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/0.10.5")]
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            // Two requests rather than one, and the same table answers both, so what decides is
532            // whether either of them is standing. gcc arranges it the same way: the asynchronous
533            // one is the default here and it implies the other, and a line that asks for a table
534            // and against an asynchronous one gets a table.
535            "-fasynchronous-unwind-tables" => opts.async_unwind_tables = true,
536            "-fno-asynchronous-unwind-tables" => opts.async_unwind_tables = false,
537            "-funwind-tables" => opts.unwind_tables = true,
538            "-fno-unwind-tables" => opts.unwind_tables = false,
539            // The other direction is a request, not a description, and it is one this compiler
540            // cannot grant, so it gets the treatment section 13.3 asks for rather than the unknown
541            // option error. Answering it by carrying on would be answering a different question:
542            // the code would still be position independent, which is correct everywhere an
543            // ordinary program runs and is wrong in a kernel, where the flag is written precisely
544            // because there is no loader to fill a global offset table in.
545            "-fno-pic" | "-fno-pie" => {
546                return Err(err(
547                    "position dependent code is not supported: an address that may be in another \
548                     object is loaded out of the global offset table, and nothing here emits the \
549                     absolute form this asks for. Use -no-pie if what you meant was how to link",
550                ));
551            }
552            // Another description of what this compiler does. A file scope declaration with no
553            // initializer is written into `.bss` as an ordinary defined symbol, not offered to the
554            // linker as a common one for it to merge, which is what `-fno-common` asks for and what
555            // gcc has done by default since 10. Nothing in the front end produces `Linkage::Common`
556            // at all.
557            "-fno-common" => {}
558            // And the request, which is the one that cannot be granted. It is a real difference and
559            // not a preference: two files each writing `int g;` link under `-fcommon` and are a
560            // duplicate definition without it, which is the whole reason the flag survives.
561            "-fcommon" => {
562                return Err(err(
563                    "a tentative definition is written into .bss as its own symbol here, and \
564                     nothing emits the common symbol this asks the linker to merge. Give the \
565                     variable a definition in one file and declare it extern in the others",
566                ));
567            }
568            // Both directions of this one are taken, which is the exception to the rule above, and
569            // the reason is which way being wrong costs something.
570            //
571            // Nothing here derives anything from the type an object is accessed through. The IR has
572            // somewhere to put a type based aliasing node and lowering fills it with nothing on
573            // every access, so no pass has one to read and the alias analysis falls back to what it
574            // can see. That makes `-fno-strict-aliasing` a description, the way `-fPIC` is.
575            //
576            // `-fstrict-aliasing` is a request to assume more than that, and this compiler assumes
577            // less. Answering a request for a weaker guarantee by giving a stronger one is safe in
578            // a way the `-fno-pic` case is not: every program that is correct under the assumption
579            // is correct without it, only slower. And `-O2` implies it, so a build that spells it
580            // out is a build that would stop on a compiler that refused it, for no gain at all.
581            "-fstrict-aliasing" | "-fno-strict-aliasing" => {}
582            // About temporary files rather than about code. There is nothing between the phases of
583            // one compilation here to write to a file in the first place.
584            "-pipe" => {}
585            // Nothing here writes colour, so all of these are the same answer, and it is the answer
586            // that costs nothing: the diagnostics come out plain either way and no build depends on
587            // an escape sequence being there. Taken rather than refused because cmake writes
588            // `-fdiagnostics-color=always` on every compile line when the generator is ninja, which
589            // makes this the second most common flag after `-fPIC` to stop a build over a question
590            // about how the text looks.
591            "-fdiagnostics-color" | "-fno-diagnostics-color" => {}
592            _ if arg.starts_with("-fdiagnostics-color=") => {}
593            // The link flags. None of them changes the compilation, which is why they are
594            // collected apart from `opts` and why `-lm` on a `-c` line is a note rather than an
595            // error: it is a thing said to a linker that is not going to run.
596            "-static" => link.is_static = true,
597            "-shared" => link.shared = true,
598            "-pie" => link.pie = Some(true),
599            "-no-pie" | "-nopie" => link.pie = Some(false),
600            "-nostdlib" => link.no_stdlib = true,
601            "-nostartfiles" => link.no_startfiles = true,
602            "-nodefaultlibs" => link.no_defaultlibs = true,
603            "-fno-builtins-lib" => link.no_builtins_lib = true,
604            "-fbuiltins-lib" => link.no_builtins_lib = false,
605            "-rdynamic" | "-export-dynamic" => link.export_dynamic = true,
606            "-s" => link.strip = true,
607            "-Xlinker" => {
608                let next = args.get(i).ok_or_else(|| err("-Xlinker requires an argument"))?;
609                i += 1;
610                link.passthrough.push(next.clone());
611            }
612            _ if arg.starts_with("-Wl,") => {
613                // Commas separate arguments rather than being part of one, which is what makes
614                // `-Wl,-rpath,/opt/lib` two words to the linker and one word here.
615                link.passthrough.extend(arg["-Wl,".len()..].split(',').map(str::to_owned));
616            }
617            _ if arg.starts_with("-fuse-ld=") => {
618                link.use_ld = Some(arg["-fuse-ld=".len()..].to_owned());
619            }
620            _ if arg.starts_with("-l") && arg.len() > 2 => {
621                inputs.push(Input::library(&arg[2..]));
622            }
623            "-l" => {
624                let next = args.get(i).ok_or_else(|| err("-l requires an argument"))?;
625                i += 1;
626                inputs.push(Input::library(next));
627            }
628            _ if arg.starts_with("-L") => {
629                link.search.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
630            }
631            _ if arg.starts_with("-B") => {
632                link.prefixes.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
633            }
634            _ if arg.starts_with("-j") => {
635                jobs = Jobs::parse(&arg[2..]).map_err(err)?;
636            }
637            _ if arg.starts_with("--sysroot=") => {
638                sysroot = Some(PathBuf::from(&arg["--sysroot=".len()..]));
639            }
640            _ if arg.starts_with("--target=") => {
641                let t = &arg["--target=".len()..];
642                opts.target = t.parse().map_err(|e| err(format!("{e}")))?;
643            }
644            _ if arg.starts_with("--emit=") => {
645                let k = &arg["--emit=".len()..];
646                opts.emit = k
647                    .parse()
648                    .map_err(|()| err(format!("unknown --emit kind `{k}`, see --help")))?;
649            }
650            // A bare `-O` is `-O1`, which is what GCC has and what a hand written makefile tends
651            // to write. `-Og` is GCC's level for a build somebody is going to step through, and
652            // it is `-O1` with the transformations that move code around left out; this compiler
653            // has no such level yet, so it is the nearest one and `--print-pipeline` says what
654            // that came to rather than the flag pretending otherwise.
655            "-O" | "-Og" => opts.opt_level = rucc_session::OptLevel::O1,
656            // The union of `-O3` and `-ffast-math`, and the second half of that changes what
657            // floating point arithmetic means. Refused rather than taken as `-O3`, because a
658            // build that asks for fast math and is quietly given ordinary arithmetic gets a
659            // slower program than it asked for and a build that is given fast math it did not
660            // ask for gets a wrong one.
661            "-Ofast" => {
662                return Err(err(
663                    "-Ofast is -O3 with fast math, and fast math is not implemented, see \
664                     spec/04-driver-and-cli.md section 4.6",
665                ));
666            }
667            _ if arg.starts_with("-O") => {
668                opts.opt_level = arg[2..]
669                    .parse()
670                    .map_err(|()| err(format!("unknown optimization level `{arg}`")))?;
671            }
672            // What every name gets when nothing in the source said, which the attribute in the
673            // source overrides rather than the other way round. Before the optimizer's `-f`
674            // family below for the reason the tier below it is.
675            _ if arg.starts_with("-fvisibility=") => {
676                let seen = &arg["-fvisibility=".len()..];
677                opts.visibility = seen.parse().map_err(|()| {
678                    err(format!(
679                        "`{seen}` is not a visibility, which is default, hidden, internal or \
680                         protected"
681                    ))
682                })?;
683            }
684            // The memory safety monitor, from section 15.4 of
685            // `spec/safe-memory/15-integration.md`. Before the optimizer's `-f` family below,
686            // because a pass that took the name `safety=detect` would otherwise be handed the
687            // flag, and the tier is not a pass.
688            _ if arg.starts_with("-fsafety=") => {
689                let tier = &arg["-fsafety=".len()..];
690                opts.safety = tier.parse().map_err(|()| {
691                    err(format!(
692                        "`{tier}` is not a safety tier, which is off, detect, enforce or kernel"
693                    ))
694                })?;
695            }
696            // The optimizer's own flags, from section 9.10 of `spec/09-optimizer.md`. These come
697            // after every `-f` the rest of the compiler answers to, so a pass can never take a
698            // name that already means something else on the command line.
699            _ if arg.starts_with("-fpass-fuel=") => {
700                let (name, count) = arg["-fpass-fuel=".len()..]
701                    .split_once('=')
702                    .ok_or_else(|| err("-fpass-fuel= is spelled <pass>=<count>"))?;
703                if rucc_opt::pass::find(name).is_none() {
704                    return Err(err(format!(
705                        "`{name}` is not a pass this compiler has, see --print-pipeline"
706                    )));
707                }
708                let count: u32 = count
709                    .parse()
710                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
711                opts.pass_fuel.push((name.to_owned(), count));
712            }
713            _ if arg.starts_with("-fpass-fuel-global=") => {
714                let count = &arg["-fpass-fuel-global=".len()..];
715                let count: u32 = count
716                    .parse()
717                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
718                opts.pass_fuel_global = Some(count);
719            }
720            // Everything from `-fopt-info` to the end of the argument, which is optional
721            // keywords joined by hyphens and an optional `=<file>`. Checked here rather than
722            // where the remarks are printed, because by then the compilation somebody wanted
723            // to hear about is over.
724            _ if arg == "-fopt-info"
725                || arg.starts_with("-fopt-info=")
726                || arg.starts_with("-fopt-info-") =>
727            {
728                let rest = &arg["-fopt-info".len()..];
729                let (kinds, file) = match rest.split_once('=') {
730                    Some((kinds, file)) => (kinds, Some(file)),
731                    None => (rest, None),
732                };
733                let kinds = kinds.strip_prefix('-').unwrap_or(kinds);
734                rucc_opt::Wants::none().add(kinds).map_err(err)?;
735                opts.opt_info.push(kinds.to_owned());
736                if let Some(file) = file {
737                    if file.is_empty() {
738                        return Err(err("-fopt-info= was given no file to write to"));
739                    }
740                    opts.opt_info_file = Some(file.to_owned());
741                }
742            }
743            _ if arg.starts_with("-fdump-ir=") => {
744                // Checked here rather than where the dumps are taken, because the compilation
745                // that would have been dumped is over by then.
746                let spec = &arg["-fdump-ir=".len()..];
747                rucc_opt::Dumps::default().add(spec).map_err(err)?;
748                opts.dump_ir.push(spec.to_owned());
749            }
750            // Before the bare `-f<pass>` below, because a pass called `enable-something` would
751            // otherwise take the flag away from the gate. Checked here rather than where the
752            // pipeline reads it, for the reason that applies to all of these: a misspelled pass
753            // name that quietly gated nothing looks exactly like a pass that is not the guilty
754            // one, and a bisection would carry on past the thing it was looking for.
755            _ if arg.starts_with("-fdisable-") || arg.starts_with("-fenable-") => {
756                let on = arg.starts_with("-fenable-");
757                let spec = &arg[if on { "-fenable-".len() } else { "-fdisable-".len() }..];
758                rucc_opt::Gates::default().add(on, spec).map_err(err)?;
759                opts.pass_gates.push((on, spec.to_owned()));
760            }
761            _ if arg.strip_prefix("-fno-").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
762                opts.passes.push((arg["-fno-".len()..].to_owned(), false));
763            }
764            _ if arg.strip_prefix("-f").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
765                opts.passes.push((arg["-f".len()..].to_owned(), true));
766            }
767            // The unstable options, spelled the way rustc spells them and carrying the same
768            // promise, which is none: one of these may change or go away in any release. They are
769            // measurements and debugging aids rather than things a build asks for, which is why
770            // none of them is in the usage text and all of them are in section 4.11 of
771            // `spec/04-driver-and-cli.md`.
772            "-Zverify-each" => opts.verify_each = true,
773            _ if arg.starts_with("-Zrule-coverage=") => {
774                let file = &arg["-Zrule-coverage=".len()..];
775                if file.is_empty() {
776                    return Err(err("-Zrule-coverage= needs a file to write to"));
777                }
778                opts.rule_coverage = Some(file.to_owned());
779            }
780            _ if arg.starts_with("-Z") => {
781                return Err(err(format!(
782                    "`{arg}` is not an unstable option this compiler has, see \
783                     spec/04-driver-and-cli.md section 4.11 for the ones it does"
784                )));
785            }
786            // The word size, which is a statement about the target and is taken as one. A build
787            // that says the size the target already has is saying nothing, and one that says the
788            // other size is asking for a target this compiler does not have, which it is told
789            // rather than being given the wrong one.
790            "-m64" | "-m32" | "-mx32" => {
791                let want: u32 = match arg {
792                    "-m64" => 64,
793                    _ => 32,
794                };
795                let have = rucc_target::TargetInfo::new(opts.target).pointer_width;
796                if have != want {
797                    return Err(err(format!(
798                        "{arg} asks for a {want} bit target and {} is {have} bit, use \
799                         --target= to name the one you mean",
800                        opts.target
801                    )));
802                }
803            }
804            // Which processor in the family to generate for. This compiler emits the base
805            // instruction set of the architecture and nothing above it, so a program built with
806            // any of these runs on the machine that was named; it is a program that could have
807            // been faster rather than a program that is wrong, which is what makes these safe to
808            // take and ignore where a flag that changed the meaning of the code would not be.
