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/0.10.4")]
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_pp::Dependency;
46use rucc_session::{Dumps, EmitKind, Options, Pic, Preinclude, SaveTemps, Session, Std, runtime};
47use rucc_target::Triple;
48
49use crate::link::LinkOptions;
50
51pub use crate::compile::{Artifact, Compiled, Temps, compile, compile_ir};
52pub use crate::phase::{Input, InputKind, Job, LinkJob, Output, Phase, Plan};
53pub use crate::preprocess::{OsFileSystem, Preprocessed, preprocess};
54pub use crate::schedule::Jobs;
55
56/// The compiler's version, taken from the workspace manifest.
57pub const VERSION: &str = env!("CARGO_PKG_VERSION");
58
59/// What the command line asked for.
60#[derive(Debug, Clone, PartialEq, Eq)]
61pub enum Action {
62    /// Print usage and exit successfully.
63    Help,
64    /// Print the version and exit successfully.
65    Version,
66    /// Print one line and exit successfully, which is what the `-dump` and `-print` family do.
67    ///
68    /// A build system asks these before it compiles anything, and what it does with the answer
69    /// is paste it into a path or into another command line, so each one is a single line with
70    /// no decoration around it.
71    Print(String),
72    /// Print the resolved configuration and exit successfully.
73    PrintConfig(Box<Options>),
74    /// Print the passes the level will run and exit successfully.
75    PrintPipeline(Box<Options>),
76    /// Print the phase plan and the link line and exit successfully, which is `-###`.
77    PrintPlan {
78        /// The resolved options, which is what says what the link line is for.
79        opts: Box<Options>,
80        /// What to do to each input, and in what order.
81        plan: Box<Plan>,
82        /// What the command line said about linking.
83        link: Box<LinkOptions>,
84    },
85    /// Compile the given inputs.
86    Compile {
87        /// The resolved options.
88        opts: Box<Options>,
89        /// What to do to each input, and in what order.
90        plan: Box<Plan>,
91        /// What the command line said about linking.
92        link: Box<LinkOptions>,
93        /// How many translation units to compile at once.
94        jobs: Jobs,
95        /// Whether `-v` asked for the plan to be printed while it runs.
96        verbose: bool,
97    },
98}
99
100/// Why a command line was rejected.
101#[derive(Debug, Clone, PartialEq, Eq)]
102pub struct CliError {
103    /// The message, lowercase and without a trailing period, in the same shape as any other
104    /// diagnostic.
105    pub message: String,
106}
107
108impl std::fmt::Display for CliError {
109    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
110        f.write_str(&self.message)
111    }
112}
113
114impl std::error::Error for CliError {}
115
116fn err(message: impl Into<String>) -> CliError {
117    CliError { message: message.into() }
118}
119
120/// A question the command line asked instead of asking for a compilation.
121///
122/// These are answered after the loop rather than where they are read, because every one of them
123/// is about the target or about the library search and the last word on both is the end of the
124/// command line.
125enum Query {
126    /// `-dumpmachine`, the triple.
127    Machine,
128    /// `-dumpversion` and `-dumpfullversion`, which are the same three numbers here.
129    Version,
130    /// `-print-multiarch`, the directory name a distribution files this target under.
131    Multiarch,
132    /// `-print-search-dirs`, in the three lines GCC prints.
133    SearchDirs,
134    /// `-print-file-name=<name>`, the full path of a library file.
135    FileName(String),
136    /// `-print-prog-name=<name>`, the full path of a program.
137    ProgName(String),
138    /// `-print-libgcc-file-name`, which is `-print-file-name=libgcc.a` under another spelling.
139    Libgcc,
140}
141
142/// Usage text.
143///
144/// Deliberately short. `spec/04-driver-and-cli.md` puts the full flag reference in the
145/// manual page, because a `--help` nobody can read in one screen is a `--help` nobody reads.
146pub const USAGE: &str = "\
147rucc, an optimizing C compiler
148
149usage: rucc [options] file...
150
151options:
152  -c                     compile and assemble, do not link
153  -S                     compile only, emit assembly
154  -E                     preprocess only
155  -o <file>              write output to <file>, or to standard output for -
156  -D <name>[=<value>], -U <name>      define a macro, or undefine one after every -D
157  -I <dir>               add <dir> to the include search path
158  -iquote -isystem -idirafter <dir>   the other chains, -nostdinc drops ours
159  -I-, -iprefix <p>, -iwithprefix[before] <dir>   the older spellings of those
160  -include <file>, -imacros <file>    read <file> first, the second for its macros only
161  --sysroot=<dir>        look for the library's headers under <dir>, -isysroot too
162  -P, -dM                with -E: leave out the markers, or dump the macros
163  -M -MM -MD -MMD        write a make rule for the source, the last two compile as well
164  -MF <file> -MT <t> -MQ <t> -MP   where the rule goes, what it builds, targets with no recipe
165  -std=<dialect>         c89 through c23, and the gnu spellings
166  -fgnuc-version=<v>     the GCC release to claim, default 7.0.0
167  -x <lang>              treat later inputs as <lang>, or none to stop
168  -O<level>              optimize: 0, 1, 2, 3, s, z
169  -fsafety=<tier>        check memory safety: off, detect, enforce, kernel
170  -f<pass> -fno-<pass> -fdump-ir=<what> -fopt-info[-<kind>][=FILE]
171  -fpass-fuel=<pass>=<n>, -fpass-fuel-global=<n>   stop a pass, or all of them, after n
172  -fdisable-<pass>[=<funcs>], -fenable-<pass>[=<funcs>]   run a pass on some functions only
173  -g -g0 -gdwarf-5, -fno-omit-frame-pointer, -mno-red-zone   debug info, frame pointer, red zone
174  -l<name>, -L <dir>, -B <dir>   link a library, where to look for one, where our own tools are
175  -fPIC -fpic -fPIE -fpie, -fno-common, -f[no-]strict-aliasing, -pipe   what it does anyway
176  -static -shared -pie -no-pie -nostdlib -nostartfiles -nodefaultlibs -rdynamic -s   how to link
177  -Wl,<arg>, -Xlinker <arg>, -fuse-ld=<name>   hand an argument to the linker, or pick one
178  -Werror -pedantic -pedantic-errors -w   how much to say, and whether it is fatal
179  -m64 -march= -mtune= -mcpu= -mabi= -mcmodel=   what machine to generate for
180  -pthread               build for more than one thread, and link the library for it
181  -dumpmachine -dumpversion -print-multiarch -print-search-dirs   what this compiler is
182  -print-file-name=<name> -print-prog-name=<name>   where a file or a program is
183  -j[n]                  compile n translation units at once, default all
184  -v, -###               print each phase as it runs, or without running any
185  -save-temps[=cwd|obj], -time   keep the .i and the .s, say how long each step took
186  --target=<triple>      generate code for <triple>
187  --emit=<kind>          exe, obj, asm, preprocessed, tast, ir, mir-final,
188                         safety-summary, type-granules
189  --print-config, --print-pipeline    print the configuration or the pipeline, and exit
190  --version              print the version and exit
191  -h, --help             print this message and exit
192
193See spec/04-driver-and-cli.md for the full flag reference.
194";
195
196/// The argument of a flag that may be joined to it or may be the next word.
197///
198/// `-DFOO` and `-D FOO` are the same thing, and `at` is where the flag's own letters end.
199fn joined_or_next(
200    arg: &str,
201    at: usize,
202    args: &[String],
203    i: &mut usize,
204) -> Result<String, CliError> {
205    if arg.len() > at {
206        return Ok(arg[at..].to_owned());
207    }
208    let next = args.get(*i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
209    *i += 1;
210    Ok(next.clone())
211}
212
213/// Parses a command line, without the program name.
214///
215/// # Errors
216///
217/// Returns the message to print when the arguments do not name a compilation this compiler
218/// can attempt.
219pub fn parse_args(args: &[String]) -> Result<Action, CliError> {
220    let host = Triple::host()
221        .ok_or_else(|| err("this host is not a supported target and no --target was given"))?;
222    let mut opts = Options::new(host);
223    let mut inputs: Vec<Input> = Vec::new();
224    let mut print_config = false;
225    let mut print_pipeline = false;
226    let mut print_plan = false;
227    let mut verbose = false;
228    let mut jobs = Jobs::default();
229    let mut nostdinc = false;
230    let mut sysroot: Option<PathBuf> = None;
231    let mut output = None;
232    let mut link = LinkOptions::default();
233    let mut query: Option<Query> = None;
234    let mut threads = false;
235    // `-x` applies to inputs that come after it and stays in effect until the next one, which
236    // is why it is tracked across the loop rather than attached to a single argument.
237    let mut forced: Option<InputKind> = None;
238    // What `-iprefix` last said, stuck on the front of every later `-iwithprefix`. It applies to
239    // the flags after it and not the ones before, so a command line may set it more than once.
240    // GCC's default is its own installed header directory with the last component taken off,
241    // which is a path a cross compiler's build system knows and passes; there is no equivalent
242    // here, so with no `-iprefix` the prefix is nothing and `-iwithprefix` names a directory
243    // outright.
244    let mut iprefix = String::new();
245
246    let mut i = 0;
247    while i < args.len() {
248        let arg = args[i].as_str();
249        i += 1;
250        match arg {
251            "-h" | "--help" => return Ok(Action::Help),
252            "--version" => return Ok(Action::Version),
253            "--print-config" => print_config = true,
254            "--print-pipeline" => print_pipeline = true,
255            "-###" => print_plan = true,
256            "-v" => verbose = true,
257            // The files a compilation goes through, kept rather than thrown away. The bare
258            // spelling means `=obj` and not `=cwd`, which is not what the manual says and is what
259            // gcc 16 does; `SaveTemps::Object` carries the measurement.
260            "-save-temps" => opts.save_temps = SaveTemps::Object,
261            _ if arg.starts_with("-save-temps=") => {
262                opts.save_temps = arg["-save-temps=".len()..].parse().map_err(err)?;
263            }
264            // How long each step took. A misspelling of this is worth rejecting rather than
265            // ignoring, since a run that says nothing looks like a compilation that took no time.
266            "-time" => opts.time = true,
267            "-c" => opts.emit = EmitKind::Object,
268            "-S" => opts.emit = EmitKind::Asm,
269            "-E" => opts.emit = EmitKind::Preprocessed,
270            "-g" => opts.debug_info = true,
271            // GCC's own levels of how much debug information to write. Zero is none and every
272            // other number is some, and this compiler has one amount, so the numbers above zero
273            // all mean the same thing here. `-ggdb` is the same flag asking for whatever the
274            // debugger on the machine prefers, which is what we emit anyway.
275            "-g0" => opts.debug_info = false,
276            "-g1" | "-g2" | "-g3" | "-ggdb" | "-ggdb1" | "-ggdb2" | "-ggdb3" => {
277                opts.debug_info = true;
278            }
279            // The version of DWARF to write. We write DWARF 5 and nothing else, so a build that
280            // asks for another version is told rather than handed a file it cannot read.
281            "-gdwarf" | "-gdwarf-5" => opts.debug_info = true,
282            _ if arg.starts_with("-gdwarf-") => {
283                return Err(err(format!(
284                    "{arg}: this compiler writes DWARF 5 and no other version, see \
285                     spec/11-debug-info.md"
286                )));
287            }
288            "-Werror" => opts.warnings_are_errors = true,
289            // Nothing that is not fatal is said at all. Read at the one place a diagnostic goes
290            // through rather than here, so that a warning `-w` dropped is not counted either.
291            "-w" => opts.warnings = false,
292            "-pedantic-errors" => {
293                opts.pedantic = true;
294                opts.warnings_are_errors = true;
295            }
296            "-P" => opts.line_markers = false,
297            // The dependency family, which section 4.4 calls required because every build system
298            // that generates its own makefiles asks for it. The two that end in `D` write a file
299            // beside the object and let the compilation happen, and the two that do not write to
300            // standard output and stop after it. Nothing here turns the system headers back on
301            // once a flag has turned them off, which is GCC's behaviour and is why `-MM -M` is
302            // `-MM`: the flag asking for fewer of them is the one with something to say.
303            "-M" => {
304                opts.deps.emit = true;
305                opts.deps.instead_of_compiling = true;
306            }
307            "-MM" => {
308                opts.deps.emit = true;
309                opts.deps.instead_of_compiling = true;
310                opts.deps.system_headers = false;
311            }
312            "-MD" => opts.deps.emit = true,
313            "-MMD" => {
314                opts.deps.emit = true;
315                opts.deps.system_headers = false;
316            }
317            "-MP" => opts.deps.phony = true,
318            // These three take a word and only in the separated form, which is how GCC spells
319            // them and how every build system writes them.
320            "-MF" | "-MT" | "-MQ" => {
321                let value =
322                    args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
323                i += 1;
324                match arg {
325                    "-MF" => opts.deps.file = Some(value.clone()),
326                    // The whole of the difference between the two. `-MT` is for a build that has
327                    // already escaped what it is passing, and `-MQ` is for one that has a name
328                    // and wants it to arrive as that name.
329                    "-MT" => opts.deps.targets.push(value.clone()),
330                    _ => opts.deps.targets.push(deps::escaped(value)),
331                }
332            }
333            // The questions a build system asks before it compiles anything. Answered after the
334            // loop, because each one is about the target or the library search and the command
335            // line has not finished saying what those are.
336            "-dumpmachine" => query = Some(Query::Machine),
337            "-dumpversion" | "-dumpfullversion" => query = Some(Query::Version),
338            "-print-multiarch" => query = Some(Query::Multiarch),
339            "-print-search-dirs" => query = Some(Query::SearchDirs),
340            "-print-libgcc-file-name" => query = Some(Query::Libgcc),
341            _ if arg.starts_with("-print-file-name=") => {
342                query = Some(Query::FileName(arg["-print-file-name=".len()..].to_owned()));
343            }
344            _ if arg.starts_with("-print-prog-name=") => {
345                query = Some(Query::ProgName(arg["-print-prog-name=".len()..].to_owned()));
346            }
347            // A program built to run in more than one thread. On every platform this compiler
348            // targets that is a macro the library's headers read and one more library on the
349            // link line, and the library is added after the loop so that it lands after the
350            // objects that refer to it.
351            "-pthread" | "-pthreads" => {
352                opts.defines.push("_REENTRANT".to_owned());
353                threads = true;
354            }
355            "-ansi" => {
356                opts.std = Std::C89;
357                opts.gnu_extensions = false;
358            }
359            // `-Wpedantic` is the same flag under the name the `-W` family gives it, which is
360            // the spelling a build system that groups its warning flags tends to write.
361            "-pedantic" | "-Wpedantic" => opts.pedantic = true,
362            // Both directions, because a build that needs this for one directory turns it back
363            // off for the next one rather than leaving it on for the whole tree.
364            "-fpermissive" => opts.permissive = true,
365            "-fno-permissive" => opts.permissive = false,
366            "-ffreestanding" => opts.hosted = false,
367            "-fhosted" => opts.hosted = true,
368            "-fno-builtin" => opts.builtins = false,
369            "-fbuiltin" => opts.builtins = true,
370            // The C89 dialects are under GNU's reading whatever this says, so turning it off
371            // there is turning off something the dialect asked for, which is accepted and does
372            // nothing. gcc refuses that command line, and there is nothing it could have meant.
373            "-fgnu89-inline" => opts.gnu89_inline = true,
374            "-fno-gnu89-inline" => opts.gnu89_inline = false,
375            // Both directions of each, because a build system that wants one of these usually
376            // writes it beside the flag that turns it back off for one directory.
377            "-fno-omit-frame-pointer" => opts.frame_pointer = true,
378            "-fomit-frame-pointer" => opts.frame_pointer = false,
379            "-mno-red-zone" => opts.red_zone = false,
380            "-mred-zone" => opts.red_zone = true,
381            // GCC drops its own include directory along with the system ones, because its
382            // headers are half of a pair with the library's and half a pair is worse than
383            // none. A build that passes this is supplying the whole set itself.
384            "-nostdinc" => nostdinc = true,
385            "-o" => {
386                output = Some(args.get(i).ok_or_else(|| err("-o requires an argument"))?.clone());
387                i += 1;
388            }
389            // The flags that take a directory only in the separated form. GCC spells them
390            // this way and nothing writes `-iquotedir`, so accepting the joined form would
391            // mean guessing at a path that starts with the flag's own letters.
