
35mod map;
36pub mod phase;
37pub mod preprocess;
38pub mod schedule;
39
40use std::fmt::Write as _;
41use std::io::Write as _;
42use std::path::PathBuf;
43
44use rucc_codegen::coverage::{self, Fired};
45use rucc_codegen::pressure::Pressure;
46use rucc_pp::Dependency;
47use rucc_session::{
48    Control, Dumps, EmitKind, Hook, Options, Pic, Preinclude, Protector, SaveTemps, Session, Std,
49    Wrapping, runtime,
50};
51use rucc_target::Triple;
52
53use crate::link::LinkOptions;
54
55pub use crate::compile::{Artifact, Compiled, Temps, compile, compile_ir};
56pub use crate::phase::{Input, InputKind, Job, LinkJob, Output, Phase, Plan};
57pub use crate::preprocess::{OsFileSystem, Preprocessed, preprocess};
58pub use crate::schedule::Jobs;
59
60/// The compiler's version, taken from the workspace manifest.
61pub const VERSION: &str = env!("CARGO_PKG_VERSION");
62
63/// What the command line asked for.
64#[derive(Debug, Clone, PartialEq, Eq)]
65pub enum Action {
66    /// Print usage and exit successfully.
67    Help,
68    /// Print the version and exit successfully.
69    Version,
70    /// Print one line and exit successfully, which is what the `-dump` and `-print` family do.
71    ///
72    /// A build system asks these before it compiles anything, and what it does with the answer
73    /// is paste it into a path or into another command line, so each one is a single line with
74    /// no decoration around it.
75    Print(String),
76    /// Print the resolved configuration and exit successfully.
77    PrintConfig(Box<Options>),
78    /// Print the passes the level will run and exit successfully.
79    PrintPipeline(Box<Options>),
80    /// Print the phase plan and the link line and exit successfully, which is `-###`.
81    PrintPlan {
82        /// The resolved options, which is what says what the link line is for.
83        opts: Box<Options>,
84        /// What to do to each input, and in what order.
85        plan: Box<Plan>,
86        /// What the command line said about linking.
87        link: Box<LinkOptions>,
88    },
89    /// Compile the given inputs.
90    Compile {
91        /// The resolved options.
92        opts: Box<Options>,
93        /// What to do to each input, and in what order.
94        plan: Box<Plan>,
95        /// What the command line said about linking.
96        link: Box<LinkOptions>,
97        /// How many translation units to compile at once.
98        jobs: Jobs,
99        /// Whether `-v` asked for the plan to be printed while it runs.
100        verbose: bool,
101    },
102}
103
104/// Why a command line was rejected.
105#[derive(Debug, Clone, PartialEq, Eq)]
106pub struct CliError {
107    /// The message, lowercase and without a trailing period, in the same shape as any other
108    /// diagnostic.
109    pub message: String,
110}
111
112impl std::fmt::Display for CliError {
113    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
114        f.write_str(&self.message)
115    }
116}
117
118impl std::error::Error for CliError {}
119
120fn err(message: impl Into<String>) -> CliError {
121    CliError { message: message.into() }
122}
123
124/// A question the command line asked instead of asking for a compilation.
125///
126/// These are answered after the loop rather than where they are read, because every one of them
127/// is about the target or about the library search and the last word on both is the end of the
128/// command line.
129enum Query {
130    /// `-dumpmachine`, the triple.
131    Machine,
132    /// `-dumpversion` and `-dumpfullversion`, which are the same three numbers here.
133    Version,
134    /// `-print-multiarch`, the directory name a distribution files this target under.
135    Multiarch,
136    /// `-print-search-dirs`, in the three lines GCC prints.
137    SearchDirs,
138    /// `-print-file-name=<name>`, the full path of a library file.
139    FileName(String),
140    /// `-print-prog-name=<name>`, the full path of a program.
141    ProgName(String),
142    /// `-print-libgcc-file-name`, which is `-print-file-name=libgcc.a` under another spelling.
143    Libgcc,
144}
145
146/// Usage text.
147///
148/// Deliberately short. `spec/04-driver-and-cli.md` puts the full flag reference in the
149/// manual page, because a `--help` nobody can read in one screen is a `--help` nobody reads.
150pub const USAGE: &str = "\
151rucc, an optimizing C compiler
152
153usage: rucc [options] file...
154
155options:
156  -c                     compile and assemble, do not link
157  -S                     compile only, emit assembly
158  -E                     preprocess only
159  -o <file>              write output to <file>, or to standard output for -
160  -D <name>[=<value>], -U <name>      define a macro, or undefine one after every -D
161  -I <dir>               add <dir> to the include search path
162  -iquote -isystem -idirafter <dir>   the other chains, -nostdinc drops ours
163  -I-, -iprefix <p>, -iwithprefix[before] <dir>   the older spellings of those
164  -include <file>, -imacros <file>    read <file> first, the second for its macros only
165  --sysroot=<dir>        look for the library's headers under <dir>, -isysroot too
166  -P, -dM                with -E: leave out the markers, or dump the macros
167  -M -MM -MD -MMD        write a make rule for the source, the last two compile as well
168  -MF <file> -MT <t> -MQ <t> -MP   where the rule goes, what it builds, targets with no recipe
169  -std=<dialect>         c89 through c23, and the gnu spellings
170  -fgnuc-version=<v>     the GCC release to claim, default 7.0.0
171  -x <lang>              treat later inputs as <lang>, or none to stop
172  -O<level>              optimize: 0, 1, 2, 3, s, z
173  -fsafety=<tier>        check memory safety: off, detect, enforce, kernel
174  -f<pass> -fno-<pass> -fdump-ir=<what> -fopt-info[-<kind>][=FILE]
175  -fpass-fuel=<pass>=<n>, -fpass-fuel-global=<n>   stop a pass, or all of them, after n
176  -fdisable-<pass>[=<funcs>], -fenable-<pass>[=<funcs>]   run a pass on some functions only
177  -g -g0 -gdwarf-5, -fno-omit-frame-pointer, -mno-red-zone   debug info, frame pointer, red zone
178  -f[no-]stack-protector[-strong|-all], -f[no-]stack-clash-protection, -fcf-protection=<edges>
179  -ffunction-sections -fdata-sections   a section per function or variable, for --gc-sections
180  -fvisibility=<what>    default, hidden, internal or protected, when nothing in the source said
181  -l<name>, -L <dir>, -B <dir>   link a library, where to look for one, where our own tools are
182  -fPIC -fpic -fPIE -fpie, -fno-common, -f[no-]strict-aliasing, -pipe   what it does anyway
183  -static -shared -pie -no-pie -nostdlib -nostartfiles -nodefaultlibs -rdynamic -s   how to link
184  -Wl,<arg>, -Xlinker <arg>, -fuse-ld=<name>   hand an argument to the linker, or pick one
185  -Werror -pedantic -pedantic-errors -w   how much to say, and whether it is fatal
186  -m64 -march= -mtune= -mcpu= -mabi= -mcmodel=   what machine to generate for
187  -pg -p, -mfentry -mno-fentry   call a profiler on the way in, and where that call goes
188  -fpatchable-function-entry=<n>[,<m>]   room at the top of every function to patch later
189  -fwrapv, -fwrapv-pointer, -fno-strict-overflow   signed or pointer overflow wraps
190  -pthread               build for more than one thread, and link the library for it
191  -dumpmachine -dumpversion -print-multiarch -print-search-dirs   what this compiler is
192  -print-file-name=<name> -print-prog-name=<name>   where a file or a program is
193  -j[n]                  compile n translation units at once, default all
194  -v, -###               print each phase as it runs, or without running any
195  -save-temps[=cwd|obj], -time   keep the .i and the .s, say how long each step took
196  --target=<triple>      generate code for <triple>
197  --emit=<kind>          exe, obj, asm, preprocessed, tast, ir, mir-final,
198                         safety-summary, type-granules
199  --print-config, --print-pipeline    print the configuration or the pipeline, and exit
200  --version              print the version and exit
201  -h, --help             print this message and exit
202
203See spec/04-driver-and-cli.md for the full flag reference.
204";
205
206/// The argument of a flag that may be joined to it or may be the next word.
207///
208/// `-DFOO` and `-D FOO` are the same thing, and `at` is where the flag's own letters end.
209fn joined_or_next(
210    arg: &str,
211    at: usize,
212    args: &[String],
213    i: &mut usize,
214) -> Result<String, CliError> {
215    if arg.len() > at {
216        return Ok(arg[at..].to_owned());
217    }
218    let next = args.get(*i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
219    *i += 1;
220    Ok(next.clone())
221}
222
223/// Parses a command line, without the program name.
224///
225/// # Errors
226///
227/// Returns the message to print when the arguments do not name a compilation this compiler
228/// can attempt.
229pub fn parse_args(args: &[String]) -> Result<Action, CliError> {
230    let host = Triple::host()
231        .ok_or_else(|| err("this host is not a supported target and no --target was given"))?;
232    let mut opts = Options::new(host);
233    let mut inputs: Vec<Input> = Vec::new();
234    let mut print_config = false;
235    let mut print_pipeline = false;
236    let mut print_plan = false;
237    let mut verbose = false;
238    let mut jobs = Jobs::default();
239    let mut nostdinc = false;
240    let mut sysroot: Option<PathBuf> = None;
241    let mut output = None;
242    let mut link = LinkOptions::default();
243    let mut query: Option<Query> = None;
244    let mut threads = false;
245    // `-x` applies to inputs that come after it and stays in effect until the next one, which
246    // is why it is tracked across the loop rather than attached to a single argument.
247    let mut forced: Option<InputKind> = None;
248    // What `-iprefix` last said, stuck on the front of every later `-iwithprefix`. It applies to
249    // the flags after it and not the ones before, so a command line may set it more than once.
250    // GCC's default is its own installed header directory with the last component taken off,
251    // which is a path a cross compiler's build system knows and passes; there is no equivalent
252    // here, so with no `-iprefix` the prefix is nothing and `-iwithprefix` names a directory
253    // outright.
254    let mut iprefix = String::new();
255
256    let mut i = 0;
257    while i < args.len() {
258        let arg = args[i].as_str();
259        i += 1;
260        match arg {
261            "-h" | "--help" => return Ok(Action::Help),
262            "--version" => return Ok(Action::Version),
263            "--print-config" => print_config = true,
264            "--print-pipeline" => print_pipeline = true,
265            "-###" => print_plan = true,
266            "-v" => verbose = true,
267            // The files a compilation goes through, kept rather than thrown away. The bare
268            // spelling means `=obj` and not `=cwd`, which is not what the manual says and is what
269            // gcc 16 does; `SaveTemps::Object` carries the measurement.
270            "-save-temps" => opts.save_temps = SaveTemps::Object,
271            _ if arg.starts_with("-save-temps=") => {
272                opts.save_temps = arg["-save-temps=".len()..].parse().map_err(err)?;
273            }
274            // How long each step took. A misspelling of this is worth rejecting rather than
275            // ignoring, since a run that says nothing looks like a compilation that took no time.
276            "-time" => opts.time = true,
277            "-c" => opts.emit = EmitKind::Object,
278            "-S" => opts.emit = EmitKind::Asm,
279            "-E" => opts.emit = EmitKind::Preprocessed,
280            "-g" => opts.debug_info = true,
281            // GCC's own levels of how much debug information to write. Zero is none and every
282            // other number is some, and this compiler has one amount, so the numbers above zero
283            // all mean the same thing here. `-ggdb` is the same flag asking for whatever the
284            // debugger on the machine prefers, which is what we emit anyway.
285            "-g0" => opts.debug_info = false,
286            "-g1" | "-g2" | "-g3" | "-ggdb" | "-ggdb1" | "-ggdb2" | "-ggdb3" => {
287                opts.debug_info = true;
288            }
289            // The version of DWARF to write. We write DWARF 5 and nothing else, so a build that
290            // asks for another version is told rather than handed a file it cannot read.
291            "-gdwarf" | "-gdwarf-5" => opts.debug_info = true,
292            _ if arg.starts_with("-gdwarf-") => {
293                return Err(err(format!(
294                    "{arg}: this compiler writes DWARF 5 and no other version, see \
295                     spec/11-debug-info.md"
296                )));
297            }
298            "-Werror" => opts.warnings_are_errors = true,
299            // Nothing that is not fatal is said at all. Read at the one place a diagnostic goes
300            // through rather than here, so that a warning `-w` dropped is not counted either.
301            "-w" => opts.warnings = false,
302            "-pedantic-errors" => {
303                opts.pedantic = true;
304                opts.warnings_are_errors = true;
305            }
306            "-P" => opts.line_markers = false,
307            // The dependency family, which section 4.4 calls required because every build system
308            // that generates its own makefiles asks for it. The two that end in `D` write a file
309            // beside the object and let the compilation happen, and the two that do not write to
310            // standard output and stop after it. Nothing here turns the system headers back on
311            // once a flag has turned them off, which is GCC's behaviour and is why `-MM -M` is
312            // `-MM`: the flag asking for fewer of them is the one with something to say.
313            "-M" => {
314                opts.deps.emit = true;
315                opts.deps.instead_of_compiling = true;
316            }
317            "-MM" => {
318                opts.deps.emit = true;
319                opts.deps.instead_of_compiling = true;
320                opts.deps.system_headers = false;
321            }
322            "-MD" => opts.deps.emit = true,
323            "-MMD" => {
324                opts.deps.emit = true;
325                opts.deps.system_headers = false;
326            }
327            "-MP" => opts.deps.phony = true,
328            // These three take a word and only in the separated form, which is how GCC spells
329            // them and how every build system writes them.
330            "-MF" | "-MT" | "-MQ" => {
331                let value =
332                    args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
333                i += 1;
334                match arg {
335                    "-MF" => opts.deps.file = Some(value.clone()),
336                    // The whole of the difference between the two. `-MT` is for a build that has
337                    // already escaped what it is passing, and `-MQ` is for one that has a name
338                    // and wants it to arrive as that name.
339                    "-MT" => opts.deps.targets.push(value.clone()),
340                    _ => opts.deps.targets.push(deps::escaped(value)),
341                }
342            }
343            // The questions a build system asks before it compiles anything. Answered after the
344            // loop, because each one is about the target or the library search and the command
345            // line has not finished saying what those are.
346            "-dumpmachine" => query = Some(Query::Machine),
347            "-dumpversion" | "-dumpfullversion" => query = Some(Query::Version),
348            "-print-multiarch" => query = Some(Query::Multiarch),
349            "-print-search-dirs" => query = Some(Query::SearchDirs),
350            "-print-libgcc-file-name" => query = Some(Query::Libgcc),
351            _ if arg.starts_with("-print-file-name=") => {
352                query = Some(Query::FileName(arg["-print-file-name=".len()..].to_owned()));
353            }
354            _ if arg.starts_with("-print-prog-name=") => {
355                query = Some(Query::ProgName(arg["-print-prog-name=".len()..].to_owned()));
356            }
357            // A program built to run in more than one thread. On every platform this compiler
358            // targets that is a macro the library's headers read and one more library on the
359            // link line, and the library is added after the loop so that it lands after the
360            // objects that refer to it.
361            "-pthread" | "-pthreads" => {
362                opts.defines.push("_REENTRANT".to_owned());
363                threads = true;
364            }
365            "-ansi" => {
366                opts.std = Std::C89;
367                opts.gnu_extensions = false;
368            }
369            // `-Wpedantic` is the same flag under the name the `-W` family gives it, which is
370            // the spelling a build system that groups its warning flags tends to write.
371            "-pedantic" | "-Wpedantic" => opts.pedantic = true,
372            // Both directions, because a build that needs this for one directory turns it back
373            // off for the next one rather than leaving it on for the whole tree.
374            "-fpermissive" => opts.permissive = true,
375            "-fno-permissive" => opts.permissive = false,
376            "-ffreestanding" => opts.hosted = false,
377            "-fhosted" => opts.hosted = true,
378            "-fno-builtin" => opts.builtins = false,
379            "-fbuiltin" => opts.builtins = true,
380            // The C89 dialects are under GNU's reading whatever this says, so turning it off
381            // there is turning off something the dialect asked for, which is accepted and does
382            // nothing. gcc refuses that command line, and there is nothing it could have meant.
383            "-fgnu89-inline" => opts.gnu89_inline = true,
384            "-fno-gnu89-inline" => opts.gnu89_inline = false,
385            // Both directions of each, because a build system that wants one of these usually
386            // writes it beside the flag that turns it back off for one directory.
387            "-fno-omit-frame-pointer" => opts.frame_pointer = true,
388            "-fomit-frame-pointer" => opts.frame_pointer = false,
389            "-mno-red-zone" => opts.red_zone = false,
390            "-mred-zone" => opts.red_zone = true,
391            // Four flags rather than one with an argument, which is how gcc spells them and how
392            // every build line writes them. Last one wins, because a package build puts
393            // `-fstack-protector-strong` in its global flags and a directory that cannot have one
394            // turns it back off on the line after.
395            "-fno-stack-protector" | "-fno-stack-protector-all" | "-fno-stack-protector-strong" => {
396                opts.protector = Protector::None;
397            }
398            "-fstack-protector" => opts.protector = Protector::Buffers,
399            "-fstack-protector-strong" => opts.protector = Protector::Strong,
400            "-fstack-protector-all" => opts.protector = Protector::All,
401            // The other half of what a hardened build asks for, and it is a question about the
402            // frame rather than about the function, so it is a switch rather than a level.
403            "-fstack-clash-protection" => opts.stack_clash = true,
404            "-fno-stack-clash-protection" => opts.stack_clash = false,
405            // The third of them, and the one that is a question with an argument rather than a
406            // family of spellings, because what it asks about is which of the two edges of a
407            // control flow transfer is checked. Bare is both of them, which is what gcc does.
408            "-fcf-protection" => opts.control = Control::Full,
409            "-fno-cf-protection" => opts.control = Control::None,
410            // Two spellings of the same request, which is what gcc has as well. `-p` was the older
411            // profiler and `-pg` the one that also recorded who called whom, and on every platform
412            // this compiler targets there is now one hook and both ask for it.
413            "-pg" | "-p" => {
414                opts.profile = true;
415                link.profile = true;
416            }
417            // Accepted on their own and doing nothing on their own, which is gcc's behaviour: they
418            // say where the call goes and a command line that asked for no call has nowhere to put
419            // one. That matters because a build system that sets `-mfentry` globally and `-pg` per
420            // directory is a build system that would otherwise fail on every other directory.
421            "-mfentry" => opts.hook = Hook::Early,
422            "-mno-fentry" => opts.hook = Hook::Late,
423            // GCC drops its own include directory along with the system ones, because its
424            // headers are half of a pair with the library's and half a pair is worse than
425            // none. A build that passes this is supplying the whole set itself.
426            "-nostdinc" => nostdinc = true,
427            "-o" => {
428                output = Some(args.get(i).ok_or_else(|| err("-o requires an argument"))?.clone());
429                i += 1;
430            }
431            // The flags that take a directory only in the separated form. GCC spells them
432            // this way and nothing writes `-iquotedir`, so accepting the joined form would
433            // mean guessing at a path that starts with the flag's own letters.
434            // Apple's spelling of `--sysroot`, and the one its own build systems pass. The
435            // two mean the same thing here: the configured directories are under there rather
436            // than under the root.