809            _ if arg.starts_with("-march=")
810                || arg.starts_with("-mtune=")
811                || arg.starts_with("-mcpu=") => {}
812            // The calling convention, which is not safe to ignore. Taken when it names the one
813            // the target already uses and refused otherwise.
814            _ if arg.starts_with("-mabi=") => {
815                let want = &arg["-mabi=".len()..];
816                let have = match opts.target.arch {
817                    rucc_target::Arch::X86_64 => "sysv",
818                    rucc_target::Arch::Aarch64 => "lp64",
819                    rucc_target::Arch::Riscv64 => "lp64d",
820                };
821                if want != have {
822                    return Err(err(format!(
823                        "{arg}: {} uses the {have} convention and this compiler has no other",
824                        opts.target
825                    )));
826                }
827            }
828            // How far apart the pieces of the program may be. The small model is what we emit and
829            // it is every hosted program's default; the kernel model is a different one and a
830            // build that asks for it and does not get it links and then does not run.
831            "-mcmodel=small" => {}
832            _ if arg.starts_with("-mcmodel=") => {
833                return Err(err(format!(
834                    "{arg}: this compiler emits the small code model and no other, see \
835                     spec/12-targets.md"
836                )));
837            }
838            // GCC's own scripting language for how the driver builds a command line.
839            // `spec/04-driver-and-cli.md` section 4.4 settles that we will not have it, so a
840            // build reaching for it is told which flags do the same job.
841            _ if arg.starts_with("-specs=") => {
842                return Err(err(
843                    "-specs= is not supported: the parts of it builds rely on are -B, -L, \
844                     -nostdlib, -nostartfiles and -Wl,, see spec/04-driver-and-cli.md \
845                     section 4.4",
846                ));
847            }
848            // Arguments meant for a separate assembler or preprocessor, which this compiler does
849            // not have: both are inside it and neither reads a command line. Refused rather than
850            // dropped, because every one of these says something about the output and a build
851            // that asked for `-Wa,--noexecstack` and was silently given an executable stack got
852            // the opposite of what it asked for.
853            _ if arg.starts_with("-Wa,") || arg.starts_with("-Wp,") => {
854                return Err(err(format!(
855                    "`{arg}` is an argument for a separate assembler or preprocessor, and both \
856                     are inside this compiler rather than programs it runs"
857                )));
858            }
859            "-Xassembler" | "-Xpreprocessor" => {
860                return Err(err(format!(
861                    "{arg} hands an argument to a separate assembler or preprocessor, and both \
862                     are inside this compiler rather than programs it runs"
863                )));
864            }
865            // Everything else in the `-W` family. `spec/04-driver-and-cli.md` section 4.1 has
866            // this one as a rule about build systems rather than about warnings: autoconf finds
867            // out whether a warning flag exists by passing it and looking at the exit status, so
868            // a compiler that refuses one it has not heard of fails a configure script written
869            // for a GCC newer than itself. The names are not checked against a list because this
870            // compiler has no warning groups for a list to be of, which #485 is about.
871            _ if arg.starts_with("-W") => {}
872            // Flags that name something this compiler does not do and would not do differently
873            // if it did. `-fno-ident` is about a comment in the output that we do not write
874            // either way, and the others are about a way of ordering the compilation that has
875            // been GCC's only way for twenty years. Section 4.1 asks for the list to be short
876            // and for adding to it to be deliberate, which is why it is written out here.
877            "-fno-ident"
878            | "-fident"
879            | "-funit-at-a-time"
880            | "-fno-unit-at-a-time"
881            | "-shared-libgcc"
882            | "-static-libgcc" => {}
883            _ if arg.starts_with('-') && arg.len() > 1 => {
884                // Silently ignoring an unknown flag is how a build ends up not doing what
885                // its author asked. spec/13-gnu-compat.md section 13.4 makes this an error
886                // for the flags that change code generation, and the safe default until the
887                // flag table is populated is to reject everything we do not know.
888                return Err(err(format!("unknown option `{arg}`")));
889            }
890            _ => inputs.push(Input { path: arg.to_owned(), forced, library: false }),
891        }
892    }
893
894    // Last, so that it lands after every `-isystem` the command line gave. That is GCC's
895    // order: a directory the user names outranks the compiler's own, and the compiler's own
896    // outranks the library's. It is pushed after the loop rather than before it because
897    // `SearchPath` appends within a group and the position is what the order is.
898    // The same directory the headers were looked for under, because a sysroot is a statement
899    // about a whole installation and not about half of one.
900    link.sysroot = sysroot.clone();
901    // After the loop rather than where `-pthread` was read, so that it lands after the objects
902    // that refer to it. A static link takes the definitions it needs from a library when it
903    // reaches it and not afterwards, so a library before the objects is a library that answers
904    // nothing.
905    if threads {
906        inputs.push(Input::library("pthread"));
907    }
908    if let Some(query) = query {
909        return Ok(Action::Print(answer(&query, &opts, &link)));
910    }
911    // `-M` and `-MM` produce the rule and nothing else, so the run stops after phase 4 whatever
912    // else the command line asked for. Read here rather than where the flag was, because a `-c`
913    // written after it has to lose and the loop cannot know that until it has ended. The output
914    // file is where the rule goes rather than where an object would have gone, and the last
915    // phase being the preprocessor is what makes that true without a second rule for it.
916    if opts.deps.instead_of_compiling {
917        opts.emit = EmitKind::Preprocessed;
918    }
919    if !nostdinc {
920        opts.search.push_system(runtime::DIR);
921        // And the library's after ours, which is the other half of the same order. They go on
922        // here rather than at the point `--target=` or `--sysroot=` was read because either
923        // one changes the answer and the last word on both is the end of the loop.
924        for dir in library::system_dirs(opts.target, sysroot.as_deref()) {
925            opts.search.push_system(dir);
926        }
927    }
928    // Once, here, rather than as each directory is pushed. A `-I` that names a system
929    // directory has to lose to the system entry and the system entry is added last, so the
930    // question cannot be answered until the whole path is known.
931    opts.search.remove_duplicates();
932
933    // The target has to be resolved before the configuration is printed, so this check comes
934    // after the loop rather than at the point `--print-config` was seen.
935    if print_config {
936        return Ok(Action::PrintConfig(Box::new(opts)));
937    }
938    if print_pipeline {
939        return Ok(Action::PrintPipeline(Box::new(opts)));
940    }
941    let plan = Plan::new(&opts, &inputs, output.as_deref()).map_err(|e| err(e.message))?;
942    if print_plan {
943        return Ok(Action::PrintPlan {
944            opts: Box::new(opts),
945            plan: Box::new(plan),
946            link: Box::new(link),
947        });
948    }
949    Ok(Action::Compile {
950        opts: Box::new(opts),
951        plan: Box::new(plan),
952        link: Box::new(link),
953        jobs,
954        verbose,
955    })
956}
957
958/// What one of the `-dump` and `-print` flags prints.
959///
960/// GCC prints the name back unchanged when it cannot find the file a `-print` flag asked about,
961/// which is what makes the answer safe to paste into a link line whether or not the file is
962/// there, and this does the same.
963fn answer(query: &Query, opts: &Options, link: &LinkOptions) -> String {
964    let found = |name: &str| {
965        link::find_in_search(link, opts.target, name)
966            .map_or_else(|| name.to_owned(), |path| path.display().to_string())
967    };
968    match query {
969        Query::Machine => opts.target.to_string(),
970        Query::Version => VERSION.to_owned(),
971        Query::Multiarch => link::multiarch(opts.target),
972        // The three lines GCC prints, in its order and with its punctuation, because what reads
973        // them is a script written against that shape. There is no installation directory to
974        // report: this compiler is one binary that works wherever it is copied, and the headers
975        // it ships are inside it, so `install` is where the binary is and nothing is under it.
976        Query::SearchDirs => {
977            let here = std::env::current_exe()
978                .ok()
979                .and_then(|p| p.parent().map(std::path::Path::to_path_buf))
980                .unwrap_or_default();
981            let list = |dirs: &[PathBuf]| {
982                dirs.iter().map(|d| d.display().to_string()).collect::<Vec<_>>().join(":")
983            };
984            let libraries = link::search_dirs(link, opts.target);
985            format!(
986                "install: {}\nprograms: ={}\nlibraries: ={}",
987                here.display(),
988                list(&link.prefixes),
989                list(&libraries)
990            )
991        }
992        Query::FileName(name) => found(name),
993        // The name GCC gives the library of routines a compiler's output calls that the C
994        // library does not have. Ours is built in and there is no file, so the answer is the
995        // name itself, which is what GCC prints when it cannot find one either.
996        Query::Libgcc => found("libgcc.a"),
997        // A program rather than a library: the linker and the archiver are the ones a build asks
998        // about, and this compiler finds them on the path or under `-B` rather than shipping
999        // them, so the name back is the honest answer unless a `-B` prefix holds one.
1000        Query::ProgName(name) => link
1001            .prefixes
1002            .iter()
1003            .map(|dir| dir.join(name))
1004            .find(|path| path.is_file())
1005            .map_or_else(|| name.clone(), |path| path.display().to_string()),
1006    }
1007}
1008
1009/// Renders the passes this level will run, in order, with what each one does.
1010///
1011/// The level is the whole of the answer unless a `-f` flag edited it, which is section 9.1 of
1012/// `spec/09-optimizer.md`: a level is a list somebody wrote down rather than something that
1013/// emerges from which flags happen to be set, and this is how that list is read.
1014#[must_use]
1015pub fn print_pipeline(opts: &Options) -> String {
1016    let mut settings = rucc_opt::Options::for_level(opts.opt_level);
1017    settings.toggles.clone_from(&opts.passes);
1018    settings.global_fuel = opts.pass_fuel_global;
1019    for (on, spec) in &opts.pass_gates {
1020        // Every spelling was checked while the arguments were parsed, so there is nothing here
1021        // this can refuse, and a listing is not the place to report it if there were.
1022        let _ = settings.gates.add(*on, spec);
1023    }
1024    rucc_opt::pipeline::print(&settings)
1025}
1026
1027/// Renders the resolved configuration.
1028///
1029/// One `key: value` per line, sorted by nothing in particular but fixed in order, because
1030/// this output is diffed across hosts in CI and a reordering would read as a change.
1031#[must_use]
1032pub fn print_config(opts: &Options) -> String {
1033    let sess = Session::new(opts.clone());
1034    let t = &sess.target;
1035    let mut out = String::new();
1036    let _ = writeln!(out, "version: {VERSION}");
1037    // The three field triple the driver was given rather than the ten field tuple it widens to,
1038    // because this output is what a build system reads to find out what it asked for. The tuple is
1039    // the compiler's model of the machine and this line is a receipt for a command line.
1040    let _ = writeln!(out, "target: {}", opts.target);
1041    let _ = writeln!(out, "arch: {}", opts.target.arch.as_str());
1042    let _ = writeln!(out, "os: {}", opts.target.os.as_str());
1043    let _ = writeln!(out, "env: {}", opts.target.env.as_str());
1044    let _ = writeln!(out, "object-format: {}", t.object_format.as_str());
1045    let _ = writeln!(out, "pointer-width: {}", t.pointer_width);
1046    let _ = writeln!(out, "long-width: {}", t.long_width);
1047    let _ = writeln!(out, "long-double-width: {}", t.long_double_width);
1048    let _ = writeln!(out, "endian: {}", if t.little_endian { "little" } else { "big" });
1049    let _ = writeln!(out, "char-signed: {}", t.char_is_signed);
1050    let _ = writeln!(out, "va-list: {}", t.va_list.map_or("none", |list| list.as_str()));
1051    // The register file as a count per class, which is enough to tell a target whose registers
1052    // are described from one whose are not without printing sixteen names nobody asked for.
1053    let regs: Vec<String> = t
1054        .regs
1055        .classes()
1056        .map(|(class, info)| format!("{} {}", info.name, t.regs.len(class)))
1057        .collect();
1058    let _ = writeln!(
1059        out,
1060        "registers: {}",
1061        if regs.is_empty() { "none".to_string() } else { regs.join(", ") }
1062    );
1063    let _ = writeln!(out, "opt-level: {}", sess.opts.opt_level);
1064    let _ = writeln!(out, "safety: {}", sess.opts.safety);
1065    let _ = writeln!(out, "emit: {}", sess.opts.emit.as_str());
1066    let _ = writeln!(out, "debug-info: {}", sess.opts.debug_info);
1067    let _ = writeln!(out, "frame-pointer: {}", sess.opts.frame_pointer);
1068    let _ = writeln!(out, "red-zone: {}", sess.opts.red_zone);
1069    // Last because it is the one key with more than one line under it, and the only one
1070    // whose value is a property of the machine rather than of the command line.
1071    for dir in sess.opts.search.dirs() {
1072        let system = if dir.is_system { " (system)" } else { "" };
1073        let _ = writeln!(out, "include: {}{system}", dir.path.display());
1074    }
1075    out
1076}
1077
1078/// The output name the make target is taken from, which is the `-o` argument or nothing.
1079///
1080/// A run that stops at the preprocessor has not named an object, whatever its `-o` says: under
1081/// `-E` that argument is the preprocessed text and under `-M` it is the rule itself, and neither
1082/// is a file `make` would rebuild by running this rule. GCC agrees and falls back to the source
1083/// name in both, which is why a `-MD -E -o out.i` writes `out.d` holding a rule for `a.o`. From
1084/// `-S` on the argument does name what the rule builds, and it is used as written.