392            // Apple's spelling of `--sysroot`, and the one its own build systems pass. The
393            // two mean the same thing here: the configured directories are under there rather
394            // than under the root.
395            "-isysroot" => {
396                let dir = args.get(i).ok_or_else(|| err("-isysroot requires an argument"))?;
397                i += 1;
398                sysroot = Some(PathBuf::from(dir));
399            }
400            "-iquote" | "-isystem" | "-idirafter" => {
401                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
402                i += 1;
403                match arg {
404                    "-iquote" => opts.search.push_quote(dir.clone()),
405                    "-isystem" => opts.search.push_system(dir.clone()),
406                    _ => opts.search.push_after(dir.clone()),
407                }
408            }
409            "-iprefix" => {
410                iprefix = args.get(i).ok_or_else(|| err("-iprefix requires an argument"))?.clone();
411                i += 1;
412            }
413            // Where GCC puts these is not where its manual says it puts them, and this is the
414            // measured answer rather than the documented one: `-iwithprefix` lands in the
415            // `-isystem` slot and not the `-idirafter` slot, and `-iwithprefixbefore` lands in
416            // the `-I` slot. A cross build that uses them is relying on the behaviour, since
417            // that is the compiler it was developed against.
418            "-iwithprefix" | "-iwithprefixbefore" => {
419                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
420                i += 1;
421                let dir = format!("{iprefix}{dir}");
422                if arg == "-iwithprefix" {
423                    opts.search.push_system(dir);
424                } else {
425                    opts.search.push_bracket(dir);
426                }
427            }
428            "-include" | "-imacros" => {
429                let name = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
430                i += 1;
431                opts.preincludes
432                    .push(Preinclude { name: name.clone(), macros_only: arg == "-imacros" });
433            }
434            // The flag `-iquote` was introduced to replace, still passed by build systems old
435            // enough to predate the replacement. It is not a directory: it says that every `-I`
436            // so far is for quoted includes only, and that a quoted include stops looking next
437            // to the file that wrote it.
438            "-I-" => opts.search.split_quote_chain(),
439            "-x" => {
440                let lang = args.get(i).ok_or_else(|| err("-x requires an argument"))?;
441                i += 1;
442                forced = if lang == "none" {
443                    None
444                } else {
445                    Some(InputKind::from_x_arg(lang).map_err(|e| err(format!("{e}")))?)
446                };
447            }
448            // Not a GCC flag. spec/03-architecture.md section 3.5 compiles several
449            // translation units in one process rather than making the build system fork, and
450            // section 3.8's determinism check compares `-j1` against `-j16`, so the knob has
451            // to exist and has to be spelled the way `make` spells it.
452            // `-DFOO`, `-D FOO` and the same for `-U` and `-I`. Both forms are in wide use
453            // and a build system may produce either, so both are read here rather than
454            // being normalised by whatever generated the command line.
455            _ if arg.starts_with("-D") => {
456                let value = joined_or_next(arg, 2, args, &mut i)?;
457                opts.defines.push(value);
458            }
459            _ if arg.starts_with("-U") => {
460                let value = joined_or_next(arg, 2, args, &mut i)?;
461                opts.undefines.push(value);
462            }
463            _ if arg.starts_with("-I") => {
464                let dir = joined_or_next(arg, 2, args, &mut i)?;
465                opts.search.push_bracket(dir);
466            }
467            _ if arg.starts_with("-std=") => {
468                let name = &arg["-std=".len()..];
469                let (std, gnu) = Std::from_flag(name)
470                    .ok_or_else(|| err(format!("unknown dialect `{name}`, see --help")))?;
471                opts.std = std;
472                opts.gnu_extensions = gnu;
473            }
474            // Section 4.5. The claim decides which half of glibc's `sys/cdefs.h` we are
475            // handed, so a differential run that does not set it is comparing two compilers
476            // that believe they are different compilers.
477            // GCC packs these into one flag, so `-dDI` is two of them. Letters in the family
478            // that we have not written yet are accepted and ignored, because a dump is a
479            // debugging aid and a build that asks for one should still compile. A letter
480            // outside the family falls through to the unknown option error, which is what
481            // keeps `-dumpversion` from being read as a dump of nothing.
482            _ if Dumps::is_family(arg) => {
483                opts.dumps.add(&arg[2..]);
484            }
485            // One name at a time, which is what a build that means its own `memcpy` and the
486            // library's everything else writes. The name is not checked against a list, because
487            // the flag is about what the program means by a name and a program is allowed to mean
488            // something by a name this compiler has never heard of.
489            _ if arg.starts_with("-fno-builtin-") => {
490                opts.no_builtin.push(arg["-fno-builtin-".len()..].to_owned());
491            }
492            _ if arg.starts_with("-fgnuc-version=") => {
493                let v = &arg["-fgnuc-version=".len()..];
494                opts.gnuc = v.parse().map_err(err)?;
495            }
496            // spec/13-gnu-compat.md section 13.3 promises this flag an error that says why rather
497            // than the unknown option one, because a build reaching for it is asking for a feature
498            // and deserves to be told it is not coming rather than told the spelling is wrong.
499            // The negative form is what this compiler does anyway, so it is taken and dropped.
500            "-fnested-functions" => {
501                return Err(err(
502                    "nested functions are not supported: a call to one goes through a trampoline \
503                     written on the stack, which no target that enforces an unexecutable stack \
504                     allows",
505                ));
506            }
507            "-fno-nested-functions" => {}
508            // Which of the two links the output is for, which is a real difference and not a
509            // description of what happens anyway. Everything here is position independent either
510            // way, and what these decide is whether a name may be one another object defines or
511            // replaces, because a link that produces an executable puts every name in the same
512            // program and a link that produces a shared library does not.
513            //
514            // It matters that they are accepted at all, whatever they then do. Every autoconf and
515            // cmake build puts `-fPIC` on the compile line, so a compiler that rejects it cannot
516            // be the `CC` of a project that has a configure script, whatever else it can do. That
517            // is how this was found: building SQLite's test fixture stopped on it.
518            "-fPIC" | "-fpic" => opts.pic = Pic::Library,
519            // Not a synonym of the pair above, which is what they were treated as until #756. The
520            // library is the expensive answer and gcc makes it the one that has to be asked for,
521            // so this is also what nothing at all means.
522            "-fPIE" | "-fpie" => opts.pic = Pic::Executable,
523            // A different question from the pair above, and the one every distribution build of a
524            // shared library answers. `-fPIC` decides how an address is reached, and this decides
525            // whether the optimizer may believe a body it can see, because an exported name is one
526            // the dynamic linker may find another definition of first. On by default, which is
527            // gcc's arrangement and is the honest answer, and off is a promise the build makes and
528            // nothing checks.
529            "-fsemantic-interposition" => opts.interposition = true,
530            "-fno-semantic-interposition" => opts.interposition = false,
531            // The other direction is a request, not a description, and it is one this compiler
532            // cannot grant, so it gets the treatment section 13.3 asks for rather than the unknown
533            // option error. Answering it by carrying on would be answering a different question:
534            // the code would still be position independent, which is correct everywhere an
535            // ordinary program runs and is wrong in a kernel, where the flag is written precisely
536            // because there is no loader to fill a global offset table in.
537            "-fno-pic" | "-fno-pie" => {
538                return Err(err(
539                    "position dependent code is not supported: an address that may be in another \
540                     object is loaded out of the global offset table, and nothing here emits the \
541                     absolute form this asks for. Use -no-pie if what you meant was how to link",
542                ));
543            }
544            // Another description of what this compiler does. A file scope declaration with no
545            // initializer is written into `.bss` as an ordinary defined symbol, not offered to the
546            // linker as a common one for it to merge, which is what `-fno-common` asks for and what
547            // gcc has done by default since 10. Nothing in the front end produces `Linkage::Common`
548            // at all.
549            "-fno-common" => {}
550            // And the request, which is the one that cannot be granted. It is a real difference and
551            // not a preference: two files each writing `int g;` link under `-fcommon` and are a
552            // duplicate definition without it, which is the whole reason the flag survives.
553            "-fcommon" => {
554                return Err(err(
555                    "a tentative definition is written into .bss as its own symbol here, and \
556                     nothing emits the common symbol this asks the linker to merge. Give the \
557                     variable a definition in one file and declare it extern in the others",
558                ));
559            }
560            // Both directions of this one are taken, which is the exception to the rule above, and
561            // the reason is which way being wrong costs something.
562            //
563            // Nothing here derives anything from the type an object is accessed through. The IR has
564            // somewhere to put a type based aliasing node and lowering fills it with nothing on
565            // every access, so no pass has one to read and the alias analysis falls back to what it
566            // can see. That makes `-fno-strict-aliasing` a description, the way `-fPIC` is.
567            //
568            // `-fstrict-aliasing` is a request to assume more than that, and this compiler assumes
569            // less. Answering a request for a weaker guarantee by giving a stronger one is safe in
570            // a way the `-fno-pic` case is not: every program that is correct under the assumption
571            // is correct without it, only slower. And `-O2` implies it, so a build that spells it
572            // out is a build that would stop on a compiler that refused it, for no gain at all.
573            "-fstrict-aliasing" | "-fno-strict-aliasing" => {}
574            // About temporary files rather than about code. There is nothing between the phases of
575            // one compilation here to write to a file in the first place.
576            "-pipe" => {}
577            // Nothing here writes colour, so all of these are the same answer, and it is the answer
578            // that costs nothing: the diagnostics come out plain either way and no build depends on
579            // an escape sequence being there. Taken rather than refused because cmake writes
580            // `-fdiagnostics-color=always` on every compile line when the generator is ninja, which
581            // makes this the second most common flag after `-fPIC` to stop a build over a question
582            // about how the text looks.
583            "-fdiagnostics-color" | "-fno-diagnostics-color" => {}
584            _ if arg.starts_with("-fdiagnostics-color=") => {}
585            // The link flags. None of them changes the compilation, which is why they are
586            // collected apart from `opts` and why `-lm` on a `-c` line is a note rather than an
587            // error: it is a thing said to a linker that is not going to run.
588            "-static" => link.is_static = true,
589            "-shared" => link.shared = true,
590            "-pie" => link.pie = Some(true),
591            "-no-pie" | "-nopie" => link.pie = Some(false),
592            "-nostdlib" => link.no_stdlib = true,
593            "-nostartfiles" => link.no_startfiles = true,
594            "-nodefaultlibs" => link.no_defaultlibs = true,
595            "-fno-builtins-lib" => link.no_builtins_lib = true,
596            "-fbuiltins-lib" => link.no_builtins_lib = false,
597            "-rdynamic" | "-export-dynamic" => link.export_dynamic = true,
598            "-s" => link.strip = true,
599            "-Xlinker" => {
600                let next = args.get(i).ok_or_else(|| err("-Xlinker requires an argument"))?;
601                i += 1;
602                link.passthrough.push(next.clone());
603            }
604            _ if arg.starts_with("-Wl,") => {
605                // Commas separate arguments rather than being part of one, which is what makes
606                // `-Wl,-rpath,/opt/lib` two words to the linker and one word here.
607                link.passthrough.extend(arg["-Wl,".len()..].split(',').map(str::to_owned));
608            }
609            _ if arg.starts_with("-fuse-ld=") => {
610                link.use_ld = Some(arg["-fuse-ld=".len()..].to_owned());
611            }
612            _ if arg.starts_with("-l") && arg.len() > 2 => {
613                inputs.push(Input::library(&arg[2..]));
614            }
615            "-l" => {
616                let next = args.get(i).ok_or_else(|| err("-l requires an argument"))?;
617                i += 1;
618                inputs.push(Input::library(next));
619            }
620            _ if arg.starts_with("-L") => {
621                link.search.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
622            }
623            _ if arg.starts_with("-B") => {
624                link.prefixes.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
625            }
626            _ if arg.starts_with("-j") => {
627                jobs = Jobs::parse(&arg[2..]).map_err(err)?;
628            }
629            _ if arg.starts_with("--sysroot=") => {
630                sysroot = Some(PathBuf::from(&arg["--sysroot=".len()..]));
631            }
632            _ if arg.starts_with("--target=") => {
633                let t = &arg["--target=".len()..];
634                opts.target = t.parse().map_err(|e| err(format!("{e}")))?;
635            }
636            _ if arg.starts_with("--emit=") => {
637                let k = &arg["--emit=".len()..];
638                opts.emit = k
639                    .parse()
640                    .map_err(|()| err(format!("unknown --emit kind `{k}`, see --help")))?;
641            }
642            // A bare `-O` is `-O1`, which is what GCC has and what a hand written makefile tends
643            // to write. `-Og` is GCC's level for a build somebody is going to step through, and
644            // it is `-O1` with the transformations that move code around left out; this compiler
645            // has no such level yet, so it is the nearest one and `--print-pipeline` says what
646            // that came to rather than the flag pretending otherwise.
647            "-O" | "-Og" => opts.opt_level = rucc_session::OptLevel::O1,
648            // The union of `-O3` and `-ffast-math`, and the second half of that changes what
649            // floating point arithmetic means. Refused rather than taken as `-O3`, because a
650            // build that asks for fast math and is quietly given ordinary arithmetic gets a
651            // slower program than it asked for and a build that is given fast math it did not
652            // ask for gets a wrong one.
653            "-Ofast" => {
654                return Err(err(
655                    "-Ofast is -O3 with fast math, and fast math is not implemented, see \
656                     spec/04-driver-and-cli.md section 4.6",
657                ));
658            }
659            _ if arg.starts_with("-O") => {
660                opts.opt_level = arg[2..]
661                    .parse()
662                    .map_err(|()| err(format!("unknown optimization level `{arg}`")))?;
663            }
664            // What every name gets when nothing in the source said, which the attribute in the
665            // source overrides rather than the other way round. Before the optimizer's `-f`
666            // family below for the reason the tier below it is.
667            _ if arg.starts_with("-fvisibility=") => {
668                let seen = &arg["-fvisibility=".len()..];
669                opts.visibility = seen.parse().map_err(|()| {
670                    err(format!(
671                        "`{seen}` is not a visibility, which is default, hidden, internal or \
672                         protected"
673                    ))
674                })?;
675            }
676            // The memory safety monitor, from section 15.4 of
677            // `spec/safe-memory/15-integration.md`. Before the optimizer's `-f` family below,
678            // because a pass that took the name `safety=detect` would otherwise be handed the
679            // flag, and the tier is not a pass.
680            _ if arg.starts_with("-fsafety=") => {
681                let tier = &arg["-fsafety=".len()..];
682                opts.safety = tier.parse().map_err(|()| {
683                    err(format!(
684                        "`{tier}` is not a safety tier, which is off, detect, enforce or kernel"
685                    ))
686                })?;
687            }
688            // The optimizer's own flags, from section 9.10 of `spec/09-optimizer.md`. These come
689            // after every `-f` the rest of the compiler answers to, so a pass can never take a
690            // name that already means something else on the command line.
691            _ if arg.starts_with("-fpass-fuel=") => {
692                let (name, count) = arg["-fpass-fuel=".len()..]
693                    .split_once('=')
694                    .ok_or_else(|| err("-fpass-fuel= is spelled <pass>=<count>"))?;
695                if rucc_opt::pass::find(name).is_none() {
696                    return Err(err(format!(
697                        "`{name}` is not a pass this compiler has, see --print-pipeline"
698                    )));
699                }
700                let count: u32 = count
701                    .parse()
702                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
703                opts.pass_fuel.push((name.to_owned(), count));
704            }
705            _ if arg.starts_with("-fpass-fuel-global=") => {
706                let count = &arg["-fpass-fuel-global=".len()..];
707                let count: u32 = count
708                    .parse()
709                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
710                opts.pass_fuel_global = Some(count);
711            }
712            // Everything from `-fopt-info` to the end of the argument, which is optional
713            // keywords joined by hyphens and an optional `=<file>`. Checked here rather than
714            // where the remarks are printed, because by then the compilation somebody wanted
715            // to hear about is over.