437            "-isysroot" => {
438                let dir = args.get(i).ok_or_else(|| err("-isysroot requires an argument"))?;
439                i += 1;
440                sysroot = Some(PathBuf::from(dir));
441            }
442            "-iquote" | "-isystem" | "-idirafter" => {
443                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
444                i += 1;
445                match arg {
446                    "-iquote" => opts.search.push_quote(dir.clone()),
447                    "-isystem" => opts.search.push_system(dir.clone()),
448                    _ => opts.search.push_after(dir.clone()),
449                }
450            }
451            "-iprefix" => {
452                iprefix = args.get(i).ok_or_else(|| err("-iprefix requires an argument"))?.clone();
453                i += 1;
454            }
455            // Where GCC puts these is not where its manual says it puts them, and this is the
456            // measured answer rather than the documented one: `-iwithprefix` lands in the
457            // `-isystem` slot and not the `-idirafter` slot, and `-iwithprefixbefore` lands in
458            // the `-I` slot. A cross build that uses them is relying on the behaviour, since
459            // that is the compiler it was developed against.
460            "-iwithprefix" | "-iwithprefixbefore" => {
461                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
462                i += 1;
463                let dir = format!("{iprefix}{dir}");
464                if arg == "-iwithprefix" {
465                    opts.search.push_system(dir);
466                } else {
467                    opts.search.push_bracket(dir);
468                }
469            }
470            "-include" | "-imacros" => {
471                let name = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
472                i += 1;
473                opts.preincludes
474                    .push(Preinclude { name: name.clone(), macros_only: arg == "-imacros" });
475            }
476            // The flag `-iquote` was introduced to replace, still passed by build systems old
477            // enough to predate the replacement. It is not a directory: it says that every `-I`
478            // so far is for quoted includes only, and that a quoted include stops looking next
479            // to the file that wrote it.
480            "-I-" => opts.search.split_quote_chain(),
481            "-x" => {
482                let lang = args.get(i).ok_or_else(|| err("-x requires an argument"))?;
483                i += 1;
484                forced = if lang == "none" {
485                    None
486                } else {
487                    Some(InputKind::from_x_arg(lang).map_err(|e| err(format!("{e}")))?)
488                };
489            }
490            // Not a GCC flag. spec/03-architecture.md section 3.5 compiles several
491            // translation units in one process rather than making the build system fork, and
492            // section 3.8's determinism check compares `-j1` against `-j16`, so the knob has
493            // to exist and has to be spelled the way `make` spells it.
494            // `-DFOO`, `-D FOO` and the same for `-U` and `-I`. Both forms are in wide use
495            // and a build system may produce either, so both are read here rather than
496            // being normalised by whatever generated the command line.
497            _ if arg.starts_with("-D") => {
498                let value = joined_or_next(arg, 2, args, &mut i)?;
499                opts.defines.push(value);
500            }
501            _ if arg.starts_with("-U") => {
502                let value = joined_or_next(arg, 2, args, &mut i)?;
503                opts.undefines.push(value);
504            }
505            _ if arg.starts_with("-I") => {
506                let dir = joined_or_next(arg, 2, args, &mut i)?;
507                opts.search.push_bracket(dir);
508            }
509            _ if arg.starts_with("-std=") => {
510                let name = &arg["-std=".len()..];
511                let (std, gnu) = Std::from_flag(name)
512                    .ok_or_else(|| err(format!("unknown dialect `{name}`, see --help")))?;
513                opts.std = std;
514                opts.gnu_extensions = gnu;
515            }
516            // Section 4.5. The claim decides which half of glibc's `sys/cdefs.h` we are
517            // handed, so a differential run that does not set it is comparing two compilers
518            // that believe they are different compilers.
519            // GCC packs these into one flag, so `-dDI` is two of them. Letters in the family
520            // that we have not written yet are accepted and ignored, because a dump is a
521            // debugging aid and a build that asks for one should still compile. A letter
522            // outside the family falls through to the unknown option error, which is what
523            // keeps `-dumpversion` from being read as a dump of nothing.
524            _ if Dumps::is_family(arg) => {
525                opts.dumps.add(&arg[2..]);
526            }
527            // One name at a time, which is what a build that means its own `memcpy` and the
528            // library's everything else writes. The name is not checked against a list, because
529            // the flag is about what the program means by a name and a program is allowed to mean
530            // something by a name this compiler has never heard of.
531            _ if arg.starts_with("-fno-builtin-") => {
532                opts.no_builtin.push(arg["-fno-builtin-".len()..].to_owned());
533            }
534            _ if arg.starts_with("-fgnuc-version=") => {
535                let v = &arg["-fgnuc-version=".len()..];
536                opts.gnuc = v.parse().map_err(err)?;
537            }
538            // spec/13-gnu-compat.md section 13.3 promises this flag an error that says why rather
539            // than the unknown option one, because a build reaching for it is asking for a feature
540            // and deserves to be told it is not coming rather than told the spelling is wrong.
541            // The negative form is what this compiler does anyway, so it is taken and dropped.
542            "-fnested-functions" => {
543                return Err(err(
544                    "nested functions are not supported: a call to one goes through a trampoline \
545                     written on the stack, which no target that enforces an unexecutable stack \
546                     allows",
547                ));
548            }
549            "-fno-nested-functions" => {}
550            // Which of the two links the output is for, which is a real difference and not a
551            // description of what happens anyway. Everything here is position independent either
552            // way, and what these decide is whether a name may be one another object defines or
553            // replaces, because a link that produces an executable puts every name in the same
554            // program and a link that produces a shared library does not.
555            //
556            // It matters that they are accepted at all, whatever they then do. Every autoconf and
557            // cmake build puts `-fPIC` on the compile line, so a compiler that rejects it cannot
558            // be the `CC` of a project that has a configure script, whatever else it can do. That
559            // is how this was found: building SQLite's test fixture stopped on it.
560            "-fPIC" | "-fpic" => opts.pic = Pic::Library,
561            // Not a synonym of the pair above, which is what they were treated as until #756. The
562            // library is the expensive answer and gcc makes it the one that has to be asked for,
563            // so this is also what nothing at all means.
564            "-fPIE" | "-fpie" => opts.pic = Pic::Executable,
565            // A different question from the pair above, and the one every distribution build of a
566            // shared library answers. `-fPIC` decides how an address is reached, and this decides
567            // whether the optimizer may believe a body it can see, because an exported name is one
568            // the dynamic linker may find another definition of first. On by default, which is
569            // gcc's arrangement and is the honest answer, and off is a promise the build makes and
570            // nothing checks.
571            "-fsemantic-interposition" => opts.interposition = true,
572            "-fno-semantic-interposition" => opts.interposition = false,
573            // Two requests rather than one, and the same table answers both, so what decides is
574            // whether either of them is standing. gcc arranges it the same way: the asynchronous
575            // one is the default here and it implies the other, and a line that asks for a table
576            // and against an asynchronous one gets a table.
577            "-fasynchronous-unwind-tables" => opts.async_unwind_tables = true,
578            "-fno-asynchronous-unwind-tables" => opts.async_unwind_tables = false,
579            "-funwind-tables" => opts.unwind_tables = true,
580            "-fno-unwind-tables" => opts.unwind_tables = false,
581            // The other direction is a request, not a description, and it is one this compiler
582            // cannot grant, so it gets the treatment section 13.3 asks for rather than the unknown
583            // option error. Answering it by carrying on would be answering a different question:
584            // the code would still be position independent, which is correct everywhere an
585            // ordinary program runs and is wrong in a kernel, where the flag is written precisely
586            // because there is no loader to fill a global offset table in.
587            "-fno-pic" | "-fno-pie" => {
588                return Err(err(
589                    "position dependent code is not supported: an address that may be in another \
590                     object is loaded out of the global offset table, and nothing here emits the \
591                     absolute form this asks for. Use -no-pie if what you meant was how to link",
592                ));
593            }
594            // A section per function and a section per variable, which is what makes
595            // `--gc-sections` able to drop anything: a linker can leave out a section nothing
596            // reaches and cannot leave out half of one. Both directions are taken, and the off
597            // one is the default rather than a refusal, since a build that writes it is asking
598            // for what happens anyway.
599            "-ffunction-sections" => opts.function_sections = true,
600            "-fno-function-sections" => opts.function_sections = false,
601            "-fdata-sections" => opts.data_sections = true,
602            "-fno-data-sections" => opts.data_sections = false,
603            // Another description of what this compiler does. A file scope declaration with no
604            // initializer is written into `.bss` as an ordinary defined symbol, not offered to the
605            // linker as a common one for it to merge, which is what `-fno-common` asks for and what
606            // gcc has done by default since 10. Nothing in the front end produces `Linkage::Common`
607            // at all.
608            "-fno-common" => {}
609            // What overflows rather than being undefined. Every one of these takes something away
610            // from the optimizer rather than asking it to do anything, which is why the negative
611            // spellings are the interesting ones and the positive spellings are the default.
612            //
613            // `-fno-strict-overflow` is both of the others, which is gcc's own reading of it: its
614            // help text for `-fstrict-overflow` says "negated as -fwrapv -fwrapv-pointer". So it is
615            // written here as the pair rather than kept as a third thing to test everywhere.
616            "-fwrapv" => opts.wrapping.signed = true,
617            "-fno-wrapv" => opts.wrapping.signed = false,
618            "-fwrapv-pointer" => opts.wrapping.pointer = true,
619            "-fno-wrapv-pointer" => opts.wrapping.pointer = false,
620            "-fno-strict-overflow" => opts.wrapping = Wrapping::ALL,
621            "-fstrict-overflow" => opts.wrapping = Wrapping::NONE,
622            // And the request, which is the one that cannot be granted. It is a real difference and
623            // not a preference: two files each writing `int g;` link under `-fcommon` and are a
624            // duplicate definition without it, which is the whole reason the flag survives.
625            "-fcommon" => {
626                return Err(err(
627                    "a tentative definition is written into .bss as its own symbol here, and \
628                     nothing emits the common symbol this asks the linker to merge. Give the \
629                     variable a definition in one file and declare it extern in the others",
630                ));
631            }
632            // Both directions of this one are taken, which is the exception to the rule above, and
633            // the reason is which way being wrong costs something.
634            //
635            // Nothing here derives anything from the type an object is accessed through. The IR has
636            // somewhere to put a type based aliasing node and lowering fills it with nothing on
637            // every access, so no pass has one to read and the alias analysis falls back to what it
638            // can see. That makes `-fno-strict-aliasing` a description, the way `-fPIC` is.
639            //
640            // `-fstrict-aliasing` is a request to assume more than that, and this compiler assumes
641            // less. Answering a request for a weaker guarantee by giving a stronger one is safe in
642            // a way the `-fno-pic` case is not: every program that is correct under the assumption
643            // is correct without it, only slower. And `-O2` implies it, so a build that spells it
644            // out is a build that would stop on a compiler that refused it, for no gain at all.
645            "-fstrict-aliasing" | "-fno-strict-aliasing" => {}
646            // About temporary files rather than about code. There is nothing between the phases of
647            // one compilation here to write to a file in the first place.
648            "-pipe" => {}
649            // Nothing here writes colour, so all of these are the same answer, and it is the answer
650            // that costs nothing: the diagnostics come out plain either way and no build depends on
651            // an escape sequence being there. Taken rather than refused because cmake writes
652            // `-fdiagnostics-color=always` on every compile line when the generator is ninja, which
653            // makes this the second most common flag after `-fPIC` to stop a build over a question
654            // about how the text looks.
655            "-fdiagnostics-color" | "-fno-diagnostics-color" => {}
656            _ if arg.starts_with("-fdiagnostics-color=") => {}
657            // The link flags. None of them changes the compilation, which is why they are
658            // collected apart from `opts` and why `-lm` on a `-c` line is a note rather than an
659            // error: it is a thing said to a linker that is not going to run.
660            "-static" => link.is_static = true,
661            "-shared" => link.shared = true,
662            "-pie" => link.pie = Some(true),
663            "-no-pie" | "-nopie" => link.pie = Some(false),
664            "-nostdlib" => link.no_stdlib = true,
665            "-nostartfiles" => link.no_startfiles = true,
666            "-nodefaultlibs" => link.no_defaultlibs = true,
667            "-fno-builtins-lib" => link.no_builtins_lib = true,
668            "-fbuiltins-lib" => link.no_builtins_lib = false,
669            "-rdynamic" | "-export-dynamic" => link.export_dynamic = true,
670            "-s" => link.strip = true,
671            "-Xlinker" => {
672                let next = args.get(i).ok_or_else(|| err("-Xlinker requires an argument"))?;
673                i += 1;
674                link.passthrough.push(next.clone());
675            }
676            _ if arg.starts_with("-Wl,") => {
677                // Commas separate arguments rather than being part of one, which is what makes
678                // `-Wl,-rpath,/opt/lib` two words to the linker and one word here.
679                link.passthrough.extend(arg["-Wl,".len()..].split(',').map(str::to_owned));
680            }
681            _ if arg.starts_with("-fuse-ld=") => {
682                link.use_ld = Some(arg["-fuse-ld=".len()..].to_owned());
683            }
684            _ if arg.starts_with("-l") && arg.len() > 2 => {
685                inputs.push(Input::library(&arg[2..]));
686            }
687            "-l" => {
688                let next = args.get(i).ok_or_else(|| err("-l requires an argument"))?;
689                i += 1;
690                inputs.push(Input::library(next));
691            }
692            _ if arg.starts_with("-L") => {
693                link.search.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
694            }
695            _ if arg.starts_with("-B") => {
696                link.prefixes.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
697            }
698            _ if arg.starts_with("-j") => {
699                jobs = Jobs::parse(&arg[2..]).map_err(err)?;
700            }
701            _ if arg.starts_with("--sysroot=") => {
702                sysroot = Some(PathBuf::from(&arg["--sysroot=".len()..]));
703            }
704            _ if arg.starts_with("--target=") => {
705                let t = &arg["--target=".len()..];
706                opts.target = t.parse().map_err(|e| err(format!("{e}")))?;
707            }
708            _ if arg.starts_with("--emit=") => {
709                let k = &arg["--emit=".len()..];
710                opts.emit = k
711                    .parse()
712                    .map_err(|()| err(format!("unknown --emit kind `{k}`, see --help")))?;
713            }
714            // A bare `-O` is `-O1`, which is what GCC has and what a hand written makefile tends
715            // to write. `-Og` is GCC's level for a build somebody is going to step through, and
716            // it is `-O1` with the transformations that move code around left out; this compiler
717            // has no such level yet, so it is the nearest one and `--print-pipeline` says what
718            // that came to rather than the flag pretending otherwise.
719            "-O" | "-Og" => opts.opt_level = rucc_session::OptLevel::O1,
720            // The union of `-O3` and `-ffast-math`, and the second half of that changes what
721            // floating point arithmetic means. Refused rather than taken as `-O3`, because a
722            // build that asks for fast math and is quietly given ordinary arithmetic gets a
723            // slower program than it asked for and a build that is given fast math it did not
724            // ask for gets a wrong one.
725            "-Ofast" => {
726                return Err(err(
727                    "-Ofast is -O3 with fast math, and fast math is not implemented, see \
728                     spec/04-driver-and-cli.md section 4.6",
729                ));
730            }
731            _ if arg.starts_with("-O") => {
732                opts.opt_level = arg[2..]
733                    .parse()
734                    .map_err(|()| err(format!("unknown optimization level `{arg}`")))?;
735            }
736            // What every name gets when nothing in the source said, which the attribute in the
737            // source overrides rather than the other way round. Before the optimizer's `-f`
738            // family below for the reason the tier below it is.
739            _ if arg.starts_with("-fvisibility=") => {
740                let seen = &arg["-fvisibility=".len()..];
741                opts.visibility = seen.parse().map_err(|()| {
742                    err(format!(
743                        "`{seen}` is not a visibility, which is default, hidden, internal or \
744                         protected"
745                    ))
746                })?;
747            }
748            // Which edges of a control flow transfer are checked. Before the optimizer's `-f`
749            // family below for the reason the two above it are, and last of the three so that the
750            // bare spelling and the negative one are matched exactly rather than by this.
751            _ if arg.starts_with("-fcf-protection=") => {
752                let edges = &arg["-fcf-protection=".len()..];
753                opts.control = edges.parse().map_err(|()| {
754                    err(format!(
755                        "`{edges}` is not a control flow protection, which is full, branch, \
756                         return, none or check"
757                    ))
758                })?;
759            }
760            // How much room every function opens with for something to be written over later.
761            // Before the optimizer's `-f` family below for the reason the ones above it are.
762            _ if arg.starts_with("-fpatchable-function-entry=") => {
763                let room = &arg["-fpatchable-function-entry=".len()..];
764                opts.patchable = room.parse().map_err(|()| {
765                    err(format!(
766                        "`{room}` is not an amount of room to reserve, which is a number of bytes                          and then, after a comma, how many of them go in front of the function's                          own label"
767                    ))
768                })?;
769            }
770            // The memory safety monitor, from section 15.4 of
771            // `spec/safe-memory/15-integration.md`. Before the optimizer's `-f` family below,
772            // because a pass that took the name `safety=detect` would otherwise be handed the
773            // flag, and the tier is not a pass.
774            _ if arg.starts_with("-fsafety=") => {
775                let tier = &arg["-fsafety=".len()..];
776                opts.safety = tier.parse().map_err(|()| {
777                    err(format!(
778                        "`{tier}` is not a safety tier, which is off, detect, enforce or kernel"
779                    ))
780                })?;
781            }
782            // The optimizer's own flags, from section 9.10 of `spec/09-optimizer.md`. These come
783            // after every `-f` the rest of the compiler answers to, so a pass can never take a
784            // name that already means something else on the command line.
785            _ if arg.starts_with("-fpass-fuel=") => {
786                let (name, count) = arg["-fpass-fuel=".len()..]
787                    .split_once('=')
788                    .ok_or_else(|| err("-fpass-fuel= is spelled <pass>=<count>"))?;
789                if rucc_opt::pass::find(name).is_none() {
790                    return Err(err(format!(
791                        "`{name}` is not a pass this compiler has, see --print-pipeline"
792                    )));
793                }
794                let count: u32 = count
795                    .parse()
796                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
797                opts.pass_fuel.push((name.to_owned(), count));
798            }
799            _ if arg.starts_with("-fpass-fuel-global=") => {
800                let count = &arg["-fpass-fuel-global=".len()..];
801                let count: u32 = count
802                    .parse()
803                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
804                opts.pass_fuel_global = Some(count);
805            }
806            // Everything from `-fopt-info` to the end of the argument, which is optional
807            // keywords joined by hyphens and an optional `=<file>`. Checked here rather than
808            // where the remarks are printed, because by then the compilation somebody wanted
809            // to hear about is over.
