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/0.10.7")]
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  -ffunction-sections -fdata-sections   a section per function or variable, for --gc-sections
175  -fvisibility=<what>    default, hidden, internal or protected, when nothing in the source said
176  -l<name>, -L <dir>, -B <dir>   link a library, where to look for one, where our own tools are
177  -fPIC -fpic -fPIE -fpie, -fno-common, -f[no-]strict-aliasing, -pipe   what it does anyway
178  -static -shared -pie -no-pie -nostdlib -nostartfiles -nodefaultlibs -rdynamic -s   how to link
179  -Wl,<arg>, -Xlinker <arg>, -fuse-ld=<name>   hand an argument to the linker, or pick one
180  -Werror -pedantic -pedantic-errors -w   how much to say, and whether it is fatal
181  -m64 -march= -mtune= -mcpu= -mabi= -mcmodel=   what machine to generate for
182  -pthread               build for more than one thread, and link the library for it
183  -dumpmachine -dumpversion -print-multiarch -print-search-dirs   what this compiler is
184  -print-file-name=<name> -print-prog-name=<name>   where a file or a program is
185  -j[n]                  compile n translation units at once, default all
186  -v, -###               print each phase as it runs, or without running any
187  -save-temps[=cwd|obj], -time   keep the .i and the .s, say how long each step took
188  --target=<triple>      generate code for <triple>
189  --emit=<kind>          exe, obj, asm, preprocessed, tast, ir, mir-final,
190                         safety-summary, type-granules
191  --print-config, --print-pipeline    print the configuration or the pipeline, and exit
192  --version              print the version and exit
193  -h, --help             print this message and exit
194
195See spec/04-driver-and-cli.md for the full flag reference.
196";
197
198/// The argument of a flag that may be joined to it or may be the next word.
199///
200/// `-DFOO` and `-D FOO` are the same thing, and `at` is where the flag's own letters end.
201fn joined_or_next(
202    arg: &str,
203    at: usize,
204    args: &[String],
205    i: &mut usize,
206) -> Result<String, CliError> {
207    if arg.len() > at {
208        return Ok(arg[at..].to_owned());
209    }
210    let next = args.get(*i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
211    *i += 1;
212    Ok(next.clone())
213}
214
215/// Parses a command line, without the program name.
216///
217/// # Errors
218///
219/// Returns the message to print when the arguments do not name a compilation this compiler
220/// can attempt.
221pub fn parse_args(args: &[String]) -> Result<Action, CliError> {
222    let host = Triple::host()
223        .ok_or_else(|| err("this host is not a supported target and no --target was given"))?;
224    let mut opts = Options::new(host);
225    let mut inputs: Vec<Input> = Vec::new();
226    let mut print_config = false;
227    let mut print_pipeline = false;
228    let mut print_plan = false;
229    let mut verbose = false;
230    let mut jobs = Jobs::default();
231    let mut nostdinc = false;
232    let mut sysroot: Option<PathBuf> = None;
233    let mut output = None;
234    let mut link = LinkOptions::default();
235    let mut query: Option<Query> = None;
236    let mut threads = false;
237    // `-x` applies to inputs that come after it and stays in effect until the next one, which
238    // is why it is tracked across the loop rather than attached to a single argument.
239    let mut forced: Option<InputKind> = None;
240    // What `-iprefix` last said, stuck on the front of every later `-iwithprefix`. It applies to
241    // the flags after it and not the ones before, so a command line may set it more than once.
242    // GCC's default is its own installed header directory with the last component taken off,
243    // which is a path a cross compiler's build system knows and passes; there is no equivalent
244    // here, so with no `-iprefix` the prefix is nothing and `-iwithprefix` names a directory
245    // outright.
246    let mut iprefix = String::new();
247
248    let mut i = 0;
249    while i < args.len() {
250        let arg = args[i].as_str();
251        i += 1;
252        match arg {
253            "-h" | "--help" => return Ok(Action::Help),
254            "--version" => return Ok(Action::Version),
255            "--print-config" => print_config = true,
256            "--print-pipeline" => print_pipeline = true,
257            "-###" => print_plan = true,
258            "-v" => verbose = true,
259            // The files a compilation goes through, kept rather than thrown away. The bare
260            // spelling means `=obj` and not `=cwd`, which is not what the manual says and is what
261            // gcc 16 does; `SaveTemps::Object` carries the measurement.
262            "-save-temps" => opts.save_temps = SaveTemps::Object,
263            _ if arg.starts_with("-save-temps=") => {
264                opts.save_temps = arg["-save-temps=".len()..].parse().map_err(err)?;
265            }
266            // How long each step took. A misspelling of this is worth rejecting rather than
267            // ignoring, since a run that says nothing looks like a compilation that took no time.
268            "-time" => opts.time = true,
269            "-c" => opts.emit = EmitKind::Object,
270            "-S" => opts.emit = EmitKind::Asm,
271            "-E" => opts.emit = EmitKind::Preprocessed,
272            "-g" => opts.debug_info = true,
273            // GCC's own levels of how much debug information to write. Zero is none and every
274            // other number is some, and this compiler has one amount, so the numbers above zero
275            // all mean the same thing here. `-ggdb` is the same flag asking for whatever the
276            // debugger on the machine prefers, which is what we emit anyway.
277            "-g0" => opts.debug_info = false,
278            "-g1" | "-g2" | "-g3" | "-ggdb" | "-ggdb1" | "-ggdb2" | "-ggdb3" => {
279                opts.debug_info = true;
280            }
281            // The version of DWARF to write. We write DWARF 5 and nothing else, so a build that
282            // asks for another version is told rather than handed a file it cannot read.
283            "-gdwarf" | "-gdwarf-5" => opts.debug_info = true,
284            _ if arg.starts_with("-gdwarf-") => {
285                return Err(err(format!(
286                    "{arg}: this compiler writes DWARF 5 and no other version, see \
287                     spec/11-debug-info.md"
288                )));
289            }
290            "-Werror" => opts.warnings_are_errors = true,
291            // Nothing that is not fatal is said at all. Read at the one place a diagnostic goes
292            // through rather than here, so that a warning `-w` dropped is not counted either.
293            "-w" => opts.warnings = false,
294            "-pedantic-errors" => {
295                opts.pedantic = true;
296                opts.warnings_are_errors = true;
297            }
298            "-P" => opts.line_markers = false,
299            // The dependency family, which section 4.4 calls required because every build system
300            // that generates its own makefiles asks for it. The two that end in `D` write a file
301            // beside the object and let the compilation happen, and the two that do not write to
302            // standard output and stop after it. Nothing here turns the system headers back on
303            // once a flag has turned them off, which is GCC's behaviour and is why `-MM -M` is
304            // `-MM`: the flag asking for fewer of them is the one with something to say.
305            "-M" => {
306                opts.deps.emit = true;
307                opts.deps.instead_of_compiling = true;
308            }
309            "-MM" => {
310                opts.deps.emit = true;
311                opts.deps.instead_of_compiling = true;
312                opts.deps.system_headers = false;
313            }
314            "-MD" => opts.deps.emit = true,
315            "-MMD" => {
316                opts.deps.emit = true;
317                opts.deps.system_headers = false;
318            }
319            "-MP" => opts.deps.phony = true,
320            // These three take a word and only in the separated form, which is how GCC spells
321            // them and how every build system writes them.
322            "-MF" | "-MT" | "-MQ" => {
323                let value =
324                    args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
325                i += 1;
326                match arg {
327                    "-MF" => opts.deps.file = Some(value.clone()),
328                    // The whole of the difference between the two. `-MT` is for a build that has
329                    // already escaped what it is passing, and `-MQ` is for one that has a name
330                    // and wants it to arrive as that name.
331                    "-MT" => opts.deps.targets.push(value.clone()),
332                    _ => opts.deps.targets.push(deps::escaped(value)),
333                }
334            }
335            // The questions a build system asks before it compiles anything. Answered after the
336            // loop, because each one is about the target or the library search and the command
337            // line has not finished saying what those are.
338            "-dumpmachine" => query = Some(Query::Machine),
339            "-dumpversion" | "-dumpfullversion" => query = Some(Query::Version),
340            "-print-multiarch" => query = Some(Query::Multiarch),
341            "-print-search-dirs" => query = Some(Query::SearchDirs),
342            "-print-libgcc-file-name" => query = Some(Query::Libgcc),
343            _ if arg.starts_with("-print-file-name=") => {
344                query = Some(Query::FileName(arg["-print-file-name=".len()..].to_owned()));
345            }
346            _ if arg.starts_with("-print-prog-name=") => {
347                query = Some(Query::ProgName(arg["-print-prog-name=".len()..].to_owned()));
348            }
349            // A program built to run in more than one thread. On every platform this compiler
350            // targets that is a macro the library's headers read and one more library on the
351            // link line, and the library is added after the loop so that it lands after the
352            // objects that refer to it.
353            "-pthread" | "-pthreads" => {
354                opts.defines.push("_REENTRANT".to_owned());
355                threads = true;
356            }
357            "-ansi" => {
358                opts.std = Std::C89;
359                opts.gnu_extensions = false;
360            }
361            // `-Wpedantic` is the same flag under the name the `-W` family gives it, which is
362            // the spelling a build system that groups its warning flags tends to write.
363            "-pedantic" | "-Wpedantic" => opts.pedantic = true,
364            // Both directions, because a build that needs this for one directory turns it back
365            // off for the next one rather than leaving it on for the whole tree.
366            "-fpermissive" => opts.permissive = true,
367            "-fno-permissive" => opts.permissive = false,
368            "-ffreestanding" => opts.hosted = false,
369            "-fhosted" => opts.hosted = true,
370            "-fno-builtin" => opts.builtins = false,
371            "-fbuiltin" => opts.builtins = true,
372            // The C89 dialects are under GNU's reading whatever this says, so turning it off
373            // there is turning off something the dialect asked for, which is accepted and does
374            // nothing. gcc refuses that command line, and there is nothing it could have meant.
375            "-fgnu89-inline" => opts.gnu89_inline = true,
376            "-fno-gnu89-inline" => opts.gnu89_inline = false,
377            // Both directions of each, because a build system that wants one of these usually
378            // writes it beside the flag that turns it back off for one directory.