716            _ if arg == "-fopt-info"
717                || arg.starts_with("-fopt-info=")
718                || arg.starts_with("-fopt-info-") =>
719            {
720                let rest = &arg["-fopt-info".len()..];
721                let (kinds, file) = match rest.split_once('=') {
722                    Some((kinds, file)) => (kinds, Some(file)),
723                    None => (rest, None),
724                };
725                let kinds = kinds.strip_prefix('-').unwrap_or(kinds);
726                rucc_opt::Wants::none().add(kinds).map_err(err)?;
727                opts.opt_info.push(kinds.to_owned());
728                if let Some(file) = file {
729                    if file.is_empty() {
730                        return Err(err("-fopt-info= was given no file to write to"));
731                    }
732                    opts.opt_info_file = Some(file.to_owned());
733                }
734            }
735            _ if arg.starts_with("-fdump-ir=") => {
736                // Checked here rather than where the dumps are taken, because the compilation
737                // that would have been dumped is over by then.
738                let spec = &arg["-fdump-ir=".len()..];
739                rucc_opt::Dumps::default().add(spec).map_err(err)?;
740                opts.dump_ir.push(spec.to_owned());
741            }
742            // Before the bare `-f<pass>` below, because a pass called `enable-something` would
743            // otherwise take the flag away from the gate. Checked here rather than where the
744            // pipeline reads it, for the reason that applies to all of these: a misspelled pass
745            // name that quietly gated nothing looks exactly like a pass that is not the guilty
746            // one, and a bisection would carry on past the thing it was looking for.
747            _ if arg.starts_with("-fdisable-") || arg.starts_with("-fenable-") => {
748                let on = arg.starts_with("-fenable-");
749                let spec = &arg[if on { "-fenable-".len() } else { "-fdisable-".len() }..];
750                rucc_opt::Gates::default().add(on, spec).map_err(err)?;
751                opts.pass_gates.push((on, spec.to_owned()));
752            }
753            _ if arg.strip_prefix("-fno-").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
754                opts.passes.push((arg["-fno-".len()..].to_owned(), false));
755            }
756            _ if arg.strip_prefix("-f").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
757                opts.passes.push((arg["-f".len()..].to_owned(), true));
758            }
759            // The unstable options, spelled the way rustc spells them and carrying the same
760            // promise, which is none: one of these may change or go away in any release. They are
761            // measurements and debugging aids rather than things a build asks for, which is why
762            // none of them is in the usage text and all of them are in section 4.11 of
763            // `spec/04-driver-and-cli.md`.
764            "-Zverify-each" => opts.verify_each = true,
765            _ if arg.starts_with("-Zrule-coverage=") => {
766                let file = &arg["-Zrule-coverage=".len()..];
767                if file.is_empty() {
768                    return Err(err("-Zrule-coverage= needs a file to write to"));
769                }
770                opts.rule_coverage = Some(file.to_owned());
771            }
772            _ if arg.starts_with("-Z") => {
773                return Err(err(format!(
774                    "`{arg}` is not an unstable option this compiler has, see \
775                     spec/04-driver-and-cli.md section 4.11 for the ones it does"
776                )));
777            }
778            // The word size, which is a statement about the target and is taken as one. A build
779            // that says the size the target already has is saying nothing, and one that says the
780            // other size is asking for a target this compiler does not have, which it is told
781            // rather than being given the wrong one.
782            "-m64" | "-m32" | "-mx32" => {
783                let want: u32 = match arg {
784                    "-m64" => 64,
785                    _ => 32,
786                };
787                let have = rucc_target::TargetInfo::new(opts.target).pointer_width;
788                if have != want {
789                    return Err(err(format!(
790                        "{arg} asks for a {want} bit target and {} is {have} bit, use \
791                         --target= to name the one you mean",
792                        opts.target
793                    )));
794                }
795            }
796            // Which processor in the family to generate for. This compiler emits the base
797            // instruction set of the architecture and nothing above it, so a program built with
798            // any of these runs on the machine that was named; it is a program that could have
799            // been faster rather than a program that is wrong, which is what makes these safe to
800            // take and ignore where a flag that changed the meaning of the code would not be.
801            _ if arg.starts_with("-march=")
802                || arg.starts_with("-mtune=")
803                || arg.starts_with("-mcpu=") => {}
804            // The calling convention, which is not safe to ignore. Taken when it names the one
805            // the target already uses and refused otherwise.
806            _ if arg.starts_with("-mabi=") => {
807                let want = &arg["-mabi=".len()..];
808                let have = match opts.target.arch {
809                    rucc_target::Arch::X86_64 => "sysv",
810                    rucc_target::Arch::Aarch64 => "lp64",
811                    rucc_target::Arch::Riscv64 => "lp64d",
812                };
813                if want != have {
814                    return Err(err(format!(
815                        "{arg}: {} uses the {have} convention and this compiler has no other",
816                        opts.target
817                    )));
818                }
819            }
820            // How far apart the pieces of the program may be. The small model is what we emit and
821            // it is every hosted program's default; the kernel model is a different one and a
822            // build that asks for it and does not get it links and then does not run.
823            "-mcmodel=small" => {}
824            _ if arg.starts_with("-mcmodel=") => {
825                return Err(err(format!(
826                    "{arg}: this compiler emits the small code model and no other, see \
827                     spec/12-targets.md"
828                )));
829            }
830            // GCC's own scripting language for how the driver builds a command line.
831            // `spec/04-driver-and-cli.md` section 4.4 settles that we will not have it, so a
832            // build reaching for it is told which flags do the same job.
833            _ if arg.starts_with("-specs=") => {
834                return Err(err(
835                    "-specs= is not supported: the parts of it builds rely on are -B, -L, \
836                     -nostdlib, -nostartfiles and -Wl,, see spec/04-driver-and-cli.md \
837                     section 4.4",
838                ));
839            }
840            // Arguments meant for a separate assembler or preprocessor, which this compiler does
841            // not have: both are inside it and neither reads a command line. Refused rather than
842            // dropped, because every one of these says something about the output and a build
843            // that asked for `-Wa,--noexecstack` and was silently given an executable stack got
844            // the opposite of what it asked for.
845            _ if arg.starts_with("-Wa,") || arg.starts_with("-Wp,") => {
846                return Err(err(format!(
847                    "`{arg}` is an argument for a separate assembler or preprocessor, and both \
848                     are inside this compiler rather than programs it runs"
849                )));
850            }
851            "-Xassembler" | "-Xpreprocessor" => {
852                return Err(err(format!(
853                    "{arg} hands an argument to a separate assembler or preprocessor, and both \
854                     are inside this compiler rather than programs it runs"
855                )));
856            }
857            // Everything else in the `-W` family. `spec/04-driver-and-cli.md` section 4.1 has
858            // this one as a rule about build systems rather than about warnings: autoconf finds
859            // out whether a warning flag exists by passing it and looking at the exit status, so
860            // a compiler that refuses one it has not heard of fails a configure script written
861            // for a GCC newer than itself. The names are not checked against a list because this
862            // compiler has no warning groups for a list to be of, which #485 is about.
863            _ if arg.starts_with("-W") => {}
864            // Flags that name something this compiler does not do and would not do differently
865            // if it did. `-fno-ident` is about a comment in the output that we do not write
866            // either way, and the others are about a way of ordering the compilation that has
867            // been GCC's only way for twenty years. Section 4.1 asks for the list to be short
868            // and for adding to it to be deliberate, which is why it is written out here.
869            "-fno-ident"
870            | "-fident"
871            | "-funit-at-a-time"
872            | "-fno-unit-at-a-time"
873            | "-shared-libgcc"
874            | "-static-libgcc" => {}
875            _ if arg.starts_with('-') && arg.len() > 1 => {
876                // Silently ignoring an unknown flag is how a build ends up not doing what
877                // its author asked. spec/13-gnu-compat.md section 13.4 makes this an error
878                // for the flags that change code generation, and the safe default until the
879                // flag table is populated is to reject everything we do not know.
880                return Err(err(format!("unknown option `{arg}`")));
881            }
882            _ => inputs.push(Input { path: arg.to_owned(), forced, library: false }),
883        }
884    }
885
886    // Last, so that it lands after every `-isystem` the command line gave. That is GCC's
887    // order: a directory the user names outranks the compiler's own, and the compiler's own
888    // outranks the library's. It is pushed after the loop rather than before it because
889    // `SearchPath` appends within a group and the position is what the order is.
890    // The same directory the headers were looked for under, because a sysroot is a statement
891    // about a whole installation and not about half of one.
892    link.sysroot = sysroot.clone();
893    // After the loop rather than where `-pthread` was read, so that it lands after the objects
894    // that refer to it. A static link takes the definitions it needs from a library when it
895    // reaches it and not afterwards, so a library before the objects is a library that answers
896    // nothing.
897    if threads {
898        inputs.push(Input::library("pthread"));
899    }
900    if let Some(query) = query {
901        return Ok(Action::Print(answer(&query, &opts, &link)));
902    }
903    // `-M` and `-MM` produce the rule and nothing else, so the run stops after phase 4 whatever
904    // else the command line asked for. Read here rather than where the flag was, because a `-c`
905    // written after it has to lose and the loop cannot know that until it has ended. The output
906    // file is where the rule goes rather than where an object would have gone, and the last
907    // phase being the preprocessor is what makes that true without a second rule for it.
908    if opts.deps.instead_of_compiling {
909        opts.emit = EmitKind::Preprocessed;
910    }
911    if !nostdinc {
912        opts.search.push_system(runtime::DIR);
913        // And the library's after ours, which is the other half of the same order. They go on
914        // here rather than at the point `--target=` or `--sysroot=` was read because either
915        // one changes the answer and the last word on both is the end of the loop.
916        for dir in library::system_dirs(opts.target, sysroot.as_deref()) {
917            opts.search.push_system(dir);
918        }
919    }
920    // Once, here, rather than as each directory is pushed. A `-I` that names a system
921    // directory has to lose to the system entry and the system entry is added last, so the
922    // question cannot be answered until the whole path is known.
923    opts.search.remove_duplicates();
924
925    // The target has to be resolved before the configuration is printed, so this check comes
926    // after the loop rather than at the point `--print-config` was seen.
927    if print_config {
928        return Ok(Action::PrintConfig(Box::new(opts)));
929    }
930    if print_pipeline {
931        return Ok(Action::PrintPipeline(Box::new(opts)));
932    }
933    let plan = Plan::new(&opts, &inputs, output.as_deref()).map_err(|e| err(e.message))?;
934    if print_plan {
935        return Ok(Action::PrintPlan {
936            opts: Box::new(opts),
937            plan: Box::new(plan),
938            link: Box::new(link),
939        });
940    }
941    Ok(Action::Compile {
942        opts: Box::new(opts),
943        plan: Box::new(plan),
944        link: Box::new(link),
945        jobs,
946        verbose,
947    })
948}
949
950/// What one of the `-dump` and `-print` flags prints.
951///
952/// GCC prints the name back unchanged when it cannot find the file a `-print` flag asked about,
953/// which is what makes the answer safe to paste into a link line whether or not the file is
954/// there, and this does the same.
955fn answer(query: &Query, opts: &Options, link: &LinkOptions) -> String {
956    let found = |name: &str| {
957        link::find_in_search(link, opts.target, name)
958            .map_or_else(|| name.to_owned(), |path| path.display().to_string())
959    };
960    match query {
961        Query::Machine => opts.target.to_string(),
962        Query::Version => VERSION.to_owned(),
963        Query::Multiarch => link::multiarch(opts.target),
964        // The three lines GCC prints, in its order and with its punctuation, because what reads
965        // them is a script written against that shape. There is no installation directory to
966        // report: this compiler is one binary that works wherever it is copied, and the headers
967        // it ships are inside it, so `install` is where the binary is and nothing is under it.
968        Query::SearchDirs => {
969            let here = std::env::current_exe()
970                .ok()
971                .and_then(|p| p.parent().map(std::path::Path::to_path_buf))
972                .unwrap_or_default();
973            let list = |dirs: &[PathBuf]| {
974                dirs.iter().map(|d| d.display().to_string()).collect::<Vec<_>>().join(":")
975            };
976            let libraries = link::search_dirs(link, opts.target);
977            format!(
978                "install: {}\nprograms: ={}\nlibraries: ={}",
979                here.display(),
980                list(&link.prefixes),
981                list(&libraries)
982            )
983        }
984        Query::FileName(name) => found(name),
985        // The name GCC gives the library of routines a compiler's output calls that the C
986        // library does not have. Ours is built in and there is no file, so the answer is the
987        // name itself, which is what GCC prints when it cannot find one either.
988        Query::Libgcc => found("libgcc.a"),
989        // A program rather than a library: the linker and the archiver are the ones a build asks
990        // about, and this compiler finds them on the path or under `-B` rather than shipping
991        // them, so the name back is the honest answer unless a `-B` prefix holds one.
992        Query::ProgName(name) => link
993            .prefixes
994            .iter()
995            .map(|dir| dir.join(name))
996            .find(|path| path.is_file())
997            .map_or_else(|| name.clone(), |path| path.display().to_string()),
998    }
999}
1000
1001/// Renders the passes this level will run, in order, with what each one does.
1002///
1003/// The level is the whole of the answer unless a `-f` flag edited it, which is section 9.1 of
1004/// `spec/09-optimizer.md`: a level is a list somebody wrote down rather than something that
1005/// emerges from which flags happen to be set, and this is how that list is read.
1006#[must_use]
1007pub fn print_pipeline(opts: &Options) -> String {
1008    let mut settings = rucc_opt::Options::for_level(opts.opt_level);
1009    settings.toggles.clone_from(&opts.passes);
1010    settings.global_fuel = opts.pass_fuel_global;
1011    for (on, spec) in &opts.pass_gates {
1012        // Every spelling was checked while the arguments were parsed, so there is nothing here
1013        // this can refuse, and a listing is not the place to report it if there were.
1014        let _ = settings.gates.add(*on, spec);
1015    }
1016    rucc_opt::pipeline::print(&settings)
1017}
1018
1019/// Renders the resolved configuration.
1020///
1021/// One `key: value` per line, sorted by nothing in particular but fixed in order, because
1022/// this output is diffed across hosts in CI and a reordering would read as a change.
1023#[must_use]
1024pub fn print_config(opts: &Options) -> String {
1025    let sess = Session::new(opts.clone());
1026    let t = &sess.target;
1027    let mut out = String::new();
1028    let _ = writeln!(out, "version: {VERSION}");
1029    // The three field triple the driver was given rather than the ten field tuple it widens to,
1030    // because this output is what a build system reads to find out what it asked for. The tuple is
1031    // the compiler's model of the machine and this line is a receipt for a command line.
1032    let _ = writeln!(out, "target: {}", opts.target);
1033    let _ = writeln!(out, "arch: {}", opts.target.arch.as_str());
1034    let _ = writeln!(out, "os: {}", opts.target.os.as_str());
1035    let _ = writeln!(out, "env: {}", opts.target.env.as_str());
1036    let _ = writeln!(out, "object-format: {}", t.object_format.as_str());
1037    let _ = writeln!(out, "pointer-width: {}", t.pointer_width);
1038    let _ = writeln!(out, "long-width: {}", t.long_width);
1039    let _ = writeln!(out, "long-double-width: {}", t.long_double_width);
1040    let _ = writeln!(out, "endian: {}", if t.little_endian { "little" } else { "big" });
1041    let _ = writeln!(out, "char-signed: {}", t.char_is_signed);
1042    let _ = writeln!(out, "va-list: {}", t.va_list.map_or("none", |list| list.as_str()));
1043    // The register file as a count per class, which is enough to tell a target whose registers
1044    // are described from one whose are not without printing sixteen names nobody asked for.
1045    let regs: Vec<String> = t
1046        .regs
1047        .classes()
1048        .map(|(class, info)| format!("{} {}", info.name, t.regs.len(class)))
1049        .collect();
1050    let _ = writeln!(
1051        out,
1052        "registers: {}",
1053        if regs.is_empty() { "none".to_string() } else { regs.join(", ") }
1054    );
1055    let _ = writeln!(out, "opt-level: {}", sess.opts.opt_level);
1056    let _ = writeln!(out, "safety: {}", sess.opts.safety);
1057    let _ = writeln!(out, "emit: {}", sess.opts.emit.as_str());
1058    let _ = writeln!(out, "debug-info: {}", sess.opts.debug_info);
1059    let _ = writeln!(out, "frame-pointer: {}", sess.opts.frame_pointer);
1060    let _ = writeln!(out, "red-zone: {}", sess.opts.red_zone);
1061    // Last because it is the one key with more than one line under it, and the only one
1062    // whose value is a property of the machine rather than of the command line.