1085fn deps_target_output<'a>(opts: &Options, plan: &'a Plan) -> Option<&'a str> {
1086    if opts.emit == EmitKind::Preprocessed { None } else { plan.output.as_deref() }
1087}
1088
1089/// Writes to a path the command line named rather than one the plan derived, where `-` is
1090/// standard output.
1091fn write_named(path: &str, bytes: &[u8]) -> Result<(), String> {
1092    if path == "-" {
1093        return write_out(&Output::Stdout, bytes);
1094    }
1095    write_out(&Output::File(path.to_owned()), bytes)
1096}
1097
1098/// Writes the make rule for one input, and reports whether it got there.
1099///
1100/// A rule with no file of its own goes where the compilation it replaced would have written,
1101/// which is what makes the usual makefile recipe work: `rucc -M $< -o $@` leaves the rule in
1102/// `$@`, and the same line with the `-o` left off puts it on standard output.
1103fn write_deps(
1104    opts: &Options,
1105    plan: &Plan,
1106    job: &Job,
1107    found: &[Dependency],
1108    stderr: &mut impl std::io::Write,
1109) -> bool {
1110    let targets = if opts.deps.targets.is_empty() {
1111        vec![deps::default_target(&job.input, deps_target_output(opts, plan))]
1112    } else {
1113        opts.deps.targets.clone()
1114    };
1115    let rule = deps::rule(&opts.deps, &targets, &job.input, found);
1116    // The file, on the other hand, is named after the `-o` in every mode that still has one to
1117    // spend, which is every mode except the two that spend it on the rule.
1118    let wrote = match deps::default_file(&opts.deps, &job.input, plan.output.as_deref()) {
1119        // A `-MF` on a run that had nowhere else to put the rule leaves the file the `-o`
1120        // named empty rather than absent, because a makefile that named it as a target of its
1121        // own is a makefile that will look for it.
1122        Some(path) => write_named(&path, rule.as_bytes()).and_then(|()| {
1123            if opts.deps.instead_of_compiling { write_out(&job.output, b"") } else { Ok(()) }
1124        }),
1125        None => write_out(&job.output, rule.as_bytes()),
1126    };
1127    if let Err(e) = wrote {
1128        let _ = writeln!(stderr, "rucc: error: {e}");
1129        return false;
1130    }
1131    true
1132}
1133
1134/// Runs phase 4 over every input that has one, and writes what came out.
1135///
1136/// One input that fails does not stop the others. A build that reports every file it could
1137/// not preprocess in one run is worth more than one that stops at the first, and the exit
1138/// status is still a failure either way.
1139fn preprocess_all(opts: &Options, plan: &Plan) -> i32 {
1140    let fs = OsFileSystem::new();
1141    let mut stderr = std::io::stderr().lock();
1142    let mut failed = false;
1143    for job in &plan.jobs {
1144        if !job.phases.first().is_some_and(|p| *p == Phase::Preprocess) {
1145            // An input that is already preprocessed, or an object file. GCC passes these
1146            // through untouched, and the plan has already said so in its notes.
1147            continue;
1148        }
1149        let started = std::time::Instant::now();
1150        let result = preprocess(opts, &job.input, &fs);
1151        if opts.time {
1152            say_time(&job.input, started.elapsed(), &mut stderr);
1153        }
1154        for message in &result.messages {
1155            let _ = writeln!(stderr, "{message}");
1156        }
1157        if result.failed() {
1158            failed = true;
1159            continue;
1160        }
1161        if opts.deps.emit {
1162            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1163            // `-M` and `-MM` asked for the rule instead of the text, so there is nothing else
1164            // to write. The other two asked for both and fall through to the text below.
1165            if opts.deps.instead_of_compiling {
1166                continue;
1167            }
1168        }
1169        if let Err(e) = write_out(&job.output, result.text.as_bytes()) {
1170            let _ = writeln!(stderr, "rucc: error: {e}");
1171            failed = true;
1172        }
1173    }
1174    i32::from(failed)
1175}
1176
1177/// Runs the front end over every input that has a compile phase, and writes what came out.
1178///
1179/// The same rule as [`preprocess_all`]: one input that fails does not stop the others, and the
1180/// exit status is a failure either way. An input that is already assembly or an object has no
1181/// compile phase and is passed over here, which the plan has already said in its notes.
1182fn compile_all(opts: &Options, plan: &Plan) -> i32 {
1183    let fs = OsFileSystem::new();
1184    let mut stderr = std::io::stderr().lock();
1185    let mut failed = false;
1186    let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1187    failed |= !ok;
1188    let mut fired = Fired::new();
1189    for job in &plan.jobs {
1190        if !job.phases.contains(&Phase::Compile) {
1191            continue;
1192        }
1193        // An input of IR is read back rather than compiled, since the C it came from is not
1194        // here any more. Everything after this is the same, so the two paths meet again at the
1195        // messages and the file the result is written to.
1196        let started = std::time::Instant::now();
1197        let result = if job.kind == InputKind::Ir {
1198            compile_ir(opts, &job.input, &fs)
1199        } else {
1200            compile(opts, &job.input, &fs)
1201        };
1202        if opts.time {
1203            say_time(&job.input, started.elapsed(), &mut stderr);
1204        }
1205        fired.merge(&result.fired);
1206        failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1207        failed |= !remarks.write(&result.remarks, &mut stderr);
1208        for message in &result.messages {
1209            let _ = writeln!(stderr, "{message}");
1210        }
1211        // Before the failure below, because a compilation that stopped in the back end is exactly
1212        // the one whose preprocessed source somebody wants to look at.
1213        failed |= !write_temps(job, &result.temps, &mut stderr);
1214        if result.failed() {
1215            failed = true;
1216            continue;
1217        }
1218        // `-MD` and `-MMD` write the rule beside the object and let the compilation happen, so
1219        // this is the one path where both files come out of the same run. An input of IR has no
1220        // dependencies to report and produces an empty list, which produces a rule naming only
1221        // itself, and that is the honest answer rather than a missing file.
1222        if opts.deps.emit {
1223            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1224        }
1225        if let Err(e) = write_out(&job.output, result.artifact.bytes()) {
1226            let _ = writeln!(stderr, "rucc: error: {e}");
1227            failed = true;
1228        }
1229    }
1230    failed |= !write_coverage(opts, &fired, &mut stderr);
1231    i32::from(failed)
1232}
1233
1234/// A directory for the object files only the link step ever sees, removed when it goes away.
1235///
1236/// `-c` writes its object where the user can see it and linking does not, which is the whole of
1237/// the difference: a `rucc a.c b.c` leaves an executable behind and nothing else, the same as
1238/// every other compiler. Removing them on drop rather than at the end of a function is so that a
1239/// link that failed leaves nothing behind either.
1240struct Scratch {
1241    /// Where the objects go.
1242    dir: PathBuf,
1243}
1244
1245impl Scratch {
1246    /// Makes one, under whatever the platform calls its temporary directory.
1247    ///
1248    /// The name carries the process id so that two compilers running at once do not share a
1249    /// directory, which they would otherwise do the moment two of them compiled a file of the
1250    /// same name.
1251    fn new() -> Result<Scratch, String> {
1252        let dir = std::env::temp_dir().join(format!("rucc-{}", std::process::id()));
1253        std::fs::create_dir_all(&dir).map_err(|e| format!("{}: {e}", dir.display()))?;
1254        Ok(Scratch { dir })
1255    }
1256}
1257
1258impl Drop for Scratch {
1259    fn drop(&mut self) {
1260        let _ = std::fs::remove_dir_all(&self.dir);
1261    }
1262}
1263
1264/// The link line the plan describes, for `-###`.
1265///
1266/// The names in it are the hints the plan carries rather than the temporaries a real compilation
1267/// would choose, because `-###` prints the line without having compiled anything and so has
1268/// nothing to point at. That also makes the printed line readable rather than naming a directory
1269/// that only exists while a compilation is running.
1270fn link_line(opts: &Options, link: &LinkOptions, job: &LinkJob) -> Result<String, link::Error> {
1271    let linker = link::find(opts.target, link)?;
1272    let args = link::line(opts.target, link, &job.inputs, &job.output)?;
1273    Ok(link::render(&linker, &args))
1274}
1275
1276/// Compiles everything, then links it.
1277///
1278/// The objects go in a directory that is removed afterwards, which is why this is not
1279/// [`compile_all`] followed by a link: the plan says an object feeding the linker is temporary
1280/// and does not say where, because where is a question that only has an answer once something is
1281/// running.
1282fn link_all(opts: &Options, plan: &Plan, link: &LinkOptions, verbose: bool) -> i32 {
1283    let Some(job) = &plan.link else {
1284        // Every path into here comes from a plan whose last phase is the link, and such a plan
1285        // has a link job. Saying so is cheaper than an unwrap that would have to be explained.
1286        let mut stderr = std::io::stderr().lock();
1287        let _ = writeln!(stderr, "rucc: error: there is nothing to link");
1288        return 1;
1289    };
1290    // Before anything is compiled, because a linker that is not on the machine is worth knowing
1291    // about in the second it takes to look rather than after the compilation.
1292    let linker = match link::find(opts.target, link) {
1293        Ok(linker) => linker,
1294        Err(why) => return complain(why),
1295    };
1296
1297    let scratch = match Scratch::new() {
1298        Ok(scratch) => scratch,
1299        Err(why) => return complain(format!("could not make a place for the object files: {why}")),
1300    };
1301
1302    let fs = OsFileSystem::new();
1303    let mut failed = false;
1304    // One per job, in job order, which is what lets the link line below be rebuilt with the real
1305    // paths in it: every job contributes exactly one file to the line and does so in this order.
1306    let mut produced: Vec<String> = Vec::with_capacity(plan.jobs.len());
1307    let mut fired = Fired::new();
1308    {
1309        let mut stderr = std::io::stderr().lock();
1310        let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1311        failed |= !ok;
1312        for (at, job) in plan.jobs.iter().enumerate() {
1313            let out = match &job.output {
1314                Output::Temporary(hint) => {
1315                    // The index because two inputs in different directories can have the same
1316                    // name, and the two objects of `rucc a/x.c b/x.c` must not be one file.
1317                    scratch.dir.join(format!("{at}-{hint}")).display().to_string()
1318                }
1319                Output::File(path) => path.clone(),
1320                // A job feeding the linker never writes to standard output, since the plan gives
1321                // it a temporary. This is here so that the match is total rather than a panic.
1322                Output::Stdout => continue,
1323            };
1324            produced.push(out.clone());
1325            if !job.phases.contains(&Phase::Compile) {
1326                continue;
1327            }
1328            let started = std::time::Instant::now();
1329            let result = if job.kind == InputKind::Ir {
1330                compile_ir(opts, &job.input, &fs)
1331            } else {
1332                compile(opts, &job.input, &fs)
1333            };
1334            if opts.time {
1335                say_time(&job.input, started.elapsed(), &mut stderr);
1336            }
1337            fired.merge(&result.fired);
1338            failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1339            failed |= !remarks.write(&result.remarks, &mut stderr);
1340            for message in &result.messages {
1341                let _ = writeln!(stderr, "{message}");
1342            }
1343            failed |= !write_temps(job, &result.temps, &mut stderr);
1344            if result.failed() {
1345                failed = true;
1346                continue;
1347            }
1348            // A `-MD` on a command line that links writes the rule next to the executable and
1349            // names the executable as its target, since that is the file this source builds
1350            // here. The object it went through is in a temporary directory and is gone by the
1351            // time `make` reads any of this.
1352            if opts.deps.emit {
1353                failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1354            }
1355            if !matches!(result.artifact, Artifact::Object(_)) {
1356                // Worth saying rather than writing whatever it is and letting the linker read it.
1357                // An empty file is a valid empty linker script, so a link handed one gets as far
1358                // as reporting every symbol of this file undefined, which is a page of messages
1359                // about something that went wrong here.
1360                let _ = writeln!(
1361                    stderr,
1362                    "rucc: internal error: {}: no object file was produced for the link",
1363                    job.input
1364                );
1365                failed = true;
1366                continue;
1367            }
1368            if let Err(e) = std::fs::write(&out, result.artifact.bytes()) {
1369                let _ = writeln!(stderr, "rucc: error: {out}: {e}");
1370                failed = true;
1371            }
1372        }
1373        failed |= !write_coverage(opts, &fired, &mut stderr);
1374    }
1375    if failed {
1376        // Nothing is linked from a compilation that did not finish. A linker run over the objects
1377        // that did compile would report every function of the file that did not as undefined,
1378        // which is a page of messages about a mistake already reported once.
1379        return 1;
1380    }
1381
1382    // The items in command line order with the temporaries filled in. A library contributes no
1383    // job and passes through, and every file item takes the next job's real output, which is
1384    // what keeps a library that was written between two objects between them here.
1385    let mut outputs = produced.into_iter();
1386    let mut items = Vec::with_capacity(job.inputs.len());
1387    for item in &job.inputs {
1388        match item {
1389            link::Item::Library(name) => items.push(link::Item::Library(name.clone())),
1390            link::Item::File(_) => match outputs.next() {
1391                Some(path) => items.push(link::Item::File(path)),
1392                None => return complain("the plan asks the linker for a file nothing produced"),
1393            },
1394        }
1395    }
1396
1397    let args = match link::line(opts.target, link, &items, &job.output) {
1398        Ok(args) => args,
1399        Err(why) => return complain(why),
1400    };
1401    if verbose {
1402        let mut stderr = std::io::stderr().lock();
1403        let _ = writeln!(stderr, "{}", link::render(&linker, &args));
1404    }
1405    let started = std::time::Instant::now();
1406    let ran = link::run(&linker, &args);
1407    if opts.time {
1408        // The one step of a compilation that really is another program, so this line is the same
1409        // measurement gcc's is and names the linker the way gcc names `collect2`.