810            _ if arg == "-fopt-info"
811                || arg.starts_with("-fopt-info=")
812                || arg.starts_with("-fopt-info-") =>
813            {
814                let rest = &arg["-fopt-info".len()..];
815                let (kinds, file) = match rest.split_once('=') {
816                    Some((kinds, file)) => (kinds, Some(file)),
817                    None => (rest, None),
818                };
819                let kinds = kinds.strip_prefix('-').unwrap_or(kinds);
820                rucc_opt::Wants::none().add(kinds).map_err(err)?;
821                opts.opt_info.push(kinds.to_owned());
822                if let Some(file) = file {
823                    if file.is_empty() {
824                        return Err(err("-fopt-info= was given no file to write to"));
825                    }
826                    opts.opt_info_file = Some(file.to_owned());
827                }
828            }
829            _ if arg.starts_with("-fdump-ir=") => {
830                // Checked here rather than where the dumps are taken, because the compilation
831                // that would have been dumped is over by then.
832                let spec = &arg["-fdump-ir=".len()..];
833                rucc_opt::Dumps::default().add(spec).map_err(err)?;
834                opts.dump_ir.push(spec.to_owned());
835            }
836            // Before the bare `-f<pass>` below, because a pass called `enable-something` would
837            // otherwise take the flag away from the gate. Checked here rather than where the
838            // pipeline reads it, for the reason that applies to all of these: a misspelled pass
839            // name that quietly gated nothing looks exactly like a pass that is not the guilty
840            // one, and a bisection would carry on past the thing it was looking for.
841            _ if arg.starts_with("-fdisable-") || arg.starts_with("-fenable-") => {
842                let on = arg.starts_with("-fenable-");
843                let spec = &arg[if on { "-fenable-".len() } else { "-fdisable-".len() }..];
844                rucc_opt::Gates::default().add(on, spec).map_err(err)?;
845                opts.pass_gates.push((on, spec.to_owned()));
846            }
847            _ if arg.strip_prefix("-fno-").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
848                opts.passes.push((arg["-fno-".len()..].to_owned(), false));
849            }
850            _ if arg.strip_prefix("-f").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
851                opts.passes.push((arg["-f".len()..].to_owned(), true));
852            }
853            // The unstable options, spelled the way rustc spells them and carrying the same
854            // promise, which is none: one of these may change or go away in any release. They are
855            // measurements and debugging aids rather than things a build asks for, which is why
856            // none of them is in the usage text and all of them are in section 4.11 of
857            // `spec/04-driver-and-cli.md`.
858            "-Zverify-each" => opts.verify_each = true,
859            _ if arg.starts_with("-Zrule-coverage=") => {
860                let file = &arg["-Zrule-coverage=".len()..];
861                if file.is_empty() {
862                    return Err(err("-Zrule-coverage= needs a file to write to"));
863                }
864                opts.rule_coverage = Some(file.to_owned());
865            }
866            _ if arg.starts_with("-Zregister-pressure=") => {
867                let file = &arg["-Zregister-pressure=".len()..];
868                if file.is_empty() {
869                    return Err(err("-Zregister-pressure= needs a file to write to"));
870                }
871                opts.register_pressure = Some(file.to_owned());
872            }
873            _ if arg.starts_with("-Z") => {
874                return Err(err(format!(
875                    "`{arg}` is not an unstable option this compiler has, see \
876                     spec/04-driver-and-cli.md section 4.11 for the ones it does"
877                )));
878            }
879            // The word size, which is a statement about the target and is taken as one. A build
880            // that says the size the target already has is saying nothing, and one that says the
881            // other size is asking for a target this compiler does not have, which it is told
882            // rather than being given the wrong one.
883            "-m64" | "-m32" | "-mx32" => {
884                let want: u32 = match arg {
885                    "-m64" => 64,
886                    _ => 32,
887                };
888                let have = rucc_target::TargetInfo::new(opts.target).pointer_width;
889                if have != want {
890                    return Err(err(format!(
891                        "{arg} asks for a {want} bit target and {} is {have} bit, use \
892                         --target= to name the one you mean",
893                        opts.target
894                    )));
895                }
896            }
897            // Which processor in the family to generate for. This compiler emits the base
898            // instruction set of the architecture and nothing above it, so a program built with
899            // any of these runs on the machine that was named; it is a program that could have
900            // been faster rather than a program that is wrong, which is what makes these safe to
901            // take and ignore where a flag that changed the meaning of the code would not be.
902            _ if arg.starts_with("-march=")
903                || arg.starts_with("-mtune=")
904                || arg.starts_with("-mcpu=") => {}
905            // The calling convention, which is not safe to ignore. Taken when it names the one
906            // the target already uses and refused otherwise.
907            _ if arg.starts_with("-mabi=") => {
908                let want = &arg["-mabi=".len()..];
909                let have = match opts.target.arch {
910                    rucc_target::Arch::X86_64 => "sysv",
911                    rucc_target::Arch::Aarch64 => "lp64",
912                    rucc_target::Arch::Riscv64 => "lp64d",
913                };
914                if want != have {
915                    return Err(err(format!(
916                        "{arg}: {} uses the {have} convention and this compiler has no other",
917                        opts.target
918                    )));
919                }
920            }
921            // How far apart the pieces of the program may be. The small model is what we emit and
922            // it is every hosted program's default; the kernel model is a different one and a
923            // build that asks for it and does not get it links and then does not run.
924            "-mcmodel=small" => {}
925            _ if arg.starts_with("-mcmodel=") => {
926                return Err(err(format!(
927                    "{arg}: this compiler emits the small code model and no other, see \
928                     spec/12-targets.md"
929                )));
930            }
931            // GCC's own scripting language for how the driver builds a command line.
932            // `spec/04-driver-and-cli.md` section 4.4 settles that we will not have it, so a
933            // build reaching for it is told which flags do the same job.
934            _ if arg.starts_with("-specs=") => {
935                return Err(err(
936                    "-specs= is not supported: the parts of it builds rely on are -B, -L, \
937                     -nostdlib, -nostartfiles and -Wl,, see spec/04-driver-and-cli.md \
938                     section 4.4",
939                ));
940            }
941            // Arguments meant for a separate assembler or preprocessor, which this compiler does
942            // not have: both are inside it and neither reads a command line. Refused rather than
943            // dropped, because every one of these says something about the output and a build
944            // that asked for `-Wa,--noexecstack` and was silently given an executable stack got
945            // the opposite of what it asked for.
946            _ if arg.starts_with("-Wa,") || arg.starts_with("-Wp,") => {
947                return Err(err(format!(
948                    "`{arg}` is an argument for a separate assembler or preprocessor, and both \
949                     are inside this compiler rather than programs it runs"
950                )));
951            }
952            "-Xassembler" | "-Xpreprocessor" => {
953                return Err(err(format!(
954                    "{arg} hands an argument to a separate assembler or preprocessor, and both \
955                     are inside this compiler rather than programs it runs"
956                )));
957            }
958            // Everything else in the `-W` family. `spec/04-driver-and-cli.md` section 4.1 has
959            // this one as a rule about build systems rather than about warnings: autoconf finds
960            // out whether a warning flag exists by passing it and looking at the exit status, so
961            // a compiler that refuses one it has not heard of fails a configure script written
962            // for a GCC newer than itself. The names are not checked against a list because this
963            // compiler has no warning groups for a list to be of, which #485 is about.
964            _ if arg.starts_with("-W") => {}
965            // Flags that name something this compiler does not do and would not do differently
966            // if it did. `-fno-ident` is about a comment in the output that we do not write
967            // either way, and the others are about a way of ordering the compilation that has
968            // been GCC's only way for twenty years. Section 4.1 asks for the list to be short
969            // and for adding to it to be deliberate, which is why it is written out here.
970            "-fno-ident"
971            | "-fident"
972            | "-funit-at-a-time"
973            | "-fno-unit-at-a-time"
974            | "-shared-libgcc"
975            | "-static-libgcc" => {}
976            _ if arg.starts_with('-') && arg.len() > 1 => {
977                // Silently ignoring an unknown flag is how a build ends up not doing what
978                // its author asked. spec/13-gnu-compat.md section 13.4 makes this an error
979                // for the flags that change code generation, and the safe default until the
980                // flag table is populated is to reject everything we do not know.
981                return Err(err(format!("unknown option `{arg}`")));
982            }
983            _ => inputs.push(Input { path: arg.to_owned(), forced, library: false }),
984        }
985    }
986
987    // Last, so that it lands after every `-isystem` the command line gave. That is GCC's
988    // order: a directory the user names outranks the compiler's own, and the compiler's own
989    // outranks the library's. It is pushed after the loop rather than before it because
990    // `SearchPath` appends within a group and the position is what the order is.
991    // The same directory the headers were looked for under, because a sysroot is a statement
992    // about a whole installation and not about half of one.
993    link.sysroot = sysroot.clone();
994    // After the loop rather than where `-pthread` was read, so that it lands after the objects
995    // that refer to it. A static link takes the definitions it needs from a library when it
996    // reaches it and not afterwards, so a library before the objects is a library that answers
997    // nothing.
998    if threads {
999        inputs.push(Input::library("pthread"));
1000    }
1001    if let Some(query) = query {
1002        return Ok(Action::Print(answer(&query, &opts, &link)));
1003    }
1004    // `-M` and `-MM` produce the rule and nothing else, so the run stops after phase 4 whatever
1005    // else the command line asked for. Read here rather than where the flag was, because a `-c`
1006    // written after it has to lose and the loop cannot know that until it has ended. The output
1007    // file is where the rule goes rather than where an object would have gone, and the last
1008    // phase being the preprocessor is what makes that true without a second rule for it.
1009    if opts.deps.instead_of_compiling {
1010        opts.emit = EmitKind::Preprocessed;
1011    }
1012    if !nostdinc {
1013        opts.search.push_system(runtime::DIR);
1014        // And the library's after ours, which is the other half of the same order. They go on
1015        // here rather than at the point `--target=` or `--sysroot=` was read because either
1016        // one changes the answer and the last word on both is the end of the loop.
1017        for dir in library::system_dirs(opts.target, sysroot.as_deref()) {
1018            opts.search.push_system(dir);
1019        }
1020    }
1021    // Once, here, rather than as each directory is pushed. A `-I` that names a system
1022    // directory has to lose to the system entry and the system entry is added last, so the
1023    // question cannot be answered until the whole path is known.
1024    opts.search.remove_duplicates();
1025
1026    // The target has to be resolved before the configuration is printed, so this check comes
1027    // after the loop rather than at the point `--print-config` was seen.
1028    if print_config {
1029        return Ok(Action::PrintConfig(Box::new(opts)));
1030    }
1031    if print_pipeline {
1032        return Ok(Action::PrintPipeline(Box::new(opts)));
1033    }
1034    let plan = Plan::new(&opts, &inputs, output.as_deref()).map_err(|e| err(e.message))?;
1035    if print_plan {
1036        return Ok(Action::PrintPlan {
1037            opts: Box::new(opts),
1038            plan: Box::new(plan),
1039            link: Box::new(link),
1040        });
1041    }
1042    Ok(Action::Compile {
1043        opts: Box::new(opts),
1044        plan: Box::new(plan),
1045        link: Box::new(link),
1046        jobs,
1047        verbose,
1048    })
1049}
1050
1051/// What one of the `-dump` and `-print` flags prints.
1052///
1053/// GCC prints the name back unchanged when it cannot find the file a `-print` flag asked about,
1054/// which is what makes the answer safe to paste into a link line whether or not the file is
1055/// there, and this does the same.
1056fn answer(query: &Query, opts: &Options, link: &LinkOptions) -> String {
1057    let found = |name: &str| {
1058        link::find_in_search(link, opts.target, name)
1059            .map_or_else(|| name.to_owned(), |path| path.display().to_string())
1060    };
1061    match query {
1062        Query::Machine => opts.target.to_string(),
1063        Query::Version => VERSION.to_owned(),
1064        Query::Multiarch => link::multiarch(opts.target),
1065        // The three lines GCC prints, in its order and with its punctuation, because what reads
1066        // them is a script written against that shape. There is no installation directory to
1067        // report: this compiler is one binary that works wherever it is copied, and the headers
1068        // it ships are inside it, so `install` is where the binary is and nothing is under it.
1069        Query::SearchDirs => {
1070            let here = std::env::current_exe()
1071                .ok()
1072                .and_then(|p| p.parent().map(std::path::Path::to_path_buf))
1073                .unwrap_or_default();
1074            let list = |dirs: &[PathBuf]| {
1075                dirs.iter().map(|d| d.display().to_string()).collect::<Vec<_>>().join(":")
1076            };
1077            let libraries = link::search_dirs(link, opts.target);
1078            format!(
1079                "install: {}\nprograms: ={}\nlibraries: ={}",
1080                here.display(),
1081                list(&link.prefixes),
1082                list(&libraries)
1083            )
1084        }
1085        Query::FileName(name) => found(name),
1086        // The name GCC gives the library of routines a compiler's output calls that the C
1087        // library does not have. Ours is built in and there is no file, so the answer is the
1088        // name itself, which is what GCC prints when it cannot find one either.
1089        Query::Libgcc => found("libgcc.a"),
1090        // A program rather than a library: the linker and the archiver are the ones a build asks
1091        // about, and this compiler finds them on the path or under `-B` rather than shipping
1092        // them, so the name back is the honest answer unless a `-B` prefix holds one.
1093        Query::ProgName(name) => link
1094            .prefixes
1095            .iter()
1096            .map(|dir| dir.join(name))
1097            .find(|path| path.is_file())
1098            .map_or_else(|| name.clone(), |path| path.display().to_string()),
1099    }
1100}
1101
1102/// Renders the passes this level will run, in order, with what each one does.
1103///
1104/// The level is the whole of the answer unless a `-f` flag edited it, which is section 9.1 of
1105/// `spec/09-optimizer.md`: a level is a list somebody wrote down rather than something that
1106/// emerges from which flags happen to be set, and this is how that list is read.
1107#[must_use]
1108pub fn print_pipeline(opts: &Options) -> String {
1109    let mut settings = rucc_opt::Options::for_level(opts.opt_level);
1110    settings.toggles.clone_from(&opts.passes);
1111    settings.global_fuel = opts.pass_fuel_global;
1112    for (on, spec) in &opts.pass_gates {
1113        // Every spelling was checked while the arguments were parsed, so there is nothing here
1114        // this can refuse, and a listing is not the place to report it if there were.
1115        let _ = settings.gates.add(*on, spec);
1116    }
1117    rucc_opt::pipeline::print(&settings)
1118}
1119
1120/// Renders the resolved configuration.
1121///
1122/// One `key: value` per line, sorted by nothing in particular but fixed in order, because
1123/// this output is diffed across hosts in CI and a reordering would read as a change.
1124#[must_use]
1125pub fn print_config(opts: &Options) -> String {
1126    let sess = Session::new(opts.clone());
1127    let t = &sess.target;
1128    let mut out = String::new();
1129    let _ = writeln!(out, "version: {VERSION}");
1130    // The three field triple the driver was given rather than the ten field tuple it widens to,
1131    // because this output is what a build system reads to find out what it asked for. The tuple is
1132    // the compiler's model of the machine and this line is a receipt for a command line.
1133    let _ = writeln!(out, "target: {}", opts.target);
1134    let _ = writeln!(out, "arch: {}", opts.target.arch.as_str());
1135    let _ = writeln!(out, "os: {}", opts.target.os.as_str());
1136    let _ = writeln!(out, "env: {}", opts.target.env.as_str());
1137    let _ = writeln!(out, "object-format: {}", t.object_format.as_str());
1138    let _ = writeln!(out, "pointer-width: {}", t.pointer_width);
1139    let _ = writeln!(out, "long-width: {}", t.long_width);
1140    let _ = writeln!(out, "long-double-width: {}", t.long_double_width);
1141    let _ = writeln!(out, "endian: {}", if t.little_endian { "little" } else { "big" });
1142    let _ = writeln!(out, "char-signed: {}", t.char_is_signed);
1143    let _ = writeln!(out, "va-list: {}", t.va_list.map_or("none", |list| list.as_str()));
1144    // The register file as a count per class, which is enough to tell a target whose registers
1145    // are described from one whose are not without printing sixteen names nobody asked for.
1146    let regs: Vec<String> = t
1147        .regs
1148        .classes()
1149        .map(|(class, info)| format!("{} {}", info.name, t.regs.len(class)))
1150        .collect();
1151    let _ = writeln!(
1152        out,
1153        "registers: {}",
1154        if regs.is_empty() { "none".to_string() } else { regs.join(", ") }
1155    );
1156    let _ = writeln!(out, "opt-level: {}", sess.opts.opt_level);
1157    let _ = writeln!(out, "safety: {}", sess.opts.safety);
1158    let _ = writeln!(out, "emit: {}", sess.opts.emit.as_str());
1159    let _ = writeln!(out, "debug-info: {}", sess.opts.debug_info);
1160    let _ = writeln!(out, "frame-pointer: {}", sess.opts.frame_pointer);
1161    let _ = writeln!(out, "red-zone: {}", sess.opts.red_zone);
1162    let _ = writeln!(out, "stack-protector: {}", sess.opts.protector);
1163    let _ = writeln!(out, "stack-clash-protection: {}", sess.opts.stack_clash);
1164    let _ = writeln!(out, "cf-protection: {}", sess.opts.control);
1165    let _ = writeln!(out, "patchable-function-entry: {}", sess.opts.patchable);
1166    let _ = writeln!(out, "profile: {}", sess.opts.profile);
1167    let _ = writeln!(out, "profile-hook: {}", sess.opts.hook);
1168    // Last because it is the one key with more than one line under it, and the only one
1169    // whose value is a property of the machine rather than of the command line.
1170    for dir in sess.opts.search.dirs() {
1171        let system = if dir.is_system { " (system)" } else { "" };
1172        let _ = writeln!(out, "include: {}{system}", dir.path.display());
1173    }
1174    out
1175}
1176
1177/// The output name the make target is taken from, which is the `-o` argument or nothing.
1178///
1179/// A run that stops at the preprocessor has not named an object, whatever its `-o` says: under
1180/// `-E` that argument is the preprocessed text and under `-M` it is the rule itself, and neither
1181/// is a file `make` would rebuild by running this rule. GCC agrees and falls back to the source
1182/// name in both, which is why a `-MD -E -o out.i` writes `out.d` holding a rule for `a.o`. From
1183/// `-S` on the argument does name what the rule builds, and it is used as written.
1184fn deps_target_output<'a>(opts: &Options, plan: &'a Plan) -> Option<&'a str> {
1185    if opts.emit == EmitKind::Preprocessed { None } else { plan.output.as_deref() }
1186}
1187
1188/// Writes to a path the command line named rather than one the plan derived, where `-` is
1189/// standard output.
1190fn write_named(path: &str, bytes: &[u8]) -> Result<(), String> {
1191    if path == "-" {
1192        return write_out(&Output::Stdout, bytes);
1193    }
1194    write_out(&Output::File(path.to_owned()), bytes)
1195}
1196
1197/// Writes the make rule for one input, and reports whether it got there.
1198///
1199/// A rule with no file of its own goes where the compilation it replaced would have written,
1200/// which is what makes the usual makefile recipe work: `rucc -M $< -o $@` leaves the rule in
1201/// `$@`, and the same line with the `-o` left off puts it on standard output.