1063    for dir in sess.opts.search.dirs() {
1064        let system = if dir.is_system { " (system)" } else { "" };
1065        let _ = writeln!(out, "include: {}{system}", dir.path.display());
1066    }
1067    out
1068}
1069
1070/// The output name the make target is taken from, which is the `-o` argument or nothing.
1071///
1072/// A run that stops at the preprocessor has not named an object, whatever its `-o` says: under
1073/// `-E` that argument is the preprocessed text and under `-M` it is the rule itself, and neither
1074/// is a file `make` would rebuild by running this rule. GCC agrees and falls back to the source
1075/// name in both, which is why a `-MD -E -o out.i` writes `out.d` holding a rule for `a.o`. From
1076/// `-S` on the argument does name what the rule builds, and it is used as written.
1077fn deps_target_output<'a>(opts: &Options, plan: &'a Plan) -> Option<&'a str> {
1078    if opts.emit == EmitKind::Preprocessed { None } else { plan.output.as_deref() }
1079}
1080
1081/// Writes to a path the command line named rather than one the plan derived, where `-` is
1082/// standard output.
1083fn write_named(path: &str, bytes: &[u8]) -> Result<(), String> {
1084    if path == "-" {
1085        return write_out(&Output::Stdout, bytes);
1086    }
1087    write_out(&Output::File(path.to_owned()), bytes)
1088}
1089
1090/// Writes the make rule for one input, and reports whether it got there.
1091///
1092/// A rule with no file of its own goes where the compilation it replaced would have written,
1093/// which is what makes the usual makefile recipe work: `rucc -M $< -o $@` leaves the rule in
1094/// `$@`, and the same line with the `-o` left off puts it on standard output.
1095fn write_deps(
1096    opts: &Options,
1097    plan: &Plan,
1098    job: &Job,
1099    found: &[Dependency],
1100    stderr: &mut impl std::io::Write,
1101) -> bool {
1102    let targets = if opts.deps.targets.is_empty() {
1103        vec![deps::default_target(&job.input, deps_target_output(opts, plan))]
1104    } else {
1105        opts.deps.targets.clone()
1106    };
1107    let rule = deps::rule(&opts.deps, &targets, &job.input, found);
1108    // The file, on the other hand, is named after the `-o` in every mode that still has one to
1109    // spend, which is every mode except the two that spend it on the rule.
1110    let wrote = match deps::default_file(&opts.deps, &job.input, plan.output.as_deref()) {
1111        // A `-MF` on a run that had nowhere else to put the rule leaves the file the `-o`
1112        // named empty rather than absent, because a makefile that named it as a target of its
1113        // own is a makefile that will look for it.
1114        Some(path) => write_named(&path, rule.as_bytes()).and_then(|()| {
1115            if opts.deps.instead_of_compiling { write_out(&job.output, b"") } else { Ok(()) }
1116        }),
1117        None => write_out(&job.output, rule.as_bytes()),
1118    };
1119    if let Err(e) = wrote {
1120        let _ = writeln!(stderr, "rucc: error: {e}");
1121        return false;
1122    }
1123    true
1124}
1125
1126/// Runs phase 4 over every input that has one, and writes what came out.
1127///
1128/// One input that fails does not stop the others. A build that reports every file it could
1129/// not preprocess in one run is worth more than one that stops at the first, and the exit
1130/// status is still a failure either way.
1131fn preprocess_all(opts: &Options, plan: &Plan) -> i32 {
1132    let fs = OsFileSystem::new();
1133    let mut stderr = std::io::stderr().lock();
1134    let mut failed = false;
1135    for job in &plan.jobs {
1136        if !job.phases.first().is_some_and(|p| *p == Phase::Preprocess) {
1137            // An input that is already preprocessed, or an object file. GCC passes these
1138            // through untouched, and the plan has already said so in its notes.
1139            continue;
1140        }
1141        let started = std::time::Instant::now();
1142        let result = preprocess(opts, &job.input, &fs);
1143        if opts.time {
1144            say_time(&job.input, started.elapsed(), &mut stderr);
1145        }
1146        for message in &result.messages {
1147            let _ = writeln!(stderr, "{message}");
1148        }
1149        if result.failed() {
1150            failed = true;
1151            continue;
1152        }
1153        if opts.deps.emit {
1154            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1155            // `-M` and `-MM` asked for the rule instead of the text, so there is nothing else
1156            // to write. The other two asked for both and fall through to the text below.
1157            if opts.deps.instead_of_compiling {
1158                continue;
1159            }
1160        }
1161        if let Err(e) = write_out(&job.output, result.text.as_bytes()) {
1162            let _ = writeln!(stderr, "rucc: error: {e}");
1163            failed = true;
1164        }
1165    }
1166    i32::from(failed)
1167}
1168
1169/// Runs the front end over every input that has a compile phase, and writes what came out.
1170///
1171/// The same rule as [`preprocess_all`]: one input that fails does not stop the others, and the
1172/// exit status is a failure either way. An input that is already assembly or an object has no
1173/// compile phase and is passed over here, which the plan has already said in its notes.
1174fn compile_all(opts: &Options, plan: &Plan) -> i32 {
1175    let fs = OsFileSystem::new();
1176    let mut stderr = std::io::stderr().lock();
1177    let mut failed = false;
1178    let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1179    failed |= !ok;
1180    let mut fired = Fired::new();
1181    for job in &plan.jobs {
1182        if !job.phases.contains(&Phase::Compile) {
1183            continue;
1184        }
1185        // An input of IR is read back rather than compiled, since the C it came from is not
1186        // here any more. Everything after this is the same, so the two paths meet again at the
1187        // messages and the file the result is written to.
1188        let started = std::time::Instant::now();
1189        let result = if job.kind == InputKind::Ir {
1190            compile_ir(opts, &job.input, &fs)
1191        } else {
1192            compile(opts, &job.input, &fs)
1193        };
1194        if opts.time {
1195            say_time(&job.input, started.elapsed(), &mut stderr);
1196        }
1197        fired.merge(&result.fired);
1198        failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1199        failed |= !remarks.write(&result.remarks, &mut stderr);
1200        for message in &result.messages {
1201            let _ = writeln!(stderr, "{message}");
1202        }
1203        // Before the failure below, because a compilation that stopped in the back end is exactly
1204        // the one whose preprocessed source somebody wants to look at.
1205        failed |= !write_temps(job, &result.temps, &mut stderr);
1206        if result.failed() {
1207            failed = true;
1208            continue;
1209        }
1210        // `-MD` and `-MMD` write the rule beside the object and let the compilation happen, so
1211        // this is the one path where both files come out of the same run. An input of IR has no
1212        // dependencies to report and produces an empty list, which produces a rule naming only
1213        // itself, and that is the honest answer rather than a missing file.
1214        if opts.deps.emit {
1215            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1216        }
1217        if let Err(e) = write_out(&job.output, result.artifact.bytes()) {
1218            let _ = writeln!(stderr, "rucc: error: {e}");
1219            failed = true;
1220        }
1221    }
1222    failed |= !write_coverage(opts, &fired, &mut stderr);
1223    i32::from(failed)
1224}
1225
1226/// A directory for the object files only the link step ever sees, removed when it goes away.
1227///
1228/// `-c` writes its object where the user can see it and linking does not, which is the whole of
1229/// the difference: a `rucc a.c b.c` leaves an executable behind and nothing else, the same as
1230/// every other compiler. Removing them on drop rather than at the end of a function is so that a
1231/// link that failed leaves nothing behind either.
1232struct Scratch {
1233    /// Where the objects go.
1234    dir: PathBuf,
1235}
1236
1237impl Scratch {
1238    /// Makes one, under whatever the platform calls its temporary directory.
1239    ///
1240    /// The name carries the process id so that two compilers running at once do not share a
1241    /// directory, which they would otherwise do the moment two of them compiled a file of the
1242    /// same name.
1243    fn new() -> Result<Scratch, String> {
1244        let dir = std::env::temp_dir().join(format!("rucc-{}", std::process::id()));
1245        std::fs::create_dir_all(&dir).map_err(|e| format!("{}: {e}", dir.display()))?;
1246        Ok(Scratch { dir })
1247    }
1248}
1249
1250impl Drop for Scratch {
1251    fn drop(&mut self) {
1252        let _ = std::fs::remove_dir_all(&self.dir);
1253    }
1254}
1255
1256/// The link line the plan describes, for `-###`.
1257///
1258/// The names in it are the hints the plan carries rather than the temporaries a real compilation
1259/// would choose, because `-###` prints the line without having compiled anything and so has
1260/// nothing to point at. That also makes the printed line readable rather than naming a directory
1261/// that only exists while a compilation is running.
1262fn link_line(opts: &Options, link: &LinkOptions, job: &LinkJob) -> Result<String, link::Error> {
1263    let linker = link::find(opts.target, link)?;
1264    let args = link::line(opts.target, link, &job.inputs, &job.output)?;
1265    Ok(link::render(&linker, &args))
1266}
1267
1268/// Compiles everything, then links it.
1269///
1270/// The objects go in a directory that is removed afterwards, which is why this is not
1271/// [`compile_all`] followed by a link: the plan says an object feeding the linker is temporary
1272/// and does not say where, because where is a question that only has an answer once something is
1273/// running.
1274fn link_all(opts: &Options, plan: &Plan, link: &LinkOptions, verbose: bool) -> i32 {
1275    let Some(job) = &plan.link else {
1276        // Every path into here comes from a plan whose last phase is the link, and such a plan
1277        // has a link job. Saying so is cheaper than an unwrap that would have to be explained.
1278        let mut stderr = std::io::stderr().lock();
1279        let _ = writeln!(stderr, "rucc: error: there is nothing to link");
1280        return 1;
1281    };
1282    // Before anything is compiled, because a linker that is not on the machine is worth knowing
1283    // about in the second it takes to look rather than after the compilation.
1284    let linker = match link::find(opts.target, link) {
1285        Ok(linker) => linker,
1286        Err(why) => return complain(why),
1287    };
1288
1289    let scratch = match Scratch::new() {
1290        Ok(scratch) => scratch,
1291        Err(why) => return complain(format!("could not make a place for the object files: {why}")),
1292    };
1293
1294    let fs = OsFileSystem::new();
1295    let mut failed = false;
1296    // One per job, in job order, which is what lets the link line below be rebuilt with the real
1297    // paths in it: every job contributes exactly one file to the line and does so in this order.
1298    let mut produced: Vec<String> = Vec::with_capacity(plan.jobs.len());
1299    let mut fired = Fired::new();
1300    {
1301        let mut stderr = std::io::stderr().lock();
1302        let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1303        failed |= !ok;
1304        for (at, job) in plan.jobs.iter().enumerate() {
1305            let out = match &job.output {
1306                Output::Temporary(hint) => {
1307                    // The index because two inputs in different directories can have the same
1308                    // name, and the two objects of `rucc a/x.c b/x.c` must not be one file.
1309                    scratch.dir.join(format!("{at}-{hint}")).display().to_string()
1310                }
1311                Output::File(path) => path.clone(),
1312                // A job feeding the linker never writes to standard output, since the plan gives
1313                // it a temporary. This is here so that the match is total rather than a panic.
1314                Output::Stdout => continue,
1315            };
1316            produced.push(out.clone());
1317            if !job.phases.contains(&Phase::Compile) {
1318                continue;
1319            }
1320            let started = std::time::Instant::now();
1321            let result = if job.kind == InputKind::Ir {
1322                compile_ir(opts, &job.input, &fs)
1323            } else {
1324                compile(opts, &job.input, &fs)
1325            };
1326            if opts.time {
1327                say_time(&job.input, started.elapsed(), &mut stderr);
1328            }
1329            fired.merge(&result.fired);
1330            failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1331            failed |= !remarks.write(&result.remarks, &mut stderr);
1332            for message in &result.messages {
1333                let _ = writeln!(stderr, "{message}");
1334            }
1335            failed |= !write_temps(job, &result.temps, &mut stderr);
1336            if result.failed() {
1337                failed = true;
1338                continue;
1339            }
1340            // A `-MD` on a command line that links writes the rule next to the executable and
1341            // names the executable as its target, since that is the file this source builds
1342            // here. The object it went through is in a temporary directory and is gone by the
1343            // time `make` reads any of this.
1344            if opts.deps.emit {
1345                failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1346            }
1347            if !matches!(result.artifact, Artifact::Object(_)) {
1348                // Worth saying rather than writing whatever it is and letting the linker read it.
1349                // An empty file is a valid empty linker script, so a link handed one gets as far
1350                // as reporting every symbol of this file undefined, which is a page of messages
1351                // about something that went wrong here.
1352                let _ = writeln!(
1353                    stderr,
1354                    "rucc: internal error: {}: no object file was produced for the link",
1355                    job.input
1356                );
1357                failed = true;
1358                continue;
1359            }
1360            if let Err(e) = std::fs::write(&out, result.artifact.bytes()) {
1361                let _ = writeln!(stderr, "rucc: error: {out}: {e}");
1362                failed = true;
1363            }
1364        }
1365        failed |= !write_coverage(opts, &fired, &mut stderr);
1366    }
1367    if failed {
1368        // Nothing is linked from a compilation that did not finish. A linker run over the objects
1369        // that did compile would report every function of the file that did not as undefined,
1370        // which is a page of messages about a mistake already reported once.
1371        return 1;
1372    }
1373
1374    // The items in command line order with the temporaries filled in. A library contributes no
1375    // job and passes through, and every file item takes the next job's real output, which is
1376    // what keeps a library that was written between two objects between them here.
1377    let mut outputs = produced.into_iter();
1378    let mut items = Vec::with_capacity(job.inputs.len());
1379    for item in &job.inputs {
1380        match item {
1381            link::Item::Library(name) => items.push(link::Item::Library(name.clone())),
1382            link::Item::File(_) => match outputs.next() {
1383                Some(path) => items.push(link::Item::File(path)),
1384                None => return complain("the plan asks the linker for a file nothing produced"),
1385            },
1386        }
1387    }
1388
1389    let args = match link::line(opts.target, link, &items, &job.output) {
1390        Ok(args) => args,
1391        Err(why) => return complain(why),
1392    };
1393    if verbose {
1394        let mut stderr = std::io::stderr().lock();
1395        let _ = writeln!(stderr, "{}", link::render(&linker, &args));
1396    }
1397    let started = std::time::Instant::now();
1398    let ran = link::run(&linker, &args);
1399    if opts.time {
1400        // The one step of a compilation that really is another program, so this line is the same
1401        // measurement gcc's is and names the linker the way gcc names `collect2`.
1402        let mut stderr = std::io::stderr().lock();
1403        say_time(&linker.name, started.elapsed(), &mut stderr);
1404    }
1405    match ran {
1406        Ok(()) => 0,
1407        // The linker has already said what was wrong on its own error output, and repeating that
1408        // linking failed would only push its message further up the screen.
1409        Err(link::Error::Refused { .. }) => 1,
1410        Err(why) => complain(why),
1411    }
1412}
1413
1414/// Prints one driver level message and gives back the exit status that goes with it.
1415fn complain(why: impl std::fmt::Display) -> i32 {
1416    let mut stderr = std::io::stderr().lock();
1417    let _ = writeln!(stderr, "rucc: error: {why}");
1418    1
1419}
1420
1421/// Writes what `-Zrule-coverage=FILE` asked for, and says whether it could.
1422///
1423/// Once for the whole command line rather than once per input, because the question is which
1424/// lowering rules this run of the compiler reached and a file per input would leave the reader
1425/// unioning files to find out something one process already knew.
1426///
1427/// A file that could not be written is a failure and not a warning. What asks for this is a
1428/// measurement run, and a measurement that quietly did not happen is worse than one that stopped.
1429fn write_coverage(opts: &Options, fired: &Fired, stderr: &mut impl std::io::Write) -> bool {
1430    let Some(path) = &opts.rule_coverage else { return true };
1431    let Some(table) = coverage::table(opts.target.arch) else {
1432        let _ = writeln!(
1433            stderr,
1434            "rucc: error: there are no lowering rules for {} yet, so there is no coverage of them \
1435             to report",
1436            opts.target
1437        );
1438        return false;
1439    };
1440    match std::fs::write(path, fired.listing(table)) {
1441        Ok(()) => true,
1442        Err(e) => {
1443            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1444            false
1445        }
1446    }
1447}
1448
1449/// Where the `-fopt-info` remarks go, and how much of the run has already gone there.