1410        let mut stderr = std::io::stderr().lock();
1411        say_time(&linker.name, started.elapsed(), &mut stderr);
1412    }
1413    match ran {
1414        Ok(()) => 0,
1415        // The linker has already said what was wrong on its own error output, and repeating that
1416        // linking failed would only push its message further up the screen.
1417        Err(link::Error::Refused { .. }) => 1,
1418        Err(why) => complain(why),
1419    }
1420}
1421
1422/// Prints one driver level message and gives back the exit status that goes with it.
1423fn complain(why: impl std::fmt::Display) -> i32 {
1424    let mut stderr = std::io::stderr().lock();
1425    let _ = writeln!(stderr, "rucc: error: {why}");
1426    1
1427}
1428
1429/// Writes what `-Zrule-coverage=FILE` asked for, and says whether it could.
1430///
1431/// Once for the whole command line rather than once per input, because the question is which
1432/// lowering rules this run of the compiler reached and a file per input would leave the reader
1433/// unioning files to find out something one process already knew.
1434///
1435/// A file that could not be written is a failure and not a warning. What asks for this is a
1436/// measurement run, and a measurement that quietly did not happen is worse than one that stopped.
1437fn write_coverage(opts: &Options, fired: &Fired, stderr: &mut impl std::io::Write) -> bool {
1438    let Some(path) = &opts.rule_coverage else { return true };
1439    let Some(table) = coverage::table(opts.target.arch) else {
1440        let _ = writeln!(
1441            stderr,
1442            "rucc: error: there are no lowering rules for {} yet, so there is no coverage of them \
1443             to report",
1444            opts.target
1445        );
1446        return false;
1447    };
1448    match std::fs::write(path, fired.listing(table)) {
1449        Ok(()) => true,
1450        Err(e) => {
1451            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1452            false
1453        }
1454    }
1455}
1456
1457/// Where the `-fopt-info` remarks go, and how much of the run has already gone there.
1458///
1459/// Standard error by default, and one file for the whole run when `-fopt-info=<file>` named one.
1460/// A file rather than the diagnostic stream is what a harness wants: the corpus in
1461/// `tamnd/rucc-corpus` matches a rejection against what the compiler said on standard error, and
1462/// a few thousand remarks mixed into that would bury it.
1463struct Remarks {
1464    /// The file, if there is one.
1465    file: Option<String>,
1466    /// Whether anything has been written to it yet, which decides between truncating and
1467    /// appending. One file holds the whole run rather than the last input in it.
1468    started: bool,
1469}
1470
1471impl Remarks {
1472    /// Prepares the destination, emptying the file if there is one.
1473    ///
1474    /// Emptied here rather than at the first remark, because a run where no pass had anything to
1475    /// say should leave an empty file and not yesterday's. An absent file and an empty one are
1476    /// different facts and something reading this will act on the difference.
1477    fn new(file: Option<&String>, stderr: &mut impl std::io::Write) -> (Self, bool) {
1478        let mut ok = true;
1479        if let Some(path) = file {
1480            if let Err(e) = std::fs::write(path, "") {
1481                let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1482                ok = false;
1483            }
1484        }
1485        (Self { file: file.cloned(), started: false }, ok)
1486    }
1487
1488    /// Writes one input's remarks, and says whether that worked.
1489    ///
1490    /// A file that cannot be written is a failure and not a warning, for the reason
1491    /// [`write_dumps`] gives: remarks that quietly did not arrive look exactly like a compilation
1492    /// where nothing happened.
1493    fn write(&mut self, text: &str, stderr: &mut impl std::io::Write) -> bool {
1494        if text.is_empty() {
1495            return true;
1496        }
1497        let Some(path) = &self.file else {
1498            let _ = write!(stderr, "{text}");
1499            return true;
1500        };
1501        let opened = std::fs::OpenOptions::new()
1502            .write(true)
1503            .append(self.started)
1504            .truncate(!self.started)
1505            .create(true)
1506            .open(path);
1507        self.started = true;
1508        let result =
1509            opened.and_then(|mut file| std::io::Write::write_all(&mut file, text.as_bytes()));
1510        if let Err(e) = result {
1511            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1512            return false;
1513        }
1514        true
1515    }
1516}
1517
1518/// Writes what `-fdump-ir=` asked to see, one file per dump.
1519///
1520/// The name is the input file with the dump's own name and `.ir` after it, so a directory listing
1521/// after a run is the passes in the order they ran, per input. They go in the working directory
1522/// rather than beside the output, because a dump is something a person asked for at a prompt and
1523/// the working directory is where that person is.
1524///
1525/// A file that could not be written is a failure and not a warning, for the reason
1526/// [`write_coverage`] gives: what asked for this is somebody debugging a pass, and a dump that
1527/// quietly did not happen looks exactly like a pass that did not run.
1528fn write_dumps(input: &str, dumps: &[rucc_opt::Dump], stderr: &mut impl std::io::Write) -> bool {
1529    let stem = std::path::Path::new(input)
1530        .file_name()
1531        .map_or_else(|| input.to_owned(), |name| name.to_string_lossy().into_owned());
1532    let mut ok = true;
1533    for dump in dumps {
1534        let path = format!("{stem}.{}.ir", dump.name);
1535        if let Err(e) = std::fs::write(&path, &dump.text) {
1536            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1537            ok = false;
1538        }
1539    }
1540    ok
1541}
1542
1543/// Writes the files `-save-temps` kept, which is nothing at all unless it was given.
1544///
1545/// A file that could not be written is a failure rather than a warning, for the reason
1546/// [`write_dumps`] gives: somebody asked for these by name, and one that quietly did not happen
1547/// looks like a compilation that never went through that step.
1548fn write_temps(job: &Job, temps: &Temps, stderr: &mut impl std::io::Write) -> bool {
1549    let mut ok = true;
1550    let kept = [(job.saved_text(), &temps.preprocessed), (job.saved_asm(), &temps.assembly)];
1551    for (path, text) in kept {
1552        // A step the compilation did not reach has nothing to keep, and a job that is not keeping
1553        // that step has nowhere to put it. Either way there is no file here.
1554        let (Some(path), Some(text)) = (path, text) else { continue };
1555        if let Err(e) = std::fs::write(&path, text) {
1556            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1557            ok = false;
1558        }
1559    }
1560    ok
1561}
1562
1563/// One line of `-time`, which is what a step was called and how long it took.
1564///
1565/// GCC's two numbers are the user and the system time of a subprocess it ran. This compiler runs
1566/// no subprocess for anything but the link, so what is measured here is the wall clock of the
1567/// step and the second column is always zero. The shape of the line is kept because a person
1568/// reading it next to gcc's should not have to work out which column is which.
1569fn say_time(name: &str, took: std::time::Duration, stderr: &mut impl std::io::Write) {
1570    let _ = writeln!(stderr, "# {name} {:.2} {:.2}", took.as_secs_f64(), 0.0);
1571}
1572
1573/// Writes one job's result where the plan said it goes.
1574///
1575/// # Errors
1576///
1577/// Returns the message to print, which names the file when there is one, because "permission
1578/// denied" on its own does not say which file was refused.
1579fn write_out(output: &Output, bytes: &[u8]) -> Result<(), String> {
1580    match output {
1581        Output::Stdout => {
1582            let mut stdout = std::io::stdout().lock();
1583            stdout.write_all(bytes).map_err(|e| format!("writing to standard output: {e}"))
1584        }
1585        Output::File(path) | Output::Temporary(path) => {
1586            std::fs::write(path, bytes).map_err(|e| format!("{path}: {e}"))
1587        }
1588    }
1589}
1590
1591/// Runs the driver and returns the process exit code.
1592///
1593/// `args` excludes the program name. Output goes to `stdout` and errors to `stderr`, which
1594/// is the one place in the compiler that is true.
1595pub fn run(args: &[String]) -> i32 {
1596    match parse_args(args) {
1597        Ok(Action::Help) => {
1598            print!("{USAGE}");
1599            0
1600        }
1601        Ok(Action::Version) => {
1602            println!("rucc {VERSION}");
1603            0
1604        }
1605        Ok(Action::Print(line)) => {
1606            println!("{line}");
1607            0
1608        }
1609        Ok(Action::PrintConfig(opts)) => {
1610            print!("{}", print_config(&opts));
1611            0
1612        }
1613        Ok(Action::PrintPipeline(opts)) => {
1614            print!("{}", print_pipeline(&opts));
1615            0
1616        }
1617        Ok(Action::PrintPlan { opts, plan, link }) => {
1618            print!("{}", plan.render());
1619            // The line as it would be typed, which is the half of `-###` that section 4.3 says
1620            // arrives with the link. It is printed even when the linker is not on this machine,
1621            // because what a build wants from `-###` is what the compiler would do.
1622            if let Some(job) = &plan.link {
1623                match link_line(&opts, &link, job) {
1624                    Ok(line) => println!("{line}"),
1625                    Err(why) => {
1626                        let mut stderr = std::io::stderr().lock();
1627                        let _ = writeln!(stderr, "rucc: error: {why}");
1628                        return 1;
1629                    }
1630                }
1631            }
1632            0
1633        }
1634        Ok(Action::Compile { opts, plan, link, jobs, verbose }) => {
1635            {
1636                let mut stderr = std::io::stderr().lock();
1637                if verbose {
1638                    let _ = write!(stderr, "{}", plan.render());
1639                    let _ = writeln!(stderr, "workers: {}", jobs.count());
1640                }
1641            }
1642            if opts.emit == EmitKind::Preprocessed {
1643                return preprocess_all(&opts, &plan);
1644            }
1645            if opts.emit != EmitKind::Executable {
1646                return compile_all(&opts, &plan);
1647            }
1648            link_all(&opts, &plan, &link, verbose)
1649        }
1650        Err(e) => {
1651            let mut stderr = std::io::stderr().lock();
1652            let _ = writeln!(stderr, "rucc: error: {e}");
1653            let _ = writeln!(stderr, "rucc: note: run `rucc --help` for usage");
1654            1
1655        }
1656    }
1657}
1658
1659#[cfg(test)]
1660mod tests {
1661    use rucc_session::{GnucVersion, IncludeForm, OptLevel, Visibility};
1662
1663    use super::*;
1664
1665    fn args(s: &[&str]) -> Vec<String> {
1666        s.iter().map(|x| (*x).to_owned()).collect()
1667    }
1668
1669    #[test]
1670    fn help_and_version_win_over_everything_else() {
1671        assert_eq!(parse_args(&args(&["-c", "--help", "x.c"])).unwrap(), Action::Help);
1672        assert_eq!(parse_args(&args(&["--version"])).unwrap(), Action::Version);
1673    }
1674
1675    fn compile(s: &[&str]) -> (Box<Options>, Box<Plan>) {
1676        match parse_args(&args(s)).expect("expected a compilation") {
1677            Action::Compile { opts, plan, .. } => (opts, plan),
1678            other => panic!("expected a compilation, got {other:?}"),
1679        }
1680    }
1681
1682    fn linking(s: &[&str]) -> (Box<LinkOptions>, Box<Plan>) {
1683        match parse_args(&args(s)).expect("expected a compilation") {
1684            Action::Compile { link, plan, .. } => (link, plan),
1685            other => panic!("expected a compilation, got {other:?}"),
1686        }
1687    }
1688
1689    #[test]
1690    fn collects_inputs_and_flags() {
1691        let (opts, plan) = compile(&["-c", "-O2", "-g", "a.c", "b.c"]);
1692        let paths: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
1693        assert_eq!(paths, vec!["a.c", "b.c"]);
1694        assert_eq!(opts.opt_level, OptLevel::O2);
1695        assert_eq!(opts.emit, EmitKind::Object);
1696        assert!(opts.debug_info);
1697    }
1698
1699    /// The unstable options, which are spelled apart from everything else on purpose: what is
1700    /// under `-Z` promises nothing, and a build that reaches for one should have had to say so.
1701    #[test]
1702    fn an_unstable_option_is_taken_and_one_that_does_not_exist_is_refused() {
1703        let (opts, _) = compile(&["-c", "-Zrule-coverage=/tmp/rules.cov", "a.c"]);
1704        assert_eq!(opts.rule_coverage.as_deref(), Some("/tmp/rules.cov"));
1705
1706        let (plain, _) = compile(&["-c", "a.c"]);
1707        assert_eq!(plain.rule_coverage, None, "nothing is measured unless it was asked for");
1708
1709        assert!(parse_args(&args(&["-Zrule-coverage=", "a.c"])).is_err(), "a file with no name");
1710        let unknown = parse_args(&args(&["-Zwhat", "a.c"])).expect_err("there is no such option");
1711        assert!(unknown.message.contains("4.11"), "{}", unknown.message);
1712    }
1713
1714    #[test]
1715    fn a_bare_dash_o_means_o1_the_way_gcc_reads_it() {
1716        let (opts, _) = compile(&["-O", "a.c"]);
1717        assert_eq!(opts.opt_level, OptLevel::O1);
1718    }
1719
1720    #[test]
1721    fn dash_x_applies_to_later_inputs_only_and_none_stops_it() {
1722        let (_, plan) = compile(&["a.o", "-x", "c", "b.txt", "-x", "none", "c.o"]);
1723        assert_eq!(plan.jobs[0].kind, InputKind::LinkerInput);
1724        assert_eq!(plan.jobs[1].kind, InputKind::C);
1725        assert_eq!(plan.jobs[2].kind, InputKind::LinkerInput);
1726    }
1727
1728    #[test]
1729    fn dash_j_reaches_the_scheduler_and_defaults_to_the_machine() {
1730        let (_, _, jobs) = match parse_args(&args(&["-j4", "a.c"])).unwrap() {
1731            Action::Compile { opts, plan, jobs, .. } => (opts, plan, jobs),
1732            other => panic!("expected a compilation, got {other:?}"),
1733        };
1734        assert_eq!(jobs.count(), 4);
1735
1736        let default = match parse_args(&args(&["a.c"])).unwrap() {
1737            Action::Compile { jobs, .. } => jobs,
1738            other => panic!("expected a compilation, got {other:?}"),
1739        };
1740        assert_eq!(default, Jobs::available());
1741        assert!(parse_args(&args(&["-j0", "a.c"])).is_err());
1742    }
1743
1744    #[test]
1745    fn triple_hash_prints_the_plan_and_runs_nothing() {
1746        let a = parse_args(&args(&["-###", "-c", "a.c"])).unwrap();
1747        let Action::PrintPlan { plan, .. } = a else { panic!("expected a plan dump") };
1748        assert!(plan.render().contains("a.c: preprocess, compile, assemble -> a.o"));
1749    }
1750
1751    #[test]
1752    fn the_flag_that_keeps_the_intermediate_files_has_three_spellings_and_two_meanings() {
1753        // The bare one is `=obj` and not `=cwd`. gcc's manual says the opposite and gcc 16 does
1754        // this, and following the compiler is what makes a build that reads either of them find
1755        // the files where they are.