1202fn write_deps(
1203    opts: &Options,
1204    plan: &Plan,
1205    job: &Job,
1206    found: &[Dependency],
1207    stderr: &mut impl std::io::Write,
1208) -> bool {
1209    let targets = if opts.deps.targets.is_empty() {
1210        vec![deps::default_target(&job.input, deps_target_output(opts, plan))]
1211    } else {
1212        opts.deps.targets.clone()
1213    };
1214    let rule = deps::rule(&opts.deps, &targets, &job.input, found);
1215    // The file, on the other hand, is named after the `-o` in every mode that still has one to
1216    // spend, which is every mode except the two that spend it on the rule.
1217    let wrote = match deps::default_file(&opts.deps, &job.input, plan.output.as_deref()) {
1218        // A `-MF` on a run that had nowhere else to put the rule leaves the file the `-o`
1219        // named empty rather than absent, because a makefile that named it as a target of its
1220        // own is a makefile that will look for it.
1221        Some(path) => write_named(&path, rule.as_bytes()).and_then(|()| {
1222            if opts.deps.instead_of_compiling { write_out(&job.output, b"") } else { Ok(()) }
1223        }),
1224        None => write_out(&job.output, rule.as_bytes()),
1225    };
1226    if let Err(e) = wrote {
1227        let _ = writeln!(stderr, "rucc: error: {e}");
1228        return false;
1229    }
1230    true
1231}
1232
1233/// Runs phase 4 over every input that has one, and writes what came out.
1234///
1235/// One input that fails does not stop the others. A build that reports every file it could
1236/// not preprocess in one run is worth more than one that stops at the first, and the exit
1237/// status is still a failure either way.
1238fn preprocess_all(opts: &Options, plan: &Plan) -> i32 {
1239    let fs = OsFileSystem::new();
1240    let mut stderr = std::io::stderr().lock();
1241    let mut failed = false;
1242    for job in &plan.jobs {
1243        if !job.phases.first().is_some_and(|p| *p == Phase::Preprocess) {
1244            // An input that is already preprocessed, or an object file. GCC passes these
1245            // through untouched, and the plan has already said so in its notes.
1246            continue;
1247        }
1248        let started = std::time::Instant::now();
1249        let result = preprocess(opts, &job.input, &fs);
1250        if opts.time {
1251            say_time(&job.input, started.elapsed(), &mut stderr);
1252        }
1253        for message in &result.messages {
1254            let _ = writeln!(stderr, "{message}");
1255        }
1256        if result.failed() {
1257            failed = true;
1258            continue;
1259        }
1260        if opts.deps.emit {
1261            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1262            // `-M` and `-MM` asked for the rule instead of the text, so there is nothing else
1263            // to write. The other two asked for both and fall through to the text below.
1264            if opts.deps.instead_of_compiling {
1265                continue;
1266            }
1267        }
1268        if let Err(e) = write_out(&job.output, result.text.as_bytes()) {
1269            let _ = writeln!(stderr, "rucc: error: {e}");
1270            failed = true;
1271        }
1272    }
1273    i32::from(failed)
1274}
1275
1276/// Runs the front end over every input that has a compile phase, and writes what came out.
1277///
1278/// The same rule as [`preprocess_all`]: one input that fails does not stop the others, and the
1279/// exit status is a failure either way. An input that is already assembly or an object has no
1280/// compile phase and is passed over here, which the plan has already said in its notes.
1281fn compile_all(opts: &Options, plan: &Plan) -> i32 {
1282    let fs = OsFileSystem::new();
1283    let mut stderr = std::io::stderr().lock();
1284    let mut failed = false;
1285    let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1286    failed |= !ok;
1287    let mut fired = Fired::new();
1288    let mut pressure = Pressure::new();
1289    for job in &plan.jobs {
1290        if !job.phases.contains(&Phase::Compile) {
1291            continue;
1292        }
1293        // An input of IR is read back rather than compiled, since the C it came from is not
1294        // here any more. Everything after this is the same, so the two paths meet again at the
1295        // messages and the file the result is written to.
1296        let started = std::time::Instant::now();
1297        let result = if job.kind == InputKind::Ir {
1298            compile_ir(opts, &job.input, &fs)
1299        } else {
1300            compile(opts, &job.input, &fs)
1301        };
1302        if opts.time {
1303            say_time(&job.input, started.elapsed(), &mut stderr);
1304        }
1305        fired.merge(&result.fired);
1306        pressure.merge(&result.pressure);
1307        failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1308        failed |= !remarks.write(&result.remarks, &mut stderr);
1309        for message in &result.messages {
1310            let _ = writeln!(stderr, "{message}");
1311        }
1312        // Before the failure below, because a compilation that stopped in the back end is exactly
1313        // the one whose preprocessed source somebody wants to look at.
1314        failed |= !write_temps(job, &result.temps, &mut stderr);
1315        if result.failed() {
1316            failed = true;
1317            continue;
1318        }
1319        // `-MD` and `-MMD` write the rule beside the object and let the compilation happen, so
1320        // this is the one path where both files come out of the same run. An input of IR has no
1321        // dependencies to report and produces an empty list, which produces a rule naming only
1322        // itself, and that is the honest answer rather than a missing file.
1323        if opts.deps.emit {
1324            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1325        }
1326        if let Err(e) = write_out(&job.output, result.artifact.bytes()) {
1327            let _ = writeln!(stderr, "rucc: error: {e}");
1328            failed = true;
1329        }
1330    }
1331    failed |= !write_coverage(opts, &fired, &mut stderr);
1332    failed |= !write_pressure(opts, &pressure, &mut stderr);
1333    i32::from(failed)
1334}
1335
1336/// A directory for the object files only the link step ever sees, removed when it goes away.
1337///
1338/// `-c` writes its object where the user can see it and linking does not, which is the whole of
1339/// the difference: a `rucc a.c b.c` leaves an executable behind and nothing else, the same as
1340/// every other compiler. Removing them on drop rather than at the end of a function is so that a
1341/// link that failed leaves nothing behind either.
1342struct Scratch {
1343    /// Where the objects go.
1344    dir: PathBuf,
1345}
1346
1347impl Scratch {
1348    /// Makes one, under whatever the platform calls its temporary directory.
1349    ///
1350    /// The name carries the process id so that two compilers running at once do not share a
1351    /// directory, which they would otherwise do the moment two of them compiled a file of the
1352    /// same name.
1353    fn new() -> Result<Scratch, String> {
1354        let dir = std::env::temp_dir().join(format!("rucc-{}", std::process::id()));
1355        std::fs::create_dir_all(&dir).map_err(|e| format!("{}: {e}", dir.display()))?;
1356        Ok(Scratch { dir })
1357    }
1358}
1359
1360impl Drop for Scratch {
1361    fn drop(&mut self) {
1362        let _ = std::fs::remove_dir_all(&self.dir);
1363    }
1364}
1365
1366/// The link line the plan describes, for `-###`.
1367///
1368/// The names in it are the hints the plan carries rather than the temporaries a real compilation
1369/// would choose, because `-###` prints the line without having compiled anything and so has
1370/// nothing to point at. That also makes the printed line readable rather than naming a directory
1371/// that only exists while a compilation is running.
1372fn link_line(opts: &Options, link: &LinkOptions, job: &LinkJob) -> Result<String, link::Error> {
1373    let linker = link::find(opts.target, link)?;
1374    let args = link::line(opts.target, link, &job.inputs, &job.output)?;
1375    Ok(link::render(&linker, &args))
1376}
1377
1378/// Compiles everything, then links it.
1379///
1380/// The objects go in a directory that is removed afterwards, which is why this is not
1381/// [`compile_all`] followed by a link: the plan says an object feeding the linker is temporary
1382/// and does not say where, because where is a question that only has an answer once something is
1383/// running.
1384fn link_all(opts: &Options, plan: &Plan, link: &LinkOptions, verbose: bool) -> i32 {
1385    let Some(job) = &plan.link else {
1386        // Every path into here comes from a plan whose last phase is the link, and such a plan
1387        // has a link job. Saying so is cheaper than an unwrap that would have to be explained.
1388        let mut stderr = std::io::stderr().lock();
1389        let _ = writeln!(stderr, "rucc: error: there is nothing to link");
1390        return 1;
1391    };
1392    // Before anything is compiled, because a linker that is not on the machine is worth knowing
1393    // about in the second it takes to look rather than after the compilation.
1394    let linker = match link::find(opts.target, link) {
1395        Ok(linker) => linker,
1396        Err(why) => return complain(why),
1397    };
1398
1399    let scratch = match Scratch::new() {
1400        Ok(scratch) => scratch,
1401        Err(why) => return complain(format!("could not make a place for the object files: {why}")),
1402    };
1403
1404    let fs = OsFileSystem::new();
1405    let mut failed = false;
1406    // One per job, in job order, which is what lets the link line below be rebuilt with the real
1407    // paths in it: every job contributes exactly one file to the line and does so in this order.
1408    let mut produced: Vec<String> = Vec::with_capacity(plan.jobs.len());
1409    let mut fired = Fired::new();
1410    let mut pressure = Pressure::new();
1411    {
1412        let mut stderr = std::io::stderr().lock();
1413        let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1414        failed |= !ok;
1415        for (at, job) in plan.jobs.iter().enumerate() {
1416            let out = match &job.output {
1417                Output::Temporary(hint) => {
1418                    // The index because two inputs in different directories can have the same
1419                    // name, and the two objects of `rucc a/x.c b/x.c` must not be one file.
1420                    scratch.dir.join(format!("{at}-{hint}")).display().to_string()
1421                }
1422                Output::File(path) => path.clone(),
1423                // A job feeding the linker never writes to standard output, since the plan gives
1424                // it a temporary. This is here so that the match is total rather than a panic.
1425                Output::Stdout => continue,
1426            };
1427            produced.push(out.clone());
1428            if !job.phases.contains(&Phase::Compile) {
1429                continue;
1430            }
1431            let started = std::time::Instant::now();
1432            let result = if job.kind == InputKind::Ir {
1433                compile_ir(opts, &job.input, &fs)
1434            } else {
1435                compile(opts, &job.input, &fs)
1436            };
1437            if opts.time {
1438                say_time(&job.input, started.elapsed(), &mut stderr);
1439            }
1440            fired.merge(&result.fired);
1441            pressure.merge(&result.pressure);
1442            failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1443            failed |= !remarks.write(&result.remarks, &mut stderr);
1444            for message in &result.messages {
1445                let _ = writeln!(stderr, "{message}");
1446            }
1447            failed |= !write_temps(job, &result.temps, &mut stderr);
1448            if result.failed() {
1449                failed = true;
1450                continue;
1451            }
1452            // A `-MD` on a command line that links writes the rule next to the executable and
1453            // names the executable as its target, since that is the file this source builds
1454            // here. The object it went through is in a temporary directory and is gone by the
1455            // time `make` reads any of this.
1456            if opts.deps.emit {
1457                failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1458            }
1459            if !matches!(result.artifact, Artifact::Object(_)) {
1460                // Worth saying rather than writing whatever it is and letting the linker read it.
1461                // An empty file is a valid empty linker script, so a link handed one gets as far
1462                // as reporting every symbol of this file undefined, which is a page of messages
1463                // about something that went wrong here.
1464                let _ = writeln!(
1465                    stderr,
1466                    "rucc: internal error: {}: no object file was produced for the link",
1467                    job.input
1468                );
1469                failed = true;
1470                continue;
1471            }
1472            if let Err(e) = std::fs::write(&out, result.artifact.bytes()) {
1473                let _ = writeln!(stderr, "rucc: error: {out}: {e}");
1474                failed = true;
1475            }
1476        }
1477        failed |= !write_coverage(opts, &fired, &mut stderr);
1478        failed |= !write_pressure(opts, &pressure, &mut stderr);
1479    }
1480    if failed {
1481        // Nothing is linked from a compilation that did not finish. A linker run over the objects
1482        // that did compile would report every function of the file that did not as undefined,
1483        // which is a page of messages about a mistake already reported once.
1484        return 1;
1485    }
1486
1487    // The items in command line order with the temporaries filled in. A library contributes no
1488    // job and passes through, and every file item takes the next job's real output, which is
1489    // what keeps a library that was written between two objects between them here.
1490    let mut outputs = produced.into_iter();
1491    let mut items = Vec::with_capacity(job.inputs.len());
1492    for item in &job.inputs {
1493        match item {
1494            link::Item::Library(name) => items.push(link::Item::Library(name.clone())),
1495            link::Item::File(_) => match outputs.next() {
1496                Some(path) => items.push(link::Item::File(path)),
1497                None => return complain("the plan asks the linker for a file nothing produced"),
1498            },
1499        }
1500    }
1501
1502    let args = match link::line(opts.target, link, &items, &job.output) {
1503        Ok(args) => args,
1504        Err(why) => return complain(why),
1505    };
1506    if verbose {
1507        let mut stderr = std::io::stderr().lock();
1508        let _ = writeln!(stderr, "{}", link::render(&linker, &args));
1509    }
1510    let started = std::time::Instant::now();
1511    let ran = link::run(&linker, &args);
1512    if opts.time {
1513        // The one step of a compilation that really is another program, so this line is the same
1514        // measurement gcc's is and names the linker the way gcc names `collect2`.
1515        let mut stderr = std::io::stderr().lock();
1516        say_time(&linker.name, started.elapsed(), &mut stderr);
1517    }
1518    match ran {
1519        Ok(()) => 0,
1520        // The linker has already said what was wrong on its own error output, and repeating that
1521        // linking failed would only push its message further up the screen.
1522        Err(link::Error::Refused { .. }) => 1,
1523        Err(why) => complain(why),
1524    }
1525}
1526
1527/// Prints one driver level message and gives back the exit status that goes with it.
1528fn complain(why: impl std::fmt::Display) -> i32 {
1529    let mut stderr = std::io::stderr().lock();
1530    let _ = writeln!(stderr, "rucc: error: {why}");
1531    1
1532}
1533
1534/// Writes what `-Zrule-coverage=FILE` asked for, and says whether it could.
1535///
1536/// Once for the whole command line rather than once per input, because the question is which
1537/// lowering rules this run of the compiler reached and a file per input would leave the reader
1538/// unioning files to find out something one process already knew.
1539///
1540/// A file that could not be written is a failure and not a warning. What asks for this is a
1541/// measurement run, and a measurement that quietly did not happen is worse than one that stopped.
1542fn write_coverage(opts: &Options, fired: &Fired, stderr: &mut impl std::io::Write) -> bool {
1543    let Some(path) = &opts.rule_coverage else { return true };
1544    let Some(table) = coverage::table(opts.target.arch) else {
1545        let _ = writeln!(
1546            stderr,
1547            "rucc: error: there are no lowering rules for {} yet, so there is no coverage of them \
1548             to report",
1549            opts.target
1550        );
1551        return false;
1552    };
1553    match std::fs::write(path, fired.listing(table)) {
1554        Ok(()) => true,
1555        Err(e) => {
1556            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1557            false
1558        }
1559    }
1560}
1561
1562/// Writes what `-Zregister-pressure=FILE` asked for, and says whether it could.
1563///
1564/// Once for the whole command line, for the reason [`write_coverage`] gives, and a file that could
1565/// not be written is a failure for the reason it gives too. There is no equivalent of the missing
1566/// rule table here, since every target this compiles for has an allocator, and a run that reached
1567/// no back end at all writes an empty listing rather than nothing: a measurement of a build that
1568/// produced no code is still an answer and it is the honest one.
1569fn write_pressure(opts: &Options, pressure: &Pressure, stderr: &mut impl std::io::Write) -> bool {
1570    let Some(path) = &opts.register_pressure else { return true };
1571    match std::fs::write(path, pressure.listing()) {
1572        Ok(()) => true,
1573        Err(e) => {
1574            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1575            false
1576        }
1577    }
1578}
1579
1580/// Where the `-fopt-info` remarks go, and how much of the run has already gone there.
1581///
1582/// Standard error by default, and one file for the whole run when `-fopt-info=<file>` named one.
1583/// A file rather than the diagnostic stream is what a harness wants: the corpus in
1584/// `tamnd/rucc-corpus` matches a rejection against what the compiler said on standard error, and
1585/// a few thousand remarks mixed into that would bury it.
1586struct Remarks {
1587    /// The file, if there is one.
1588    file: Option<String>,
1589    /// Whether anything has been written to it yet, which decides between truncating and
1590    /// appending. One file holds the whole run rather than the last input in it.
1591    started: bool,
1592}
1593
1594impl Remarks {
1595    /// Prepares the destination, emptying the file if there is one.
1596    ///
1597    /// Emptied here rather than at the first remark, because a run where no pass had anything to
1598    /// say should leave an empty file and not yesterday's. An absent file and an empty one are
1599    /// different facts and something reading this will act on the difference.
1600    fn new(file: Option<&String>, stderr: &mut impl std::io::Write) -> (Self, bool) {
1601        let mut ok = true;
1602        if let Some(path) = file {
1603            if let Err(e) = std::fs::write(path, "") {
1604                let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1605                ok = false;
1606            }
1607        }
1608        (Self { file: file.cloned(), started: false }, ok)
1609    }
1610
1611    /// Writes one input's remarks, and says whether that worked.
1612    ///
1613    /// A file that cannot be written is a failure and not a warning, for the reason
1614    /// [`write_dumps`] gives: remarks that quietly did not arrive look exactly like a compilation
1615    /// where nothing happened.
1616    fn write(&mut self, text: &str, stderr: &mut impl std::io::Write) -> bool {
1617        if text.is_empty() {
1618            return true;
1619        }
1620        let Some(path) = &self.file else {
1621            let _ = write!(stderr, "{text}");
1622            return true;
1623        };
1624        let opened = std::fs::OpenOptions::new()
1625            .write(true)
1626            .append(self.started)
1627            .truncate(!self.started)
1628            .create(true)
1629            .open(path);
1630        self.started = true;
1631        let result =
1632            opened.and_then(|mut file| std::io::Write::write_all(&mut file, text.as_bytes()));
1633        if let Err(e) = result {
1634            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1635            return false;
1636        }
1637        true
1638    }
1639}
1640
1641/// Writes what `-fdump-ir=` asked to see, one file per dump.
1642///
1643/// The name is the input file with the dump's own name and `.ir` after it, so a directory listing
1644/// after a run is the passes in the order they ran, per input. They go in the working directory
1645/// rather than beside the output, because a dump is something a person asked for at a prompt and
1646/// the working directory is where that person is.
1647///
1648/// A file that could not be written is a failure and not a warning, for the reason
1649/// [`write_coverage`] gives: what asked for this is somebody debugging a pass, and a dump that
1650/// quietly did not happen looks exactly like a pass that did not run.
1651fn write_dumps(input: &str, dumps: &[rucc_opt::Dump], stderr: &mut impl std::io::Write) -> bool {
1652    let stem = std::path::Path::new(input)
1653        .file_name()
1654        .map_or_else(|| input.to_owned(), |name| name.to_string_lossy().into_owned());
1655    let mut ok = true;
1656    for dump in dumps {
1657        let path = format!("{stem}.{}.ir", dump.name);
1658        if let Err(e) = std::fs::write(&path, &dump.text) {
1659            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1660            ok = false;
1661        }
1662    }
1663    ok
1664}
1665
1666/// Writes the files `-save-temps` kept, which is nothing at all unless it was given.
1667///
1668/// A file that could not be written is a failure rather than a warning, for the reason
1669/// [`write_dumps`] gives: somebody asked for these by name, and one that quietly did not happen
1670/// looks like a compilation that never went through that step.