1450///
1451/// Standard error by default, and one file for the whole run when `-fopt-info=<file>` named one.
1452/// A file rather than the diagnostic stream is what a harness wants: the corpus in
1453/// `tamnd/rucc-corpus` matches a rejection against what the compiler said on standard error, and
1454/// a few thousand remarks mixed into that would bury it.
1455struct Remarks {
1456    /// The file, if there is one.
1457    file: Option<String>,
1458    /// Whether anything has been written to it yet, which decides between truncating and
1459    /// appending. One file holds the whole run rather than the last input in it.
1460    started: bool,
1461}
1462
1463impl Remarks {
1464    /// Prepares the destination, emptying the file if there is one.
1465    ///
1466    /// Emptied here rather than at the first remark, because a run where no pass had anything to
1467    /// say should leave an empty file and not yesterday's. An absent file and an empty one are
1468    /// different facts and something reading this will act on the difference.
1469    fn new(file: Option<&String>, stderr: &mut impl std::io::Write) -> (Self, bool) {
1470        let mut ok = true;
1471        if let Some(path) = file {
1472            if let Err(e) = std::fs::write(path, "") {
1473                let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1474                ok = false;
1475            }
1476        }
1477        (Self { file: file.cloned(), started: false }, ok)
1478    }
1479
1480    /// Writes one input's remarks, and says whether that worked.
1481    ///
1482    /// A file that cannot be written is a failure and not a warning, for the reason
1483    /// [`write_dumps`] gives: remarks that quietly did not arrive look exactly like a compilation
1484    /// where nothing happened.
1485    fn write(&mut self, text: &str, stderr: &mut impl std::io::Write) -> bool {
1486        if text.is_empty() {
1487            return true;
1488        }
1489        let Some(path) = &self.file else {
1490            let _ = write!(stderr, "{text}");
1491            return true;
1492        };
1493        let opened = std::fs::OpenOptions::new()
1494            .write(true)
1495            .append(self.started)
1496            .truncate(!self.started)
1497            .create(true)
1498            .open(path);
1499        self.started = true;
1500        let result =
1501            opened.and_then(|mut file| std::io::Write::write_all(&mut file, text.as_bytes()));
1502        if let Err(e) = result {
1503            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1504            return false;
1505        }
1506        true
1507    }
1508}
1509
1510/// Writes what `-fdump-ir=` asked to see, one file per dump.
1511///
1512/// The name is the input file with the dump's own name and `.ir` after it, so a directory listing
1513/// after a run is the passes in the order they ran, per input. They go in the working directory
1514/// rather than beside the output, because a dump is something a person asked for at a prompt and
1515/// the working directory is where that person is.
1516///
1517/// A file that could not be written is a failure and not a warning, for the reason
1518/// [`write_coverage`] gives: what asked for this is somebody debugging a pass, and a dump that
1519/// quietly did not happen looks exactly like a pass that did not run.
1520fn write_dumps(input: &str, dumps: &[rucc_opt::Dump], stderr: &mut impl std::io::Write) -> bool {
1521    let stem = std::path::Path::new(input)
1522        .file_name()
1523        .map_or_else(|| input.to_owned(), |name| name.to_string_lossy().into_owned());
1524    let mut ok = true;
1525    for dump in dumps {
1526        let path = format!("{stem}.{}.ir", dump.name);
1527        if let Err(e) = std::fs::write(&path, &dump.text) {
1528            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1529            ok = false;
1530        }
1531    }
1532    ok
1533}
1534
1535/// Writes the files `-save-temps` kept, which is nothing at all unless it was given.
1536///
1537/// A file that could not be written is a failure rather than a warning, for the reason
1538/// [`write_dumps`] gives: somebody asked for these by name, and one that quietly did not happen
1539/// looks like a compilation that never went through that step.
1540fn write_temps(job: &Job, temps: &Temps, stderr: &mut impl std::io::Write) -> bool {
1541    let mut ok = true;
1542    let kept = [(job.saved_text(), &temps.preprocessed), (job.saved_asm(), &temps.assembly)];
1543    for (path, text) in kept {
1544        // A step the compilation did not reach has nothing to keep, and a job that is not keeping
1545        // that step has nowhere to put it. Either way there is no file here.
1546        let (Some(path), Some(text)) = (path, text) else { continue };
1547        if let Err(e) = std::fs::write(&path, text) {
1548            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1549            ok = false;
1550        }
1551    }
1552    ok
1553}
1554
1555/// One line of `-time`, which is what a step was called and how long it took.
1556///
1557/// GCC's two numbers are the user and the system time of a subprocess it ran. This compiler runs
1558/// no subprocess for anything but the link, so what is measured here is the wall clock of the
1559/// step and the second column is always zero. The shape of the line is kept because a person
1560/// reading it next to gcc's should not have to work out which column is which.
1561fn say_time(name: &str, took: std::time::Duration, stderr: &mut impl std::io::Write) {
1562    let _ = writeln!(stderr, "# {name} {:.2} {:.2}", took.as_secs_f64(), 0.0);
1563}
1564
1565/// Writes one job's result where the plan said it goes.
1566///
1567/// # Errors
1568///
1569/// Returns the message to print, which names the file when there is one, because "permission
1570/// denied" on its own does not say which file was refused.
1571fn write_out(output: &Output, bytes: &[u8]) -> Result<(), String> {
1572    match output {
1573        Output::Stdout => {
1574            let mut stdout = std::io::stdout().lock();
1575            stdout.write_all(bytes).map_err(|e| format!("writing to standard output: {e}"))
1576        }
1577        Output::File(path) | Output::Temporary(path) => {
1578            std::fs::write(path, bytes).map_err(|e| format!("{path}: {e}"))
1579        }
1580    }
1581}
1582
1583/// Runs the driver and returns the process exit code.
1584///
1585/// `args` excludes the program name. Output goes to `stdout` and errors to `stderr`, which
1586/// is the one place in the compiler that is true.
1587pub fn run(args: &[String]) -> i32 {
1588    match parse_args(args) {
1589        Ok(Action::Help) => {
1590            print!("{USAGE}");
1591            0
1592        }
1593        Ok(Action::Version) => {
1594            println!("rucc {VERSION}");
1595            0
1596        }
1597        Ok(Action::Print(line)) => {
1598            println!("{line}");
1599            0
1600        }
1601        Ok(Action::PrintConfig(opts)) => {
1602            print!("{}", print_config(&opts));
1603            0
1604        }
1605        Ok(Action::PrintPipeline(opts)) => {
1606            print!("{}", print_pipeline(&opts));
1607            0
1608        }
1609        Ok(Action::PrintPlan { opts, plan, link }) => {
1610            print!("{}", plan.render());
1611            // The line as it would be typed, which is the half of `-###` that section 4.3 says
1612            // arrives with the link. It is printed even when the linker is not on this machine,
1613            // because what a build wants from `-###` is what the compiler would do.
1614            if let Some(job) = &plan.link {
1615                match link_line(&opts, &link, job) {
1616                    Ok(line) => println!("{line}"),
1617                    Err(why) => {
1618                        let mut stderr = std::io::stderr().lock();
1619                        let _ = writeln!(stderr, "rucc: error: {why}");
1620                        return 1;
1621                    }
1622                }
1623            }
1624            0
1625        }
1626        Ok(Action::Compile { opts, plan, link, jobs, verbose }) => {
1627            {
1628                let mut stderr = std::io::stderr().lock();
1629                if verbose {
1630                    let _ = write!(stderr, "{}", plan.render());
1631                    let _ = writeln!(stderr, "workers: {}", jobs.count());
1632                }
1633            }
1634            if opts.emit == EmitKind::Preprocessed {
1635                return preprocess_all(&opts, &plan);
1636            }
1637            if opts.emit != EmitKind::Executable {
1638                return compile_all(&opts, &plan);
1639            }
1640            link_all(&opts, &plan, &link, verbose)
1641        }
1642        Err(e) => {
1643            let mut stderr = std::io::stderr().lock();
1644            let _ = writeln!(stderr, "rucc: error: {e}");
1645            let _ = writeln!(stderr, "rucc: note: run `rucc --help` for usage");
1646            1
1647        }
1648    }
1649}
1650
1651#[cfg(test)]
1652mod tests {
1653    use rucc_session::{GnucVersion, IncludeForm, OptLevel, Visibility};
1654
1655    use super::*;
1656
1657    fn args(s: &[&str]) -> Vec<String> {
1658        s.iter().map(|x| (*x).to_owned()).collect()
1659    }
1660
1661    #[test]
1662    fn help_and_version_win_over_everything_else() {
1663        assert_eq!(parse_args(&args(&["-c", "--help", "x.c"])).unwrap(), Action::Help);
1664        assert_eq!(parse_args(&args(&["--version"])).unwrap(), Action::Version);
1665    }
1666
1667    fn compile(s: &[&str]) -> (Box<Options>, Box<Plan>) {
1668        match parse_args(&args(s)).expect("expected a compilation") {
1669            Action::Compile { opts, plan, .. } => (opts, plan),
1670            other => panic!("expected a compilation, got {other:?}"),
1671        }
1672    }
1673
1674    fn linking(s: &[&str]) -> (Box<LinkOptions>, Box<Plan>) {
1675        match parse_args(&args(s)).expect("expected a compilation") {
1676            Action::Compile { link, plan, .. } => (link, plan),
1677            other => panic!("expected a compilation, got {other:?}"),
1678        }
1679    }
1680
1681    #[test]
1682    fn collects_inputs_and_flags() {
1683        let (opts, plan) = compile(&["-c", "-O2", "-g", "a.c", "b.c"]);
1684        let paths: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
1685        assert_eq!(paths, vec!["a.c", "b.c"]);
1686        assert_eq!(opts.opt_level, OptLevel::O2);
1687        assert_eq!(opts.emit, EmitKind::Object);
1688        assert!(opts.debug_info);
1689    }
1690
1691    /// The unstable options, which are spelled apart from everything else on purpose: what is
1692    /// under `-Z` promises nothing, and a build that reaches for one should have had to say so.
1693    #[test]
1694    fn an_unstable_option_is_taken_and_one_that_does_not_exist_is_refused() {
1695        let (opts, _) = compile(&["-c", "-Zrule-coverage=/tmp/rules.cov", "a.c"]);
1696        assert_eq!(opts.rule_coverage.as_deref(), Some("/tmp/rules.cov"));
1697
1698        let (plain, _) = compile(&["-c", "a.c"]);
1699        assert_eq!(plain.rule_coverage, None, "nothing is measured unless it was asked for");
1700
1701        assert!(parse_args(&args(&["-Zrule-coverage=", "a.c"])).is_err(), "a file with no name");
1702        let unknown = parse_args(&args(&["-Zwhat", "a.c"])).expect_err("there is no such option");
1703        assert!(unknown.message.contains("4.11"), "{}", unknown.message);
1704    }
1705
1706    #[test]
1707    fn a_bare_dash_o_means_o1_the_way_gcc_reads_it() {
1708        let (opts, _) = compile(&["-O", "a.c"]);
1709        assert_eq!(opts.opt_level, OptLevel::O1);
1710    }
1711
1712    #[test]
1713    fn dash_x_applies_to_later_inputs_only_and_none_stops_it() {
1714        let (_, plan) = compile(&["a.o", "-x", "c", "b.txt", "-x", "none", "c.o"]);
1715        assert_eq!(plan.jobs[0].kind, InputKind::LinkerInput);
1716        assert_eq!(plan.jobs[1].kind, InputKind::C);
1717        assert_eq!(plan.jobs[2].kind, InputKind::LinkerInput);
1718    }
1719
1720    #[test]
1721    fn dash_j_reaches_the_scheduler_and_defaults_to_the_machine() {
1722        let (_, _, jobs) = match parse_args(&args(&["-j4", "a.c"])).unwrap() {
1723            Action::Compile { opts, plan, jobs, .. } => (opts, plan, jobs),
1724            other => panic!("expected a compilation, got {other:?}"),
1725        };
1726        assert_eq!(jobs.count(), 4);
1727
1728        let default = match parse_args(&args(&["a.c"])).unwrap() {
1729            Action::Compile { jobs, .. } => jobs,
1730            other => panic!("expected a compilation, got {other:?}"),
1731        };
1732        assert_eq!(default, Jobs::available());
1733        assert!(parse_args(&args(&["-j0", "a.c"])).is_err());
1734    }
1735
1736    #[test]
1737    fn triple_hash_prints_the_plan_and_runs_nothing() {
1738        let a = parse_args(&args(&["-###", "-c", "a.c"])).unwrap();
1739        let Action::PrintPlan { plan, .. } = a else { panic!("expected a plan dump") };
1740        assert!(plan.render().contains("a.c: preprocess, compile, assemble -> a.o"));
1741    }
1742
1743    #[test]
1744    fn the_flag_that_keeps_the_intermediate_files_has_three_spellings_and_two_meanings() {
1745        // The bare one is `=obj` and not `=cwd`. gcc's manual says the opposite and gcc 16 does
1746        // this, and following the compiler is what makes a build that reads either of them find
1747        // the files where they are.
1748        assert_eq!(compile(&["-c", "-save-temps", "a.c"]).0.save_temps, SaveTemps::Object);
1749        assert_eq!(compile(&["-c", "-save-temps=obj", "a.c"]).0.save_temps, SaveTemps::Object);
1750        assert_eq!(compile(&["-c", "-save-temps=cwd", "a.c"]).0.save_temps, SaveTemps::Cwd);
1751        assert_eq!(compile(&["-c", "a.c"]).0.save_temps, SaveTemps::No);
1752        // The last one on the line decides, the way it does for every other flag with an
1753        // argument, and a keyword that is neither is fatal rather than ignored: a run that kept
1754        // nothing and said nothing looks exactly like one where the files were not produced.
1755        let (opts, _) = compile(&["-c", "-save-temps", "-save-temps=cwd", "a.c"]);
1756        assert_eq!(opts.save_temps, SaveTemps::Cwd);
1757        let e = parse_args(&args(&["-c", "-save-temps=nowhere", "a.c"])).unwrap_err();
1758        assert!(e.message.contains("accepted: cwd, obj"), "{}", e.message);
1759    }
1760
1761    #[test]
1762    fn the_flag_that_times_each_step_reaches_the_options_and_changes_nothing_else() {
1763        let (opts, plan) = compile(&["-c", "-time", "a.c"]);
1764        let (plain, without) = compile(&["-c", "a.c"]);
1765        assert!(opts.time);
1766        assert!(!plain.time);
1767        // Against the same line without the flag rather than against a spelling of the object's
1768        // name, since what the object is called is the host's business and this is not about that.
1769        assert_eq!(plan.jobs[0].output, without.jobs[0].output);
1770    }
1771
1772    #[test]
1773    fn dash_x_names_what_it_accepts_when_it_does_not_know_a_language() {
1774        let e = parse_args(&args(&["-x", "fortran", "a.c"])).unwrap_err();
1775        assert!(e.message.contains("assembler-with-cpp"), "{}", e.message);
1776    }
1777
1778    #[test]
1779    fn an_unknown_flag_is_an_error_rather_than_a_shrug() {
1780        let e = parse_args(&args(&["-fno-such-thing", "a.c"])).unwrap_err();
1781        assert!(e.message.contains("unknown option"), "{}", e.message);
1782    }
1783
1784    /// `-fpermissive` and the flag that turns it back off, which a build writes beside it when
1785    /// one directory needs the older rules and the rest of the tree does not.
1786    #[test]
1787    fn permissive_reads_in_both_directions_and_the_last_one_wins() {
1788        let (opts, _) = compile(&["-c", "a.c"]);
1789        assert!(!opts.permissive, "off unless it is asked for");
1790
1791        let (opts, _) = compile(&["-c", "-fpermissive", "a.c"]);
1792        assert!(opts.permissive);
1793
1794        let (opts, _) = compile(&["-c", "-fpermissive", "-fno-permissive", "a.c"]);
1795        assert!(!opts.permissive);
1796    }
1797
1798    #[test]
1799    fn asking_for_nested_functions_is_told_why_it_is_not_coming() {
1800        let e = parse_args(&args(&["-fnested-functions", "a.c"])).unwrap_err();
1801        assert!(e.message.contains("trampoline"), "{}", e.message);
1802        assert!(parse_args(&args(&["-fno-nested-functions", "a.c"])).is_ok());
1803    }
1804
1805    #[test]
1806    fn the_flag_every_configure_script_writes_is_taken() {
1807        // All four spellings, because a build writes whichever one its macros picked and a
1808        // compiler that takes three of them is a compiler that fails on the fourth.