1756        assert_eq!(compile(&["-c", "-save-temps", "a.c"]).0.save_temps, SaveTemps::Object);
1757        assert_eq!(compile(&["-c", "-save-temps=obj", "a.c"]).0.save_temps, SaveTemps::Object);
1758        assert_eq!(compile(&["-c", "-save-temps=cwd", "a.c"]).0.save_temps, SaveTemps::Cwd);
1759        assert_eq!(compile(&["-c", "a.c"]).0.save_temps, SaveTemps::No);
1760        // The last one on the line decides, the way it does for every other flag with an
1761        // argument, and a keyword that is neither is fatal rather than ignored: a run that kept
1762        // nothing and said nothing looks exactly like one where the files were not produced.
1763        let (opts, _) = compile(&["-c", "-save-temps", "-save-temps=cwd", "a.c"]);
1764        assert_eq!(opts.save_temps, SaveTemps::Cwd);
1765        let e = parse_args(&args(&["-c", "-save-temps=nowhere", "a.c"])).unwrap_err();
1766        assert!(e.message.contains("accepted: cwd, obj"), "{}", e.message);
1767    }
1768
1769    #[test]
1770    fn the_flag_that_times_each_step_reaches_the_options_and_changes_nothing_else() {
1771        let (opts, plan) = compile(&["-c", "-time", "a.c"]);
1772        let (plain, without) = compile(&["-c", "a.c"]);
1773        assert!(opts.time);
1774        assert!(!plain.time);
1775        // Against the same line without the flag rather than against a spelling of the object's
1776        // name, since what the object is called is the host's business and this is not about that.
1777        assert_eq!(plan.jobs[0].output, without.jobs[0].output);
1778    }
1779
1780    #[test]
1781    fn dash_x_names_what_it_accepts_when_it_does_not_know_a_language() {
1782        let e = parse_args(&args(&["-x", "fortran", "a.c"])).unwrap_err();
1783        assert!(e.message.contains("assembler-with-cpp"), "{}", e.message);
1784    }
1785
1786    #[test]
1787    fn an_unknown_flag_is_an_error_rather_than_a_shrug() {
1788        let e = parse_args(&args(&["-fno-such-thing", "a.c"])).unwrap_err();
1789        assert!(e.message.contains("unknown option"), "{}", e.message);
1790    }
1791
1792    /// `-fpermissive` and the flag that turns it back off, which a build writes beside it when
1793    /// one directory needs the older rules and the rest of the tree does not.
1794    #[test]
1795    fn permissive_reads_in_both_directions_and_the_last_one_wins() {
1796        let (opts, _) = compile(&["-c", "a.c"]);
1797        assert!(!opts.permissive, "off unless it is asked for");
1798
1799        let (opts, _) = compile(&["-c", "-fpermissive", "a.c"]);
1800        assert!(opts.permissive);
1801
1802        let (opts, _) = compile(&["-c", "-fpermissive", "-fno-permissive", "a.c"]);
1803        assert!(!opts.permissive);
1804    }
1805
1806    #[test]
1807    fn asking_for_nested_functions_is_told_why_it_is_not_coming() {
1808        let e = parse_args(&args(&["-fnested-functions", "a.c"])).unwrap_err();
1809        assert!(e.message.contains("trampoline"), "{}", e.message);
1810        assert!(parse_args(&args(&["-fno-nested-functions", "a.c"])).is_ok());
1811    }
1812
1813    #[test]
1814    fn the_flag_every_configure_script_writes_is_taken() {
1815        // All four spellings, because a build writes whichever one its macros picked and a
1816        // compiler that takes three of them is a compiler that fails on the fourth.
1817        for flag in ["-fPIC", "-fpic", "-fPIE", "-fpie"] {
1818            let (opts, _) = compile(&["-c", flag, "a.c"]);
1819            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1820        }
1821    }
1822
1823    #[test]
1824    fn a_table_is_written_unless_the_build_says_nothing_will_walk_it() {
1825        let (opts, _) = compile(&["-c", "a.c"]);
1826        assert!(opts.unwinds(), "the default is off");
1827        let (opts, _) = compile(&["-c", "-fno-asynchronous-unwind-tables", "a.c"]);
1828        assert!(!opts.unwinds(), "the build was not taken at its word");
1829        let (opts, _) = compile(&[
1830            "-c",
1831            "-fno-asynchronous-unwind-tables",
1832            "-fasynchronous-unwind-tables",
1833            "a.c",
1834        ]);
1835        assert!(opts.unwinds(), "the last flag did not win");
1836        // The weaker request, which the same table answers, so a line that asks for a table and
1837        // against an asynchronous one gets one. That is gcc's arrangement and it turns up when a
1838        // build turns the asynchronous one off globally and a directory asks for a table back.
1839        let (opts, _) =
1840            compile(&["-c", "-fno-asynchronous-unwind-tables", "-funwind-tables", "a.c"]);
1841        assert!(opts.unwinds(), "the weaker request was dropped");
1842        let (opts, _) = compile(&["-c", "-fno-unwind-tables", "a.c"]);
1843        assert!(opts.unwinds(), "the weaker negative turned off the stronger request");
1844        let (opts, _) =
1845            compile(&["-c", "-fno-unwind-tables", "-fno-asynchronous-unwind-tables", "a.c"]);
1846        assert!(!opts.unwinds(), "both were turned off and one stayed on");
1847    }
1848
1849    #[test]
1850    fn the_flags_that_describe_what_this_compiler_already_does_are_taken() {
1851        // Every one of these is on a real build line somewhere and every one of them was an
1852        // unknown option. What they have in common is that the answer rucc gives is the answer
1853        // they ask for, so there is nothing to implement and nothing to refuse.
1854        for flag in [
1855            "-fno-common",
1856            "-fstrict-aliasing",
1857            "-fno-strict-aliasing",
1858            "-pipe",
1859            "-fdiagnostics-color",
1860            "-fno-diagnostics-color",
1861            "-fdiagnostics-color=always",
1862            "-fdiagnostics-color=never",
1863            "-fdiagnostics-color=auto",
1864        ] {
1865            let (opts, _) = compile(&["-c", flag, "a.c"]);
1866            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1867        }
1868    }
1869
1870    #[test]
1871    fn asking_the_linker_to_merge_tentative_definitions_is_told_why_it_is_not_coming() {
1872        // The one of that family that is a request rather than a description, and it is a real
1873        // difference: two files each writing `int g;` link under it and do not without it.
1874        let e = parse_args(&args(&["-fcommon", "a.c"])).unwrap_err();
1875        assert!(e.message.contains(".bss"), "{}", e.message);
1876        assert!(e.message.contains("extern"), "the way out is worth saying: {}", e.message);
1877    }
1878
1879    #[test]
1880    fn asking_for_position_dependent_code_is_told_why_it_is_not_coming() {
1881        for flag in ["-fno-pic", "-fno-pie"] {
1882            let e = parse_args(&args(&[flag, "a.c"])).unwrap_err();
1883            assert!(e.message.contains("global offset table"), "{flag}: {}", e.message);
1884            // The one it may have meant, since the two are a letter apart and one of them is
1885            // about linking and is taken.
1886            assert!(e.message.contains("-no-pie"), "{flag}: {}", e.message);
1887        }
1888    }
1889
1890    #[test]
1891    fn an_unsupported_target_names_itself() {
1892        let e = parse_args(&args(&["--target=sparc64-linux-gnu", "a.c"])).unwrap_err();
1893        assert!(e.message.contains("sparc64"), "{}", e.message);
1894    }
1895
1896    #[test]
1897    fn no_inputs_is_an_error_but_print_config_needs_none() {
1898        assert!(parse_args(&args(&[])).is_err());
1899        assert!(matches!(parse_args(&args(&["--print-config"])), Ok(Action::PrintConfig(_))));
1900    }
1901
1902    #[test]
1903    fn print_config_reports_the_target_it_was_given_not_the_host() {
1904        let a = parse_args(&args(&["--print-config", "--target=riscv64-linux-musl"])).unwrap();
1905        let Action::PrintConfig(opts) = a else { panic!("expected a configuration dump") };
1906        let text = print_config(&opts);
1907        assert!(text.contains("target: riscv64-unknown-linux-musl"), "{text}");
1908        assert!(text.contains("char-signed: false"), "{text}");
1909        assert!(text.contains("object-format: elf"), "{text}");
1910        assert!(text.contains("va-list: void-pointer"), "{text}");
1911        // RISC-V has a register file and this compiler has not written it down yet, and the
1912        // dump says which of those two it is rather than leaving the line out.
1913        assert!(text.contains("registers: none"), "{text}");
1914    }
1915
1916    #[test]
1917    fn print_config_has_one_key_per_line_and_a_fixed_order() {
1918        let opts = Options::new("x86_64-unknown-linux-gnu".parse().unwrap());
1919        let text = print_config(&opts);
1920        let keys: Vec<&str> =
1921            text.lines().map(|l| l.split(':').next().unwrap_or_default()).collect();
1922        assert_eq!(keys[0], "version");
1923        assert_eq!(keys[1], "target");
1924        assert_eq!(keys.len(), 19);
1925        assert!(text.ends_with('\n'));
1926    }
1927
1928    #[test]
1929    fn the_safety_tier_is_read_off_the_command_line_and_a_wrong_one_is_refused() {
1930        let (opts, _) = compile(&["a.c"]);
1931        assert_eq!(opts.safety, rucc_session::Safety::Off);
1932
1933        for (flag, tier) in [
1934            ("-fsafety=detect", rucc_session::Safety::Detect),
1935            ("-fsafety=enforce", rucc_session::Safety::Enforce),
1936            ("-fsafety=kernel", rucc_session::Safety::Kernel),
1937            ("-fsafety=off", rucc_session::Safety::Off),
1938        ] {
1939            let (opts, _) = compile(&[flag, "a.c"]);
1940            assert_eq!(opts.safety, tier, "{flag}");
1941        }
1942
1943        // The last one wins, the way every other repeated flag on this command line does.
1944        let (opts, _) = compile(&["-fsafety=enforce", "-fsafety=off", "a.c"]);
1945        assert_eq!(opts.safety, rucc_session::Safety::Off);
1946
1947        // A misspelled tier is refused rather than ignored. Silently compiling without the
1948        // monitor a build asked for is the one failure mode this feature cannot have.
1949        let e = parse_args(&args(&["-fsafety=on", "a.c"])).unwrap_err();
1950        assert!(e.message.contains("is not a safety tier"), "{}", e.message);
1951        assert!(parse_args(&args(&["-fsafety", "a.c"])).is_err());
1952    }
1953
1954    #[test]
1955    fn print_pipeline_answers_with_the_passes_the_level_asked_for() {
1956        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
1957        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1958        let text = print_pipeline(&opts);
1959        assert!(text.starts_with("level: -O2\n"), "{text}");
1960        assert!(text.contains("fold"), "{text}");
1961
1962        let a = parse_args(&args(&["--print-pipeline"])).unwrap();
1963        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1964        // One pass runs at `-O0` and it is the one that removes code nothing reaches, which is
1965        // not an optimization. See issue 359.
1966        assert!(print_pipeline(&opts).contains("1: simplify-cfg,"), "{}", print_pipeline(&opts));
1967
1968        let a = parse_args(&args(&["--print-pipeline", "-fno-simplify-cfg"])).unwrap();
1969        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1970        // And with that one turned off there is nothing left, which the dump says rather than
1971        // printing an empty list.
1972        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
1973    }
1974
1975    #[test]
1976    fn print_pipeline_takes_the_toggles_into_account() {
1977        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fno-fold"])).unwrap();
1978        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1979        let text = print_pipeline(&opts);
1980        // The one that was named is gone and the rest of the level is not, which is the whole
1981        // of what a toggle promises.
1982        assert!(!text.contains("fold"), "{text}");
1983        assert!(text.contains("dce"), "{text}");
1984
1985        // Every pass the compiler has, named off. Built from the registry rather than written
1986        // out, so a pass added later is turned off here too and this keeps testing the thing it
1987        // is about, which is that the toggles can empty a level.
1988        let mut off = vec!["--print-pipeline".to_owned(), "-O2".to_owned()];
1989        off.extend(rucc_opt::PASSES.iter().map(|p| format!("-fno-{}", p.name())));
1990        let spelled: Vec<&str> = off.iter().map(String::as_str).collect();
1991        let a = parse_args(&args(&spelled)).unwrap();
1992        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1993        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
1994    }
1995
1996    #[test]
1997    fn print_pipeline_says_when_a_budget_will_stop_the_run_short() {
1998        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
1999        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2000        assert!(!print_pipeline(&opts).contains("global fuel"));
2001
2002        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fpass-fuel-global=4"])).unwrap();
2003        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2004        let text = print_pipeline(&opts);
2005        // Because the listing is the answer to what this compilation will do, and a run that
2006        // stops after four rewrites is not doing what the level says it does.