1671fn write_temps(job: &Job, temps: &Temps, stderr: &mut impl std::io::Write) -> bool {
1672    let mut ok = true;
1673    let kept = [(job.saved_text(), &temps.preprocessed), (job.saved_asm(), &temps.assembly)];
1674    for (path, text) in kept {
1675        // A step the compilation did not reach has nothing to keep, and a job that is not keeping
1676        // that step has nowhere to put it. Either way there is no file here.
1677        let (Some(path), Some(text)) = (path, text) else { continue };
1678        if let Err(e) = std::fs::write(&path, text) {
1679            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1680            ok = false;
1681        }
1682    }
1683    ok
1684}
1685
1686/// One line of `-time`, which is what a step was called and how long it took.
1687///
1688/// GCC's two numbers are the user and the system time of a subprocess it ran. This compiler runs
1689/// no subprocess for anything but the link, so what is measured here is the wall clock of the
1690/// step and the second column is always zero. The shape of the line is kept because a person
1691/// reading it next to gcc's should not have to work out which column is which.
1692fn say_time(name: &str, took: std::time::Duration, stderr: &mut impl std::io::Write) {
1693    let _ = writeln!(stderr, "# {name} {:.2} {:.2}", took.as_secs_f64(), 0.0);
1694}
1695
1696/// Writes one job's result where the plan said it goes.
1697///
1698/// # Errors
1699///
1700/// Returns the message to print, which names the file when there is one, because "permission
1701/// denied" on its own does not say which file was refused.
1702fn write_out(output: &Output, bytes: &[u8]) -> Result<(), String> {
1703    match output {
1704        Output::Stdout => {
1705            let mut stdout = std::io::stdout().lock();
1706            stdout.write_all(bytes).map_err(|e| format!("writing to standard output: {e}"))
1707        }
1708        Output::File(path) | Output::Temporary(path) => {
1709            std::fs::write(path, bytes).map_err(|e| format!("{path}: {e}"))
1710        }
1711    }
1712}
1713
1714/// Runs the driver and returns the process exit code.
1715///
1716/// `args` excludes the program name. Output goes to `stdout` and errors to `stderr`, which
1717/// is the one place in the compiler that is true.
1718pub fn run(args: &[String]) -> i32 {
1719    match parse_args(args) {
1720        Ok(Action::Help) => {
1721            print!("{USAGE}");
1722            0
1723        }
1724        Ok(Action::Version) => {
1725            println!("rucc {VERSION}");
1726            0
1727        }
1728        Ok(Action::Print(line)) => {
1729            println!("{line}");
1730            0
1731        }
1732        Ok(Action::PrintConfig(opts)) => {
1733            print!("{}", print_config(&opts));
1734            0
1735        }
1736        Ok(Action::PrintPipeline(opts)) => {
1737            print!("{}", print_pipeline(&opts));
1738            0
1739        }
1740        Ok(Action::PrintPlan { opts, plan, link }) => {
1741            print!("{}", plan.render());
1742            // The line as it would be typed, which is the half of `-###` that section 4.3 says
1743            // arrives with the link. It is printed even when the linker is not on this machine,
1744            // because what a build wants from `-###` is what the compiler would do.
1745            if let Some(job) = &plan.link {
1746                match link_line(&opts, &link, job) {
1747                    Ok(line) => println!("{line}"),
1748                    Err(why) => {
1749                        let mut stderr = std::io::stderr().lock();
1750                        let _ = writeln!(stderr, "rucc: error: {why}");
1751                        return 1;
1752                    }
1753                }
1754            }
1755            0
1756        }
1757        Ok(Action::Compile { opts, plan, link, jobs, verbose }) => {
1758            {
1759                let mut stderr = std::io::stderr().lock();
1760                if verbose {
1761                    let _ = write!(stderr, "{}", plan.render());
1762                    let _ = writeln!(stderr, "workers: {}", jobs.count());
1763                }
1764            }
1765            if opts.emit == EmitKind::Preprocessed {
1766                return preprocess_all(&opts, &plan);
1767            }
1768            if opts.emit != EmitKind::Executable {
1769                return compile_all(&opts, &plan);
1770            }
1771            link_all(&opts, &plan, &link, verbose)
1772        }
1773        Err(e) => {
1774            let mut stderr = std::io::stderr().lock();
1775            let _ = writeln!(stderr, "rucc: error: {e}");
1776            let _ = writeln!(stderr, "rucc: note: run `rucc --help` for usage");
1777            1
1778        }
1779    }
1780}
1781
1782#[cfg(test)]
1783mod tests {
1784    use rucc_session::{GnucVersion, IncludeForm, OptLevel, Patchable, Visibility};
1785
1786    use super::*;
1787
1788    fn args(s: &[&str]) -> Vec<String> {
1789        s.iter().map(|x| (*x).to_owned()).collect()
1790    }
1791
1792    #[test]
1793    fn help_and_version_win_over_everything_else() {
1794        assert_eq!(parse_args(&args(&["-c", "--help", "x.c"])).unwrap(), Action::Help);
1795        assert_eq!(parse_args(&args(&["--version"])).unwrap(), Action::Version);
1796    }
1797
1798    fn compile(s: &[&str]) -> (Box<Options>, Box<Plan>) {
1799        match parse_args(&args(s)).expect("expected a compilation") {
1800            Action::Compile { opts, plan, .. } => (opts, plan),
1801            other => panic!("expected a compilation, got {other:?}"),
1802        }
1803    }
1804
1805    fn linking(s: &[&str]) -> (Box<LinkOptions>, Box<Plan>) {
1806        match parse_args(&args(s)).expect("expected a compilation") {
1807            Action::Compile { link, plan, .. } => (link, plan),
1808            other => panic!("expected a compilation, got {other:?}"),
1809        }
1810    }
1811
1812    #[test]
1813    fn collects_inputs_and_flags() {
1814        let (opts, plan) = compile(&["-c", "-O2", "-g", "a.c", "b.c"]);
1815        let paths: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
1816        assert_eq!(paths, vec!["a.c", "b.c"]);
1817        assert_eq!(opts.opt_level, OptLevel::O2);
1818        assert_eq!(opts.emit, EmitKind::Object);
1819        assert!(opts.debug_info);
1820    }
1821
1822    /// The unstable options, which are spelled apart from everything else on purpose: what is
1823    /// under `-Z` promises nothing, and a build that reaches for one should have had to say so.
1824    #[test]
1825    fn an_unstable_option_is_taken_and_one_that_does_not_exist_is_refused() {
1826        let (opts, _) = compile(&["-c", "-Zrule-coverage=/tmp/rules.cov", "a.c"]);
1827        assert_eq!(opts.rule_coverage.as_deref(), Some("/tmp/rules.cov"));
1828
1829        let (plain, _) = compile(&["-c", "a.c"]);
1830        assert_eq!(plain.rule_coverage, None, "nothing is measured unless it was asked for");
1831
1832        assert!(parse_args(&args(&["-Zrule-coverage=", "a.c"])).is_err(), "a file with no name");
1833        let unknown = parse_args(&args(&["-Zwhat", "a.c"])).expect_err("there is no such option");
1834        assert!(unknown.message.contains("4.11"), "{}", unknown.message);
1835    }
1836
1837    /// The other measurement written to a file, which reads the same way and fails the same way.
1838    #[test]
1839    fn where_the_register_pressure_goes_is_asked_for_the_same_way() {
1840        let (opts, _) = compile(&["-c", "-O2", "-Zregister-pressure=/tmp/spills.txt", "a.c"]);
1841        assert_eq!(opts.register_pressure.as_deref(), Some("/tmp/spills.txt"));
1842
1843        let (plain, _) = compile(&["-c", "a.c"]);
1844        assert_eq!(plain.register_pressure, None, "nothing is measured unless it was asked for");
1845
1846        assert!(parse_args(&args(&["-Zregister-pressure=", "a.c"])).is_err(), "no file named");
1847    }
1848
1849    #[test]
1850    fn a_bare_dash_o_means_o1_the_way_gcc_reads_it() {
1851        let (opts, _) = compile(&["-O", "a.c"]);
1852        assert_eq!(opts.opt_level, OptLevel::O1);
1853    }
1854
1855    #[test]
1856    fn dash_x_applies_to_later_inputs_only_and_none_stops_it() {
1857        let (_, plan) = compile(&["a.o", "-x", "c", "b.txt", "-x", "none", "c.o"]);
1858        assert_eq!(plan.jobs[0].kind, InputKind::LinkerInput);
1859        assert_eq!(plan.jobs[1].kind, InputKind::C);
1860        assert_eq!(plan.jobs[2].kind, InputKind::LinkerInput);
1861    }
1862
1863    #[test]
1864    fn dash_j_reaches_the_scheduler_and_defaults_to_the_machine() {
1865        let (_, _, jobs) = match parse_args(&args(&["-j4", "a.c"])).unwrap() {
1866            Action::Compile { opts, plan, jobs, .. } => (opts, plan, jobs),
1867            other => panic!("expected a compilation, got {other:?}"),
1868        };
1869        assert_eq!(jobs.count(), 4);
1870
1871        let default = match parse_args(&args(&["a.c"])).unwrap() {
1872            Action::Compile { jobs, .. } => jobs,
1873            other => panic!("expected a compilation, got {other:?}"),
1874        };
1875        assert_eq!(default, Jobs::available());
1876        assert!(parse_args(&args(&["-j0", "a.c"])).is_err());
1877    }
1878
1879    #[test]
1880    fn triple_hash_prints_the_plan_and_runs_nothing() {
1881        let a = parse_args(&args(&["-###", "-c", "a.c"])).unwrap();
1882        let Action::PrintPlan { plan, .. } = a else { panic!("expected a plan dump") };
1883        assert!(plan.render().contains("a.c: preprocess, compile, assemble -> a.o"));
1884    }
1885
1886    #[test]
1887    fn the_flag_that_keeps_the_intermediate_files_has_three_spellings_and_two_meanings() {
1888        // The bare one is `=obj` and not `=cwd`. gcc's manual says the opposite and gcc 16 does
1889        // this, and following the compiler is what makes a build that reads either of them find
1890        // the files where they are.
1891        assert_eq!(compile(&["-c", "-save-temps", "a.c"]).0.save_temps, SaveTemps::Object);
1892        assert_eq!(compile(&["-c", "-save-temps=obj", "a.c"]).0.save_temps, SaveTemps::Object);
1893        assert_eq!(compile(&["-c", "-save-temps=cwd", "a.c"]).0.save_temps, SaveTemps::Cwd);
1894        assert_eq!(compile(&["-c", "a.c"]).0.save_temps, SaveTemps::No);
1895        // The last one on the line decides, the way it does for every other flag with an
1896        // argument, and a keyword that is neither is fatal rather than ignored: a run that kept
1897        // nothing and said nothing looks exactly like one where the files were not produced.
1898        let (opts, _) = compile(&["-c", "-save-temps", "-save-temps=cwd", "a.c"]);
1899        assert_eq!(opts.save_temps, SaveTemps::Cwd);
1900        let e = parse_args(&args(&["-c", "-save-temps=nowhere", "a.c"])).unwrap_err();
1901        assert!(e.message.contains("accepted: cwd, obj"), "{}", e.message);
1902    }
1903
1904    #[test]
1905    fn the_flag_that_times_each_step_reaches_the_options_and_changes_nothing_else() {
1906        let (opts, plan) = compile(&["-c", "-time", "a.c"]);
1907        let (plain, without) = compile(&["-c", "a.c"]);
1908        assert!(opts.time);
1909        assert!(!plain.time);
1910        // Against the same line without the flag rather than against a spelling of the object's
1911        // name, since what the object is called is the host's business and this is not about that.
1912        assert_eq!(plan.jobs[0].output, without.jobs[0].output);
1913    }
1914
1915    #[test]
1916    fn dash_x_names_what_it_accepts_when_it_does_not_know_a_language() {
1917        let e = parse_args(&args(&["-x", "fortran", "a.c"])).unwrap_err();
1918        assert!(e.message.contains("assembler-with-cpp"), "{}", e.message);
1919    }
1920
1921    #[test]
1922    fn an_unknown_flag_is_an_error_rather_than_a_shrug() {
1923        let e = parse_args(&args(&["-fno-such-thing", "a.c"])).unwrap_err();
1924        assert!(e.message.contains("unknown option"), "{}", e.message);
1925    }
1926
1927    /// `-fpermissive` and the flag that turns it back off, which a build writes beside it when
1928    /// one directory needs the older rules and the rest of the tree does not.
1929    #[test]
1930    fn permissive_reads_in_both_directions_and_the_last_one_wins() {
1931        let (opts, _) = compile(&["-c", "a.c"]);
1932        assert!(!opts.permissive, "off unless it is asked for");
1933
1934        let (opts, _) = compile(&["-c", "-fpermissive", "a.c"]);
1935        assert!(opts.permissive);
1936
1937        let (opts, _) = compile(&["-c", "-fpermissive", "-fno-permissive", "a.c"]);
1938        assert!(!opts.permissive);
1939    }
1940
1941    #[test]
1942    fn asking_for_nested_functions_is_told_why_it_is_not_coming() {
1943        let e = parse_args(&args(&["-fnested-functions", "a.c"])).unwrap_err();
1944        assert!(e.message.contains("trampoline"), "{}", e.message);
1945        assert!(parse_args(&args(&["-fno-nested-functions", "a.c"])).is_ok());
1946    }
1947
1948    #[test]
1949    fn the_flag_every_configure_script_writes_is_taken() {
1950        // All four spellings, because a build writes whichever one its macros picked and a
1951        // compiler that takes three of them is a compiler that fails on the fourth.
1952        for flag in ["-fPIC", "-fpic", "-fPIE", "-fpie"] {
1953            let (opts, _) = compile(&["-c", flag, "a.c"]);
1954            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1955        }
1956    }
1957
1958    #[test]
1959    fn a_table_is_written_unless_the_build_says_nothing_will_walk_it() {
1960        let (opts, _) = compile(&["-c", "a.c"]);
1961        assert!(opts.unwinds(), "the default is off");
1962        let (opts, _) = compile(&["-c", "-fno-asynchronous-unwind-tables", "a.c"]);
1963        assert!(!opts.unwinds(), "the build was not taken at its word");
1964        let (opts, _) = compile(&[
1965            "-c",
1966            "-fno-asynchronous-unwind-tables",
1967            "-fasynchronous-unwind-tables",
1968            "a.c",
1969        ]);
1970        assert!(opts.unwinds(), "the last flag did not win");
1971        // The weaker request, which the same table answers, so a line that asks for a table and
1972        // against an asynchronous one gets one. That is gcc's arrangement and it turns up when a
1973        // build turns the asynchronous one off globally and a directory asks for a table back.
1974        let (opts, _) =
1975            compile(&["-c", "-fno-asynchronous-unwind-tables", "-funwind-tables", "a.c"]);
1976        assert!(opts.unwinds(), "the weaker request was dropped");
1977        let (opts, _) = compile(&["-c", "-fno-unwind-tables", "a.c"]);
1978        assert!(opts.unwinds(), "the weaker negative turned off the stronger request");
1979        let (opts, _) =
1980            compile(&["-c", "-fno-unwind-tables", "-fno-asynchronous-unwind-tables", "a.c"]);
1981        assert!(!opts.unwinds(), "both were turned off and one stayed on");
1982    }
1983
1984    #[test]
1985    fn the_flags_that_describe_what_this_compiler_already_does_are_taken() {
1986        // Every one of these is on a real build line somewhere and every one of them was an
1987        // unknown option. What they have in common is that the answer rucc gives is the answer
1988        // they ask for, so there is nothing to implement and nothing to refuse.
1989        for flag in [
1990            "-fno-common",
1991            "-fstrict-aliasing",
1992            "-fno-strict-aliasing",
1993            "-pipe",
1994            "-fdiagnostics-color",
1995            "-fno-diagnostics-color",
1996            "-fdiagnostics-color=always",
1997            "-fdiagnostics-color=never",
1998            "-fdiagnostics-color=auto",
1999        ] {
2000            let (opts, _) = compile(&["-c", flag, "a.c"]);
2001            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
2002        }
2003    }
2004
2005    #[test]
2006    fn asking_the_linker_to_merge_tentative_definitions_is_told_why_it_is_not_coming() {
2007        // The one of that family that is a request rather than a description, and it is a real
2008        // difference: two files each writing `int g;` link under it and do not without it.
2009        let e = parse_args(&args(&["-fcommon", "a.c"])).unwrap_err();
2010        assert!(e.message.contains(".bss"), "{}", e.message);
2011        assert!(e.message.contains("extern"), "the way out is worth saying: {}", e.message);
2012    }
2013
2014    #[test]
2015    fn asking_for_position_dependent_code_is_told_why_it_is_not_coming() {
2016        for flag in ["-fno-pic", "-fno-pie"] {
2017            let e = parse_args(&args(&[flag, "a.c"])).unwrap_err();
2018            assert!(e.message.contains("global offset table"), "{flag}: {}", e.message);
2019            // The one it may have meant, since the two are a letter apart and one of them is
2020            // about linking and is taken.
2021            assert!(e.message.contains("-no-pie"), "{flag}: {}", e.message);
2022        }
2023    }
2024
2025    #[test]
2026    fn an_unsupported_target_names_itself() {
2027        let e = parse_args(&args(&["--target=sparc64-linux-gnu", "a.c"])).unwrap_err();
2028        assert!(e.message.contains("sparc64"), "{}", e.message);
2029    }
2030
2031    #[test]
2032    fn no_inputs_is_an_error_but_print_config_needs_none() {
2033        assert!(parse_args(&args(&[])).is_err());
2034        assert!(matches!(parse_args(&args(&["--print-config"])), Ok(Action::PrintConfig(_))));
2035    }
2036
2037    #[test]
2038    fn print_config_reports_the_target_it_was_given_not_the_host() {
2039        let a = parse_args(&args(&["--print-config", "--target=riscv64-linux-musl"])).unwrap();
2040        let Action::PrintConfig(opts) = a else { panic!("expected a configuration dump") };
2041        let text = print_config(&opts);
2042        assert!(text.contains("target: riscv64-unknown-linux-musl"), "{text}");
2043        assert!(text.contains("char-signed: false"), "{text}");
2044        assert!(text.contains("object-format: elf"), "{text}");
2045        assert!(text.contains("va-list: void-pointer"), "{text}");
2046        // RISC-V has a register file and this compiler has not written it down yet, and the
2047        // dump says which of those two it is rather than leaving the line out.