1809        for flag in ["-fPIC", "-fpic", "-fPIE", "-fpie"] {
1810            let (opts, _) = compile(&["-c", flag, "a.c"]);
1811            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1812        }
1813    }
1814
1815    #[test]
1816    fn the_flags_that_describe_what_this_compiler_already_does_are_taken() {
1817        // Every one of these is on a real build line somewhere and every one of them was an
1818        // unknown option. What they have in common is that the answer rucc gives is the answer
1819        // they ask for, so there is nothing to implement and nothing to refuse.
1820        for flag in [
1821            "-fno-common",
1822            "-fstrict-aliasing",
1823            "-fno-strict-aliasing",
1824            "-pipe",
1825            "-fdiagnostics-color",
1826            "-fno-diagnostics-color",
1827            "-fdiagnostics-color=always",
1828            "-fdiagnostics-color=never",
1829            "-fdiagnostics-color=auto",
1830        ] {
1831            let (opts, _) = compile(&["-c", flag, "a.c"]);
1832            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1833        }
1834    }
1835
1836    #[test]
1837    fn asking_the_linker_to_merge_tentative_definitions_is_told_why_it_is_not_coming() {
1838        // The one of that family that is a request rather than a description, and it is a real
1839        // difference: two files each writing `int g;` link under it and do not without it.
1840        let e = parse_args(&args(&["-fcommon", "a.c"])).unwrap_err();
1841        assert!(e.message.contains(".bss"), "{}", e.message);
1842        assert!(e.message.contains("extern"), "the way out is worth saying: {}", e.message);
1843    }
1844
1845    #[test]
1846    fn asking_for_position_dependent_code_is_told_why_it_is_not_coming() {
1847        for flag in ["-fno-pic", "-fno-pie"] {
1848            let e = parse_args(&args(&[flag, "a.c"])).unwrap_err();
1849            assert!(e.message.contains("global offset table"), "{flag}: {}", e.message);
1850            // The one it may have meant, since the two are a letter apart and one of them is
1851            // about linking and is taken.
1852            assert!(e.message.contains("-no-pie"), "{flag}: {}", e.message);
1853        }
1854    }
1855
1856    #[test]
1857    fn an_unsupported_target_names_itself() {
1858        let e = parse_args(&args(&["--target=sparc64-linux-gnu", "a.c"])).unwrap_err();
1859        assert!(e.message.contains("sparc64"), "{}", e.message);
1860    }
1861
1862    #[test]
1863    fn no_inputs_is_an_error_but_print_config_needs_none() {
1864        assert!(parse_args(&args(&[])).is_err());
1865        assert!(matches!(parse_args(&args(&["--print-config"])), Ok(Action::PrintConfig(_))));
1866    }
1867
1868    #[test]
1869    fn print_config_reports_the_target_it_was_given_not_the_host() {
1870        let a = parse_args(&args(&["--print-config", "--target=riscv64-linux-musl"])).unwrap();
1871        let Action::PrintConfig(opts) = a else { panic!("expected a configuration dump") };
1872        let text = print_config(&opts);
1873        assert!(text.contains("target: riscv64-unknown-linux-musl"), "{text}");
1874        assert!(text.contains("char-signed: false"), "{text}");
1875        assert!(text.contains("object-format: elf"), "{text}");
1876        assert!(text.contains("va-list: void-pointer"), "{text}");
1877        // RISC-V has a register file and this compiler has not written it down yet, and the
1878        // dump says which of those two it is rather than leaving the line out.
1879        assert!(text.contains("registers: none"), "{text}");
1880    }
1881
1882    #[test]
1883    fn print_config_has_one_key_per_line_and_a_fixed_order() {
1884        let opts = Options::new("x86_64-unknown-linux-gnu".parse().unwrap());
1885        let text = print_config(&opts);
1886        let keys: Vec<&str> =
1887            text.lines().map(|l| l.split(':').next().unwrap_or_default()).collect();
1888        assert_eq!(keys[0], "version");
1889        assert_eq!(keys[1], "target");
1890        assert_eq!(keys.len(), 19);
1891        assert!(text.ends_with('\n'));
1892    }
1893
1894    #[test]
1895    fn the_safety_tier_is_read_off_the_command_line_and_a_wrong_one_is_refused() {
1896        let (opts, _) = compile(&["a.c"]);
1897        assert_eq!(opts.safety, rucc_session::Safety::Off);
1898
1899        for (flag, tier) in [
1900            ("-fsafety=detect", rucc_session::Safety::Detect),
1901            ("-fsafety=enforce", rucc_session::Safety::Enforce),
1902            ("-fsafety=kernel", rucc_session::Safety::Kernel),
1903            ("-fsafety=off", rucc_session::Safety::Off),
1904        ] {
1905            let (opts, _) = compile(&[flag, "a.c"]);
1906            assert_eq!(opts.safety, tier, "{flag}");
1907        }
1908
1909        // The last one wins, the way every other repeated flag on this command line does.
1910        let (opts, _) = compile(&["-fsafety=enforce", "-fsafety=off", "a.c"]);
1911        assert_eq!(opts.safety, rucc_session::Safety::Off);
1912
1913        // A misspelled tier is refused rather than ignored. Silently compiling without the
1914        // monitor a build asked for is the one failure mode this feature cannot have.
1915        let e = parse_args(&args(&["-fsafety=on", "a.c"])).unwrap_err();
1916        assert!(e.message.contains("is not a safety tier"), "{}", e.message);
1917        assert!(parse_args(&args(&["-fsafety", "a.c"])).is_err());
1918    }
1919
1920    #[test]
1921    fn print_pipeline_answers_with_the_passes_the_level_asked_for() {
1922        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
1923        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1924        let text = print_pipeline(&opts);
1925        assert!(text.starts_with("level: -O2\n"), "{text}");
1926        assert!(text.contains("fold"), "{text}");
1927
1928        let a = parse_args(&args(&["--print-pipeline"])).unwrap();
1929        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1930        // One pass runs at `-O0` and it is the one that removes code nothing reaches, which is
1931        // not an optimization. See issue 359.
1932        assert!(print_pipeline(&opts).contains("1: simplify-cfg,"), "{}", print_pipeline(&opts));
1933
1934        let a = parse_args(&args(&["--print-pipeline", "-fno-simplify-cfg"])).unwrap();
1935        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1936        // And with that one turned off there is nothing left, which the dump says rather than
1937        // printing an empty list.
1938        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
1939    }
1940
1941    #[test]
1942    fn print_pipeline_takes_the_toggles_into_account() {
1943        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fno-fold"])).unwrap();
1944        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1945        let text = print_pipeline(&opts);
1946        // The one that was named is gone and the rest of the level is not, which is the whole
1947        // of what a toggle promises.
1948        assert!(!text.contains("fold"), "{text}");
1949        assert!(text.contains("dce"), "{text}");
1950
1951        // Every pass the compiler has, named off. Built from the registry rather than written
1952        // out, so a pass added later is turned off here too and this keeps testing the thing it
1953        // is about, which is that the toggles can empty a level.
1954        let mut off = vec!["--print-pipeline".to_owned(), "-O2".to_owned()];
1955        off.extend(rucc_opt::PASSES.iter().map(|p| format!("-fno-{}", p.name())));
1956        let spelled: Vec<&str> = off.iter().map(String::as_str).collect();
1957        let a = parse_args(&args(&spelled)).unwrap();
1958        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1959        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
1960    }
1961
1962    #[test]
1963    fn print_pipeline_says_when_a_budget_will_stop_the_run_short() {
1964        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
1965        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1966        assert!(!print_pipeline(&opts).contains("global fuel"));
1967
1968        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fpass-fuel-global=4"])).unwrap();
1969        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1970        let text = print_pipeline(&opts);
1971        // Because the listing is the answer to what this compilation will do, and a run that
1972        // stops after four rewrites is not doing what the level says it does.
1973        assert!(text.contains("global fuel: 4"), "{text}");
1974    }
1975
1976    /// A pass is turned on and off by its own name, and the order the flags were given in is
1977    /// kept, because the last spelling of a name is the one that decides.
1978    #[test]
1979    fn a_pass_is_named_by_dash_f_and_unnamed_by_dash_f_no() {
1980        let (opts, _) = compile(&["-c", "-O0", "-ffold", "-fno-fold", "-ffold", "a.c"]);
1981        assert_eq!(
1982            opts.passes,
1983            [("fold".to_owned(), true), ("fold".to_owned(), false), ("fold".to_owned(), true)]
1984        );
1985
1986        let e = parse_args(&args(&["-fno-such-pass", "a.c"])).unwrap_err();
1987        assert!(e.message.contains("unknown option"), "{}", e.message);
1988    }
1989
1990    #[test]
1991    fn pass_fuel_names_a_pass_and_a_count_and_refuses_anything_else() {
1992        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel=fold=3", "a.c"]);
1993        assert_eq!(opts.pass_fuel, [("fold".to_owned(), 3)]);
1994
1995        let e = parse_args(&args(&["-fpass-fuel=fold", "a.c"])).unwrap_err();
1996        assert!(e.message.contains("<pass>=<count>"), "{}", e.message);
1997        let e = parse_args(&args(&["-fpass-fuel=nosuch=3", "a.c"])).unwrap_err();
1998        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
1999        let e = parse_args(&args(&["-fpass-fuel=fold=lots", "a.c"])).unwrap_err();
2000        assert!(e.message.contains("not a number"), "{}", e.message);
2001    }
2002
2003    #[test]
2004    fn global_pass_fuel_is_a_count_on_its_own_and_defaults_to_no_limit() {
2005        let (opts, _) = compile(&["-c", "-O2", "a.c"]);
2006        assert_eq!(opts.pass_fuel_global, None);
2007
2008        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel-global=12", "a.c"]);
2009        assert_eq!(opts.pass_fuel_global, Some(12));
2010        // And it is not the per pass flag with a longer name, so neither spelling swallows the
2011        // other.
2012        assert!(opts.pass_fuel.is_empty());
2013
2014        let e = parse_args(&args(&["-fpass-fuel-global=lots", "a.c"])).unwrap_err();
2015        assert!(e.message.contains("not a number"), "{}", e.message);
2016    }
2017
2018    #[test]
2019    fn a_gate_names_a_pass_and_optionally_the_functions_it_covers() {
2020        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold", "-fenable-fold=2-4,main", "a.c"]);
2021        assert_eq!(
2022            opts.pass_gates,
2023            [(false, "fold".to_owned()), (true, "fold=2-4,main".to_owned())],
2024            "the order is what decides, so it has to survive the parse"
2025        );
2026
2027        let e = parse_args(&args(&["-fdisable-nosuch", "a.c"])).unwrap_err();
2028        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2029        let e = parse_args(&args(&["-fenable-fold=9-2", "a.c"])).unwrap_err();
2030        assert!(e.message.contains("ends before it starts"), "{}", e.message);
2031        let e = parse_args(&args(&["-fdisable-fold=", "a.c"])).unwrap_err();
2032        assert!(e.message.contains("is empty"), "{}", e.message);
2033    }
2034
2035    #[test]
2036    fn the_pipeline_listing_says_which_passes_a_gate_touched() {
2037        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold=main", "a.c"]);
2038        let text = print_pipeline(&opts);
2039        assert!(text.contains("fold, "), "{text}");
2040        assert!(text.contains("[off for main]"), "{text}");
2041    }
2042
2043    /// The spelling is checked while the arguments are read, because a dump that names a pass
2044    /// this compiler does not have is a typo, and a typo found after the compilation has run is
2045    /// found too late to be any use.
2046    #[test]
2047    fn a_dump_is_checked_when_it_is_asked_for_rather_than_when_it_is_taken() {
2048        let (opts, _) = compile(&["-c", "-O2", "-fdump-ir=all", "-fdump-ir=after-fold", "a.c"]);
2049        assert_eq!(opts.dump_ir, ["all", "after-fold"]);
2050
2051        let e = parse_args(&args(&["-fdump-ir=after-nosuch", "a.c"])).unwrap_err();
2052        assert!(e.message.contains("nosuch"), "{}", e.message);
2053        assert!(parse_args(&args(&["-fdump-ir=sideways-fold", "a.c"])).is_err());
2054    }
2055
2056    /// Every spelling `-fopt-info` takes, and the one it does not.
2057    ///
2058    /// The keywords are checked here for the same reason a dump's pass name is: a person who
2059    /// misspelled one gets no output, and no output is also what a compilation where nothing
2060    /// happened looks like. Telling those two apart is the entire reason to reach for this flag.
2061    #[test]
2062    fn opt_info_takes_kinds_and_a_file_and_refuses_a_kind_it_does_not_have() {
2063        let (opts, _) = compile(&["-c", "-O2", "-fopt-info", "a.c"]);
2064        assert_eq!(opts.opt_info, [""], "a bare flag asks for the rewrites");
2065        assert_eq!(opts.opt_info_file, None, "and goes to standard error");
2066
2067        let (opts, _) = compile(&["-c", "-O2", "-fopt-info-missed-note", "a.c"]);
2068        assert_eq!(opts.opt_info, ["missed-note"]);
2069
2070        // Two flags add up rather than the second replacing the first, and the file is the last
2071        // one that named a file, which is how GCC treats both.
2072        let (opts, _) =
2073            compile(&["-c", "-O2", "-fopt-info-missed=one.txt", "-fopt-info-all=two.txt", "a.c"]);
2074        assert_eq!(opts.opt_info, ["missed", "all"]);
2075        assert_eq!(opts.opt_info_file.as_deref(), Some("two.txt"));
2076
2077        let e = parse_args(&args(&["-fopt-info-vectorized", "a.c"])).unwrap_err();
2078        assert!(e.message.contains("vectorized"), "{}", e.message);
2079        assert!(e.message.contains("`missed`"), "{}", e.message);
2080        let e = parse_args(&args(&["-fopt-info-missed=", "a.c"])).unwrap_err();
2081        assert!(e.message.contains("no file"), "{}", e.message);
2082    }
2083
2084    #[test]
2085    fn verify_each_is_unstable_and_off_unless_it_was_asked_for() {
2086        let (opts, _) = compile(&["-c", "-Zverify-each", "a.c"]);
2087        assert!(opts.verify_each);
2088        assert!(!USAGE.contains("verify-each"), "an unstable option stays out of the usage text");
2089    }
2090
2091    #[test]
2092    fn dash_o_needs_an_argument() {
2093        let e = parse_args(&args(&["a.c", "-o"])).unwrap_err();
2094        assert_eq!(e.message, "-o requires an argument");
2095    }
2096
2097    #[test]
2098    fn dash_d_and_dash_u_are_read_joined_or_separated_and_keep_their_order() {
2099        let (opts, _) = compile(&["-DFOO=1", "-D", "BAR", "-UBAZ", "-U", "QUX", "a.c"]);
2100        assert_eq!(opts.defines, ["FOO=1", "BAR"]);
2101        assert_eq!(opts.undefines, ["BAZ", "QUX"]);
2102    }
2103
2104    #[test]
2105    fn the_include_flags_land_on_the_chain_each_one_names() {
2106        // A sysroot with nothing under it, so that the library's own directories are the
2107        // same on every machine this test runs on, which is none of them.
2108        let (opts, _) = compile(&[
2109            "-Ii",
2110            "-iquote",
2111            "q",
2112            "-isystem",
2113            "sys",
2114            "-idirafter",
2115            "after",
2116            "--sysroot=/nowhere-at-all",
2117            "a.c",
2118        ]);
2119        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2120        // The compiler's own headers sit after every `-isystem` and before `-idirafter`,
2121        // which is where GCC puts its own: a directory the user named outranks ours.
2122        assert_eq!(dirs, ["q", "i", "sys", runtime::DIR, "after"]);
2123        assert!(!opts.search.dirs()[1].is_system);
2124        assert!(opts.search.dirs()[2].is_system);
2125    }
2126
2127    #[test]
2128    fn the_librarys_headers_come_after_the_compilers_own_and_go_away_with_them() {
2129        // Which machine this runs on decides what is on the path, so the test is about the
2130        // order rather than about the names: ours is on it, the library's follow it, and
2131        // `-nostdinc` is the one flag that takes both halves of the pair off at once.