2007        assert!(text.contains("global fuel: 4"), "{text}");
2008    }
2009
2010    /// A pass is turned on and off by its own name, and the order the flags were given in is
2011    /// kept, because the last spelling of a name is the one that decides.
2012    #[test]
2013    fn a_pass_is_named_by_dash_f_and_unnamed_by_dash_f_no() {
2014        let (opts, _) = compile(&["-c", "-O0", "-ffold", "-fno-fold", "-ffold", "a.c"]);
2015        assert_eq!(
2016            opts.passes,
2017            [("fold".to_owned(), true), ("fold".to_owned(), false), ("fold".to_owned(), true)]
2018        );
2019
2020        let e = parse_args(&args(&["-fno-such-pass", "a.c"])).unwrap_err();
2021        assert!(e.message.contains("unknown option"), "{}", e.message);
2022    }
2023
2024    #[test]
2025    fn pass_fuel_names_a_pass_and_a_count_and_refuses_anything_else() {
2026        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel=fold=3", "a.c"]);
2027        assert_eq!(opts.pass_fuel, [("fold".to_owned(), 3)]);
2028
2029        let e = parse_args(&args(&["-fpass-fuel=fold", "a.c"])).unwrap_err();
2030        assert!(e.message.contains("<pass>=<count>"), "{}", e.message);
2031        let e = parse_args(&args(&["-fpass-fuel=nosuch=3", "a.c"])).unwrap_err();
2032        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2033        let e = parse_args(&args(&["-fpass-fuel=fold=lots", "a.c"])).unwrap_err();
2034        assert!(e.message.contains("not a number"), "{}", e.message);
2035    }
2036
2037    #[test]
2038    fn global_pass_fuel_is_a_count_on_its_own_and_defaults_to_no_limit() {
2039        let (opts, _) = compile(&["-c", "-O2", "a.c"]);
2040        assert_eq!(opts.pass_fuel_global, None);
2041
2042        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel-global=12", "a.c"]);
2043        assert_eq!(opts.pass_fuel_global, Some(12));
2044        // And it is not the per pass flag with a longer name, so neither spelling swallows the
2045        // other.
2046        assert!(opts.pass_fuel.is_empty());
2047
2048        let e = parse_args(&args(&["-fpass-fuel-global=lots", "a.c"])).unwrap_err();
2049        assert!(e.message.contains("not a number"), "{}", e.message);
2050    }
2051
2052    #[test]
2053    fn a_gate_names_a_pass_and_optionally_the_functions_it_covers() {
2054        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold", "-fenable-fold=2-4,main", "a.c"]);
2055        assert_eq!(
2056            opts.pass_gates,
2057            [(false, "fold".to_owned()), (true, "fold=2-4,main".to_owned())],
2058            "the order is what decides, so it has to survive the parse"
2059        );
2060
2061        let e = parse_args(&args(&["-fdisable-nosuch", "a.c"])).unwrap_err();
2062        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2063        let e = parse_args(&args(&["-fenable-fold=9-2", "a.c"])).unwrap_err();
2064        assert!(e.message.contains("ends before it starts"), "{}", e.message);
2065        let e = parse_args(&args(&["-fdisable-fold=", "a.c"])).unwrap_err();
2066        assert!(e.message.contains("is empty"), "{}", e.message);
2067    }
2068
2069    #[test]
2070    fn the_pipeline_listing_says_which_passes_a_gate_touched() {
2071        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold=main", "a.c"]);
2072        let text = print_pipeline(&opts);
2073        assert!(text.contains("fold, "), "{text}");
2074        assert!(text.contains("[off for main]"), "{text}");
2075    }
2076
2077    /// The spelling is checked while the arguments are read, because a dump that names a pass
2078    /// this compiler does not have is a typo, and a typo found after the compilation has run is
2079    /// found too late to be any use.
2080    #[test]
2081    fn a_dump_is_checked_when_it_is_asked_for_rather_than_when_it_is_taken() {
2082        let (opts, _) = compile(&["-c", "-O2", "-fdump-ir=all", "-fdump-ir=after-fold", "a.c"]);
2083        assert_eq!(opts.dump_ir, ["all", "after-fold"]);
2084
2085        let e = parse_args(&args(&["-fdump-ir=after-nosuch", "a.c"])).unwrap_err();
2086        assert!(e.message.contains("nosuch"), "{}", e.message);
2087        assert!(parse_args(&args(&["-fdump-ir=sideways-fold", "a.c"])).is_err());
2088    }
2089
2090    /// Every spelling `-fopt-info` takes, and the one it does not.
2091    ///
2092    /// The keywords are checked here for the same reason a dump's pass name is: a person who
2093    /// misspelled one gets no output, and no output is also what a compilation where nothing
2094    /// happened looks like. Telling those two apart is the entire reason to reach for this flag.
2095    #[test]
2096    fn opt_info_takes_kinds_and_a_file_and_refuses_a_kind_it_does_not_have() {
2097        let (opts, _) = compile(&["-c", "-O2", "-fopt-info", "a.c"]);
2098        assert_eq!(opts.opt_info, [""], "a bare flag asks for the rewrites");
2099        assert_eq!(opts.opt_info_file, None, "and goes to standard error");
2100
2101        let (opts, _) = compile(&["-c", "-O2", "-fopt-info-missed-note", "a.c"]);
2102        assert_eq!(opts.opt_info, ["missed-note"]);
2103
2104        // Two flags add up rather than the second replacing the first, and the file is the last
2105        // one that named a file, which is how GCC treats both.
2106        let (opts, _) =
2107            compile(&["-c", "-O2", "-fopt-info-missed=one.txt", "-fopt-info-all=two.txt", "a.c"]);
2108        assert_eq!(opts.opt_info, ["missed", "all"]);
2109        assert_eq!(opts.opt_info_file.as_deref(), Some("two.txt"));
2110
2111        let e = parse_args(&args(&["-fopt-info-vectorized", "a.c"])).unwrap_err();
2112        assert!(e.message.contains("vectorized"), "{}", e.message);
2113        assert!(e.message.contains("`missed`"), "{}", e.message);
2114        let e = parse_args(&args(&["-fopt-info-missed=", "a.c"])).unwrap_err();
2115        assert!(e.message.contains("no file"), "{}", e.message);
2116    }
2117
2118    #[test]
2119    fn verify_each_is_unstable_and_off_unless_it_was_asked_for() {
2120        let (opts, _) = compile(&["-c", "-Zverify-each", "a.c"]);
2121        assert!(opts.verify_each);
2122        assert!(!USAGE.contains("verify-each"), "an unstable option stays out of the usage text");
2123    }
2124
2125    #[test]
2126    fn dash_o_needs_an_argument() {
2127        let e = parse_args(&args(&["a.c", "-o"])).unwrap_err();
2128        assert_eq!(e.message, "-o requires an argument");
2129    }
2130
2131    #[test]
2132    fn dash_d_and_dash_u_are_read_joined_or_separated_and_keep_their_order() {
2133        let (opts, _) = compile(&["-DFOO=1", "-D", "BAR", "-UBAZ", "-U", "QUX", "a.c"]);
2134        assert_eq!(opts.defines, ["FOO=1", "BAR"]);
2135        assert_eq!(opts.undefines, ["BAZ", "QUX"]);
2136    }
2137
2138    #[test]
2139    fn the_include_flags_land_on_the_chain_each_one_names() {
2140        // A sysroot with nothing under it, so that the library's own directories are the
2141        // same on every machine this test runs on, which is none of them.
2142        let (opts, _) = compile(&[
2143            "-Ii",
2144            "-iquote",
2145            "q",
2146            "-isystem",
2147            "sys",
2148            "-idirafter",
2149            "after",
2150            "--sysroot=/nowhere-at-all",
2151            "a.c",
2152        ]);
2153        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2154        // The compiler's own headers sit after every `-isystem` and before `-idirafter`,
2155        // which is where GCC puts its own: a directory the user named outranks ours.
2156        assert_eq!(dirs, ["q", "i", "sys", runtime::DIR, "after"]);
2157        assert!(!opts.search.dirs()[1].is_system);
2158        assert!(opts.search.dirs()[2].is_system);
2159    }
2160
2161    #[test]
2162    fn the_librarys_headers_come_after_the_compilers_own_and_go_away_with_them() {
2163        // Which machine this runs on decides what is on the path, so the test is about the
2164        // order rather than about the names: ours is on it, the library's follow it, and
2165        // `-nostdinc` is the one flag that takes both halves of the pair off at once.
2166        let (opts, _) = compile(&["a.c"]);
2167        let dirs = opts.search.dirs();
2168        let ours = dirs.iter().position(|d| d.path.to_str() == Some(runtime::DIR));
2169        assert_eq!(ours, Some(0), "{dirs:?}");
2170        assert!(dirs[1..].iter().all(|d| d.is_system), "{dirs:?}");
2171        let (bare, _) = compile(&["-nostdinc", "a.c"]);
2172        assert!(bare.search.dirs().is_empty(), "{:?}", bare.search.dirs());
2173    }
2174
2175    #[test]
2176    fn a_sysroot_moves_the_librarys_directories_and_nothing_else() {
2177        let (opts, _) = compile(&["-isystem", "sys", "--sysroot=/nowhere-at-all", "a.c"]);
2178        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2179        assert_eq!(dirs, ["sys", runtime::DIR]);
2180    }
2181
2182    #[test]
2183    fn dash_i_dash_moves_the_bracket_directories_into_the_quoted_chain() {
2184        let (opts, _) =
2185            compile(&["-Iinc1", "-iquote", "inc2", "-I-", "-Iinc3", "-nostdinc", "a.c"]);
2186        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2187        assert_eq!(dirs, ["inc1", "inc2", "inc3"]);
2188        // An angled include sees only what came after the flag.
2189        assert_eq!(opts.search.start(IncludeForm::Angled), 2);
2190        assert!(!opts.search.searches_current_dir());
2191    }
2192
2193    #[test]
2194    fn the_prefix_flags_stick_what_iprefix_said_on_the_front_of_what_follows_it() {
2195        let (opts, _) = compile(&[
2196            "-iprefix",
2197            "/tools/",
2198            "-iwithprefix",
2199            "late",
2200            "-iwithprefixbefore",
2201            "early",
2202            "-iprefix",
2203            "/other/",
2204            "-iwithprefix",
2205            "last",
2206            "-nostdinc",
2207            "a.c",
2208        ]);
2209        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2210        // `-iwithprefixbefore` is an `-I` and the other two are `-isystem`, which is where GCC
2211        // puts them rather than where its manual says it does.
2212        assert_eq!(dirs, ["/tools/early", "/tools/late", "/other/last"]);
2213        assert!(!opts.search.dirs()[0].is_system);
2214        assert!(opts.search.dirs()[1].is_system);
2215    }
2216
2217    #[test]
2218    fn the_files_named_on_the_command_line_keep_their_order_and_which_flag_named_them() {
2219        let (opts, _) =
2220            compile(&["-include", "one.h", "-imacros", "two.h", "-include", "3.h", "a.c"]);
2221        let names: Vec<&str> = opts.preincludes.iter().map(|p| p.name.as_str()).collect();
2222        assert_eq!(names, ["one.h", "two.h", "3.h"]);
2223        assert_eq!(opts.preincludes.iter().filter(|p| p.macros_only).count(), 1);
2224    }
2225
2226    #[test]
2227    fn nostdinc_takes_the_compilers_own_headers_off_the_path() {
2228        let (opts, _) = compile(&["-Ii", "-nostdinc", "a.c"]);
2229        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2230        assert_eq!(dirs, ["i"]);
2231    }
2232
2233    #[test]
2234    fn the_dialect_flags_set_the_language_and_the_extensions_separately() {
2235        let (opts, _) = compile(&["-std=gnu11", "a.c"]);
2236        assert_eq!(opts.std, Std::C11);
2237        assert!(opts.gnu_extensions);
2238
2239        let (opts, _) = compile(&["-std=iso9899:1999", "a.c"]);
2240        assert_eq!(opts.std, Std::C99);
2241        assert!(!opts.gnu_extensions);
2242
2243        let (opts, _) = compile(&["-ansi", "a.c"]);
2244        assert_eq!(opts.std, Std::C89);
2245        assert!(!opts.gnu_extensions);
2246
2247        let e = parse_args(&args(&["-std=c94jr", "a.c"])).unwrap_err();
2248        assert!(e.message.contains("unknown dialect"), "{}", e.message);
2249    }
2250
2251    #[test]
2252    fn the_dump_letters_are_a_family_and_everything_else_beginning_with_d_is_not() {
2253        let (opts, _) = compile(&["-dM", "a.c"]);
2254        assert!(opts.dumps.macros);
2255
2256        // Packed, the way GCC takes them, and a letter in the family we have not written yet
2257        // is accepted and does nothing rather than failing a build.
2258        let (opts, _) = compile(&["-dDM", "a.c"]);
2259        assert!(opts.dumps.macros);
2260        let (opts, _) = compile(&["-dD", "a.c"]);
2261        assert!(!opts.dumps.macros);
2262
2263        let (opts, _) = compile(&["a.c"]);
2264        assert!(!opts.dumps.any());
2265
2266        // `-dumpversion` is a different flag that happens to start the same way, and it is read
2267        // as itself rather than as a dump of nothing.