379            "-fno-omit-frame-pointer" => opts.frame_pointer = true,
380            "-fomit-frame-pointer" => opts.frame_pointer = false,
381            "-mno-red-zone" => opts.red_zone = false,
382            "-mred-zone" => opts.red_zone = true,
383            // GCC drops its own include directory along with the system ones, because its
384            // headers are half of a pair with the library's and half a pair is worse than
385            // none. A build that passes this is supplying the whole set itself.
386            "-nostdinc" => nostdinc = true,
387            "-o" => {
388                output = Some(args.get(i).ok_or_else(|| err("-o requires an argument"))?.clone());
389                i += 1;
390            }
391            // The flags that take a directory only in the separated form. GCC spells them
392            // this way and nothing writes `-iquotedir`, so accepting the joined form would
393            // mean guessing at a path that starts with the flag's own letters.
394            // Apple's spelling of `--sysroot`, and the one its own build systems pass. The
395            // two mean the same thing here: the configured directories are under there rather
396            // than under the root.
397            "-isysroot" => {
398                let dir = args.get(i).ok_or_else(|| err("-isysroot requires an argument"))?;
399                i += 1;
400                sysroot = Some(PathBuf::from(dir));
401            }
402            "-iquote" | "-isystem" | "-idirafter" => {
403                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
404                i += 1;
405                match arg {
406                    "-iquote" => opts.search.push_quote(dir.clone()),
407                    "-isystem" => opts.search.push_system(dir.clone()),
408                    _ => opts.search.push_after(dir.clone()),
409                }
410            }
411            "-iprefix" => {
412                iprefix = args.get(i).ok_or_else(|| err("-iprefix requires an argument"))?.clone();
413                i += 1;
414            }
415            // Where GCC puts these is not where its manual says it puts them, and this is the
416            // measured answer rather than the documented one: `-iwithprefix` lands in the
417            // `-isystem` slot and not the `-idirafter` slot, and `-iwithprefixbefore` lands in
418            // the `-I` slot. A cross build that uses them is relying on the behaviour, since
419            // that is the compiler it was developed against.
420            "-iwithprefix" | "-iwithprefixbefore" => {
421                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
422                i += 1;
423                let dir = format!("{iprefix}{dir}");
424                if arg == "-iwithprefix" {
425                    opts.search.push_system(dir);
426                } else {
427                    opts.search.push_bracket(dir);
428                }
429            }
430            "-include" | "-imacros" => {
431                let name = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
432                i += 1;
433                opts.preincludes
434                    .push(Preinclude { name: name.clone(), macros_only: arg == "-imacros" });
435            }
436            // The flag `-iquote` was introduced to replace, still passed by build systems old
437            // enough to predate the replacement. It is not a directory: it says that every `-I`
438            // so far is for quoted includes only, and that a quoted include stops looking next
439            // to the file that wrote it.
440            "-I-" => opts.search.split_quote_chain(),
441            "-x" => {
442                let lang = args.get(i).ok_or_else(|| err("-x requires an argument"))?;
443                i += 1;
444                forced = if lang == "none" {
445                    None
446                } else {
447                    Some(InputKind::from_x_arg(lang).map_err(|e| err(format!("{e}")))?)
448                };
449            }
450            // Not a GCC flag. spec/03-architecture.md section 3.5 compiles several
451            // translation units in one process rather than making the build system fork, and
452            // section 3.8's determinism check compares `-j1` against `-j16`, so the knob has
453            // to exist and has to be spelled the way `make` spells it.
454            // `-DFOO`, `-D FOO` and the same for `-U` and `-I`. Both forms are in wide use
455            // and a build system may produce either, so both are read here rather than
456            // being normalised by whatever generated the command line.
457            _ if arg.starts_with("-D") => {
458                let value = joined_or_next(arg, 2, args, &mut i)?;
459                opts.defines.push(value);
460            }
461            _ if arg.starts_with("-U") => {
462                let value = joined_or_next(arg, 2, args, &mut i)?;
463                opts.undefines.push(value);
464            }
465            _ if arg.starts_with("-I") => {
466                let dir = joined_or_next(arg, 2, args, &mut i)?;
467                opts.search.push_bracket(dir);
468            }
469            _ if arg.starts_with("-std=") => {
470                let name = &arg["-std=".len()..];
471                let (std, gnu) = Std::from_flag(name)
472                    .ok_or_else(|| err(format!("unknown dialect `{name}`, see --help")))?;
473                opts.std = std;
474                opts.gnu_extensions = gnu;
475            }
476            // Section 4.5. The claim decides which half of glibc's `sys/cdefs.h` we are
477            // handed, so a differential run that does not set it is comparing two compilers
478            // that believe they are different compilers.
479            // GCC packs these into one flag, so `-dDI` is two of them. Letters in the family
480            // that we have not written yet are accepted and ignored, because a dump is a
481            // debugging aid and a build that asks for one should still compile. A letter
482            // outside the family falls through to the unknown option error, which is what
483            // keeps `-dumpversion` from being read as a dump of nothing.
484            _ if Dumps::is_family(arg) => {
485                opts.dumps.add(&arg[2..]);
486            }
487            // One name at a time, which is what a build that means its own `memcpy` and the
488            // library's everything else writes. The name is not checked against a list, because
489            // the flag is about what the program means by a name and a program is allowed to mean
490            // something by a name this compiler has never heard of.
491            _ if arg.starts_with("-fno-builtin-") => {
492                opts.no_builtin.push(arg["-fno-builtin-".len()..].to_owned());
493            }
494            _ if arg.starts_with("-fgnuc-version=") => {
495                let v = &arg["-fgnuc-version=".len()..];
496                opts.gnuc = v.parse().map_err(err)?;
497            }
498            // spec/13-gnu-compat.md section 13.3 promises this flag an error that says why rather
499            // than the unknown option one, because a build reaching for it is asking for a feature
500            // and deserves to be told it is not coming rather than told the spelling is wrong.
501            // The negative form is what this compiler does anyway, so it is taken and dropped.
502            "-fnested-functions" => {
503                return Err(err(
504                    "nested functions are not supported: a call to one goes through a trampoline \
505                     written on the stack, which no target that enforces an unexecutable stack \
506                     allows",
507                ));
508            }
509            "-fno-nested-functions" => {}
510            // Which of the two links the output is for, which is a real difference and not a
511            // description of what happens anyway. Everything here is position independent either
512            // way, and what these decide is whether a name may be one another object defines or
513            // replaces, because a link that produces an executable puts every name in the same
514            // program and a link that produces a shared library does not.
515            //
516            // It matters that they are accepted at all, whatever they then do. Every autoconf and
517            // cmake build puts `-fPIC` on the compile line, so a compiler that rejects it cannot
518            // be the `CC` of a project that has a configure script, whatever else it can do. That
519            // is how this was found: building SQLite's test fixture stopped on it.
520            "-fPIC" | "-fpic" => opts.pic = Pic::Library,
521            // Not a synonym of the pair above, which is what they were treated as until #756. The
522            // library is the expensive answer and gcc makes it the one that has to be asked for,
523            // so this is also what nothing at all means.
524            "-fPIE" | "-fpie" => opts.pic = Pic::Executable,
525            // A different question from the pair above, and the one every distribution build of a
526            // shared library answers. `-fPIC` decides how an address is reached, and this decides
527            // whether the optimizer may believe a body it can see, because an exported name is one
528            // the dynamic linker may find another definition of first. On by default, which is
529            // gcc's arrangement and is the honest answer, and off is a promise the build makes and
530            // nothing checks.
531            "-fsemantic-interposition" => opts.interposition = true,
532            "-fno-semantic-interposition" => opts.interposition = false,
533            // Two requests rather than one, and the same table answers both, so what decides is
534            // whether either of them is standing. gcc arranges it the same way: the asynchronous
535            // one is the default here and it implies the other, and a line that asks for a table
536            // and against an asynchronous one gets a table.
537            "-fasynchronous-unwind-tables" => opts.async_unwind_tables = true,
538            "-fno-asynchronous-unwind-tables" => opts.async_unwind_tables = false,
539            "-funwind-tables" => opts.unwind_tables = true,
540            "-fno-unwind-tables" => opts.unwind_tables = false,
541            // The other direction is a request, not a description, and it is one this compiler
542            // cannot grant, so it gets the treatment section 13.3 asks for rather than the unknown
543            // option error. Answering it by carrying on would be answering a different question:
544            // the code would still be position independent, which is correct everywhere an
545            // ordinary program runs and is wrong in a kernel, where the flag is written precisely
546            // because there is no loader to fill a global offset table in.
547            "-fno-pic" | "-fno-pie" => {
548                return Err(err(
549                    "position dependent code is not supported: an address that may be in another \
550                     object is loaded out of the global offset table, and nothing here emits the \
551                     absolute form this asks for. Use -no-pie if what you meant was how to link",
552                ));
553            }
554            // A section per function and a section per variable, which is what makes
555            // `--gc-sections` able to drop anything: a linker can leave out a section nothing
556            // reaches and cannot leave out half of one. Both directions are taken, and the off
557            // one is the default rather than a refusal, since a build that writes it is asking
558            // for what happens anyway.
559            "-ffunction-sections" => opts.function_sections = true,
560            "-fno-function-sections" => opts.function_sections = false,
561            "-fdata-sections" => opts.data_sections = true,
562            "-fno-data-sections" => opts.data_sections = false,
563            // Another description of what this compiler does. A file scope declaration with no
564            // initializer is written into `.bss` as an ordinary defined symbol, not offered to the
565            // linker as a common one for it to merge, which is what `-fno-common` asks for and what
566            // gcc has done by default since 10. Nothing in the front end produces `Linkage::Common`
567            // at all.
568            "-fno-common" => {}
569            // And the request, which is the one that cannot be granted. It is a real difference and
570            // not a preference: two files each writing `int g;` link under `-fcommon` and are a
571            // duplicate definition without it, which is the whole reason the flag survives.
572            "-fcommon" => {
573                return Err(err(
574                    "a tentative definition is written into .bss as its own symbol here, and \
575                     nothing emits the common symbol this asks the linker to merge. Give the \
576                     variable a definition in one file and declare it extern in the others",
577                ));
578            }
579            // Both directions of this one are taken, which is the exception to the rule above, and
580            // the reason is which way being wrong costs something.