2132        let (opts, _) = compile(&["a.c"]);
2133        let dirs = opts.search.dirs();
2134        let ours = dirs.iter().position(|d| d.path.to_str() == Some(runtime::DIR));
2135        assert_eq!(ours, Some(0), "{dirs:?}");
2136        assert!(dirs[1..].iter().all(|d| d.is_system), "{dirs:?}");
2137        let (bare, _) = compile(&["-nostdinc", "a.c"]);
2138        assert!(bare.search.dirs().is_empty(), "{:?}", bare.search.dirs());
2139    }
2140
2141    #[test]
2142    fn a_sysroot_moves_the_librarys_directories_and_nothing_else() {
2143        let (opts, _) = compile(&["-isystem", "sys", "--sysroot=/nowhere-at-all", "a.c"]);
2144        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2145        assert_eq!(dirs, ["sys", runtime::DIR]);
2146    }
2147
2148    #[test]
2149    fn dash_i_dash_moves_the_bracket_directories_into_the_quoted_chain() {
2150        let (opts, _) =
2151            compile(&["-Iinc1", "-iquote", "inc2", "-I-", "-Iinc3", "-nostdinc", "a.c"]);
2152        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2153        assert_eq!(dirs, ["inc1", "inc2", "inc3"]);
2154        // An angled include sees only what came after the flag.
2155        assert_eq!(opts.search.start(IncludeForm::Angled), 2);
2156        assert!(!opts.search.searches_current_dir());
2157    }
2158
2159    #[test]
2160    fn the_prefix_flags_stick_what_iprefix_said_on_the_front_of_what_follows_it() {
2161        let (opts, _) = compile(&[
2162            "-iprefix",
2163            "/tools/",
2164            "-iwithprefix",
2165            "late",
2166            "-iwithprefixbefore",
2167            "early",
2168            "-iprefix",
2169            "/other/",
2170            "-iwithprefix",
2171            "last",
2172            "-nostdinc",
2173            "a.c",
2174        ]);
2175        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2176        // `-iwithprefixbefore` is an `-I` and the other two are `-isystem`, which is where GCC
2177        // puts them rather than where its manual says it does.
2178        assert_eq!(dirs, ["/tools/early", "/tools/late", "/other/last"]);
2179        assert!(!opts.search.dirs()[0].is_system);
2180        assert!(opts.search.dirs()[1].is_system);
2181    }
2182
2183    #[test]
2184    fn the_files_named_on_the_command_line_keep_their_order_and_which_flag_named_them() {
2185        let (opts, _) =
2186            compile(&["-include", "one.h", "-imacros", "two.h", "-include", "3.h", "a.c"]);
2187        let names: Vec<&str> = opts.preincludes.iter().map(|p| p.name.as_str()).collect();
2188        assert_eq!(names, ["one.h", "two.h", "3.h"]);
2189        assert_eq!(opts.preincludes.iter().filter(|p| p.macros_only).count(), 1);
2190    }
2191
2192    #[test]
2193    fn nostdinc_takes_the_compilers_own_headers_off_the_path() {
2194        let (opts, _) = compile(&["-Ii", "-nostdinc", "a.c"]);
2195        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2196        assert_eq!(dirs, ["i"]);
2197    }
2198
2199    #[test]
2200    fn the_dialect_flags_set_the_language_and_the_extensions_separately() {
2201        let (opts, _) = compile(&["-std=gnu11", "a.c"]);
2202        assert_eq!(opts.std, Std::C11);
2203        assert!(opts.gnu_extensions);
2204
2205        let (opts, _) = compile(&["-std=iso9899:1999", "a.c"]);
2206        assert_eq!(opts.std, Std::C99);
2207        assert!(!opts.gnu_extensions);
2208
2209        let (opts, _) = compile(&["-ansi", "a.c"]);
2210        assert_eq!(opts.std, Std::C89);
2211        assert!(!opts.gnu_extensions);
2212
2213        let e = parse_args(&args(&["-std=c94jr", "a.c"])).unwrap_err();
2214        assert!(e.message.contains("unknown dialect"), "{}", e.message);
2215    }
2216
2217    #[test]
2218    fn the_dump_letters_are_a_family_and_everything_else_beginning_with_d_is_not() {
2219        let (opts, _) = compile(&["-dM", "a.c"]);
2220        assert!(opts.dumps.macros);
2221
2222        // Packed, the way GCC takes them, and a letter in the family we have not written yet
2223        // is accepted and does nothing rather than failing a build.
2224        let (opts, _) = compile(&["-dDM", "a.c"]);
2225        assert!(opts.dumps.macros);
2226        let (opts, _) = compile(&["-dD", "a.c"]);
2227        assert!(!opts.dumps.macros);
2228
2229        let (opts, _) = compile(&["a.c"]);
2230        assert!(!opts.dumps.any());
2231
2232        // `-dumpversion` is a different flag that happens to start the same way, and it is read
2233        // as itself rather than as a dump of nothing.
2234        assert_eq!(printed(&["-dumpversion", "a.c"]), VERSION);
2235    }
2236
2237    #[test]
2238    fn the_gcc_version_claimed_is_a_flag_and_the_short_spellings_are_the_ones_people_write() {
2239        let (opts, _) = compile(&["a.c"]);
2240        assert_eq!(
2241            opts.gnuc,
2242            GnucVersion { major: 7, minor: 0, patch: 0 },
2243            "the lowest claim a modern glibc gives its own declarations to"
2244        );
2245
2246        let (opts, _) = compile(&["-fgnuc-version=15.1.0", "a.c"]);
2247        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 1, patch: 0 });
2248
2249        // A missing component is zero. `gcc -dumpversion` says `15` on a release with no
2250        // patchlevel and a harness that pastes that back has to be understood.
2251        let (opts, _) = compile(&["-fgnuc-version=15", "a.c"]);
2252        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 0, patch: 0 });
2253
2254        let (opts, _) = compile(&["-fgnuc-version=13.2", "a.c"]);
2255        assert_eq!(opts.gnuc, GnucVersion { major: 13, minor: 2, patch: 0 });
2256
2257        let e = parse_args(&args(&["-fgnuc-version=15.x", "a.c"])).unwrap_err();
2258        assert!(e.message.contains("minor that is not a number"), "{}", e.message);
2259
2260        let e = parse_args(&args(&["-fgnuc-version=1.2.3.4", "a.c"])).unwrap_err();
2261        assert!(e.message.contains("more than three"), "{}", e.message);
2262    }
2263
2264    #[test]
2265    fn pedantic_has_two_spellings_and_is_not_the_same_knob_as_the_dialect() {
2266        let (opts, _) = compile(&["-std=c17", "-pedantic", "a.c"]);
2267        assert!(opts.pedantic);
2268        assert_eq!(opts.std, Std::C17);
2269
2270        // The `-W` family's name for it, which is what a build that groups its warning flags
2271        // tends to write.
2272        let (opts, _) = compile(&["-Wpedantic", "a.c"]);
2273        assert!(opts.pedantic);
2274
2275        let (opts, _) = compile(&["-std=c17", "a.c"]);
2276        assert!(!opts.pedantic, "a dialect on its own does not diagnose an extension");
2277    }
2278
2279    #[test]
2280    fn dash_p_and_dash_ffreestanding_reach_the_options() {
2281        let (opts, _) = compile(&["-E", "-P", "-ffreestanding", "a.c"]);
2282        assert!(!opts.line_markers);
2283        assert!(!opts.hosted);
2284        assert_eq!(opts.emit, EmitKind::Preprocessed);
2285    }
2286
2287    /// The two ways a build says it means its own function by a name the C library also has.
2288    ///
2289    /// `-fno-builtin` is all of them and `-fno-builtin-<name>` is one, and the second is what a
2290    /// build writes when it means its own `memcpy` and the library's everything else. The name is
2291    /// kept as it was written and not checked against anything, because a program is allowed to
2292    /// mean something by a name this compiler has never heard of.
2293    #[test]
2294    fn the_builtin_flags_are_read_in_both_directions_and_one_name_at_a_time() {
2295        let (opts, _) = compile(&["-c", "a.c"]);
2296        assert!(opts.builtins, "a library name means the library function by default");
2297        assert!(opts.no_builtin.is_empty());
2298
2299        let (opts, _) = compile(&["-c", "-fno-builtin", "a.c"]);
2300        assert!(!opts.builtins);
2301
2302        let (opts, _) = compile(&["-c", "-fno-builtin", "-fbuiltin", "a.c"]);
2303        assert!(opts.builtins, "the last mention decides");
2304
2305        let (opts, _) = compile(&["-c", "-fno-builtin-memcpy", "-fno-builtin-nonesuch", "a.c"]);
2306        assert!(opts.builtins, "one name is not the family");
2307        assert_eq!(opts.no_builtin, vec!["memcpy".to_owned(), "nonesuch".to_owned()]);
2308    }
2309
2310    /// `-fvisibility=`, which is on every cmake project that cares about which names it exports
2311    /// and which was refused as an unknown option until now.
2312    ///
2313    /// Four spellings and three answers. `internal` is hidden plus a promise about never taking
2314    /// the address across a component boundary, and nothing derives anything from that promise
2315    /// here, so it comes out as the weaker of the two rather than as a refusal that stops a build
2316    /// over a distinction this compiler does not make.
2317    #[test]
2318    fn visibility_takes_the_four_spellings_gcc_takes_and_refuses_the_rest() {
2319        let (opts, _) = compile(&["-c", "a.c"]);
2320        assert_eq!(opts.visibility, Visibility::Default, "exported unless something says not");
2321
2322        for (written, wanted) in [
2323            ("default", Visibility::Default),
2324            ("hidden", Visibility::Hidden),
2325            ("internal", Visibility::Hidden),
2326            ("protected", Visibility::Protected),
2327        ] {
2328            let (opts, _) = compile(&["-c", &format!("-fvisibility={written}"), "a.c"]);
2329            assert_eq!(opts.visibility, wanted, "{written}");
2330        }
2331
2332        // The last mention decides, which is what every other flag of this shape does and what a
2333        // build that turns something off for one directory relies on.
2334        let (opts, _) = compile(&["-c", "-fvisibility=hidden", "-fvisibility=default", "a.c"]);
2335        assert_eq!(opts.visibility, Visibility::Default, "the last mention decides");
2336
2337        // A spelling gcc does not take is refused rather than read as the default, because a
2338        // build that meant hidden and got exported is a library with the wrong interface and
2339        // nothing said about it anywhere.
2340        let failed = parse_args(&args(&["-fvisibility=none", "a.c"])).expect_err("refused");
2341        assert!(failed.to_string().contains("is not a visibility"), "{failed}");
2342    }
2343
2344    /// `-fgnu89-inline`, which is off by default and is not implied by anything on the command
2345    /// line, since the dialect asks for GNU's reading further in rather than through this.
2346    #[test]
2347    fn gnu89_inline_is_off_until_it_is_asked_for_and_the_last_mention_decides() {
2348        let (opts, _) = compile(&["-c", "a.c"]);
2349        assert!(!opts.gnu89_inline, "C's reading of inline by default");
2350
2351        let (opts, _) = compile(&["-c", "-fgnu89-inline", "a.c"]);
2352        assert!(opts.gnu89_inline);
2353
2354        let (opts, _) = compile(&["-c", "-fgnu89-inline", "-fno-gnu89-inline", "a.c"]);
2355        assert!(!opts.gnu89_inline, "the last mention decides");
2356
2357        // The C89 dialects are under GNU's reading whether this was written or not, so the flag
2358        // stays off there and the dialect is what the checker and the macro set both ask. That is
2359        // also why `-std=c89 -fno-gnu89-inline` needs no diagnostic: it asks for the reading the
2360        // dialect already has. gcc refuses that command line, which is measured in the issue.
2361        let (opts, _) = compile(&["-c", "-std=c89", "a.c"]);
2362        assert!(!opts.gnu89_inline);
2363    }
2364
2365    /// Both spellings of both frame flags, since a build that wants one usually writes the
2366    /// other beside it for the one file that has to be compiled the ordinary way.
2367    #[test]
2368    fn the_two_frame_flags_are_read_in_both_directions() {
2369        let (opts, _) = compile(&["-c", "a.c"]);
2370        assert!(!opts.frame_pointer, "gcc omits it above -O0 and so does this");
2371        assert!(opts.red_zone, "the psABI has one and nothing said not to use it");
2372
2373        let (opts, _) = compile(&["-c", "-fno-omit-frame-pointer", "-mno-red-zone", "a.c"]);
2374        assert!(opts.frame_pointer);
2375        assert!(!opts.red_zone);
2376
2377        let (opts, _) = compile(&[
2378            "-c",
2379            "-fno-omit-frame-pointer",
2380            "-fomit-frame-pointer",
2381            "-mno-red-zone",
2382            "-mred-zone",
2383            "a.c",
2384        ]);
2385        assert!(!opts.frame_pointer, "the last one wins, as it does in gcc");
2386        assert!(opts.red_zone);
2387    }
2388
2389    #[test]
2390    fn the_link_flags_are_collected_apart_from_the_compilation() {
2391        let (link, _) = linking(&[
2392            "-static",
2393            "-nostartfiles",
2394            "-rdynamic",
2395            "-s",
2396            "-fuse-ld=mold",
2397            "-L/opt/lib",
2398            "-B",
2399            "/opt/tools",
2400            "a.c",
2401        ]);
2402        assert!(link.is_static);
2403        assert!(link.no_startfiles);
2404        assert!(link.export_dynamic);
2405        assert!(link.strip);
2406        assert_eq!(link.use_ld.as_deref(), Some("mold"));
2407        assert_eq!(link.search, vec![PathBuf::from("/opt/lib")]);
2408        assert_eq!(link.prefixes, vec![PathBuf::from("/opt/tools")]);
2409    }
2410
2411    #[test]
2412    fn a_comma_in_dash_wl_separates_two_arguments() {
2413        let (link, _) = linking(&["-Wl,-rpath,/opt/lib", "-Xlinker", "--as-needed", "a.c"]);
2414        assert_eq!(link.passthrough, vec!["-rpath", "/opt/lib", "--as-needed"]);
2415    }
2416
2417    #[test]
2418    fn a_library_keeps_its_place_between_the_objects() {
2419        // Link order is semantic: `-lm` written between two files resolves for the one before
2420        // it and not for the one after, so a library cannot be collected into a list of its own.
2421        // The target is named because the suffix of an object is the target's and this asserts
2422        // on the names: the same command line on a Windows host plans two `.obj` files.
2423        let (_, plan) = linking(&["--target=x86_64-unknown-linux-gnu", "a.c", "-lm", "b.c"]);
2424        let link = plan.link.expect("expected a link step");
2425        assert_eq!(
2426            link.inputs,
2427            vec![
2428                link::Item::File("a.o".into()),
2429                link::Item::Library("m".into()),
2430                link::Item::File("b.o".into()),
2431            ]
2432        );
2433        // And it is not a job, because there is nothing to compile in a library.
2434        assert_eq!(plan.jobs.len(), 2);
2435    }
2436
2437    #[test]
2438    fn a_library_on_a_dash_c_line_is_a_note_rather_than_an_error() {
2439        let (_, plan) = linking(&["-c", "-lm", "a.c"]);
2440        assert!(plan.link.is_none());
2441        assert!(plan.notes.iter().any(|n| n.contains("-lm")), "{:?}", plan.notes);
2442    }
2443
2444    #[test]
2445    fn the_sysroot_reaches_the_linker_as_well_as_the_headers() {
2446        let (link, _) = linking(&["--sysroot=/opt/root", "a.c"]);
2447        assert_eq!(link.sysroot, Some(PathBuf::from("/opt/root")));
2448    }
2449
2450    fn printed(s: &[&str]) -> String {
2451        match parse_args(&args(s)).expect("expected an answer") {
2452            Action::Print(line) => line,
2453            other => panic!("expected an answer, got {other:?}"),
2454        }
2455    }
2456
2457    fn refused(s: &[&str]) -> String {
2458        parse_args(&args(s)).expect_err("expected a refusal").message
2459    }
2460
2461    #[test]
2462    fn a_warning_flag_this_compiler_has_not_heard_of_is_taken_rather_than_refused() {
2463        // The rule in section 4.1, and the reason for it is autoconf: a configure script finds
2464        // out whether a warning flag exists by passing it and looking at the exit status, so a
2465        // compiler that refuses one it does not know fails a script written for a newer GCC.