2268        assert_eq!(printed(&["-dumpversion", "a.c"]), VERSION);
2269    }
2270
2271    #[test]
2272    fn the_gcc_version_claimed_is_a_flag_and_the_short_spellings_are_the_ones_people_write() {
2273        let (opts, _) = compile(&["a.c"]);
2274        assert_eq!(
2275            opts.gnuc,
2276            GnucVersion { major: 7, minor: 0, patch: 0 },
2277            "the lowest claim a modern glibc gives its own declarations to"
2278        );
2279
2280        let (opts, _) = compile(&["-fgnuc-version=15.1.0", "a.c"]);
2281        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 1, patch: 0 });
2282
2283        // A missing component is zero. `gcc -dumpversion` says `15` on a release with no
2284        // patchlevel and a harness that pastes that back has to be understood.
2285        let (opts, _) = compile(&["-fgnuc-version=15", "a.c"]);
2286        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 0, patch: 0 });
2287
2288        let (opts, _) = compile(&["-fgnuc-version=13.2", "a.c"]);
2289        assert_eq!(opts.gnuc, GnucVersion { major: 13, minor: 2, patch: 0 });
2290
2291        let e = parse_args(&args(&["-fgnuc-version=15.x", "a.c"])).unwrap_err();
2292        assert!(e.message.contains("minor that is not a number"), "{}", e.message);
2293
2294        let e = parse_args(&args(&["-fgnuc-version=1.2.3.4", "a.c"])).unwrap_err();
2295        assert!(e.message.contains("more than three"), "{}", e.message);
2296    }
2297
2298    #[test]
2299    fn pedantic_has_two_spellings_and_is_not_the_same_knob_as_the_dialect() {
2300        let (opts, _) = compile(&["-std=c17", "-pedantic", "a.c"]);
2301        assert!(opts.pedantic);
2302        assert_eq!(opts.std, Std::C17);
2303
2304        // The `-W` family's name for it, which is what a build that groups its warning flags
2305        // tends to write.
2306        let (opts, _) = compile(&["-Wpedantic", "a.c"]);
2307        assert!(opts.pedantic);
2308
2309        let (opts, _) = compile(&["-std=c17", "a.c"]);
2310        assert!(!opts.pedantic, "a dialect on its own does not diagnose an extension");
2311    }
2312
2313    #[test]
2314    fn dash_p_and_dash_ffreestanding_reach_the_options() {
2315        let (opts, _) = compile(&["-E", "-P", "-ffreestanding", "a.c"]);
2316        assert!(!opts.line_markers);
2317        assert!(!opts.hosted);
2318        assert_eq!(opts.emit, EmitKind::Preprocessed);
2319    }
2320
2321    /// The two ways a build says it means its own function by a name the C library also has.
2322    ///
2323    /// `-fno-builtin` is all of them and `-fno-builtin-<name>` is one, and the second is what a
2324    /// build writes when it means its own `memcpy` and the library's everything else. The name is
2325    /// kept as it was written and not checked against anything, because a program is allowed to
2326    /// mean something by a name this compiler has never heard of.
2327    #[test]
2328    fn the_builtin_flags_are_read_in_both_directions_and_one_name_at_a_time() {
2329        let (opts, _) = compile(&["-c", "a.c"]);
2330        assert!(opts.builtins, "a library name means the library function by default");
2331        assert!(opts.no_builtin.is_empty());
2332
2333        let (opts, _) = compile(&["-c", "-fno-builtin", "a.c"]);
2334        assert!(!opts.builtins);
2335
2336        let (opts, _) = compile(&["-c", "-fno-builtin", "-fbuiltin", "a.c"]);
2337        assert!(opts.builtins, "the last mention decides");
2338
2339        let (opts, _) = compile(&["-c", "-fno-builtin-memcpy", "-fno-builtin-nonesuch", "a.c"]);
2340        assert!(opts.builtins, "one name is not the family");
2341        assert_eq!(opts.no_builtin, vec!["memcpy".to_owned(), "nonesuch".to_owned()]);
2342    }
2343
2344    /// `-fvisibility=`, which is on every cmake project that cares about which names it exports
2345    /// and which was refused as an unknown option until now.
2346    ///
2347    /// Four spellings and three answers. `internal` is hidden plus a promise about never taking
2348    /// the address across a component boundary, and nothing derives anything from that promise
2349    /// here, so it comes out as the weaker of the two rather than as a refusal that stops a build
2350    /// over a distinction this compiler does not make.
2351    #[test]
2352    fn visibility_takes_the_four_spellings_gcc_takes_and_refuses_the_rest() {
2353        let (opts, _) = compile(&["-c", "a.c"]);
2354        assert_eq!(opts.visibility, Visibility::Default, "exported unless something says not");
2355
2356        for (written, wanted) in [
2357            ("default", Visibility::Default),
2358            ("hidden", Visibility::Hidden),
2359            ("internal", Visibility::Hidden),
2360            ("protected", Visibility::Protected),
2361        ] {
2362            let (opts, _) = compile(&["-c", &format!("-fvisibility={written}"), "a.c"]);
2363            assert_eq!(opts.visibility, wanted, "{written}");
2364        }
2365
2366        // The last mention decides, which is what every other flag of this shape does and what a
2367        // build that turns something off for one directory relies on.
2368        let (opts, _) = compile(&["-c", "-fvisibility=hidden", "-fvisibility=default", "a.c"]);
2369        assert_eq!(opts.visibility, Visibility::Default, "the last mention decides");
2370
2371        // A spelling gcc does not take is refused rather than read as the default, because a
2372        // build that meant hidden and got exported is a library with the wrong interface and
2373        // nothing said about it anywhere.
2374        let failed = parse_args(&args(&["-fvisibility=none", "a.c"])).expect_err("refused");
2375        assert!(failed.to_string().contains("is not a visibility"), "{failed}");
2376    }
2377
2378    /// `-fgnu89-inline`, which is off by default and is not implied by anything on the command
2379    /// line, since the dialect asks for GNU's reading further in rather than through this.
2380    #[test]
2381    fn gnu89_inline_is_off_until_it_is_asked_for_and_the_last_mention_decides() {
2382        let (opts, _) = compile(&["-c", "a.c"]);
2383        assert!(!opts.gnu89_inline, "C's reading of inline by default");
2384
2385        let (opts, _) = compile(&["-c", "-fgnu89-inline", "a.c"]);
2386        assert!(opts.gnu89_inline);
2387
2388        let (opts, _) = compile(&["-c", "-fgnu89-inline", "-fno-gnu89-inline", "a.c"]);
2389        assert!(!opts.gnu89_inline, "the last mention decides");
2390
2391        // The C89 dialects are under GNU's reading whether this was written or not, so the flag
2392        // stays off there and the dialect is what the checker and the macro set both ask. That is
2393        // also why `-std=c89 -fno-gnu89-inline` needs no diagnostic: it asks for the reading the
2394        // dialect already has. gcc refuses that command line, which is measured in the issue.
2395        let (opts, _) = compile(&["-c", "-std=c89", "a.c"]);
2396        assert!(!opts.gnu89_inline);
2397    }
2398
2399    /// Both spellings of both frame flags, since a build that wants one usually writes the
2400    /// other beside it for the one file that has to be compiled the ordinary way.
2401    #[test]
2402    fn the_two_frame_flags_are_read_in_both_directions() {
2403        let (opts, _) = compile(&["-c", "a.c"]);
2404        assert!(!opts.frame_pointer, "gcc omits it above -O0 and so does this");
2405        assert!(opts.red_zone, "the psABI has one and nothing said not to use it");
2406
2407        let (opts, _) = compile(&["-c", "-fno-omit-frame-pointer", "-mno-red-zone", "a.c"]);
2408        assert!(opts.frame_pointer);
2409        assert!(!opts.red_zone);
2410
2411        let (opts, _) = compile(&[
2412            "-c",
2413            "-fno-omit-frame-pointer",
2414            "-fomit-frame-pointer",
2415            "-mno-red-zone",
2416            "-mred-zone",
2417            "a.c",
2418        ]);
2419        assert!(!opts.frame_pointer, "the last one wins, as it does in gcc");
2420        assert!(opts.red_zone);
2421    }
2422
2423    #[test]
2424    fn the_link_flags_are_collected_apart_from_the_compilation() {
2425        let (link, _) = linking(&[
2426            "-static",
2427            "-nostartfiles",
2428            "-rdynamic",
2429            "-s",
2430            "-fuse-ld=mold",
2431            "-L/opt/lib",
2432            "-B",
2433            "/opt/tools",
2434            "a.c",
2435        ]);
2436        assert!(link.is_static);
2437        assert!(link.no_startfiles);
2438        assert!(link.export_dynamic);
2439        assert!(link.strip);
2440        assert_eq!(link.use_ld.as_deref(), Some("mold"));
2441        assert_eq!(link.search, vec![PathBuf::from("/opt/lib")]);
2442        assert_eq!(link.prefixes, vec![PathBuf::from("/opt/tools")]);
2443    }
2444
2445    #[test]
2446    fn a_comma_in_dash_wl_separates_two_arguments() {
2447        let (link, _) = linking(&["-Wl,-rpath,/opt/lib", "-Xlinker", "--as-needed", "a.c"]);
2448        assert_eq!(link.passthrough, vec!["-rpath", "/opt/lib", "--as-needed"]);
2449    }
2450
2451    #[test]
2452    fn a_library_keeps_its_place_between_the_objects() {
2453        // Link order is semantic: `-lm` written between two files resolves for the one before
2454        // it and not for the one after, so a library cannot be collected into a list of its own.
2455        // The target is named because the suffix of an object is the target's and this asserts
2456        // on the names: the same command line on a Windows host plans two `.obj` files.
2457        let (_, plan) = linking(&["--target=x86_64-unknown-linux-gnu", "a.c", "-lm", "b.c"]);
2458        let link = plan.link.expect("expected a link step");
2459        assert_eq!(
2460            link.inputs,
2461            vec![
2462                link::Item::File("a.o".into()),
2463                link::Item::Library("m".into()),
2464                link::Item::File("b.o".into()),
2465            ]
2466        );
2467        // And it is not a job, because there is nothing to compile in a library.
2468        assert_eq!(plan.jobs.len(), 2);
2469    }
2470
2471    #[test]
2472    fn a_library_on_a_dash_c_line_is_a_note_rather_than_an_error() {
2473        let (_, plan) = linking(&["-c", "-lm", "a.c"]);
2474        assert!(plan.link.is_none());
2475        assert!(plan.notes.iter().any(|n| n.contains("-lm")), "{:?}", plan.notes);
2476    }
2477
2478    #[test]
2479    fn the_sysroot_reaches_the_linker_as_well_as_the_headers() {
2480        let (link, _) = linking(&["--sysroot=/opt/root", "a.c"]);
2481        assert_eq!(link.sysroot, Some(PathBuf::from("/opt/root")));
2482    }
2483
2484    fn printed(s: &[&str]) -> String {
2485        match parse_args(&args(s)).expect("expected an answer") {
2486            Action::Print(line) => line,
2487            other => panic!("expected an answer, got {other:?}"),
2488        }
2489    }
2490
2491    fn refused(s: &[&str]) -> String {
2492        parse_args(&args(s)).expect_err("expected a refusal").message
2493    }
2494
2495    #[test]
2496    fn a_warning_flag_this_compiler_has_not_heard_of_is_taken_rather_than_refused() {
2497        // The rule in section 4.1, and the reason for it is autoconf: a configure script finds
2498        // out whether a warning flag exists by passing it and looking at the exit status, so a
2499        // compiler that refuses one it does not know fails a script written for a newer GCC.
2500        let (opts, _) = compile(&["-Wall", "-Wextra", "-Wno-format-truncation", "-c", "a.c"]);
2501        assert!(!opts.warnings_are_errors);
2502        assert!(opts.warnings);
2503        // The two spellings that do mean something are still read.
2504        let (opts, _) = compile(&["-Werror", "-c", "a.c"]);
2505        assert!(opts.warnings_are_errors);
2506        let (opts, _) = compile(&["-w", "-c", "a.c"]);
2507        assert!(!opts.warnings);
2508        let (opts, _) = compile(&["-pedantic-errors", "-c", "a.c"]);
2509        assert!(opts.pedantic && opts.warnings_are_errors);
2510    }
2511
2512    #[test]
2513    fn an_argument_for_a_separate_tool_is_refused_rather_than_dropped() {
2514        // Every one of these says something about the output, so the wrong answer is silence.
2515        assert!(refused(&["-Wa,--noexecstack", "-c", "a.c"]).contains("separate assembler"));
2516        assert!(refused(&["-Wp,-DX", "-c", "a.c"]).contains("separate assembler"));
2517        assert!(refused(&["-specs=/x", "a.c"]).contains("-specs= is not supported"));
2518        assert!(refused(&["-mcmodel=kernel", "-c", "a.c"]).contains("small code model"));
2519        assert!(refused(&["-gdwarf-4", "-c", "a.c"]).contains("DWARF 5"));
2520        assert!(refused(&["-Ofast", "-c", "a.c"]).contains("fast math"));
2521        // The word size the target does not have, which is a target this compiler was not asked
2522        // for rather than a flag it does not know.