581            //
582            // Nothing here derives anything from the type an object is accessed through. The IR has
583            // somewhere to put a type based aliasing node and lowering fills it with nothing on
584            // every access, so no pass has one to read and the alias analysis falls back to what it
585            // can see. That makes `-fno-strict-aliasing` a description, the way `-fPIC` is.
586            //
587            // `-fstrict-aliasing` is a request to assume more than that, and this compiler assumes
588            // less. Answering a request for a weaker guarantee by giving a stronger one is safe in
589            // a way the `-fno-pic` case is not: every program that is correct under the assumption
590            // is correct without it, only slower. And `-O2` implies it, so a build that spells it
591            // out is a build that would stop on a compiler that refused it, for no gain at all.
592            "-fstrict-aliasing" | "-fno-strict-aliasing" => {}
593            // About temporary files rather than about code. There is nothing between the phases of
594            // one compilation here to write to a file in the first place.
595            "-pipe" => {}
596            // Nothing here writes colour, so all of these are the same answer, and it is the answer
597            // that costs nothing: the diagnostics come out plain either way and no build depends on
598            // an escape sequence being there. Taken rather than refused because cmake writes
599            // `-fdiagnostics-color=always` on every compile line when the generator is ninja, which
600            // makes this the second most common flag after `-fPIC` to stop a build over a question
601            // about how the text looks.
602            "-fdiagnostics-color" | "-fno-diagnostics-color" => {}
603            _ if arg.starts_with("-fdiagnostics-color=") => {}
604            // The link flags. None of them changes the compilation, which is why they are
605            // collected apart from `opts` and why `-lm` on a `-c` line is a note rather than an
606            // error: it is a thing said to a linker that is not going to run.
607            "-static" => link.is_static = true,
608            "-shared" => link.shared = true,
609            "-pie" => link.pie = Some(true),
610            "-no-pie" | "-nopie" => link.pie = Some(false),
611            "-nostdlib" => link.no_stdlib = true,
612            "-nostartfiles" => link.no_startfiles = true,
613            "-nodefaultlibs" => link.no_defaultlibs = true,
614            "-fno-builtins-lib" => link.no_builtins_lib = true,
615            "-fbuiltins-lib" => link.no_builtins_lib = false,
616            "-rdynamic" | "-export-dynamic" => link.export_dynamic = true,
617            "-s" => link.strip = true,
618            "-Xlinker" => {
619                let next = args.get(i).ok_or_else(|| err("-Xlinker requires an argument"))?;
620                i += 1;
621                link.passthrough.push(next.clone());
622            }
623            _ if arg.starts_with("-Wl,") => {
624                // Commas separate arguments rather than being part of one, which is what makes
625                // `-Wl,-rpath,/opt/lib` two words to the linker and one word here.
626                link.passthrough.extend(arg["-Wl,".len()..].split(',').map(str::to_owned));
627            }
628            _ if arg.starts_with("-fuse-ld=") => {
629                link.use_ld = Some(arg["-fuse-ld=".len()..].to_owned());
630            }
631            _ if arg.starts_with("-l") && arg.len() > 2 => {
632                inputs.push(Input::library(&arg[2..]));
633            }
634            "-l" => {
635                let next = args.get(i).ok_or_else(|| err("-l requires an argument"))?;
636                i += 1;
637                inputs.push(Input::library(next));
638            }
639            _ if arg.starts_with("-L") => {
640                link.search.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
641            }
642            _ if arg.starts_with("-B") => {
643                link.prefixes.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
644            }
645            _ if arg.starts_with("-j") => {
646                jobs = Jobs::parse(&arg[2..]).map_err(err)?;
647            }
648            _ if arg.starts_with("--sysroot=") => {
649                sysroot = Some(PathBuf::from(&arg["--sysroot=".len()..]));
650            }
651            _ if arg.starts_with("--target=") => {
652                let t = &arg["--target=".len()..];
653                opts.target = t.parse().map_err(|e| err(format!("{e}")))?;
654            }
655            _ if arg.starts_with("--emit=") => {
656                let k = &arg["--emit=".len()..];
657                opts.emit = k
658                    .parse()
659                    .map_err(|()| err(format!("unknown --emit kind `{k}`, see --help")))?;
660            }
661            // A bare `-O` is `-O1`, which is what GCC has and what a hand written makefile tends
662            // to write. `-Og` is GCC's level for a build somebody is going to step through, and
663            // it is `-O1` with the transformations that move code around left out; this compiler
664            // has no such level yet, so it is the nearest one and `--print-pipeline` says what
665            // that came to rather than the flag pretending otherwise.
666            "-O" | "-Og" => opts.opt_level = rucc_session::OptLevel::O1,
667            // The union of `-O3` and `-ffast-math`, and the second half of that changes what
668            // floating point arithmetic means. Refused rather than taken as `-O3`, because a
669            // build that asks for fast math and is quietly given ordinary arithmetic gets a
670            // slower program than it asked for and a build that is given fast math it did not
671            // ask for gets a wrong one.
672            "-Ofast" => {
673                return Err(err(
674                    "-Ofast is -O3 with fast math, and fast math is not implemented, see \
675                     spec/04-driver-and-cli.md section 4.6",
676                ));
677            }
678            _ if arg.starts_with("-O") => {
679                opts.opt_level = arg[2..]
680                    .parse()
681                    .map_err(|()| err(format!("unknown optimization level `{arg}`")))?;
682            }
683            // What every name gets when nothing in the source said, which the attribute in the
684            // source overrides rather than the other way round. Before the optimizer's `-f`
685            // family below for the reason the tier below it is.
686            _ if arg.starts_with("-fvisibility=") => {
687                let seen = &arg["-fvisibility=".len()..];
688                opts.visibility = seen.parse().map_err(|()| {
689                    err(format!(
690                        "`{seen}` is not a visibility, which is default, hidden, internal or \
691                         protected"
692                    ))
693                })?;
694            }
695            // The memory safety monitor, from section 15.4 of
696            // `spec/safe-memory/15-integration.md`. Before the optimizer's `-f` family below,
697            // because a pass that took the name `safety=detect` would otherwise be handed the
698            // flag, and the tier is not a pass.
699            _ if arg.starts_with("-fsafety=") => {
700                let tier = &arg["-fsafety=".len()..];
701                opts.safety = tier.parse().map_err(|()| {
702                    err(format!(
703                        "`{tier}` is not a safety tier, which is off, detect, enforce or kernel"
704                    ))
705                })?;
706            }
707            // The optimizer's own flags, from section 9.10 of `spec/09-optimizer.md`. These come
708            // after every `-f` the rest of the compiler answers to, so a pass can never take a
709            // name that already means something else on the command line.
710            _ if arg.starts_with("-fpass-fuel=") => {
711                let (name, count) = arg["-fpass-fuel=".len()..]
712                    .split_once('=')
713                    .ok_or_else(|| err("-fpass-fuel= is spelled <pass>=<count>"))?;
714                if rucc_opt::pass::find(name).is_none() {
715                    return Err(err(format!(
716                        "`{name}` is not a pass this compiler has, see --print-pipeline"
717                    )));
718                }
719                let count: u32 = count
720                    .parse()
721                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
722                opts.pass_fuel.push((name.to_owned(), count));
723            }
724            _ if arg.starts_with("-fpass-fuel-global=") => {
725                let count = &arg["-fpass-fuel-global=".len()..];
726                let count: u32 = count
727                    .parse()
728                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
729                opts.pass_fuel_global = Some(count);
730            }
731            // Everything from `-fopt-info` to the end of the argument, which is optional
732            // keywords joined by hyphens and an optional `=<file>`. Checked here rather than
733            // where the remarks are printed, because by then the compilation somebody wanted
734            // to hear about is over.
735            _ if arg == "-fopt-info"
736                || arg.starts_with("-fopt-info=")
737                || arg.starts_with("-fopt-info-") =>
738            {
739                let rest = &arg["-fopt-info".len()..];
740                let (kinds, file) = match rest.split_once('=') {
741                    Some((kinds, file)) => (kinds, Some(file)),
742                    None => (rest, None),
743                };
744                let kinds = kinds.strip_prefix('-').unwrap_or(kinds);
745                rucc_opt::Wants::none().add(kinds).map_err(err)?;
746                opts.opt_info.push(kinds.to_owned());
747                if let Some(file) = file {
748                    if file.is_empty() {
749                        return Err(err("-fopt-info= was given no file to write to"));
750                    }
751                    opts.opt_info_file = Some(file.to_owned());
752                }
753            }
754            _ if arg.starts_with("-fdump-ir=") => {
755                // Checked here rather than where the dumps are taken, because the compilation
756                // that would have been dumped is over by then.
757                let spec = &arg["-fdump-ir=".len()..];
758                rucc_opt::Dumps::default().add(spec).map_err(err)?;
759                opts.dump_ir.push(spec.to_owned());
760            }
761            // Before the bare `-f<pass>` below, because a pass called `enable-something` would
762            // otherwise take the flag away from the gate. Checked here rather than where the
763            // pipeline reads it, for the reason that applies to all of these: a misspelled pass
764            // name that quietly gated nothing looks exactly like a pass that is not the guilty
765            // one, and a bisection would carry on past the thing it was looking for.
766            _ if arg.starts_with("-fdisable-") || arg.starts_with("-fenable-") => {
767                let on = arg.starts_with("-fenable-");
768                let spec = &arg[if on { "-fenable-".len() } else { "-fdisable-".len() }..];
769                rucc_opt::Gates::default().add(on, spec).map_err(err)?;
770                opts.pass_gates.push((on, spec.to_owned()));
771            }
772            _ if arg.strip_prefix("-fno-").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
773                opts.passes.push((arg["-fno-".len()..].to_owned(), false));
774            }
775            _ if arg.strip_prefix("-f").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
776                opts.passes.push((arg["-f".len()..].to_owned(), true));
777            }
778            // The unstable options, spelled the way rustc spells them and carrying the same
779            // promise, which is none: one of these may change or go away in any release. They are
780            // measurements and debugging aids rather than things a build asks for, which is why
781            // none of them is in the usage text and all of them are in section 4.11 of
782            // `spec/04-driver-and-cli.md`.