2048        assert!(text.contains("registers: none"), "{text}");
2049    }
2050
2051    #[test]
2052    fn print_config_has_one_key_per_line_and_a_fixed_order() {
2053        let opts = Options::new("x86_64-unknown-linux-gnu".parse().unwrap());
2054        let text = print_config(&opts);
2055        let keys: Vec<&str> =
2056            text.lines().map(|l| l.split(':').next().unwrap_or_default()).collect();
2057        assert_eq!(keys[0], "version");
2058        assert_eq!(keys[1], "target");
2059        assert_eq!(keys.len(), 25);
2060        assert!(text.ends_with('\n'));
2061    }
2062
2063    #[test]
2064    fn the_safety_tier_is_read_off_the_command_line_and_a_wrong_one_is_refused() {
2065        let (opts, _) = compile(&["a.c"]);
2066        assert_eq!(opts.safety, rucc_session::Safety::Off);
2067
2068        for (flag, tier) in [
2069            ("-fsafety=detect", rucc_session::Safety::Detect),
2070            ("-fsafety=enforce", rucc_session::Safety::Enforce),
2071            ("-fsafety=kernel", rucc_session::Safety::Kernel),
2072            ("-fsafety=off", rucc_session::Safety::Off),
2073        ] {
2074            let (opts, _) = compile(&[flag, "a.c"]);
2075            assert_eq!(opts.safety, tier, "{flag}");
2076        }
2077
2078        // The last one wins, the way every other repeated flag on this command line does.
2079        let (opts, _) = compile(&["-fsafety=enforce", "-fsafety=off", "a.c"]);
2080        assert_eq!(opts.safety, rucc_session::Safety::Off);
2081
2082        // A misspelled tier is refused rather than ignored. Silently compiling without the
2083        // monitor a build asked for is the one failure mode this feature cannot have.
2084        let e = parse_args(&args(&["-fsafety=on", "a.c"])).unwrap_err();
2085        assert!(e.message.contains("is not a safety tier"), "{}", e.message);
2086        assert!(parse_args(&args(&["-fsafety", "a.c"])).is_err());
2087    }
2088
2089    #[test]
2090    fn print_pipeline_answers_with_the_passes_the_level_asked_for() {
2091        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2092        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2093        let text = print_pipeline(&opts);
2094        assert!(text.starts_with("level: -O2\n"), "{text}");
2095        assert!(text.contains("fold"), "{text}");
2096
2097        let a = parse_args(&args(&["--print-pipeline"])).unwrap();
2098        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2099        // One pass runs at `-O0` and it is the one that removes code nothing reaches, which is
2100        // not an optimization. See issue 359.
2101        assert!(print_pipeline(&opts).contains("1: simplify-cfg,"), "{}", print_pipeline(&opts));
2102
2103        let a = parse_args(&args(&["--print-pipeline", "-fno-simplify-cfg"])).unwrap();
2104        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2105        // And with that one turned off there is nothing left, which the dump says rather than
2106        // printing an empty list.
2107        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2108    }
2109
2110    #[test]
2111    fn print_pipeline_takes_the_toggles_into_account() {
2112        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fno-fold"])).unwrap();
2113        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2114        let text = print_pipeline(&opts);
2115        // The one that was named is gone and the rest of the level is not, which is the whole
2116        // of what a toggle promises.
2117        assert!(!text.contains("fold"), "{text}");
2118        assert!(text.contains("dce"), "{text}");
2119
2120        // Every pass the compiler has, named off. Built from the registry rather than written
2121        // out, so a pass added later is turned off here too and this keeps testing the thing it
2122        // is about, which is that the toggles can empty a level.
2123        let mut off = vec!["--print-pipeline".to_owned(), "-O2".to_owned()];
2124        off.extend(rucc_opt::PASSES.iter().map(|p| format!("-fno-{}", p.name())));
2125        let spelled: Vec<&str> = off.iter().map(String::as_str).collect();
2126        let a = parse_args(&args(&spelled)).unwrap();
2127        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2128        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2129    }
2130
2131    #[test]
2132    fn print_pipeline_says_when_a_budget_will_stop_the_run_short() {
2133        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2134        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2135        assert!(!print_pipeline(&opts).contains("global fuel"));
2136
2137        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fpass-fuel-global=4"])).unwrap();
2138        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2139        let text = print_pipeline(&opts);
2140        // Because the listing is the answer to what this compilation will do, and a run that
2141        // stops after four rewrites is not doing what the level says it does.
2142        assert!(text.contains("global fuel: 4"), "{text}");
2143    }
2144
2145    /// A pass is turned on and off by its own name, and the order the flags were given in is
2146    /// kept, because the last spelling of a name is the one that decides.
2147    #[test]
2148    fn a_pass_is_named_by_dash_f_and_unnamed_by_dash_f_no() {
2149        let (opts, _) = compile(&["-c", "-O0", "-ffold", "-fno-fold", "-ffold", "a.c"]);
2150        assert_eq!(
2151            opts.passes,
2152            [("fold".to_owned(), true), ("fold".to_owned(), false), ("fold".to_owned(), true)]
2153        );
2154
2155        let e = parse_args(&args(&["-fno-such-pass", "a.c"])).unwrap_err();
2156        assert!(e.message.contains("unknown option"), "{}", e.message);
2157    }
2158
2159    #[test]
2160    fn pass_fuel_names_a_pass_and_a_count_and_refuses_anything_else() {
2161        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel=fold=3", "a.c"]);
2162        assert_eq!(opts.pass_fuel, [("fold".to_owned(), 3)]);
2163
2164        let e = parse_args(&args(&["-fpass-fuel=fold", "a.c"])).unwrap_err();
2165        assert!(e.message.contains("<pass>=<count>"), "{}", e.message);
2166        let e = parse_args(&args(&["-fpass-fuel=nosuch=3", "a.c"])).unwrap_err();
2167        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2168        let e = parse_args(&args(&["-fpass-fuel=fold=lots", "a.c"])).unwrap_err();
2169        assert!(e.message.contains("not a number"), "{}", e.message);
2170    }
2171
2172    #[test]
2173    fn global_pass_fuel_is_a_count_on_its_own_and_defaults_to_no_limit() {
2174        let (opts, _) = compile(&["-c", "-O2", "a.c"]);
2175        assert_eq!(opts.pass_fuel_global, None);
2176
2177        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel-global=12", "a.c"]);
2178        assert_eq!(opts.pass_fuel_global, Some(12));
2179        // And it is not the per pass flag with a longer name, so neither spelling swallows the
2180        // other.
2181        assert!(opts.pass_fuel.is_empty());
2182
2183        let e = parse_args(&args(&["-fpass-fuel-global=lots", "a.c"])).unwrap_err();
2184        assert!(e.message.contains("not a number"), "{}", e.message);
2185    }
2186
2187    #[test]
2188    fn a_gate_names_a_pass_and_optionally_the_functions_it_covers() {
2189        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold", "-fenable-fold=2-4,main", "a.c"]);
2190        assert_eq!(
2191            opts.pass_gates,
2192            [(false, "fold".to_owned()), (true, "fold=2-4,main".to_owned())],
2193            "the order is what decides, so it has to survive the parse"
2194        );
2195
2196        let e = parse_args(&args(&["-fdisable-nosuch", "a.c"])).unwrap_err();
2197        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2198        let e = parse_args(&args(&["-fenable-fold=9-2", "a.c"])).unwrap_err();
2199        assert!(e.message.contains("ends before it starts"), "{}", e.message);
2200        let e = parse_args(&args(&["-fdisable-fold=", "a.c"])).unwrap_err();
2201        assert!(e.message.contains("is empty"), "{}", e.message);
2202    }
2203
2204    #[test]
2205    fn the_pipeline_listing_says_which_passes_a_gate_touched() {
2206        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold=main", "a.c"]);
2207        let text = print_pipeline(&opts);
2208        assert!(text.contains("fold, "), "{text}");
2209        assert!(text.contains("[off for main]"), "{text}");
2210    }
2211
2212    /// The spelling is checked while the arguments are read, because a dump that names a pass
2213    /// this compiler does not have is a typo, and a typo found after the compilation has run is
2214    /// found too late to be any use.
2215    #[test]
2216    fn a_dump_is_checked_when_it_is_asked_for_rather_than_when_it_is_taken() {
2217        let (opts, _) = compile(&["-c", "-O2", "-fdump-ir=all", "-fdump-ir=after-fold", "a.c"]);
2218        assert_eq!(opts.dump_ir, ["all", "after-fold"]);
2219
2220        let e = parse_args(&args(&["-fdump-ir=after-nosuch", "a.c"])).unwrap_err();
2221        assert!(e.message.contains("nosuch"), "{}", e.message);
2222        assert!(parse_args(&args(&["-fdump-ir=sideways-fold", "a.c"])).is_err());
2223    }
2224
2225    /// Every spelling `-fopt-info` takes, and the one it does not.
2226    ///
2227    /// The keywords are checked here for the same reason a dump's pass name is: a person who
2228    /// misspelled one gets no output, and no output is also what a compilation where nothing
2229    /// happened looks like. Telling those two apart is the entire reason to reach for this flag.
2230    #[test]
2231    fn opt_info_takes_kinds_and_a_file_and_refuses_a_kind_it_does_not_have() {
2232        let (opts, _) = compile(&["-c", "-O2", "-fopt-info", "a.c"]);
2233        assert_eq!(opts.opt_info, [""], "a bare flag asks for the rewrites");
2234        assert_eq!(opts.opt_info_file, None, "and goes to standard error");
2235
2236        let (opts, _) = compile(&["-c", "-O2", "-fopt-info-missed-note", "a.c"]);
2237        assert_eq!(opts.opt_info, ["missed-note"]);
2238
2239        // Two flags add up rather than the second replacing the first, and the file is the last
2240        // one that named a file, which is how GCC treats both.
2241        let (opts, _) =
2242            compile(&["-c", "-O2", "-fopt-info-missed=one.txt", "-fopt-info-all=two.txt", "a.c"]);
2243        assert_eq!(opts.opt_info, ["missed", "all"]);
2244        assert_eq!(opts.opt_info_file.as_deref(), Some("two.txt"));
2245
2246        let e = parse_args(&args(&["-fopt-info-vectorized", "a.c"])).unwrap_err();
2247        assert!(e.message.contains("vectorized"), "{}", e.message);
2248        assert!(e.message.contains("`missed`"), "{}", e.message);
2249        let e = parse_args(&args(&["-fopt-info-missed=", "a.c"])).unwrap_err();
2250        assert!(e.message.contains("no file"), "{}", e.message);
2251    }
2252
2253    #[test]
2254    fn verify_each_is_unstable_and_off_unless_it_was_asked_for() {
2255        let (opts, _) = compile(&["-c", "-Zverify-each", "a.c"]);
2256        assert!(opts.verify_each);
2257        assert!(!USAGE.contains("verify-each"), "an unstable option stays out of the usage text");
2258    }
2259
2260    #[test]
2261    fn dash_o_needs_an_argument() {
2262        let e = parse_args(&args(&["a.c", "-o"])).unwrap_err();
2263        assert_eq!(e.message, "-o requires an argument");
2264    }
2265
2266    #[test]
2267    fn dash_d_and_dash_u_are_read_joined_or_separated_and_keep_their_order() {
2268        let (opts, _) = compile(&["-DFOO=1", "-D", "BAR", "-UBAZ", "-U", "QUX", "a.c"]);
2269        assert_eq!(opts.defines, ["FOO=1", "BAR"]);
2270        assert_eq!(opts.undefines, ["BAZ", "QUX"]);
2271    }
2272
2273    #[test]
2274    fn the_include_flags_land_on_the_chain_each_one_names() {
2275        // A sysroot with nothing under it, so that the library's own directories are the
2276        // same on every machine this test runs on, which is none of them.
2277        let (opts, _) = compile(&[
2278            "-Ii",
2279            "-iquote",
2280            "q",
2281            "-isystem",
2282            "sys",
2283            "-idirafter",
2284            "after",
2285            "--sysroot=/nowhere-at-all",
2286            "a.c",
2287        ]);
2288        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2289        // The compiler's own headers sit after every `-isystem` and before `-idirafter`,
2290        // which is where GCC puts its own: a directory the user named outranks ours.
2291        assert_eq!(dirs, ["q", "i", "sys", runtime::DIR, "after"]);
2292        assert!(!opts.search.dirs()[1].is_system);
2293        assert!(opts.search.dirs()[2].is_system);
2294    }
2295
2296    #[test]
2297    fn the_librarys_headers_come_after_the_compilers_own_and_go_away_with_them() {
2298        // Which machine this runs on decides what is on the path, so the test is about the
2299        // order rather than about the names: ours is on it, the library's follow it, and
2300        // `-nostdinc` is the one flag that takes both halves of the pair off at once.
2301        let (opts, _) = compile(&["a.c"]);
2302        let dirs = opts.search.dirs();
2303        let ours = dirs.iter().position(|d| d.path.to_str() == Some(runtime::DIR));
2304        assert_eq!(ours, Some(0), "{dirs:?}");
2305        assert!(dirs[1..].iter().all(|d| d.is_system), "{dirs:?}");
2306        let (bare, _) = compile(&["-nostdinc", "a.c"]);
2307        assert!(bare.search.dirs().is_empty(), "{:?}", bare.search.dirs());
2308    }
2309
2310    #[test]
2311    fn a_sysroot_moves_the_librarys_directories_and_nothing_else() {
2312        let (opts, _) = compile(&["-isystem", "sys", "--sysroot=/nowhere-at-all", "a.c"]);
2313        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2314        assert_eq!(dirs, ["sys", runtime::DIR]);
2315    }
2316
2317    #[test]
2318    fn dash_i_dash_moves_the_bracket_directories_into_the_quoted_chain() {
2319        let (opts, _) =
2320            compile(&["-Iinc1", "-iquote", "inc2", "-I-", "-Iinc3", "-nostdinc", "a.c"]);
2321        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2322        assert_eq!(dirs, ["inc1", "inc2", "inc3"]);
2323        // An angled include sees only what came after the flag.
2324        assert_eq!(opts.search.start(IncludeForm::Angled), 2);
2325        assert!(!opts.search.searches_current_dir());
2326    }
2327
2328    #[test]
2329    fn the_prefix_flags_stick_what_iprefix_said_on_the_front_of_what_follows_it() {
2330        let (opts, _) = compile(&[
2331            "-iprefix",
2332            "/tools/",
2333            "-iwithprefix",
2334            "late",
2335            "-iwithprefixbefore",
2336            "early",
2337            "-iprefix",
2338            "/other/",
2339            "-iwithprefix",
2340            "last",
2341            "-nostdinc",
2342            "a.c",
2343        ]);
2344        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2345        // `-iwithprefixbefore` is an `-I` and the other two are `-isystem`, which is where GCC
2346        // puts them rather than where its manual says it does.
2347        assert_eq!(dirs, ["/tools/early", "/tools/late", "/other/last"]);
2348        assert!(!opts.search.dirs()[0].is_system);
2349        assert!(opts.search.dirs()[1].is_system);
2350    }
2351
2352    #[test]
2353    fn the_files_named_on_the_command_line_keep_their_order_and_which_flag_named_them() {
2354        let (opts, _) =
2355            compile(&["-include", "one.h", "-imacros", "two.h", "-include", "3.h", "a.c"]);
2356        let names: Vec<&str> = opts.preincludes.iter().map(|p| p.name.as_str()).collect();
2357        assert_eq!(names, ["one.h", "two.h", "3.h"]);
2358        assert_eq!(opts.preincludes.iter().filter(|p| p.macros_only).count(), 1);
2359    }
2360
2361    #[test]
2362    fn nostdinc_takes_the_compilers_own_headers_off_the_path() {
2363        let (opts, _) = compile(&["-Ii", "-nostdinc", "a.c"]);
2364        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2365        assert_eq!(dirs, ["i"]);
2366    }
2367
2368    #[test]
2369    fn the_dialect_flags_set_the_language_and_the_extensions_separately() {
2370        let (opts, _) = compile(&["-std=gnu11", "a.c"]);
2371        assert_eq!(opts.std, Std::C11);
2372        assert!(opts.gnu_extensions);
2373
2374        let (opts, _) = compile(&["-std=iso9899:1999", "a.c"]);
2375        assert_eq!(opts.std, Std::C99);
2376        assert!(!opts.gnu_extensions);
2377
2378        let (opts, _) = compile(&["-ansi", "a.c"]);
2379        assert_eq!(opts.std, Std::C89);
2380        assert!(!opts.gnu_extensions);
2381
2382        let e = parse_args(&args(&["-std=c94jr", "a.c"])).unwrap_err();
2383        assert!(e.message.contains("unknown dialect"), "{}", e.message);
2384    }
2385
2386    #[test]
2387    fn the_dump_letters_are_a_family_and_everything_else_beginning_with_d_is_not() {
2388        let (opts, _) = compile(&["-dM", "a.c"]);
2389        assert!(opts.dumps.macros);
2390
2391        // Packed, the way GCC takes them, and a letter in the family we have not written yet
2392        // is accepted and does nothing rather than failing a build.
2393        let (opts, _) = compile(&["-dDM", "a.c"]);
2394        assert!(opts.dumps.macros);
2395        let (opts, _) = compile(&["-dD", "a.c"]);
2396        assert!(!opts.dumps.macros);
2397
2398        let (opts, _) = compile(&["a.c"]);
2399        assert!(!opts.dumps.any());
2400
2401        // `-dumpversion` is a different flag that happens to start the same way, and it is read
2402        // as itself rather than as a dump of nothing.
2403        assert_eq!(printed(&["-dumpversion", "a.c"]), VERSION);
2404    }
2405
2406    #[test]
2407    fn the_gcc_version_claimed_is_a_flag_and_the_short_spellings_are_the_ones_people_write() {
2408        let (opts, _) = compile(&["a.c"]);
2409        assert_eq!(
2410            opts.gnuc,
2411            GnucVersion { major: 7, minor: 0, patch: 0 },
2412            "the lowest claim a modern glibc gives its own declarations to"
2413        );
2414
2415        let (opts, _) = compile(&["-fgnuc-version=15.1.0", "a.c"]);
2416        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 1, patch: 0 });
2417
2418        // A missing component is zero. `gcc -dumpversion` says `15` on a release with no
2419        // patchlevel and a harness that pastes that back has to be understood.
2420        let (opts, _) = compile(&["-fgnuc-version=15", "a.c"]);
2421        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 0, patch: 0 });
2422
2423        let (opts, _) = compile(&["-fgnuc-version=13.2", "a.c"]);
2424        assert_eq!(opts.gnuc, GnucVersion { major: 13, minor: 2, patch: 0 });
2425
2426        let e = parse_args(&args(&["-fgnuc-version=15.x", "a.c"])).unwrap_err();
2427        assert!(e.message.contains("minor that is not a number"), "{}", e.message);
2428
2429        let e = parse_args(&args(&["-fgnuc-version=1.2.3.4", "a.c"])).unwrap_err();
2430        assert!(e.message.contains("more than three"), "{}", e.message);
2431    }
2432
2433    #[test]
2434    fn pedantic_has_two_spellings_and_is_not_the_same_knob_as_the_dialect() {
2435        let (opts, _) = compile(&["-std=c17", "-pedantic", "a.c"]);
2436        assert!(opts.pedantic);
2437        assert_eq!(opts.std, Std::C17);
2438
2439        // The `-W` family's name for it, which is what a build that groups its warning flags
2440        // tends to write.
2441        let (opts, _) = compile(&["-Wpedantic", "a.c"]);
2442        assert!(opts.pedantic);
2443
2444        let (opts, _) = compile(&["-std=c17", "a.c"]);
2445        assert!(!opts.pedantic, "a dialect on its own does not diagnose an extension");
2446    }
2447
2448    #[test]
2449    fn dash_p_and_dash_ffreestanding_reach_the_options() {
2450        let (opts, _) = compile(&["-E", "-P", "-ffreestanding", "a.c"]);
2451        assert!(!opts.line_markers);
2452        assert!(!opts.hosted);
2453        assert_eq!(opts.emit, EmitKind::Preprocessed);
2454    }
2455
2456    /// The two ways a build says it means its own function by a name the C library also has.