2466        let (opts, _) = compile(&["-Wall", "-Wextra", "-Wno-format-truncation", "-c", "a.c"]);
2467        assert!(!opts.warnings_are_errors);
2468        assert!(opts.warnings);
2469        // The two spellings that do mean something are still read.
2470        let (opts, _) = compile(&["-Werror", "-c", "a.c"]);
2471        assert!(opts.warnings_are_errors);
2472        let (opts, _) = compile(&["-w", "-c", "a.c"]);
2473        assert!(!opts.warnings);
2474        let (opts, _) = compile(&["-pedantic-errors", "-c", "a.c"]);
2475        assert!(opts.pedantic && opts.warnings_are_errors);
2476    }
2477
2478    #[test]
2479    fn an_argument_for_a_separate_tool_is_refused_rather_than_dropped() {
2480        // Every one of these says something about the output, so the wrong answer is silence.
2481        assert!(refused(&["-Wa,--noexecstack", "-c", "a.c"]).contains("separate assembler"));
2482        assert!(refused(&["-Wp,-DX", "-c", "a.c"]).contains("separate assembler"));
2483        assert!(refused(&["-specs=/x", "a.c"]).contains("-specs= is not supported"));
2484        assert!(refused(&["-mcmodel=kernel", "-c", "a.c"]).contains("small code model"));
2485        assert!(refused(&["-gdwarf-4", "-c", "a.c"]).contains("DWARF 5"));
2486        assert!(refused(&["-Ofast", "-c", "a.c"]).contains("fast math"));
2487        // The word size the target does not have, which is a target this compiler was not asked
2488        // for rather than a flag it does not know.
2489        let no32 = refused(&["--target=x86_64-unknown-linux-gnu", "-m32", "-c", "a.c"]);
2490        assert!(no32.contains("32 bit target"), "{no32}");
2491    }
2492
2493    #[test]
2494    fn the_levels_gcc_spells_differently_are_the_levels_they_mean() {
2495        assert_eq!(compile(&["-O", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2496        assert_eq!(compile(&["-Og", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2497        assert_eq!(compile(&["-O2", "-c", "a.c"]).0.opt_level, OptLevel::O2);
2498    }
2499
2500    #[test]
2501    fn the_machine_flags_that_name_what_we_already_do_are_taken_and_the_rest_are_not() {
2502        let line = ["--target=x86_64-unknown-linux-gnu", "-m64", "-march=x86-64-v3"];
2503        let (opts, _) =
2504            compile(&[&line[..], &["-mtune=native", "-mabi=sysv", "-c", "a.c"]].concat());
2505        assert_eq!(opts.target.to_string(), "x86_64-unknown-linux-gnu");
2506        let wrong = refused(&["--target=x86_64-unknown-linux-gnu", "-mabi=ms", "-c", "a.c"]);
2507        assert!(wrong.contains("sysv convention"), "{wrong}");
2508    }
2509
2510    #[test]
2511    fn the_thread_flag_is_a_macro_and_a_library_and_the_library_goes_last() {
2512        let (opts, plan) = compile(&["-pthread", "-c", "a.c"]);
2513        assert!(opts.defines.iter().any(|d| d == "_REENTRANT"));
2514        // After the input, because a static link takes what it needs from a library when it
2515        // reaches it and not afterwards.
2516        let names: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
2517        assert_eq!(names, vec!["a.c"]);
2518    }
2519
2520    #[test]
2521    fn the_questions_a_build_system_asks_before_it_compiles_anything() {
2522        let target = "--target=x86_64-unknown-linux-gnu";
2523        assert_eq!(printed(&[target, "-dumpmachine"]), "x86_64-unknown-linux-gnu");
2524        assert_eq!(printed(&[target, "-dumpversion"]), VERSION);
2525        assert_eq!(printed(&[target, "-dumpfullversion"]), VERSION);
2526        assert_eq!(printed(&[target, "-print-multiarch"]), "x86_64-linux-gnu");
2527        // A name nothing holds comes back unchanged, which is GCC's rule and is what makes the
2528        // answer safe to paste into a link line whether or not the file is there.
2529        assert_eq!(printed(&[target, "-print-file-name=no-such-library.a"]), "no-such-library.a");
2530        assert_eq!(printed(&[target, "-print-prog-name=ld"]), "ld");
2531        let dirs = printed(&[target, "-print-search-dirs"]);
2532        assert!(dirs.starts_with("install: "), "{dirs}");
2533        assert!(dirs.contains("\nlibraries: ="), "{dirs}");
2534    }
2535
2536    #[test]
2537    fn the_two_dependency_flags_that_stop_after_the_rule_stop_after_the_rule() {
2538        let (opts, _) = compile(&["-M", "a.c"]);
2539        assert!(opts.deps.emit && opts.deps.instead_of_compiling);
2540        assert!(opts.deps.system_headers, "plain -M lists them");
2541        assert_eq!(opts.emit, EmitKind::Preprocessed);
2542
2543        // Even where a later flag asked for something else, because the family is a mode and
2544        // the mode is what the run is for.
2545        let (opts, _) = compile(&["-M", "-c", "a.c"]);
2546        assert_eq!(opts.emit, EmitKind::Preprocessed);
2547
2548        let (opts, _) = compile(&["-MM", "a.c"]);
2549        assert!(!opts.deps.system_headers);
2550    }
2551
2552    #[test]
2553    fn the_two_that_end_in_d_leave_the_compilation_alone() {
2554        let (opts, _) = compile(&["-MD", "-c", "a.c"]);
2555        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2556        assert!(opts.deps.system_headers);
2557        assert_eq!(opts.emit, EmitKind::Object);
2558
2559        let (opts, _) = compile(&["-MMD", "-c", "a.c"]);
2560        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2561        assert!(!opts.deps.system_headers);
2562    }
2563
2564    #[test]
2565    fn nothing_puts_the_system_headers_back_once_a_flag_has_taken_them_out() {
2566        // GCC's rule, and not an oversight in it. The flag asking for fewer of them is read as
2567        // the answer, because the other one never asked the question.
2568        let (opts, _) = compile(&["-MM", "-M", "a.c"]);
2569        assert!(!opts.deps.system_headers);
2570        let (opts, _) = compile(&["-MD", "-MMD", "-c", "a.c"]);
2571        assert!(!opts.deps.system_headers);
2572        let (opts, _) = compile(&["-MMD", "-MD", "-c", "a.c"]);
2573        assert!(!opts.deps.system_headers);
2574    }
2575
2576    #[test]
2577    fn a_target_arrives_escaped_from_one_flag_and_untouched_from_the_other() {
2578        let (opts, _) = compile(&["-MM", "-MT", "a b.o", "-MQ", "a b.o", "a.c"]);
2579        assert_eq!(opts.deps.targets, vec!["a b.o".to_owned(), "a\\ b.o".to_owned()]);
2580    }
2581
2582    #[test]
2583    fn the_rest_of_the_family_is_a_file_and_a_switch() {
2584        let (opts, _) = compile(&["-MM", "-MF", "dep.d", "-MP", "a.c"]);
2585        assert_eq!(opts.deps.file.as_deref(), Some("dep.d"));
2586        assert!(opts.deps.phony);
2587
2588        for flag in ["-MF", "-MT", "-MQ"] {
2589            let e = parse_args(&args(&[flag])).unwrap_err();
2590            assert!(e.message.contains("requires an argument"), "{}", e.message);
2591        }
2592    }
2593
2594    /// A directory of sources for one test, removed when the test is done with it.
2595    struct TempTree(PathBuf);
2596
2597    impl Drop for TempTree {
2598        fn drop(&mut self) {
2599            let _ = std::fs::remove_dir_all(&self.0);
2600        }
2601    }
2602
2603    impl TempTree {
2604        fn new(name: &str, files: &[(&str, &str)]) -> TempTree {
2605            let dir = std::env::temp_dir().join(format!("rucc-deps-{}-{name}", std::process::id()));
2606            let _ = std::fs::remove_dir_all(&dir);
2607            std::fs::create_dir_all(&dir).expect("temporary directory should be writable");
2608            for (path, text) in files {
2609                let at = dir.join(path);
2610                if let Some(parent) = at.parent() {
2611                    std::fs::create_dir_all(parent).expect("creating a subdirectory should work");
2612                }
2613                std::fs::write(&at, text).expect("writing a temporary file should work");
2614            }
2615            TempTree(dir)
2616        }
2617
2618        fn path(&self, name: &str) -> String {
2619            self.0.join(name).to_string_lossy().into_owned()
2620        }
2621    }
2622
2623    #[test]
2624    fn the_rule_names_what_the_includes_found_and_names_each_of_them_once() {
2625        // End to end, because the list comes from the preprocessor and the format comes from
2626        // somewhere else, and a test of either half on its own would pass with the two of them
2627        // wired up backwards.
2628        let tree = TempTree::new(
2629            "found",
2630            &[
2631                ("a.c", "#include \"one.h\"\n#include \"two.h\"\nint main(void) { return X; }\n"),
2632                ("one.h", "#define X 0\n"),
2633                ("two.h", "#include \"one.h\"\n"),
2634            ],
2635        );
2636        let out = tree.path("dep.d");
2637        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2638        assert_eq!(code, 0);
2639
2640        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2641        let names: Vec<&str> = text.split_whitespace().collect();
2642        // The target, the source, and each header once however many times it was reached.
2643        assert_eq!(names.first(), Some(&"a.o:"), "{text}");
2644        assert_eq!(names.iter().filter(|n| n.ends_with("one.h")).count(), 1, "{text}");
2645        assert_eq!(names.iter().filter(|n| n.ends_with("two.h")).count(), 1, "{text}");
2646        // And the `-o` went to the file the rule replaced, which is left empty rather than
2647        // absent because a makefile that named it as a target will look for it.
2648        assert_eq!(std::fs::read(tree.path("a.i")).expect("the output should exist"), b"");
2649    }
2650
2651    #[test]
2652    fn a_header_that_is_only_reached_under_a_guard_is_still_a_dependency() {
2653        // The multiple-include optimization means the second reach never opens the file. It is
2654        // still a file this translation unit was built from, so it is still in the rule.
2655        let tree = TempTree::new(
2656            "guarded",
2657            &[
2658                ("a.c", "#include \"g.h\"\n#include \"g.h\"\nint main(void) { return 0; }\n"),
2659                ("g.h", "#ifndef G\n#define G\n#endif\n"),
2660            ],
2661        );
2662        let out = tree.path("dep.d");
2663        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2664        assert_eq!(code, 0);
2665        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2666        assert_eq!(text.split_whitespace().filter(|n| n.ends_with("g.h")).count(), 1, "{text}");
2667    }
2668
2669    #[test]
2670    fn every_imacros_file_is_read_before_every_include_file_whatever_order_they_were_written() {
2671        // Measured against GCC rather than read: the two flags the other way round produce the
2672        // same output byte for byte, so the command line order between the two families does not
2673        // decide anything and the order within one does. The `-include` file here can only see
2674        // the definition if the `-imacros` file that was written after it ran first.
2675        let tree = TempTree::new(
2676            "preinclude",
2677            &[
2678                ("a.c", "int main(void) { return 0; }\n"),
2679                ("i.h", "#ifdef FROM_MACROS\nint saw_it;\n#else\nint missed_it;\n#endif\n"),
2680                ("m.h", "#define FROM_MACROS 1\nint macros_text;\n"),
2681            ],
2682        );
2683        let out = tree.path("a.i");
2684        let code = run(&args(&[
2685            "-E",
2686            "-include",
2687            &tree.path("i.h"),
2688            "-imacros",
2689            &tree.path("m.h"),
2690            "-o",
2691            &out,
2692            &tree.path("a.c"),
2693        ]));
2694        assert_eq!(code, 0);
2695        let text = std::fs::read_to_string(&out).expect("the output should have been written");
2696        assert!(text.contains("saw_it"), "{text}");
2697        // And the text of the `-imacros` file is thrown away, which is the whole difference
2698        // between the two flags.
2699        assert!(!text.contains("macros_text"), "{text}");
2700    }
2701
2702    #[test]
2703    fn a_file_the_command_line_named_is_a_prerequisite_the_same_as_one_a_directive_named() {
2704        let tree = TempTree::new(
2705            "preinclude-deps",
2706            &[
2707                ("a.c", "int main(void) { return 0; }\n"),
2708                ("i.h", "int from_include;\n"),
2709                ("m.h", "#define M 1\n"),
2710            ],
2711        );
2712        let out = tree.path("dep.d");
2713        let code = run(&args(&[
2714            "-MM",
2715            "-MF",
2716            &out,
2717            "-include",
2718            &tree.path("i.h"),
2719            "-imacros",
2720            &tree.path("m.h"),
2721            "-o",
2722            &tree.path("a.i"),
2723            &tree.path("a.c"),
2724        ]));
2725        assert_eq!(code, 0);
2726        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2727        assert!(text.contains("i.h"), "{text}");
2728        assert!(text.contains("m.h"), "{text}");
2729    }
2730
2731    #[test]
2732    fn a_command_line_include_that_is_nowhere_on_the_path_is_an_error_and_not_a_warning() {
2733        // Including the directory of the source file, which is not on the path for these: the
2734        // command line was not written there, so a name in it is relative to where the compiler
2735        // was run rather than to where the source sits.
2736        let tree = TempTree::new(
2737            "preinclude-missing",
2738            &[("sub/a.c", "int main(void) { return 0; }\n"), ("sub/beside.h", "int x;\n")],
2739        );
2740        let code = run(&args(&["-E", "-include", "beside.h", "-o", "-", &tree.path("sub/a.c")]));
2741        assert_eq!(code, 1);
2742    }
2743
2744    #[test]
2745    fn a_command_line_that_links_names_the_executable_and_not_the_object_it_went_through() {
2746        // The object a link goes through is in a temporary directory and is gone before `make`
2747        // reads any of this, so the rule that named it would be a rule for a file that is never
2748        // there. The target and the file are both the `-o`, which is the executable.
2749        let (opts, plan) = compile(&["-MD", "sub/a.c", "-o", "prog"]);
2750        assert_eq!(plan.output.as_deref(), Some("prog"));
2751        assert_eq!(deps::default_target("sub/a.c", deps_target_output(&opts, &plan)), "prog");
2752        assert_eq!(
2753            deps::default_file(&opts.deps, "sub/a.c", plan.output.as_deref()).as_deref(),
2754            Some("prog.d")
2755        );
2756    }
2757
2758    #[test]
2759    fn the_plan_keeps_the_output_name_because_the_rule_is_written_from_it() {
2760        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c", "-o", "obj/x.o"]);
2761        assert_eq!(plan.output.as_deref(), Some("obj/x.o"));
2762        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c"]);
2763        assert_eq!(plan.output, None);
2764    }
2765
2766    #[test]
2767    fn usage_fits_on_a_screen() {
2768        // Not a style preference. A help text that scrolls is one nobody reads, and this is
2769        // the cheapest way to keep it honest as flags accumulate. The number goes up only when
2770        // a family of flags arrives that has nowhere to share a line, which the two pass gates
2771        // were and which the two fuel flags and `-fsafety=` now are, and it goes up by exactly
2772        // the lines that family took. The four it went up by last are the flags a build system
2773        // passes without being asked to: how much to say, what machine to generate for, threads,
2774        // and the questions `configure` asks before it compiles anything. The one it went up by
2775        // last is the second line of `--emit`, whose kinds are a family that has now outgrown
2776        // one line and has nowhere else to go. The two it went up by last are the dependency
2777        // family, which is eight flags that share nothing with anything above them. The one it
2778        // went up by last is the four spellings of position independent code, which every
2779        // configure script writes and which could only have shared the link line, and that line
2780        // is already four characters short of the limit. The two it went up by last are the rest
2781        // of the include family, which is six more flags that change where a header is looked for
2782        // and two that name a header outright. The one it went up by last is the pair that keeps
2783        // the intermediate files and times the steps, which belong next to the two flags above
2784        // them that are also about watching a compilation rather than changing one.
2785        assert!(USAGE.lines().count() < 48, "usage text has grown past one screen");
2786    }
2787}