783            "-Zverify-each" => opts.verify_each = true,
784            _ if arg.starts_with("-Zrule-coverage=") => {
785                let file = &arg["-Zrule-coverage=".len()..];
786                if file.is_empty() {
787                    return Err(err("-Zrule-coverage= needs a file to write to"));
788                }
789                opts.rule_coverage = Some(file.to_owned());
790            }
791            _ if arg.starts_with("-Z") => {
792                return Err(err(format!(
793                    "`{arg}` is not an unstable option this compiler has, see \
794                     spec/04-driver-and-cli.md section 4.11 for the ones it does"
795                )));
796            }
797            // The word size, which is a statement about the target and is taken as one. A build
798            // that says the size the target already has is saying nothing, and one that says the
799            // other size is asking for a target this compiler does not have, which it is told
800            // rather than being given the wrong one.
801            "-m64" | "-m32" | "-mx32" => {
802                let want: u32 = match arg {
803                    "-m64" => 64,
804                    _ => 32,
805                };
806                let have = rucc_target::TargetInfo::new(opts.target).pointer_width;
807                if have != want {
808                    return Err(err(format!(
809                        "{arg} asks for a {want} bit target and {} is {have} bit, use \
810                         --target= to name the one you mean",
811                        opts.target
812                    )));
813                }
814            }
815            // Which processor in the family to generate for. This compiler emits the base
816            // instruction set of the architecture and nothing above it, so a program built with
817            // any of these runs on the machine that was named; it is a program that could have
818            // been faster rather than a program that is wrong, which is what makes these safe to
819            // take and ignore where a flag that changed the meaning of the code would not be.
820            _ if arg.starts_with("-march=")
821                || arg.starts_with("-mtune=")
822                || arg.starts_with("-mcpu=") => {}
823            // The calling convention, which is not safe to ignore. Taken when it names the one
824            // the target already uses and refused otherwise.
825            _ if arg.starts_with("-mabi=") => {
826                let want = &arg["-mabi=".len()..];
827                let have = match opts.target.arch {
828                    rucc_target::Arch::X86_64 => "sysv",
829                    rucc_target::Arch::Aarch64 => "lp64",
830                    rucc_target::Arch::Riscv64 => "lp64d",
831                };
832                if want != have {
833                    return Err(err(format!(
834                        "{arg}: {} uses the {have} convention and this compiler has no other",
835                        opts.target
836                    )));
837                }
838            }
839            // How far apart the pieces of the program may be. The small model is what we emit and
840            // it is every hosted program's default; the kernel model is a different one and a
841            // build that asks for it and does not get it links and then does not run.
842            "-mcmodel=small" => {}
843            _ if arg.starts_with("-mcmodel=") => {
844                return Err(err(format!(
845                    "{arg}: this compiler emits the small code model and no other, see \
846                     spec/12-targets.md"
847                )));
848            }
849            // GCC's own scripting language for how the driver builds a command line.
850            // `spec/04-driver-and-cli.md` section 4.4 settles that we will not have it, so a
851            // build reaching for it is told which flags do the same job.
852            _ if arg.starts_with("-specs=") => {
853                return Err(err(
854                    "-specs= is not supported: the parts of it builds rely on are -B, -L, \
855                     -nostdlib, -nostartfiles and -Wl,, see spec/04-driver-and-cli.md \
856                     section 4.4",
857                ));
858            }
859            // Arguments meant for a separate assembler or preprocessor, which this compiler does
860            // not have: both are inside it and neither reads a command line. Refused rather than
861            // dropped, because every one of these says something about the output and a build
862            // that asked for `-Wa,--noexecstack` and was silently given an executable stack got
863            // the opposite of what it asked for.
864            _ if arg.starts_with("-Wa,") || arg.starts_with("-Wp,") => {
865                return Err(err(format!(
866                    "`{arg}` is an argument for a separate assembler or preprocessor, and both \
867                     are inside this compiler rather than programs it runs"
868                )));
869            }
870            "-Xassembler" | "-Xpreprocessor" => {
871                return Err(err(format!(
872                    "{arg} hands an argument to a separate assembler or preprocessor, and both \
873                     are inside this compiler rather than programs it runs"
874                )));
875            }
876            // Everything else in the `-W` family. `spec/04-driver-and-cli.md` section 4.1 has
877            // this one as a rule about build systems rather than about warnings: autoconf finds
878            // out whether a warning flag exists by passing it and looking at the exit status, so
879            // a compiler that refuses one it has not heard of fails a configure script written
880            // for a GCC newer than itself. The names are not checked against a list because this
881            // compiler has no warning groups for a list to be of, which #485 is about.
882            _ if arg.starts_with("-W") => {}
883            // Flags that name something this compiler does not do and would not do differently
884            // if it did. `-fno-ident` is about a comment in the output that we do not write
885            // either way, and the others are about a way of ordering the compilation that has
886            // been GCC's only way for twenty years. Section 4.1 asks for the list to be short
887            // and for adding to it to be deliberate, which is why it is written out here.
888            "-fno-ident"
889            | "-fident"
890            | "-funit-at-a-time"
891            | "-fno-unit-at-a-time"
892            | "-shared-libgcc"
893            | "-static-libgcc" => {}
894            _ if arg.starts_with('-') && arg.len() > 1 => {
895                // Silently ignoring an unknown flag is how a build ends up not doing what
896                // its author asked. spec/13-gnu-compat.md section 13.4 makes this an error
897                // for the flags that change code generation, and the safe default until the
898                // flag table is populated is to reject everything we do not know.
899                return Err(err(format!("unknown option `{arg}`")));
900            }
901            _ => inputs.push(Input { path: arg.to_owned(), forced, library: false }),
902        }
903    }
904
905    // Last, so that it lands after every `-isystem` the command line gave. That is GCC's
906    // order: a directory the user names outranks the compiler's own, and the compiler's own
907    // outranks the library's. It is pushed after the loop rather than before it because
908    // `SearchPath` appends within a group and the position is what the order is.
909    // The same directory the headers were looked for under, because a sysroot is a statement
910    // about a whole installation and not about half of one.
911    link.sysroot = sysroot.clone();
912    // After the loop rather than where `-pthread` was read, so that it lands after the objects
913    // that refer to it. A static link takes the definitions it needs from a library when it
914    // reaches it and not afterwards, so a library before the objects is a library that answers
915    // nothing.
916    if threads {
917        inputs.push(Input::library("pthread"));
918    }
919    if let Some(query) = query {
920        return Ok(Action::Print(answer(&query, &opts, &link)));
921    }
922    // `-M` and `-MM` produce the rule and nothing else, so the run stops after phase 4 whatever
923    // else the command line asked for. Read here rather than where the flag was, because a `-c`
924    // written after it has to lose and the loop cannot know that until it has ended. The output
925    // file is where the rule goes rather than where an object would have gone, and the last
926    // phase being the preprocessor is what makes that true without a second rule for it.
927    if opts.deps.instead_of_compiling {
928        opts.emit = EmitKind::Preprocessed;
929    }
930    if !nostdinc {
931        opts.search.push_system(runtime::DIR);
932        // And the library's after ours, which is the other half of the same order. They go on
933        // here rather than at the point `--target=` or `--sysroot=` was read because either
934        // one changes the answer and the last word on both is the end of the loop.
935        for dir in library::system_dirs(opts.target, sysroot.as_deref()) {
936            opts.search.push_system(dir);
937        }
938    }
939    // Once, here, rather than as each directory is pushed. A `-I` that names a system
940    // directory has to lose to the system entry and the system entry is added last, so the
941    // question cannot be answered until the whole path is known.
942    opts.search.remove_duplicates();
943
944    // The target has to be resolved before the configuration is printed, so this check comes
945    // after the loop rather than at the point `--print-config` was seen.
946    if print_config {
947        return Ok(Action::PrintConfig(Box::new(opts)));
948    }
949    if print_pipeline {
950        return Ok(Action::PrintPipeline(Box::new(opts)));
951    }
952    let plan = Plan::new(&opts, &inputs, output.as_deref()).map_err(|e| err(e.message))?;
953    if print_plan {
954        return Ok(Action::PrintPlan {
955            opts: Box::new(opts),
956            plan: Box::new(plan),
957            link: Box::new(link),
958        });
959    }
960    Ok(Action::Compile {
961        opts: Box::new(opts),
962        plan: Box::new(plan),
963        link: Box::new(link),
964        jobs,
965        verbose,
966    })
967}
968
969/// What one of the `-dump` and `-print` flags prints.
970///
971/// GCC prints the name back unchanged when it cannot find the file a `-print` flag asked about,
972/// which is what makes the answer safe to paste into a link line whether or not the file is
973/// there, and this does the same.
974fn answer(query: &Query, opts: &Options, link: &LinkOptions) -> String {
975    let found = |name: &str| {
976        link::find_in_search(link, opts.target, name)
977            .map_or_else(|| name.to_owned(), |path| path.display().to_string())
978    };
979    match query {
980        Query::Machine => opts.target.to_string(),
981        Query::Version => VERSION.to_owned(),
982        Query::Multiarch => link::multiarch(opts.target),
983        // The three lines GCC prints, in its order and with its punctuation, because what reads
984        // them is a script written against that shape. There is no installation directory to
985        // report: this compiler is one binary that works wherever it is copied, and the headers
986        // it ships are inside it, so `install` is where the binary is and nothing is under it.
987        Query::SearchDirs => {
988            let here = std::env::current_exe()
989                .ok()
990                .and_then(|p| p.parent().map(std::path::Path::to_path_buf))
991                .unwrap_or_default();
992            let list = |dirs: &[PathBuf]| {
993                dirs.iter().map(|d| d.display().to_string()).collect::<Vec<_>>().join(":")
994            };
995            let libraries = link::search_dirs(link, opts.target);
996            format!(
997                "install: {}\nprograms: ={}\nlibraries: ={}",
998                here.display(),
999                list(&link.prefixes),
1000                list(&libraries)
1001            )
1002        }
1003        Query::FileName(name) => found(name),
1004        // The name GCC gives the library of routines a compiler's output calls that the C
1005        // library does not have. Ours is built in and there is no file, so the answer is the
1006        // name itself, which is what GCC prints when it cannot find one either.