2457    ///
2458    /// `-fno-builtin` is all of them and `-fno-builtin-<name>` is one, and the second is what a
2459    /// build writes when it means its own `memcpy` and the library's everything else. The name is
2460    /// kept as it was written and not checked against anything, because a program is allowed to
2461    /// mean something by a name this compiler has never heard of.
2462    #[test]
2463    fn the_builtin_flags_are_read_in_both_directions_and_one_name_at_a_time() {
2464        let (opts, _) = compile(&["-c", "a.c"]);
2465        assert!(opts.builtins, "a library name means the library function by default");
2466        assert!(opts.no_builtin.is_empty());
2467
2468        let (opts, _) = compile(&["-c", "-fno-builtin", "a.c"]);
2469        assert!(!opts.builtins);
2470
2471        let (opts, _) = compile(&["-c", "-fno-builtin", "-fbuiltin", "a.c"]);
2472        assert!(opts.builtins, "the last mention decides");
2473
2474        let (opts, _) = compile(&["-c", "-fno-builtin-memcpy", "-fno-builtin-nonesuch", "a.c"]);
2475        assert!(opts.builtins, "one name is not the family");
2476        assert_eq!(opts.no_builtin, vec!["memcpy".to_owned(), "nonesuch".to_owned()]);
2477    }
2478
2479    /// `-fvisibility=`, which is on every cmake project that cares about which names it exports
2480    /// and which was refused as an unknown option until now.
2481    ///
2482    /// Four spellings and three answers. `internal` is hidden plus a promise about never taking
2483    /// the address across a component boundary, and nothing derives anything from that promise
2484    /// here, so it comes out as the weaker of the two rather than as a refusal that stops a build
2485    /// over a distinction this compiler does not make.
2486    #[test]
2487    fn visibility_takes_the_four_spellings_gcc_takes_and_refuses_the_rest() {
2488        let (opts, _) = compile(&["-c", "a.c"]);
2489        assert_eq!(opts.visibility, Visibility::Default, "exported unless something says not");
2490
2491        for (written, wanted) in [
2492            ("default", Visibility::Default),
2493            ("hidden", Visibility::Hidden),
2494            ("internal", Visibility::Hidden),
2495            ("protected", Visibility::Protected),
2496        ] {
2497            let (opts, _) = compile(&["-c", &format!("-fvisibility={written}"), "a.c"]);
2498            assert_eq!(opts.visibility, wanted, "{written}");
2499        }
2500
2501        // The last mention decides, which is what every other flag of this shape does and what a
2502        // build that turns something off for one directory relies on.
2503        let (opts, _) = compile(&["-c", "-fvisibility=hidden", "-fvisibility=default", "a.c"]);
2504        assert_eq!(opts.visibility, Visibility::Default, "the last mention decides");
2505
2506        // A spelling gcc does not take is refused rather than read as the default, because a
2507        // build that meant hidden and got exported is a library with the wrong interface and
2508        // nothing said about it anywhere.
2509        let failed = parse_args(&args(&["-fvisibility=none", "a.c"])).expect_err("refused");
2510        assert!(failed.to_string().contains("is not a visibility"), "{failed}");
2511    }
2512
2513    /// `-ffunction-sections` and `-fdata-sections`, which are what make `--gc-sections` able to
2514    /// drop anything: a linker can leave out a section nothing reaches and cannot leave out half of
2515    /// one. A kernel and an embedded image are both linked that way.
2516    ///
2517    /// Two flags rather than one because gcc has two, and a build that asks for one of them and not
2518    /// the other is a build that measured something: splitting the code is nearly free at link time
2519    /// and splitting the data can defeat the linker's ordering of what is next to what.
2520    #[test]
2521    fn a_section_per_function_and_a_section_per_variable_are_asked_for_one_at_a_time() {
2522        let (opts, _) = compile(&["-c", "a.c"]);
2523        assert!(!opts.function_sections, "one text section unless something says otherwise");
2524        assert!(!opts.data_sections);
2525
2526        let (opts, _) = compile(&["-c", "-ffunction-sections", "a.c"]);
2527        assert!(opts.function_sections);
2528        assert!(!opts.data_sections, "one flag is not the other");
2529
2530        let (opts, _) = compile(&["-c", "-fdata-sections", "a.c"]);
2531        assert!(opts.data_sections);
2532        assert!(!opts.function_sections);
2533
2534        // Both directions taken, and the off one is what happens anyway rather than a refusal,
2535        // since a build that writes it is asking for the default.
2536        let (opts, _) = compile(&[
2537            "-c",
2538            "-ffunction-sections",
2539            "-fno-function-sections",
2540            "-fdata-sections",
2541            "-fno-data-sections",
2542            "a.c",
2543        ]);
2544        assert!(!opts.function_sections, "the last mention decides");
2545        assert!(!opts.data_sections, "the last mention decides");
2546    }
2547
2548    /// `-fgnu89-inline`, which is off by default and is not implied by anything on the command
2549    /// line, since the dialect asks for GNU's reading further in rather than through this.
2550    #[test]
2551    fn gnu89_inline_is_off_until_it_is_asked_for_and_the_last_mention_decides() {
2552        let (opts, _) = compile(&["-c", "a.c"]);
2553        assert!(!opts.gnu89_inline, "C's reading of inline by default");
2554
2555        let (opts, _) = compile(&["-c", "-fgnu89-inline", "a.c"]);
2556        assert!(opts.gnu89_inline);
2557
2558        let (opts, _) = compile(&["-c", "-fgnu89-inline", "-fno-gnu89-inline", "a.c"]);
2559        assert!(!opts.gnu89_inline, "the last mention decides");
2560
2561        // The C89 dialects are under GNU's reading whether this was written or not, so the flag
2562        // stays off there and the dialect is what the checker and the macro set both ask. That is
2563        // also why `-std=c89 -fno-gnu89-inline` needs no diagnostic: it asks for the reading the
2564        // dialect already has. gcc refuses that command line, which is measured in the issue.
2565        let (opts, _) = compile(&["-c", "-std=c89", "a.c"]);
2566        assert!(!opts.gnu89_inline);
2567    }
2568
2569    /// Both spellings of both frame flags, since a build that wants one usually writes the
2570    /// other beside it for the one file that has to be compiled the ordinary way.
2571    #[test]
2572    fn the_two_frame_flags_are_read_in_both_directions() {
2573        let (opts, _) = compile(&["-c", "a.c"]);
2574        assert!(!opts.frame_pointer, "gcc omits it above -O0 and so does this");
2575        assert!(opts.red_zone, "the psABI has one and nothing said not to use it");
2576
2577        let (opts, _) = compile(&["-c", "-fno-omit-frame-pointer", "-mno-red-zone", "a.c"]);
2578        assert!(opts.frame_pointer);
2579        assert!(!opts.red_zone);
2580
2581        let (opts, _) = compile(&[
2582            "-c",
2583            "-fno-omit-frame-pointer",
2584            "-fomit-frame-pointer",
2585            "-mno-red-zone",
2586            "-mred-zone",
2587            "a.c",
2588        ]);
2589        assert!(!opts.frame_pointer, "the last one wins, as it does in gcc");
2590        assert!(opts.red_zone);
2591    }
2592
2593    /// Four flags rather than one with an argument, which is how gcc spells them, and the negative
2594    /// spelled three ways because a build that turns one off writes whichever it turned on.
2595    #[test]
2596    fn the_stack_protector_is_four_flags_and_the_last_one_wins() {
2597        let (opts, _) = compile(&["-c", "a.c"]);
2598        assert_eq!(opts.protector, Protector::None, "gcc protects nothing unless it was asked");
2599
2600        for (flag, want) in [
2601            ("-fstack-protector", Protector::Buffers),
2602            ("-fstack-protector-strong", Protector::Strong),
2603            ("-fstack-protector-all", Protector::All),
2604        ] {
2605            let (opts, _) = compile(&["-c", flag, "a.c"]);
2606            assert_eq!(opts.protector, want, "{flag}");
2607        }
2608
2609        // What a package build does: the strong one in the global flags and one directory that
2610        // cannot have a protector turning it off on the line after.
2611        for off in ["-fno-stack-protector", "-fno-stack-protector-strong"] {
2612            let (opts, _) = compile(&["-c", "-fstack-protector-strong", off, "a.c"]);
2613            assert_eq!(opts.protector, Protector::None, "{off}");
2614        }
2615        let (opts, _) = compile(&["-c", "-fno-stack-protector", "-fstack-protector-all", "a.c"]);
2616        assert_eq!(opts.protector, Protector::All, "the last one wins either way round");
2617    }
2618
2619    /// A switch rather than a level, because how a frame is taken is one question and which
2620    /// functions get a canary is another, and gcc spells it that way for the same reason.
2621    #[test]
2622    fn taking_a_frame_a_page_at_a_time_is_off_until_it_is_asked_for() {
2623        let (opts, _) = compile(&["-c", "a.c"]);
2624        assert!(!opts.stack_clash, "gcc takes a frame in one subtraction unless it was asked");
2625
2626        let (opts, _) = compile(&["-c", "-fstack-clash-protection", "a.c"]);
2627        assert!(opts.stack_clash);
2628
2629        // The same shape a package build uses for the protector: on in the global flags and off
2630        // for the one directory that cannot have it.
2631        let (opts, _) =
2632            compile(&["-c", "-fstack-clash-protection", "-fno-stack-clash-protection", "a.c"]);
2633        assert!(!opts.stack_clash);
2634        let (opts, _) =
2635            compile(&["-c", "-fno-stack-clash-protection", "-fstack-clash-protection", "a.c"]);
2636        assert!(opts.stack_clash, "the last one wins either way round");
2637
2638        // The two are independent, since one is about the frame and the other about the function.
2639        let (opts, _) =
2640            compile(&["-c", "-fstack-clash-protection", "-fstack-protector-strong", "a.c"]);
2641        assert!(opts.stack_clash);
2642        assert_eq!(opts.protector, Protector::Strong);
2643    }
2644
2645    /// One flag with an argument rather than a family of spellings, because what it asks about is
2646    /// which of the two edges of a control flow transfer is checked and the two are not separate
2647    /// questions to the hardware.
2648    #[test]
2649    fn which_control_flow_edges_are_checked_is_asked_for_by_name() {
2650        let (opts, _) = compile(&["-c", "a.c"]);
2651        assert_eq!(opts.control, Control::None, "gcc's default on the targets this compiler has");
2652
2653        for (arg, want) in [
2654            ("-fcf-protection", Control::Full),
2655            ("-fcf-protection=full", Control::Full),
2656            ("-fcf-protection=branch", Control::Branch),
2657            ("-fcf-protection=return", Control::Return),
2658            ("-fcf-protection=none", Control::None),
2659            ("-fcf-protection=check", Control::Check),
2660        ] {
2661            let (opts, _) = compile(&["-c", arg, "a.c"]);
2662            assert_eq!(opts.control, want, "{arg}");
2663        }
2664
2665        // The shape a package build uses: on in the global flags and off for the one directory
2666        // that cannot have it, whichever of the two spellings of off it reaches for.
2667        let (opts, _) = compile(&["-c", "-fcf-protection=full", "-fno-cf-protection", "a.c"]);
2668        assert_eq!(opts.control, Control::None);
2669        let (opts, _) = compile(&["-c", "-fno-cf-protection", "-fcf-protection=branch", "a.c"]);
2670        assert_eq!(opts.control, Control::Branch, "the last one wins either way round");
2671    }
2672
2673    /// The profiler is asked for by two spellings, and where its hook goes by two more.
2674    ///
2675    /// The two halves are separate on purpose. `-mfentry` on its own says where a call would go and
2676    /// asks for no call, which is what gcc does with it, and a build system that sets it globally
2677    /// and asks for the profile per directory needs that to be true rather than an error.
2678    ///
2679    /// The link is asserted alongside, because the flag changes it too and a build that compiled
2680    /// with it and linked without it is a program that calls the hook everywhere and never writes a
2681    /// profile.
2682    #[test]
2683    fn the_profiler_and_where_its_hook_goes_are_two_separate_questions() {
2684        let (opts, _) = compile(&["-c", "a.c"]);
2685        assert!(!opts.profile);
2686        assert_eq!(opts.hook, Hook::Platform, "neither was named, so the target decides");
2687
2688        for arg in ["-pg", "-p"] {
2689            let (opts, _) = compile(&["-c", arg, "a.c"]);
2690            assert!(opts.profile, "{arg}");
2691            let (link, _) = linking(&[arg, "a.c"]);
2692            assert!(link.profile, "{arg} changes the link as well");
2693        }
2694
2695        for (arg, want) in [("-mfentry", Hook::Early), ("-mno-fentry", Hook::Late)] {
2696            let (opts, _) = compile(&["-c", arg, "a.c"]);
2697            assert_eq!(opts.hook, want, "{arg}");
2698            assert!(!opts.profile, "{arg} asks for no call of its own");
2699        }
2700
2701        let (opts, _) = compile(&["-c", "-mfentry", "-mno-fentry", "-pg", "a.c"]);
2702        assert_eq!(opts.hook, Hook::Late, "the last one wins");
2703        assert!(opts.profile);
2704    }
2705
2706    /// How much room a patcher is promised, which is one number or two.
2707    ///
2708    /// A command line that did not ask is asserted alongside, because the flag has to be written to
2709    /// mean anything and a build that reserved room nobody asked for would grow every function in
2710    /// it for nothing.
2711    #[test]
2712    fn the_room_a_patcher_is_promised_is_a_number_of_bytes_and_where_they_go() {
2713        let (opts, _) = compile(&["-c", "a.c"]);
2714        assert_eq!(opts.patchable, Patchable::default());
2715        assert!(!opts.patchable.any(), "nothing is reserved unless it was asked for");
2716
2717        let (opts, _) = compile(&["-c", "-fpatchable-function-entry=16", "a.c"]);
2718        assert_eq!(opts.patchable, Patchable { total: 16, before: 0 });
2719
2720        let (opts, _) = compile(&["-c", "-fpatchable-function-entry=5,3", "a.c"]);
2721        assert_eq!(opts.patchable, Patchable { total: 5, before: 3 });
2722        assert_eq!(opts.patchable.after(), 2);
2723
2724        // The last one wins, which is what every other flag of this shape does and what a build
2725        // that adds one to a command line it did not write is relying on.
2726        let (opts, _) = compile(&[
2727            "-c",
2728            "-fpatchable-function-entry=5,3",
2729            "-fpatchable-function-entry=2",
2730            "a.c",
2731        ]);
2732        assert_eq!(opts.patchable, Patchable { total: 2, before: 0 });
2733    }
2734
2735    /// And a request nothing could satisfy is refused rather than rounded into one that can be.
2736    #[test]
2737    fn room_in_front_of_the_label_that_is_more_than_the_room_asked_for_is_refused() {
2738        for arg in ["-fpatchable-function-entry=1,2", "-fpatchable-function-entry=x"] {
2739            let e = parse_args(&args(&["-c", arg, "a.c"])).unwrap_err();
2740            assert!(e.message.contains("is not an amount of room to reserve"), "{}", e.message);
2741        }
2742    }
2743
2744    /// What wraps rather than being undefined, which is two questions and three flags.
2745    ///
2746    /// The older flag is the pair of the newer two, which is gcc's own reading of it, so a build
2747    /// that writes `-fno-strict-overflow` gets both and a build that writes one of the others gets
2748    /// only what it asked for.
2749    #[test]
2750    fn what_overflows_rather_than_being_undefined_is_asked_for_two_ways() {
2751        let (opts, _) = compile(&["-c", "a.c"]);
2752        assert_eq!(opts.wrapping, Wrapping::NONE, "nothing wraps unless it was asked for");
2753
2754        let (opts, _) = compile(&["-c", "-fwrapv", "a.c"]);
2755        assert_eq!(opts.wrapping, Wrapping { signed: true, pointer: false });
2756
2757        let (opts, _) = compile(&["-c", "-fwrapv-pointer", "a.c"]);
2758        assert_eq!(opts.wrapping, Wrapping { signed: false, pointer: true });
2759
2760        let (opts, _) = compile(&["-c", "-fno-strict-overflow", "a.c"]);
2761        assert_eq!(opts.wrapping, Wrapping::ALL);
2762
2763        // And the last one wins, in both directions. A build that turns one of these on globally
2764        // and off for one directory is relying on that, and so is one that writes the pair and
2765        // then takes half of it back.
2766        let (opts, _) = compile(&["-c", "-fwrapv", "-fno-wrapv", "a.c"]);
2767        assert_eq!(opts.wrapping, Wrapping::NONE);
2768
2769        let (opts, _) = compile(&["-c", "-fno-strict-overflow", "-fstrict-overflow", "a.c"]);
2770        assert_eq!(opts.wrapping, Wrapping::NONE);
2771
2772        let (opts, _) = compile(&["-c", "-fno-strict-overflow", "-fno-wrapv-pointer", "a.c"]);
2773        assert_eq!(opts.wrapping, Wrapping { signed: true, pointer: false });
2774    }
2775
2776    /// And a value nothing means is refused rather than taken for the nearest thing it looks like.
2777    ///
2778    /// `-fcf-protection=all` is the spelling somebody writes from memory, and a compiler that read
2779    /// it as `full` would be guessing, while one that let it fall through to the optimizer's `-f`
2780    /// family would report it as an unknown pass. Neither is the news the build wants.
2781    #[test]
2782    fn a_control_flow_protection_nothing_means_is_refused() {
2783        let e = parse_args(&args(&["-c", "-fcf-protection=all", "a.c"])).unwrap_err();
2784        assert!(e.message.contains("is not a control flow protection"), "{}", e.message);
2785        assert!(e.message.contains("full, branch, return, none or check"), "{}", e.message);
2786    }
2787
2788    #[test]
2789    fn the_link_flags_are_collected_apart_from_the_compilation() {
2790        let (link, _) = linking(&[
2791            "-static",
2792            "-nostartfiles",
2793            "-rdynamic",
2794            "-s",
2795            "-fuse-ld=mold",
2796            "-L/opt/lib",
2797            "-B",
2798            "/opt/tools",
2799            "a.c",
2800        ]);
2801        assert!(link.is_static);
2802        assert!(link.no_startfiles);
2803        assert!(link.export_dynamic);
2804        assert!(link.strip);
2805        assert_eq!(link.use_ld.as_deref(), Some("mold"));
2806        assert_eq!(link.search, vec![PathBuf::from("/opt/lib")]);
2807        assert_eq!(link.prefixes, vec![PathBuf::from("/opt/tools")]);
2808    }
2809
2810    #[test]
2811    fn a_comma_in_dash_wl_separates_two_arguments() {
2812        let (link, _) = linking(&["-Wl,-rpath,/opt/lib", "-Xlinker", "--as-needed", "a.c"]);
2813        assert_eq!(link.passthrough, vec!["-rpath", "/opt/lib", "--as-needed"]);
2814    }
2815
2816    #[test]
2817    fn a_library_keeps_its_place_between_the_objects() {
2818        // Link order is semantic: `-lm` written between two files resolves for the one before
2819        // it and not for the one after, so a library cannot be collected into a list of its own.