2523        let no32 = refused(&["--target=x86_64-unknown-linux-gnu", "-m32", "-c", "a.c"]);
2524        assert!(no32.contains("32 bit target"), "{no32}");
2525    }
2526
2527    #[test]
2528    fn the_levels_gcc_spells_differently_are_the_levels_they_mean() {
2529        assert_eq!(compile(&["-O", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2530        assert_eq!(compile(&["-Og", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2531        assert_eq!(compile(&["-O2", "-c", "a.c"]).0.opt_level, OptLevel::O2);
2532    }
2533
2534    #[test]
2535    fn the_machine_flags_that_name_what_we_already_do_are_taken_and_the_rest_are_not() {
2536        let line = ["--target=x86_64-unknown-linux-gnu", "-m64", "-march=x86-64-v3"];
2537        let (opts, _) =
2538            compile(&[&line[..], &["-mtune=native", "-mabi=sysv", "-c", "a.c"]].concat());
2539        assert_eq!(opts.target.to_string(), "x86_64-unknown-linux-gnu");
2540        let wrong = refused(&["--target=x86_64-unknown-linux-gnu", "-mabi=ms", "-c", "a.c"]);
2541        assert!(wrong.contains("sysv convention"), "{wrong}");
2542    }
2543
2544    #[test]
2545    fn the_thread_flag_is_a_macro_and_a_library_and_the_library_goes_last() {
2546        let (opts, plan) = compile(&["-pthread", "-c", "a.c"]);
2547        assert!(opts.defines.iter().any(|d| d == "_REENTRANT"));
2548        // After the input, because a static link takes what it needs from a library when it
2549        // reaches it and not afterwards.
2550        let names: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
2551        assert_eq!(names, vec!["a.c"]);
2552    }
2553
2554    #[test]
2555    fn the_questions_a_build_system_asks_before_it_compiles_anything() {
2556        let target = "--target=x86_64-unknown-linux-gnu";
2557        assert_eq!(printed(&[target, "-dumpmachine"]), "x86_64-unknown-linux-gnu");
2558        assert_eq!(printed(&[target, "-dumpversion"]), VERSION);
2559        assert_eq!(printed(&[target, "-dumpfullversion"]), VERSION);
2560        assert_eq!(printed(&[target, "-print-multiarch"]), "x86_64-linux-gnu");
2561        // A name nothing holds comes back unchanged, which is GCC's rule and is what makes the
2562        // answer safe to paste into a link line whether or not the file is there.
2563        assert_eq!(printed(&[target, "-print-file-name=no-such-library.a"]), "no-such-library.a");
2564        assert_eq!(printed(&[target, "-print-prog-name=ld"]), "ld");
2565        let dirs = printed(&[target, "-print-search-dirs"]);
2566        assert!(dirs.starts_with("install: "), "{dirs}");
2567        assert!(dirs.contains("\nlibraries: ="), "{dirs}");
2568    }
2569
2570    #[test]
2571    fn the_two_dependency_flags_that_stop_after_the_rule_stop_after_the_rule() {
2572        let (opts, _) = compile(&["-M", "a.c"]);
2573        assert!(opts.deps.emit && opts.deps.instead_of_compiling);
2574        assert!(opts.deps.system_headers, "plain -M lists them");
2575        assert_eq!(opts.emit, EmitKind::Preprocessed);
2576
2577        // Even where a later flag asked for something else, because the family is a mode and
2578        // the mode is what the run is for.
2579        let (opts, _) = compile(&["-M", "-c", "a.c"]);
2580        assert_eq!(opts.emit, EmitKind::Preprocessed);
2581
2582        let (opts, _) = compile(&["-MM", "a.c"]);
2583        assert!(!opts.deps.system_headers);
2584    }
2585
2586    #[test]
2587    fn the_two_that_end_in_d_leave_the_compilation_alone() {
2588        let (opts, _) = compile(&["-MD", "-c", "a.c"]);
2589        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2590        assert!(opts.deps.system_headers);
2591        assert_eq!(opts.emit, EmitKind::Object);
2592
2593        let (opts, _) = compile(&["-MMD", "-c", "a.c"]);
2594        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2595        assert!(!opts.deps.system_headers);
2596    }
2597
2598    #[test]
2599    fn nothing_puts_the_system_headers_back_once_a_flag_has_taken_them_out() {
2600        // GCC's rule, and not an oversight in it. The flag asking for fewer of them is read as
2601        // the answer, because the other one never asked the question.
2602        let (opts, _) = compile(&["-MM", "-M", "a.c"]);
2603        assert!(!opts.deps.system_headers);
2604        let (opts, _) = compile(&["-MD", "-MMD", "-c", "a.c"]);
2605        assert!(!opts.deps.system_headers);
2606        let (opts, _) = compile(&["-MMD", "-MD", "-c", "a.c"]);
2607        assert!(!opts.deps.system_headers);
2608    }
2609
2610    #[test]
2611    fn a_target_arrives_escaped_from_one_flag_and_untouched_from_the_other() {
2612        let (opts, _) = compile(&["-MM", "-MT", "a b.o", "-MQ", "a b.o", "a.c"]);
2613        assert_eq!(opts.deps.targets, vec!["a b.o".to_owned(), "a\\ b.o".to_owned()]);
2614    }
2615
2616    #[test]
2617    fn the_rest_of_the_family_is_a_file_and_a_switch() {
2618        let (opts, _) = compile(&["-MM", "-MF", "dep.d", "-MP", "a.c"]);
2619        assert_eq!(opts.deps.file.as_deref(), Some("dep.d"));
2620        assert!(opts.deps.phony);
2621
2622        for flag in ["-MF", "-MT", "-MQ"] {
2623            let e = parse_args(&args(&[flag])).unwrap_err();
2624            assert!(e.message.contains("requires an argument"), "{}", e.message);
2625        }
2626    }
2627
2628    /// A directory of sources for one test, removed when the test is done with it.
2629    struct TempTree(PathBuf);
2630
2631    impl Drop for TempTree {
2632        fn drop(&mut self) {
2633            let _ = std::fs::remove_dir_all(&self.0);
2634        }
2635    }
2636
2637    impl TempTree {
2638        fn new(name: &str, files: &[(&str, &str)]) -> TempTree {
2639            let dir = std::env::temp_dir().join(format!("rucc-deps-{}-{name}", std::process::id()));
2640            let _ = std::fs::remove_dir_all(&dir);
2641            std::fs::create_dir_all(&dir).expect("temporary directory should be writable");
2642            for (path, text) in files {
2643                let at = dir.join(path);
2644                if let Some(parent) = at.parent() {
2645                    std::fs::create_dir_all(parent).expect("creating a subdirectory should work");
2646                }
2647                std::fs::write(&at, text).expect("writing a temporary file should work");
2648            }
2649            TempTree(dir)
2650        }
2651
2652        fn path(&self, name: &str) -> String {
2653            self.0.join(name).to_string_lossy().into_owned()
2654        }
2655    }
2656
2657    #[test]
2658    fn the_rule_names_what_the_includes_found_and_names_each_of_them_once() {
2659        // End to end, because the list comes from the preprocessor and the format comes from
2660        // somewhere else, and a test of either half on its own would pass with the two of them
2661        // wired up backwards.
2662        let tree = TempTree::new(
2663            "found",
2664            &[
2665                ("a.c", "#include \"one.h\"\n#include \"two.h\"\nint main(void) { return X; }\n"),
2666                ("one.h", "#define X 0\n"),
2667                ("two.h", "#include \"one.h\"\n"),
2668            ],
2669        );
2670        let out = tree.path("dep.d");
2671        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2672        assert_eq!(code, 0);
2673
2674        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2675        let names: Vec<&str> = text.split_whitespace().collect();
2676        // The target, the source, and each header once however many times it was reached.
2677        assert_eq!(names.first(), Some(&"a.o:"), "{text}");
2678        assert_eq!(names.iter().filter(|n| n.ends_with("one.h")).count(), 1, "{text}");
2679        assert_eq!(names.iter().filter(|n| n.ends_with("two.h")).count(), 1, "{text}");
2680        // And the `-o` went to the file the rule replaced, which is left empty rather than
2681        // absent because a makefile that named it as a target will look for it.
2682        assert_eq!(std::fs::read(tree.path("a.i")).expect("the output should exist"), b"");
2683    }
2684
2685    #[test]
2686    fn a_header_that_is_only_reached_under_a_guard_is_still_a_dependency() {
2687        // The multiple-include optimization means the second reach never opens the file. It is
2688        // still a file this translation unit was built from, so it is still in the rule.
2689        let tree = TempTree::new(
2690            "guarded",
2691            &[
2692                ("a.c", "#include \"g.h\"\n#include \"g.h\"\nint main(void) { return 0; }\n"),
2693                ("g.h", "#ifndef G\n#define G\n#endif\n"),
2694            ],
2695        );
2696        let out = tree.path("dep.d");
2697        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2698        assert_eq!(code, 0);
2699        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2700        assert_eq!(text.split_whitespace().filter(|n| n.ends_with("g.h")).count(), 1, "{text}");
2701    }
2702
2703    #[test]
2704    fn every_imacros_file_is_read_before_every_include_file_whatever_order_they_were_written() {
2705        // Measured against GCC rather than read: the two flags the other way round produce the
2706        // same output byte for byte, so the command line order between the two families does not
2707        // decide anything and the order within one does. The `-include` file here can only see
2708        // the definition if the `-imacros` file that was written after it ran first.
2709        let tree = TempTree::new(
2710            "preinclude",
2711            &[
2712                ("a.c", "int main(void) { return 0; }\n"),
2713                ("i.h", "#ifdef FROM_MACROS\nint saw_it;\n#else\nint missed_it;\n#endif\n"),
2714                ("m.h", "#define FROM_MACROS 1\nint macros_text;\n"),
2715            ],
2716        );
2717        let out = tree.path("a.i");
2718        let code = run(&args(&[
2719            "-E",
2720            "-include",
2721            &tree.path("i.h"),
2722            "-imacros",
2723            &tree.path("m.h"),
2724            "-o",
2725            &out,
2726            &tree.path("a.c"),
2727        ]));
2728        assert_eq!(code, 0);
2729        let text = std::fs::read_to_string(&out).expect("the output should have been written");
2730        assert!(text.contains("saw_it"), "{text}");
2731        // And the text of the `-imacros` file is thrown away, which is the whole difference
2732        // between the two flags.
2733        assert!(!text.contains("macros_text"), "{text}");
2734    }
2735
2736    #[test]
2737    fn a_file_the_command_line_named_is_a_prerequisite_the_same_as_one_a_directive_named() {
2738        let tree = TempTree::new(
2739            "preinclude-deps",
2740            &[
2741                ("a.c", "int main(void) { return 0; }\n"),
2742                ("i.h", "int from_include;\n"),
2743                ("m.h", "#define M 1\n"),
2744            ],
2745        );
2746        let out = tree.path("dep.d");
2747        let code = run(&args(&[
2748            "-MM",
2749            "-MF",
2750            &out,
2751            "-include",
2752            &tree.path("i.h"),
2753            "-imacros",
2754            &tree.path("m.h"),
2755            "-o",
2756            &tree.path("a.i"),
2757            &tree.path("a.c"),
2758        ]));
2759        assert_eq!(code, 0);
2760        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2761        assert!(text.contains("i.h"), "{text}");
2762        assert!(text.contains("m.h"), "{text}");
2763    }
2764
2765    #[test]
2766    fn a_command_line_include_that_is_nowhere_on_the_path_is_an_error_and_not_a_warning() {
2767        // Including the directory of the source file, which is not on the path for these: the
2768        // command line was not written there, so a name in it is relative to where the compiler
2769        // was run rather than to where the source sits.
2770        let tree = TempTree::new(
2771            "preinclude-missing",
2772            &[("sub/a.c", "int main(void) { return 0; }\n"), ("sub/beside.h", "int x;\n")],
2773        );
2774        let code = run(&args(&["-E", "-include", "beside.h", "-o", "-", &tree.path("sub/a.c")]));
2775        assert_eq!(code, 1);
2776    }
2777
2778    #[test]
2779    fn a_command_line_that_links_names_the_executable_and_not_the_object_it_went_through() {
2780        // The object a link goes through is in a temporary directory and is gone before `make`
2781        // reads any of this, so the rule that named it would be a rule for a file that is never
2782        // there. The target and the file are both the `-o`, which is the executable.
2783        let (opts, plan) = compile(&["-MD", "sub/a.c", "-o", "prog"]);
2784        assert_eq!(plan.output.as_deref(), Some("prog"));
2785        assert_eq!(deps::default_target("sub/a.c", deps_target_output(&opts, &plan)), "prog");
2786        assert_eq!(
2787            deps::default_file(&opts.deps, "sub/a.c", plan.output.as_deref()).as_deref(),
2788            Some("prog.d")
2789        );
2790    }
2791
2792    #[test]
2793    fn the_plan_keeps_the_output_name_because_the_rule_is_written_from_it() {
2794        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c", "-o", "obj/x.o"]);
2795        assert_eq!(plan.output.as_deref(), Some("obj/x.o"));
2796        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c"]);
2797        assert_eq!(plan.output, None);
2798    }
2799
2800    #[test]
2801    fn usage_fits_on_a_screen() {
2802        // Not a style preference. A help text that scrolls is one nobody reads, and this is
2803        // the cheapest way to keep it honest as flags accumulate. The number goes up only when
2804        // a family of flags arrives that has nowhere to share a line, which the two pass gates
2805        // were and which the two fuel flags and `-fsafety=` now are, and it goes up by exactly
2806        // the lines that family took. The four it went up by last are the flags a build system
2807        // passes without being asked to: how much to say, what machine to generate for, threads,
2808        // and the questions `configure` asks before it compiles anything. The one it went up by
2809        // last is the second line of `--emit`, whose kinds are a family that has now outgrown
2810        // one line and has nowhere else to go. The two it went up by last are the dependency
2811        // family, which is eight flags that share nothing with anything above them. The one it
2812        // went up by last is the four spellings of position independent code, which every
2813        // configure script writes and which could only have shared the link line, and that line
2814        // is already four characters short of the limit. The two it went up by last are the rest
2815        // of the include family, which is six more flags that change where a header is looked for
2816        // and two that name a header outright. The one it went up by last is the pair that keeps
2817        // the intermediate files and times the steps, which belong next to the two flags above
2818        // them that are also about watching a compilation rather than changing one.
2819        assert!(USAGE.lines().count() < 48, "usage text has grown past one screen");
2820    }
2821}