1007        Query::Libgcc => found("libgcc.a"),
1008        // A program rather than a library: the linker and the archiver are the ones a build asks
1009        // about, and this compiler finds them on the path or under `-B` rather than shipping
1010        // them, so the name back is the honest answer unless a `-B` prefix holds one.
1011        Query::ProgName(name) => link
1012            .prefixes
1013            .iter()
1014            .map(|dir| dir.join(name))
1015            .find(|path| path.is_file())
1016            .map_or_else(|| name.clone(), |path| path.display().to_string()),
1017    }
1018}
1019
1020/// Renders the passes this level will run, in order, with what each one does.
1021///
1022/// The level is the whole of the answer unless a `-f` flag edited it, which is section 9.1 of
1023/// `spec/09-optimizer.md`: a level is a list somebody wrote down rather than something that
1024/// emerges from which flags happen to be set, and this is how that list is read.
1025#[must_use]
1026pub fn print_pipeline(opts: &Options) -> String {
1027    let mut settings = rucc_opt::Options::for_level(opts.opt_level);
1028    settings.toggles.clone_from(&opts.passes);
1029    settings.global_fuel = opts.pass_fuel_global;
1030    for (on, spec) in &opts.pass_gates {
1031        // Every spelling was checked while the arguments were parsed, so there is nothing here
1032        // this can refuse, and a listing is not the place to report it if there were.
1033        let _ = settings.gates.add(*on, spec);
1034    }
1035    rucc_opt::pipeline::print(&settings)
1036}
1037
1038/// Renders the resolved configuration.
1039///
1040/// One `key: value` per line, sorted by nothing in particular but fixed in order, because
1041/// this output is diffed across hosts in CI and a reordering would read as a change.
1042#[must_use]
1043pub fn print_config(opts: &Options) -> String {
1044    let sess = Session::new(opts.clone());
1045    let t = &sess.target;
1046    let mut out = String::new();
1047    let _ = writeln!(out, "version: {VERSION}");
1048    // The three field triple the driver was given rather than the ten field tuple it widens to,
1049    // because this output is what a build system reads to find out what it asked for. The tuple is
1050    // the compiler's model of the machine and this line is a receipt for a command line.
1051    let _ = writeln!(out, "target: {}", opts.target);
1052    let _ = writeln!(out, "arch: {}", opts.target.arch.as_str());
1053    let _ = writeln!(out, "os: {}", opts.target.os.as_str());
1054    let _ = writeln!(out, "env: {}", opts.target.env.as_str());
1055    let _ = writeln!(out, "object-format: {}", t.object_format.as_str());
1056    let _ = writeln!(out, "pointer-width: {}", t.pointer_width);
1057    let _ = writeln!(out, "long-width: {}", t.long_width);
1058    let _ = writeln!(out, "long-double-width: {}", t.long_double_width);
1059    let _ = writeln!(out, "endian: {}", if t.little_endian { "little" } else { "big" });
1060    let _ = writeln!(out, "char-signed: {}", t.char_is_signed);
1061    let _ = writeln!(out, "va-list: {}", t.va_list.map_or("none", |list| list.as_str()));
1062    // The register file as a count per class, which is enough to tell a target whose registers
1063    // are described from one whose are not without printing sixteen names nobody asked for.
1064    let regs: Vec<String> = t
1065        .regs
1066        .classes()
1067        .map(|(class, info)| format!("{} {}", info.name, t.regs.len(class)))
1068        .collect();
1069    let _ = writeln!(
1070        out,
1071        "registers: {}",
1072        if regs.is_empty() { "none".to_string() } else { regs.join(", ") }
1073    );
1074    let _ = writeln!(out, "opt-level: {}", sess.opts.opt_level);
1075    let _ = writeln!(out, "safety: {}", sess.opts.safety);
1076    let _ = writeln!(out, "emit: {}", sess.opts.emit.as_str());
1077    let _ = writeln!(out, "debug-info: {}", sess.opts.debug_info);
1078    let _ = writeln!(out, "frame-pointer: {}", sess.opts.frame_pointer);
1079    let _ = writeln!(out, "red-zone: {}", sess.opts.red_zone);
1080    // Last because it is the one key with more than one line under it, and the only one
1081    // whose value is a property of the machine rather than of the command line.
1082    for dir in sess.opts.search.dirs() {
1083        let system = if dir.is_system { " (system)" } else { "" };
1084        let _ = writeln!(out, "include: {}{system}", dir.path.display());
1085    }
1086    out
1087}
1088
1089/// The output name the make target is taken from, which is the `-o` argument or nothing.
1090///
1091/// A run that stops at the preprocessor has not named an object, whatever its `-o` says: under
1092/// `-E` that argument is the preprocessed text and under `-M` it is the rule itself, and neither
1093/// is a file `make` would rebuild by running this rule. GCC agrees and falls back to the source
1094/// name in both, which is why a `-MD -E -o out.i` writes `out.d` holding a rule for `a.o`. From
1095/// `-S` on the argument does name what the rule builds, and it is used as written.
1096fn deps_target_output<'a>(opts: &Options, plan: &'a Plan) -> Option<&'a str> {
1097    if opts.emit == EmitKind::Preprocessed { None } else { plan.output.as_deref() }
1098}
1099
1100/// Writes to a path the command line named rather than one the plan derived, where `-` is
1101/// standard output.
1102fn write_named(path: &str, bytes: &[u8]) -> Result<(), String> {
1103    if path == "-" {
1104        return write_out(&Output::Stdout, bytes);
1105    }
1106    write_out(&Output::File(path.to_owned()), bytes)
1107}
1108
1109/// Writes the make rule for one input, and reports whether it got there.
1110///
1111/// A rule with no file of its own goes where the compilation it replaced would have written,
1112/// which is what makes the usual makefile recipe work: `rucc -M $< -o $@` leaves the rule in
1113/// `$@`, and the same line with the `-o` left off puts it on standard output.
1114fn write_deps(
1115    opts: &Options,
1116    plan: &Plan,
1117    job: &Job,
1118    found: &[Dependency],
1119    stderr: &mut impl std::io::Write,
1120) -> bool {
1121    let targets = if opts.deps.targets.is_empty() {
1122        vec![deps::default_target(&job.input, deps_target_output(opts, plan))]
1123    } else {
1124        opts.deps.targets.clone()
1125    };
1126    let rule = deps::rule(&opts.deps, &targets, &job.input, found);
1127    // The file, on the other hand, is named after the `-o` in every mode that still has one to
1128    // spend, which is every mode except the two that spend it on the rule.
1129    let wrote = match deps::default_file(&opts.deps, &job.input, plan.output.as_deref()) {
1130        // A `-MF` on a run that had nowhere else to put the rule leaves the file the `-o`
1131        // named empty rather than absent, because a makefile that named it as a target of its
1132        // own is a makefile that will look for it.
1133        Some(path) => write_named(&path, rule.as_bytes()).and_then(|()| {
1134            if opts.deps.instead_of_compiling { write_out(&job.output, b"") } else { Ok(()) }
1135        }),
1136        None => write_out(&job.output, rule.as_bytes()),
1137    };
1138    if let Err(e) = wrote {
1139        let _ = writeln!(stderr, "rucc: error: {e}");
1140        return false;
1141    }
1142    true
1143}
1144
1145/// Runs phase 4 over every input that has one, and writes what came out.
1146///
1147/// One input that fails does not stop the others. A build that reports every file it could
1148/// not preprocess in one run is worth more than one that stops at the first, and the exit
1149/// status is still a failure either way.
1150fn preprocess_all(opts: &Options, plan: &Plan) -> i32 {
1151    let fs = OsFileSystem::new();
1152    let mut stderr = std::io::stderr().lock();
1153    let mut failed = false;
1154    for job in &plan.jobs {
1155        if !job.phases.first().is_some_and(|p| *p == Phase::Preprocess) {
1156            // An input that is already preprocessed, or an object file. GCC passes these
1157            // through untouched, and the plan has already said so in its notes.
1158            continue;
1159        }
1160        let started = std::time::Instant::now();
1161        let result = preprocess(opts, &job.input, &fs);
1162        if opts.time {
1163            say_time(&job.input, started.elapsed(), &mut stderr);
1164        }
1165        for message in &result.messages {
1166            let _ = writeln!(stderr, "{message}");
1167        }
1168        if result.failed() {
1169            failed = true;
1170            continue;
1171        }
1172        if opts.deps.emit {
1173            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1174            // `-M` and `-MM` asked for the rule instead of the text, so there is nothing else
1175            // to write. The other two asked for both and fall through to the text below.
1176            if opts.deps.instead_of_compiling {
1177                continue;
1178            }
1179        }
1180        if let Err(e) = write_out(&job.output, result.text.as_bytes()) {
1181            let _ = writeln!(stderr, "rucc: error: {e}");
1182            failed = true;
1183        }
1184    }
1185    i32::from(failed)
1186}
1187
1188/// Runs the front end over every input that has a compile phase, and writes what came out.
1189///
1190/// The same rule as [`preprocess_all`]: one input that fails does not stop the others, and the
1191/// exit status is a failure either way. An input that is already assembly or an object has no
1192/// compile phase and is passed over here, which the plan has already said in its notes.
1193fn compile_all(opts: &Options, plan: &Plan) -> i32 {
1194    let fs = OsFileSystem::new();
1195    let mut stderr = std::io::stderr().lock();
1196    let mut failed = false;
1197    let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1198    failed |= !ok;
1199    let mut fired = Fired::new();
1200    for job in &plan.jobs {
1201        if !job.phases.contains(&Phase::Compile) {
1202            continue;
1203        }
1204        // An input of IR is read back rather than compiled, since the C it came from is not
1205        // here any more. Everything after this is the same, so the two paths meet again at the
1206        // messages and the file the result is written to.