2820        // The target is named because the suffix of an object is the target's and this asserts
2821        // on the names: the same command line on a Windows host plans two `.obj` files.
2822        let (_, plan) = linking(&["--target=x86_64-unknown-linux-gnu", "a.c", "-lm", "b.c"]);
2823        let link = plan.link.expect("expected a link step");
2824        assert_eq!(
2825            link.inputs,
2826            vec![
2827                link::Item::File("a.o".into()),
2828                link::Item::Library("m".into()),
2829                link::Item::File("b.o".into()),
2830            ]
2831        );
2832        // And it is not a job, because there is nothing to compile in a library.
2833        assert_eq!(plan.jobs.len(), 2);
2834    }
2835
2836    #[test]
2837    fn a_library_on_a_dash_c_line_is_a_note_rather_than_an_error() {
2838        let (_, plan) = linking(&["-c", "-lm", "a.c"]);
2839        assert!(plan.link.is_none());
2840        assert!(plan.notes.iter().any(|n| n.contains("-lm")), "{:?}", plan.notes);
2841    }
2842
2843    #[test]
2844    fn the_sysroot_reaches_the_linker_as_well_as_the_headers() {
2845        let (link, _) = linking(&["--sysroot=/opt/root", "a.c"]);
2846        assert_eq!(link.sysroot, Some(PathBuf::from("/opt/root")));
2847    }
2848
2849    fn printed(s: &[&str]) -> String {
2850        match parse_args(&args(s)).expect("expected an answer") {
2851            Action::Print(line) => line,
2852            other => panic!("expected an answer, got {other:?}"),
2853        }
2854    }
2855
2856    fn refused(s: &[&str]) -> String {
2857        parse_args(&args(s)).expect_err("expected a refusal").message
2858    }
2859
2860    #[test]
2861    fn a_warning_flag_this_compiler_has_not_heard_of_is_taken_rather_than_refused() {
2862        // The rule in section 4.1, and the reason for it is autoconf: a configure script finds
2863        // out whether a warning flag exists by passing it and looking at the exit status, so a
2864        // compiler that refuses one it does not know fails a script written for a newer GCC.
2865        let (opts, _) = compile(&["-Wall", "-Wextra", "-Wno-format-truncation", "-c", "a.c"]);
2866        assert!(!opts.warnings_are_errors);
2867        assert!(opts.warnings);
2868        // The two spellings that do mean something are still read.
2869        let (opts, _) = compile(&["-Werror", "-c", "a.c"]);
2870        assert!(opts.warnings_are_errors);
2871        let (opts, _) = compile(&["-w", "-c", "a.c"]);
2872        assert!(!opts.warnings);
2873        let (opts, _) = compile(&["-pedantic-errors", "-c", "a.c"]);
2874        assert!(opts.pedantic && opts.warnings_are_errors);
2875    }
2876
2877    #[test]
2878    fn an_argument_for_a_separate_tool_is_refused_rather_than_dropped() {
2879        // Every one of these says something about the output, so the wrong answer is silence.
2880        assert!(refused(&["-Wa,--noexecstack", "-c", "a.c"]).contains("separate assembler"));
2881        assert!(refused(&["-Wp,-DX", "-c", "a.c"]).contains("separate assembler"));
2882        assert!(refused(&["-specs=/x", "a.c"]).contains("-specs= is not supported"));
2883        assert!(refused(&["-mcmodel=kernel", "-c", "a.c"]).contains("small code model"));
2884        assert!(refused(&["-gdwarf-4", "-c", "a.c"]).contains("DWARF 5"));
2885        assert!(refused(&["-Ofast", "-c", "a.c"]).contains("fast math"));
2886        // The word size the target does not have, which is a target this compiler was not asked
2887        // for rather than a flag it does not know.
2888        let no32 = refused(&["--target=x86_64-unknown-linux-gnu", "-m32", "-c", "a.c"]);
2889        assert!(no32.contains("32 bit target"), "{no32}");
2890    }
2891
2892    #[test]
2893    fn the_levels_gcc_spells_differently_are_the_levels_they_mean() {
2894        assert_eq!(compile(&["-O", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2895        assert_eq!(compile(&["-Og", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2896        assert_eq!(compile(&["-O2", "-c", "a.c"]).0.opt_level, OptLevel::O2);
2897    }
2898
2899    #[test]
2900    fn the_machine_flags_that_name_what_we_already_do_are_taken_and_the_rest_are_not() {
2901        let line = ["--target=x86_64-unknown-linux-gnu", "-m64", "-march=x86-64-v3"];
2902        let (opts, _) =
2903            compile(&[&line[..], &["-mtune=native", "-mabi=sysv", "-c", "a.c"]].concat());
2904        assert_eq!(opts.target.to_string(), "x86_64-unknown-linux-gnu");
2905        let wrong = refused(&["--target=x86_64-unknown-linux-gnu", "-mabi=ms", "-c", "a.c"]);
2906        assert!(wrong.contains("sysv convention"), "{wrong}");
2907    }
2908
2909    #[test]
2910    fn the_thread_flag_is_a_macro_and_a_library_and_the_library_goes_last() {
2911        let (opts, plan) = compile(&["-pthread", "-c", "a.c"]);
2912        assert!(opts.defines.iter().any(|d| d == "_REENTRANT"));
2913        // After the input, because a static link takes what it needs from a library when it
2914        // reaches it and not afterwards.
2915        let names: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
2916        assert_eq!(names, vec!["a.c"]);
2917    }
2918
2919    #[test]
2920    fn the_questions_a_build_system_asks_before_it_compiles_anything() {
2921        let target = "--target=x86_64-unknown-linux-gnu";
2922        assert_eq!(printed(&[target, "-dumpmachine"]), "x86_64-unknown-linux-gnu");
2923        assert_eq!(printed(&[target, "-dumpversion"]), VERSION);
2924        assert_eq!(printed(&[target, "-dumpfullversion"]), VERSION);
2925        assert_eq!(printed(&[target, "-print-multiarch"]), "x86_64-linux-gnu");
2926        // A name nothing holds comes back unchanged, which is GCC's rule and is what makes the
2927        // answer safe to paste into a link line whether or not the file is there.
2928        assert_eq!(printed(&[target, "-print-file-name=no-such-library.a"]), "no-such-library.a");
2929        assert_eq!(printed(&[target, "-print-prog-name=ld"]), "ld");
2930        let dirs = printed(&[target, "-print-search-dirs"]);
2931        assert!(dirs.starts_with("install: "), "{dirs}");
2932        assert!(dirs.contains("\nlibraries: ="), "{dirs}");
2933    }
2934
2935    #[test]
2936    fn the_two_dependency_flags_that_stop_after_the_rule_stop_after_the_rule() {
2937        let (opts, _) = compile(&["-M", "a.c"]);
2938        assert!(opts.deps.emit && opts.deps.instead_of_compiling);
2939        assert!(opts.deps.system_headers, "plain -M lists them");
2940        assert_eq!(opts.emit, EmitKind::Preprocessed);
2941
2942        // Even where a later flag asked for something else, because the family is a mode and
2943        // the mode is what the run is for.
2944        let (opts, _) = compile(&["-M", "-c", "a.c"]);
2945        assert_eq!(opts.emit, EmitKind::Preprocessed);
2946
2947        let (opts, _) = compile(&["-MM", "a.c"]);
2948        assert!(!opts.deps.system_headers);
2949    }
2950
2951    #[test]
2952    fn the_two_that_end_in_d_leave_the_compilation_alone() {
2953        let (opts, _) = compile(&["-MD", "-c", "a.c"]);
2954        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2955        assert!(opts.deps.system_headers);
2956        assert_eq!(opts.emit, EmitKind::Object);
2957
2958        let (opts, _) = compile(&["-MMD", "-c", "a.c"]);
2959        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2960        assert!(!opts.deps.system_headers);
2961    }
2962
2963    #[test]
2964    fn nothing_puts_the_system_headers_back_once_a_flag_has_taken_them_out() {
2965        // GCC's rule, and not an oversight in it. The flag asking for fewer of them is read as
2966        // the answer, because the other one never asked the question.
2967        let (opts, _) = compile(&["-MM", "-M", "a.c"]);
2968        assert!(!opts.deps.system_headers);
2969        let (opts, _) = compile(&["-MD", "-MMD", "-c", "a.c"]);
2970        assert!(!opts.deps.system_headers);
2971        let (opts, _) = compile(&["-MMD", "-MD", "-c", "a.c"]);
2972        assert!(!opts.deps.system_headers);
2973    }
2974
2975    #[test]
2976    fn a_target_arrives_escaped_from_one_flag_and_untouched_from_the_other() {
2977        let (opts, _) = compile(&["-MM", "-MT", "a b.o", "-MQ", "a b.o", "a.c"]);
2978        assert_eq!(opts.deps.targets, vec!["a b.o".to_owned(), "a\\ b.o".to_owned()]);
2979    }
2980
2981    #[test]
2982    fn the_rest_of_the_family_is_a_file_and_a_switch() {
2983        let (opts, _) = compile(&["-MM", "-MF", "dep.d", "-MP", "a.c"]);
2984        assert_eq!(opts.deps.file.as_deref(), Some("dep.d"));
2985        assert!(opts.deps.phony);
2986
2987        for flag in ["-MF", "-MT", "-MQ"] {
2988            let e = parse_args(&args(&[flag])).unwrap_err();
2989            assert!(e.message.contains("requires an argument"), "{}", e.message);
2990        }
2991    }
2992
2993    /// A directory of sources for one test, removed when the test is done with it.
2994    struct TempTree(PathBuf);
2995
2996    impl Drop for TempTree {
2997        fn drop(&mut self) {
2998            let _ = std::fs::remove_dir_all(&self.0);
2999        }
3000    }
3001
3002    impl TempTree {
3003        fn new(name: &str, files: &[(&str, &str)]) -> TempTree {
3004            let dir = std::env::temp_dir().join(format!("rucc-deps-{}-{name}", std::process::id()));
3005            let _ = std::fs::remove_dir_all(&dir);
3006            std::fs::create_dir_all(&dir).expect("temporary directory should be writable");
3007            for (path, text) in files {
3008                let at = dir.join(path);
3009                if let Some(parent) = at.parent() {
3010                    std::fs::create_dir_all(parent).expect("creating a subdirectory should work");
3011                }
3012                std::fs::write(&at, text).expect("writing a temporary file should work");
3013            }
3014            TempTree(dir)
3015        }
3016
3017        fn path(&self, name: &str) -> String {
3018            self.0.join(name).to_string_lossy().into_owned()
3019        }
3020    }
3021
3022    #[test]
3023    fn the_rule_names_what_the_includes_found_and_names_each_of_them_once() {
3024        // End to end, because the list comes from the preprocessor and the format comes from
3025        // somewhere else, and a test of either half on its own would pass with the two of them
3026        // wired up backwards.
3027        let tree = TempTree::new(
3028            "found",
3029            &[
3030                ("a.c", "#include \"one.h\"\n#include \"two.h\"\nint main(void) { return X; }\n"),
3031                ("one.h", "#define X 0\n"),
3032                ("two.h", "#include \"one.h\"\n"),
3033            ],
3034        );
3035        let out = tree.path("dep.d");
3036        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
3037        assert_eq!(code, 0);
3038
3039        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
3040        let names: Vec<&str> = text.split_whitespace().collect();
3041        // The target, the source, and each header once however many times it was reached.
3042        assert_eq!(names.first(), Some(&"a.o:"), "{text}");
3043        assert_eq!(names.iter().filter(|n| n.ends_with("one.h")).count(), 1, "{text}");
3044        assert_eq!(names.iter().filter(|n| n.ends_with("two.h")).count(), 1, "{text}");
3045        // And the `-o` went to the file the rule replaced, which is left empty rather than
3046        // absent because a makefile that named it as a target will look for it.
3047        assert_eq!(std::fs::read(tree.path("a.i")).expect("the output should exist"), b"");
3048    }
3049
3050    #[test]
3051    fn a_header_that_is_only_reached_under_a_guard_is_still_a_dependency() {
3052        // The multiple-include optimization means the second reach never opens the file. It is
3053        // still a file this translation unit was built from, so it is still in the rule.
3054        let tree = TempTree::new(
3055            "guarded",
3056            &[
3057                ("a.c", "#include \"g.h\"\n#include \"g.h\"\nint main(void) { return 0; }\n"),
3058                ("g.h", "#ifndef G\n#define G\n#endif\n"),
3059            ],
3060        );
3061        let out = tree.path("dep.d");
3062        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
3063        assert_eq!(code, 0);
3064        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
3065        assert_eq!(text.split_whitespace().filter(|n| n.ends_with("g.h")).count(), 1, "{text}");
3066    }
3067
3068    #[test]
3069    fn every_imacros_file_is_read_before_every_include_file_whatever_order_they_were_written() {
3070        // Measured against GCC rather than read: the two flags the other way round produce the
3071        // same output byte for byte, so the command line order between the two families does not
3072        // decide anything and the order within one does. The `-include` file here can only see
3073        // the definition if the `-imacros` file that was written after it ran first.
3074        let tree = TempTree::new(
3075            "preinclude",
3076            &[
3077                ("a.c", "int main(void) { return 0; }\n"),
3078                ("i.h", "#ifdef FROM_MACROS\nint saw_it;\n#else\nint missed_it;\n#endif\n"),
3079                ("m.h", "#define FROM_MACROS 1\nint macros_text;\n"),
3080            ],
3081        );
3082        let out = tree.path("a.i");
3083        let code = run(&args(&[
3084            "-E",
3085            "-include",
3086            &tree.path("i.h"),
3087            "-imacros",
3088            &tree.path("m.h"),
3089            "-o",
3090            &out,
3091            &tree.path("a.c"),
3092        ]));
3093        assert_eq!(code, 0);
3094        let text = std::fs::read_to_string(&out).expect("the output should have been written");
3095        assert!(text.contains("saw_it"), "{text}");
3096        // And the text of the `-imacros` file is thrown away, which is the whole difference
3097        // between the two flags.
3098        assert!(!text.contains("macros_text"), "{text}");
3099    }
3100
3101    #[test]
3102    fn a_file_the_command_line_named_is_a_prerequisite_the_same_as_one_a_directive_named() {
3103        let tree = TempTree::new(
3104            "preinclude-deps",
3105            &[
3106                ("a.c", "int main(void) { return 0; }\n"),
3107                ("i.h", "int from_include;\n"),
3108                ("m.h", "#define M 1\n"),
3109            ],
3110        );
3111        let out = tree.path("dep.d");
3112        let code = run(&args(&[
3113            "-MM",
3114            "-MF",
3115            &out,
3116            "-include",
3117            &tree.path("i.h"),
3118            "-imacros",
3119            &tree.path("m.h"),
3120            "-o",
3121            &tree.path("a.i"),
3122            &tree.path("a.c"),
3123        ]));
3124        assert_eq!(code, 0);
3125        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
3126        assert!(text.contains("i.h"), "{text}");
3127        assert!(text.contains("m.h"), "{text}");
3128    }
3129
3130    #[test]
3131    fn a_command_line_include_that_is_nowhere_on_the_path_is_an_error_and_not_a_warning() {
3132        // Including the directory of the source file, which is not on the path for these: the
3133        // command line was not written there, so a name in it is relative to where the compiler
3134        // was run rather than to where the source sits.
3135        let tree = TempTree::new(
3136            "preinclude-missing",
3137            &[("sub/a.c", "int main(void) { return 0; }\n"), ("sub/beside.h", "int x;\n")],
3138        );
3139        let code = run(&args(&["-E", "-include", "beside.h", "-o", "-", &tree.path("sub/a.c")]));
3140        assert_eq!(code, 1);
3141    }
3142
3143    #[test]
3144    fn a_command_line_that_links_names_the_executable_and_not_the_object_it_went_through() {
3145        // The object a link goes through is in a temporary directory and is gone before `make`
3146        // reads any of this, so the rule that named it would be a rule for a file that is never
3147        // there. The target and the file are both the `-o`, which is the executable.
3148        let (opts, plan) = compile(&["-MD", "sub/a.c", "-o", "prog"]);
3149        assert_eq!(plan.output.as_deref(), Some("prog"));
3150        assert_eq!(deps::default_target("sub/a.c", deps_target_output(&opts, &plan)), "prog");
3151        assert_eq!(
3152            deps::default_file(&opts.deps, "sub/a.c", plan.output.as_deref()).as_deref(),
3153            Some("prog.d")
3154        );
3155    }
3156
3157    #[test]
3158    fn the_plan_keeps_the_output_name_because_the_rule_is_written_from_it() {
3159        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c", "-o", "obj/x.o"]);
3160        assert_eq!(plan.output.as_deref(), Some("obj/x.o"));
3161        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c"]);
3162        assert_eq!(plan.output, None);
3163    }
3164
3165    #[test]
3166    fn usage_fits_on_a_screen() {
3167        // Not a style preference. A help text that scrolls is one nobody reads, and this is
3168        // the cheapest way to keep it honest as flags accumulate. The number goes up only when
3169        // a family of flags arrives that has nowhere to share a line, which the two pass gates
3170        // were and which the two fuel flags and `-fsafety=` now are, and it goes up by exactly
3171        // the lines that family took. The four it went up by last are the flags a build system
3172        // passes without being asked to: how much to say, what machine to generate for, threads,
3173        // and the questions `configure` asks before it compiles anything. The one it went up by
3174        // last is the second line of `--emit`, whose kinds are a family that has now outgrown
3175        // one line and has nowhere else to go. The two it went up by last are the dependency
3176        // family, which is eight flags that share nothing with anything above them. The one it
3177        // went up by last is the four spellings of position independent code, which every
3178        // configure script writes and which could only have shared the link line, and that line
3179        // is already four characters short of the limit. The two it went up by last are the rest
3180        // of the include family, which is six more flags that change where a header is looked for
3181        // and two that name a header outright. The one it went up by last is the pair that keeps
3182        // the intermediate files and times the steps, which belong next to the two flags above
3183        // them that are also about watching a compilation rather than changing one. The two it
3184        // went up by last are the section flags and the visibility flag, which are what a build
3185        // that cares about the size of what it ships and about which names it exports writes, and
3186        // the second of them was already taken and only missing from here. The one it went up by
3187        // last is the stack protector, which is four spellings of one question and which every
3188        // distribution puts on every command line it issues, so a build that reads this list
3189        // looking for it and does not find it has to go and read the specification instead. The one
3190        // it went up by last is the profiler, which is two spellings of the request and two of
3191        // where the call goes, and which is about watching a program run rather than about what is
3192        // generated, so it shares its subject with nothing above it. The one it went up by last is
3193        // the room a function opens with for something to be written over it later, which takes an
3194        // argument of its own shape and is what a kernel build asks for, so it fits beside the
3195        // profiler and nothing else. The one it went up by last is what overflows rather than being
3196        // undefined, which is three spellings of two questions and which a kernel build and a great
3197        // deal of code written before the standard settled both pass.
3198        assert!(USAGE.lines().count() < 54, "usage text has grown past one screen");
3199    }
3200}