1207        let started = std::time::Instant::now();
1208        let result = if job.kind == InputKind::Ir {
1209            compile_ir(opts, &job.input, &fs)
1210        } else {
1211            compile(opts, &job.input, &fs)
1212        };
1213        if opts.time {
1214            say_time(&job.input, started.elapsed(), &mut stderr);
1215        }
1216        fired.merge(&result.fired);
1217        failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1218        failed |= !remarks.write(&result.remarks, &mut stderr);
1219        for message in &result.messages {
1220            let _ = writeln!(stderr, "{message}");
1221        }
1222        // Before the failure below, because a compilation that stopped in the back end is exactly
1223        // the one whose preprocessed source somebody wants to look at.
1224        failed |= !write_temps(job, &result.temps, &mut stderr);
1225        if result.failed() {
1226            failed = true;
1227            continue;
1228        }
1229        // `-MD` and `-MMD` write the rule beside the object and let the compilation happen, so
1230        // this is the one path where both files come out of the same run. An input of IR has no
1231        // dependencies to report and produces an empty list, which produces a rule naming only
1232        // itself, and that is the honest answer rather than a missing file.
1233        if opts.deps.emit {
1234            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1235        }
1236        if let Err(e) = write_out(&job.output, result.artifact.bytes()) {
1237            let _ = writeln!(stderr, "rucc: error: {e}");
1238            failed = true;
1239        }
1240    }
1241    failed |= !write_coverage(opts, &fired, &mut stderr);
1242    i32::from(failed)
1243}
1244
1245/// A directory for the object files only the link step ever sees, removed when it goes away.
1246///
1247/// `-c` writes its object where the user can see it and linking does not, which is the whole of
1248/// the difference: a `rucc a.c b.c` leaves an executable behind and nothing else, the same as
1249/// every other compiler. Removing them on drop rather than at the end of a function is so that a
1250/// link that failed leaves nothing behind either.
1251struct Scratch {
1252    /// Where the objects go.
1253    dir: PathBuf,
1254}
1255
1256impl Scratch {
1257    /// Makes one, under whatever the platform calls its temporary directory.
1258    ///
1259    /// The name carries the process id so that two compilers running at once do not share a
1260    /// directory, which they would otherwise do the moment two of them compiled a file of the
1261    /// same name.
1262    fn new() -> Result<Scratch, String> {
1263        let dir = std::env::temp_dir().join(format!("rucc-{}", std::process::id()));
1264        std::fs::create_dir_all(&dir).map_err(|e| format!("{}: {e}", dir.display()))?;
1265        Ok(Scratch { dir })
1266    }
1267}
1268
1269impl Drop for Scratch {
1270    fn drop(&mut self) {
1271        let _ = std::fs::remove_dir_all(&self.dir);
1272    }
1273}
1274
1275/// The link line the plan describes, for `-###`.
1276///
1277/// The names in it are the hints the plan carries rather than the temporaries a real compilation
1278/// would choose, because `-###` prints the line without having compiled anything and so has
1279/// nothing to point at. That also makes the printed line readable rather than naming a directory
1280/// that only exists while a compilation is running.
1281fn link_line(opts: &Options, link: &LinkOptions, job: &LinkJob) -> Result<String, link::Error> {
1282    let linker = link::find(opts.target, link)?;
1283    let args = link::line(opts.target, link, &job.inputs, &job.output)?;
1284    Ok(link::render(&linker, &args))
1285}
1286
1287/// Compiles everything, then links it.
1288///
1289/// The objects go in a directory that is removed afterwards, which is why this is not
1290/// [`compile_all`] followed by a link: the plan says an object feeding the linker is temporary
1291/// and does not say where, because where is a question that only has an answer once something is
1292/// running.
1293fn link_all(opts: &Options, plan: &Plan, link: &LinkOptions, verbose: bool) -> i32 {
1294    let Some(job) = &plan.link else {
1295        // Every path into here comes from a plan whose last phase is the link, and such a plan
1296        // has a link job. Saying so is cheaper than an unwrap that would have to be explained.
1297        let mut stderr = std::io::stderr().lock();
1298        let _ = writeln!(stderr, "rucc: error: there is nothing to link");
1299        return 1;
1300    };
1301    // Before anything is compiled, because a linker that is not on the machine is worth knowing
1302    // about in the second it takes to look rather than after the compilation.
1303    let linker = match link::find(opts.target, link) {
1304        Ok(linker) => linker,
1305        Err(why) => return complain(why),
1306    };
1307
1308    let scratch = match Scratch::new() {
1309        Ok(scratch) => scratch,
1310        Err(why) => return complain(format!("could not make a place for the object files: {why}")),
1311    };
1312
1313    let fs = OsFileSystem::new();
1314    let mut failed = false;
1315    // One per job, in job order, which is what lets the link line below be rebuilt with the real
1316    // paths in it: every job contributes exactly one file to the line and does so in this order.
1317    let mut produced: Vec<String> = Vec::with_capacity(plan.jobs.len());
1318    let mut fired = Fired::new();
1319    {
1320        let mut stderr = std::io::stderr().lock();
1321        let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1322        failed |= !ok;
1323        for (at, job) in plan.jobs.iter().enumerate() {
1324            let out = match &job.output {
1325                Output::Temporary(hint) => {
1326                    // The index because two inputs in different directories can have the same
1327                    // name, and the two objects of `rucc a/x.c b/x.c` must not be one file.
1328                    scratch.dir.join(format!("{at}-{hint}")).display().to_string()
1329                }
1330                Output::File(path) => path.clone(),
1331                // A job feeding the linker never writes to standard output, since the plan gives
1332                // it a temporary. This is here so that the match is total rather than a panic.
1333                Output::Stdout => continue,
1334            };
1335            produced.push(out.clone());
1336            if !job.phases.contains(&Phase::Compile) {
1337                continue;
1338            }
1339            let started = std::time::Instant::now();
1340            let result = if job.kind == InputKind::Ir {
1341                compile_ir(opts, &job.input, &fs)
1342            } else {
1343                compile(opts, &job.input, &fs)
1344            };
1345            if opts.time {
1346                say_time(&job.input, started.elapsed(), &mut stderr);
1347            }
1348            fired.merge(&result.fired);
1349            failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1350            failed |= !remarks.write(&result.remarks, &mut stderr);
1351            for message in &result.messages {
1352                let _ = writeln!(stderr, "{message}");
1353            }
1354            failed |= !write_temps(job, &result.temps, &mut stderr);
1355            if result.failed() {
1356                failed = true;
1357                continue;
1358            }
1359            // A `-MD` on a command line that links writes the rule next to the executable and
1360            // names the executable as its target, since that is the file this source builds
1361            // here. The object it went through is in a temporary directory and is gone by the
1362            // time `make` reads any of this.
1363            if opts.deps.emit {
1364                failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1365            }
1366            if !matches!(result.artifact, Artifact::Object(_)) {
1367                // Worth saying rather than writing whatever it is and letting the linker read it.
1368                // An empty file is a valid empty linker script, so a link handed one gets as far
1369                // as reporting every symbol of this file undefined, which is a page of messages
1370                // about something that went wrong here.
1371                let _ = writeln!(
1372                    stderr,
1373                    "rucc: internal error: {}: no object file was produced for the link",
1374                    job.input
1375                );
1376                failed = true;
1377                continue;
1378            }
1379            if let Err(e) = std::fs::write(&out, result.artifact.bytes()) {
1380                let _ = writeln!(stderr, "rucc: error: {out}: {e}");
1381                failed = true;
1382            }
1383        }
1384        failed |= !write_coverage(opts, &fired, &mut stderr);
1385    }
1386    if failed {
1387        // Nothing is linked from a compilation that did not finish. A linker run over the objects
1388        // that did compile would report every function of the file that did not as undefined,
1389        // which is a page of messages about a mistake already reported once.
1390        return 1;
1391    }
1392
1393    // The items in command line order with the temporaries filled in. A library contributes no
1394    // job and passes through, and every file item takes the next job's real output, which is
1395    // what keeps a library that was written between two objects between them here.
1396    let mut outputs = produced.into_iter();
1397    let mut items = Vec::with_capacity(job.inputs.len());
1398    for item in &job.inputs {
1399        match item {
1400            link::Item::Library(name) => items.push(link::Item::Library(name.clone())),
1401            link::Item::File(_) => match outputs.next() {
1402                Some(path) => items.push(link::Item::File(path)),
1403                None => return complain("the plan asks the linker for a file nothing produced"),
1404            },
1405        }
1406    }
1407
1408    let args = match link::line(opts.target, link, &items, &job.output) {
1409        Ok(args) => args,
1410        Err(why) => return complain(why),
1411    };
1412    if verbose {
1413        let mut stderr = std::io::stderr().lock();
1414        let _ = writeln!(stderr, "{}", link::render(&linker, &args));
1415    }
1416    let started = std::time::Instant::now();
1417    let ran = link::run(&linker, &args);
1418    if opts.time {
1419        // The one step of a compilation that really is another program, so this line is the same
1420        // measurement gcc's is and names the linker the way gcc names `collect2`.
1421        let mut stderr = std::io::stderr().lock();
1422        say_time(&linker.name, started.elapsed(), &mut stderr);
1423    }
1424    match ran {
1425        Ok(()) => 0,
1426        // The linker has already said what was wrong on its own error output, and repeating that
1427        // linking failed would only push its message further up the screen.
1428        Err(link::Error::Refused { .. }) => 1,
1429        Err(why) => complain(why),
1430    }
1431}
1432
1433/// Prints one driver level message and gives back the exit status that goes with it.
1434fn complain(why: impl std::fmt::Display) -> i32 {
1435    let mut stderr = std::io::stderr().lock();
1436    let _ = writeln!(stderr, "rucc: error: {why}");
1437    1
1438}
1439
1440/// Writes what `-Zrule-coverage=FILE` asked for, and says whether it could.
1441///
1442/// Once for the whole command line rather than once per input, because the question is which
1443/// lowering rules this run of the compiler reached and a file per input would leave the reader
1444/// unioning files to find out something one process already knew.
1445///
1446/// A file that could not be written is a failure and not a warning. What asks for this is a
1447/// measurement run, and a measurement that quietly did not happen is worse than one that stopped.
1448fn write_coverage(opts: &Options, fired: &Fired, stderr: &mut impl std::io::Write) -> bool {
1449    let Some(path) = &opts.rule_coverage else { return true };
1450    let Some(table) = coverage::table(opts.target.arch) else {
1451        let _ = writeln!(
1452            stderr,
1453            "rucc: error: there are no lowering rules for {} yet, so there is no coverage of them \
1454             to report",
1455            opts.target
1456        );
1457        return false;
1458    };
1459    match std::fs::write(path, fired.listing(table)) {
1460        Ok(()) => true,
1461        Err(e) => {
1462            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1463            false
1464        }
1465    }
1466}
1467
1468/// Where the `-fopt-info` remarks go, and how much of the run has already gone there.
1469///
1470/// Standard error by default, and one file for the whole run when `-fopt-info=<file>` named one.
1471/// A file rather than the diagnostic stream is what a harness wants: the corpus in
1472/// `tamnd/rucc-corpus` matches a rejection against what the compiler said on standard error, and
1473/// a few thousand remarks mixed into that would bury it.
1474struct Remarks {
1475    /// The file, if there is one.
1476    file: Option<String>,
1477    /// Whether anything has been written to it yet, which decides between truncating and
1478    /// appending. One file holds the whole run rather than the last input in it.
1479    started: bool,
1480}
1481
1482impl Remarks {
1483    /// Prepares the destination, emptying the file if there is one.
1484    ///
1485    /// Emptied here rather than at the first remark, because a run where no pass had anything to
1486    /// say should leave an empty file and not yesterday's. An absent file and an empty one are
1487    /// different facts and something reading this will act on the difference.
1488    fn new(file: Option<&String>, stderr: &mut impl std::io::Write) -> (Self, bool) {
1489        let mut ok = true;
1490        if let Some(path) = file {
1491            if let Err(e) = std::fs::write(path, "") {
1492                let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1493                ok = false;
1494            }
1495        }
1496        (Self { file: file.cloned(), started: false }, ok)
1497    }
1498
1499    /// Writes one input's remarks, and says whether that worked.
1500    ///
1501    /// A file that cannot be written is a failure and not a warning, for the reason
1502    /// [`write_dumps`] gives: remarks that quietly did not arrive look exactly like a compilation
1503    /// where nothing happened.
1504    fn write(&mut self, text: &str, stderr: &mut impl std::io::Write) -> bool {
1505        if text.is_empty() {
1506            return true;
1507        }
1508        let Some(path) = &self.file else {
1509            let _ = write!(stderr, "{text}");
1510            return true;
1511        };
1512        let opened = std::fs::OpenOptions::new()
1513            .write(true)
1514            .append(self.started)
1515            .truncate(!self.started)
1516            .create(true)
1517            .open(path);
1518        self.started = true;
1519        let result =
1520            opened.and_then(|mut file| std::io::Write::write_all(&mut file, text.as_bytes()));
1521        if let Err(e) = result {
1522            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1523            return false;
1524        }
1525        true
1526    }
1527}
1528
1529/// Writes what `-fdump-ir=` asked to see, one file per dump.
1530///
1531/// The name is the input file with the dump's own name and `.ir` after it, so a directory listing
1532/// after a run is the passes in the order they ran, per input. They go in the working directory
1533/// rather than beside the output, because a dump is something a person asked for at a prompt and
1534/// the working directory is where that person is.
1535///
1536/// A file that could not be written is a failure and not a warning, for the reason
1537/// [`write_coverage`] gives: what asked for this is somebody debugging a pass, and a dump that
1538/// quietly did not happen looks exactly like a pass that did not run.
1539fn write_dumps(input: &str, dumps: &[rucc_opt::Dump], stderr: &mut impl std::io::Write) -> bool {
1540    let stem = std::path::Path::new(input)
1541        .file_name()
1542        .map_or_else(|| input.to_owned(), |name| name.to_string_lossy().into_owned());
1543    let mut ok = true;
1544    for dump in dumps {
1545        let path = format!("{stem}.{}.ir", dump.name);
1546        if let Err(e) = std::fs::write(&path, &dump.text) {
1547            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1548            ok = false;
1549        }
1550    }
1551    ok
1552}
1553
1554/// Writes the files `-save-temps` kept, which is nothing at all unless it was given.
1555///
1556/// A file that could not be written is a failure rather than a warning, for the reason
1557/// [`write_dumps`] gives: somebody asked for these by name, and one that quietly did not happen
1558/// looks like a compilation that never went through that step.
1559fn write_temps(job: &Job, temps: &Temps, stderr: &mut impl std::io::Write) -> bool {
1560    let mut ok = true;
1561    let kept = [(job.saved_text(), &temps.preprocessed), (job.saved_asm(), &temps.assembly)];
1562    for (path, text) in kept {
1563        // A step the compilation did not reach has nothing to keep, and a job that is not keeping
1564        // that step has nowhere to put it. Either way there is no file here.
1565        let (Some(path), Some(text)) = (path, text) else { continue };
1566        if let Err(e) = std::fs::write(&path, text) {
1567            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1568            ok = false;
1569        }
1570    }
1571    ok
1572}
1573
1574/// One line of `-time`, which is what a step was called and how long it took.
1575///
1576/// GCC's two numbers are the user and the system time of a subprocess it ran. This compiler runs
1577/// no subprocess for anything but the link, so what is measured here is the wall clock of the
1578/// step and the second column is always zero. The shape of the line is kept because a person
1579/// reading it next to gcc's should not have to work out which column is which.
1580fn say_time(name: &str, took: std::time::Duration, stderr: &mut impl std::io::Write) {
1581    let _ = writeln!(stderr, "# {name} {:.2} {:.2}", took.as_secs_f64(), 0.0);
1582}
1583
1584/// Writes one job's result where the plan said it goes.
1585///
1586/// # Errors
1587///
1588/// Returns the message to print, which names the file when there is one, because "permission
1589/// denied" on its own does not say which file was refused.
1590fn write_out(output: &Output, bytes: &[u8]) -> Result<(), String> {
1591    match output {
1592        Output::Stdout => {
1593            let mut stdout = std::io::stdout().lock();
1594            stdout.write_all(bytes).map_err(|e| format!("writing to standard output: {e}"))
1595        }
1596        Output::File(path) | Output::Temporary(path) => {
1597            std::fs::write(path, bytes).map_err(|e| format!("{path}: {e}"))
1598        }
1599    }
1600}
1601
1602/// Runs the driver and returns the process exit code.
1603///
1604/// `args` excludes the program name. Output goes to `stdout` and errors to `stderr`, which
1605/// is the one place in the compiler that is true.
1606pub fn run(args: &[String]) -> i32 {
1607    match parse_args(args) {
1608        Ok(Action::Help) => {
1609            print!("{USAGE}");
1610            0
1611        }
1612        Ok(Action::Version) => {
1613            println!("rucc {VERSION}");
1614            0
1615        }
1616        Ok(Action::Print(line)) => {
1617            println!("{line}");
1618            0
1619        }
1620        Ok(Action::PrintConfig(opts)) => {
1621            print!("{}", print_config(&opts));
1622            0
1623        }
1624        Ok(Action::PrintPipeline(opts)) => {
1625            print!("{}", print_pipeline(&opts));
1626            0
1627        }
1628        Ok(Action::PrintPlan { opts, plan, link }) => {
1629            print!("{}", plan.render());
1630            // The line as it would be typed, which is the half of `-###` that section 4.3 says
1631            // arrives with the link. It is printed even when the linker is not on this machine,
1632            // because what a build wants from `-###` is what the compiler would do.
1633            if let Some(job) = &plan.link {
1634                match link_line(&opts, &link, job) {
1635                    Ok(line) => println!("{line}"),
1636                    Err(why) => {
1637                        let mut stderr = std::io::stderr().lock();
1638                        let _ = writeln!(stderr, "rucc: error: {why}");
1639                        return 1;
1640                    }
1641                }
1642            }
1643            0
1644        }
1645        Ok(Action::Compile { opts, plan, link, jobs, verbose }) => {
1646            {
1647                let mut stderr = std::io::stderr().lock();
1648                if verbose {
1649                    let _ = write!(stderr, "{}", plan.render());
1650                    let _ = writeln!(stderr, "workers: {}", jobs.count());
1651                }
1652            }
1653            if opts.emit == EmitKind::Preprocessed {
1654                return preprocess_all(&opts, &plan);
1655            }
1656            if opts.emit != EmitKind::Executable {
1657                return compile_all(&opts, &plan);
1658            }
1659            link_all(&opts, &plan, &link, verbose)
1660        }
1661        Err(e) => {
1662            let mut stderr = std::io::stderr().lock();
1663            let _ = writeln!(stderr, "rucc: error: {e}");
1664            let _ = writeln!(stderr, "rucc: note: run `rucc --help` for usage");
1665            1
1666        }
1667    }
1668}
1669
1670#[cfg(test)]
1671mod tests {
1672    use rucc_session::{GnucVersion, IncludeForm, OptLevel, Visibility};
1673
1674    use super::*;
1675
1676    fn args(s: &[&str]) -> Vec<String> {
1677        s.iter().map(|x| (*x).to_owned()).collect()
1678    }
1679
1680    #[test]
1681    fn help_and_version_win_over_everything_else() {
1682        assert_eq!(parse_args(&args(&["-c", "--help", "x.c"])).unwrap(), Action::Help);
1683        assert_eq!(parse_args(&args(&["--version"])).unwrap(), Action::Version);
1684    }
1685
1686    fn compile(s: &[&str]) -> (Box<Options>, Box<Plan>) {
1687        match parse_args(&args(s)).expect("expected a compilation") {
1688            Action::Compile { opts, plan, .. } => (opts, plan),
1689            other => panic!("expected a compilation, got {other:?}"),
1690        }
1691    }
1692
1693    fn linking(s: &[&str]) -> (Box<LinkOptions>, Box<Plan>) {
1694        match parse_args(&args(s)).expect("expected a compilation") {
1695            Action::Compile { link, plan, .. } => (link, plan),
1696            other => panic!("expected a compilation, got {other:?}"),
1697        }
1698    }
1699
1700    #[test]
1701    fn collects_inputs_and_flags() {
1702        let (opts, plan) = compile(&["-c", "-O2", "-g", "a.c", "b.c"]);
1703        let paths: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
1704        assert_eq!(paths, vec!["a.c", "b.c"]);
1705        assert_eq!(opts.opt_level, OptLevel::O2);
1706        assert_eq!(opts.emit, EmitKind::Object);
1707        assert!(opts.debug_info);
1708    }
1709
1710    /// The unstable options, which are spelled apart from everything else on purpose: what is
1711    /// under `-Z` promises nothing, and a build that reaches for one should have had to say so.
1712    #[test]
1713    fn an_unstable_option_is_taken_and_one_that_does_not_exist_is_refused() {
1714        let (opts, _) = compile(&["-c", "-Zrule-coverage=/tmp/rules.cov", "a.c"]);
1715        assert_eq!(opts.rule_coverage.as_deref(), Some("/tmp/rules.cov"));
1716
1717        let (plain, _) = compile(&["-c", "a.c"]);
1718        assert_eq!(plain.rule_coverage, None, "nothing is measured unless it was asked for");
1719
1720        assert!(parse_args(&args(&["-Zrule-coverage=", "a.c"])).is_err(), "a file with no name");
1721        let unknown = parse_args(&args(&["-Zwhat", "a.c"])).expect_err("there is no such option");
1722        assert!(unknown.message.contains("4.11"), "{}", unknown.message);
1723    }
1724
1725    #[test]
1726    fn a_bare_dash_o_means_o1_the_way_gcc_reads_it() {
1727        let (opts, _) = compile(&["-O", "a.c"]);
1728        assert_eq!(opts.opt_level, OptLevel::O1);
1729    }
1730
1731    #[test]
1732    fn dash_x_applies_to_later_inputs_only_and_none_stops_it() {
1733        let (_, plan) = compile(&["a.o", "-x", "c", "b.txt", "-x", "none", "c.o"]);
1734        assert_eq!(plan.jobs[0].kind, InputKind::LinkerInput);
1735        assert_eq!(plan.jobs[1].kind, InputKind::C);
1736        assert_eq!(plan.jobs[2].kind, InputKind::LinkerInput);
1737    }
1738
1739    #[test]
1740    fn dash_j_reaches_the_scheduler_and_defaults_to_the_machine() {
1741        let (_, _, jobs) = match parse_args(&args(&["-j4", "a.c"])).unwrap() {
1742            Action::Compile { opts, plan, jobs, .. } => (opts, plan, jobs),
1743            other => panic!("expected a compilation, got {other:?}"),
1744        };
1745        assert_eq!(jobs.count(), 4);
1746
1747        let default = match parse_args(&args(&["a.c"])).unwrap() {
1748            Action::Compile { jobs, .. } => jobs,
1749            other => panic!("expected a compilation, got {other:?}"),
1750        };
1751        assert_eq!(default, Jobs::available());
1752        assert!(parse_args(&args(&["-j0", "a.c"])).is_err());
1753    }
1754
1755    #[test]
1756    fn triple_hash_prints_the_plan_and_runs_nothing() {
1757        let a = parse_args(&args(&["-###", "-c", "a.c"])).unwrap();
1758        let Action::PrintPlan { plan, .. } = a else { panic!("expected a plan dump") };
1759        assert!(plan.render().contains("a.c: preprocess, compile, assemble -> a.o"));
1760    }
1761
1762    #[test]
1763    fn the_flag_that_keeps_the_intermediate_files_has_three_spellings_and_two_meanings() {
1764        // The bare one is `=obj` and not `=cwd`. gcc's manual says the opposite and gcc 16 does
1765        // this, and following the compiler is what makes a build that reads either of them find
1766        // the files where they are.
1767        assert_eq!(compile(&["-c", "-save-temps", "a.c"]).0.save_temps, SaveTemps::Object);
1768        assert_eq!(compile(&["-c", "-save-temps=obj", "a.c"]).0.save_temps, SaveTemps::Object);
1769        assert_eq!(compile(&["-c", "-save-temps=cwd", "a.c"]).0.save_temps, SaveTemps::Cwd);
1770        assert_eq!(compile(&["-c", "a.c"]).0.save_temps, SaveTemps::No);
1771        // The last one on the line decides, the way it does for every other flag with an
1772        // argument, and a keyword that is neither is fatal rather than ignored: a run that kept
1773        // nothing and said nothing looks exactly like one where the files were not produced.
1774        let (opts, _) = compile(&["-c", "-save-temps", "-save-temps=cwd", "a.c"]);
1775        assert_eq!(opts.save_temps, SaveTemps::Cwd);
1776        let e = parse_args(&args(&["-c", "-save-temps=nowhere", "a.c"])).unwrap_err();
1777        assert!(e.message.contains("accepted: cwd, obj"), "{}", e.message);
1778    }
1779
1780    #[test]
1781    fn the_flag_that_times_each_step_reaches_the_options_and_changes_nothing_else() {
1782        let (opts, plan) = compile(&["-c", "-time", "a.c"]);
1783        let (plain, without) = compile(&["-c", "a.c"]);
1784        assert!(opts.time);
1785        assert!(!plain.time);
1786        // Against the same line without the flag rather than against a spelling of the object's
1787        // name, since what the object is called is the host's business and this is not about that.
1788        assert_eq!(plan.jobs[0].output, without.jobs[0].output);
1789    }
1790
1791    #[test]
1792    fn dash_x_names_what_it_accepts_when_it_does_not_know_a_language() {
1793        let e = parse_args(&args(&["-x", "fortran", "a.c"])).unwrap_err();
1794        assert!(e.message.contains("assembler-with-cpp"), "{}", e.message);
1795    }
1796
1797    #[test]
1798    fn an_unknown_flag_is_an_error_rather_than_a_shrug() {
1799        let e = parse_args(&args(&["-fno-such-thing", "a.c"])).unwrap_err();
1800        assert!(e.message.contains("unknown option"), "{}", e.message);
1801    }
1802
1803    /// `-fpermissive` and the flag that turns it back off, which a build writes beside it when
1804    /// one directory needs the older rules and the rest of the tree does not.
1805    #[test]
1806    fn permissive_reads_in_both_directions_and_the_last_one_wins() {
1807        let (opts, _) = compile(&["-c", "a.c"]);
1808        assert!(!opts.permissive, "off unless it is asked for");
1809
1810        let (opts, _) = compile(&["-c", "-fpermissive", "a.c"]);
1811        assert!(opts.permissive);
1812
1813        let (opts, _) = compile(&["-c", "-fpermissive", "-fno-permissive", "a.c"]);
1814        assert!(!opts.permissive);
1815    }
1816
1817    #[test]
1818    fn asking_for_nested_functions_is_told_why_it_is_not_coming() {
1819        let e = parse_args(&args(&["-fnested-functions", "a.c"])).unwrap_err();
1820        assert!(e.message.contains("trampoline"), "{}", e.message);
1821        assert!(parse_args(&args(&["-fno-nested-functions", "a.c"])).is_ok());
1822    }
1823
1824    #[test]
1825    fn the_flag_every_configure_script_writes_is_taken() {
1826        // All four spellings, because a build writes whichever one its macros picked and a
1827        // compiler that takes three of them is a compiler that fails on the fourth.
1828        for flag in ["-fPIC", "-fpic", "-fPIE", "-fpie"] {
1829            let (opts, _) = compile(&["-c", flag, "a.c"]);
1830            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1831        }
1832    }
1833
1834    #[test]
1835    fn a_table_is_written_unless_the_build_says_nothing_will_walk_it() {
1836        let (opts, _) = compile(&["-c", "a.c"]);
1837        assert!(opts.unwinds(), "the default is off");
1838        let (opts, _) = compile(&["-c", "-fno-asynchronous-unwind-tables", "a.c"]);
1839        assert!(!opts.unwinds(), "the build was not taken at its word");
1840        let (opts, _) = compile(&[
1841            "-c",
1842            "-fno-asynchronous-unwind-tables",
1843            "-fasynchronous-unwind-tables",
1844            "a.c",
1845        ]);
1846        assert!(opts.unwinds(), "the last flag did not win");
1847        // The weaker request, which the same table answers, so a line that asks for a table and
1848        // against an asynchronous one gets one. That is gcc's arrangement and it turns up when a
1849        // build turns the asynchronous one off globally and a directory asks for a table back.
1850        let (opts, _) =
1851            compile(&["-c", "-fno-asynchronous-unwind-tables", "-funwind-tables", "a.c"]);
1852        assert!(opts.unwinds(), "the weaker request was dropped");
1853        let (opts, _) = compile(&["-c", "-fno-unwind-tables", "a.c"]);
1854        assert!(opts.unwinds(), "the weaker negative turned off the stronger request");
1855        let (opts, _) =
1856            compile(&["-c", "-fno-unwind-tables", "-fno-asynchronous-unwind-tables", "a.c"]);
1857        assert!(!opts.unwinds(), "both were turned off and one stayed on");
1858    }
1859
1860    #[test]
1861    fn the_flags_that_describe_what_this_compiler_already_does_are_taken() {
1862        // Every one of these is on a real build line somewhere and every one of them was an
1863        // unknown option. What they have in common is that the answer rucc gives is the answer
1864        // they ask for, so there is nothing to implement and nothing to refuse.
1865        for flag in [
1866            "-fno-common",
1867            "-fstrict-aliasing",
1868            "-fno-strict-aliasing",
1869            "-pipe",
1870            "-fdiagnostics-color",
1871            "-fno-diagnostics-color",
1872            "-fdiagnostics-color=always",
1873            "-fdiagnostics-color=never",
1874            "-fdiagnostics-color=auto",
1875        ] {
1876            let (opts, _) = compile(&["-c", flag, "a.c"]);
1877            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1878        }
1879    }
1880
1881    #[test]
1882    fn asking_the_linker_to_merge_tentative_definitions_is_told_why_it_is_not_coming() {
1883        // The one of that family that is a request rather than a description, and it is a real
1884        // difference: two files each writing `int g;` link under it and do not without it.
1885        let e = parse_args(&args(&["-fcommon", "a.c"])).unwrap_err();
1886        assert!(e.message.contains(".bss"), "{}", e.message);
1887        assert!(e.message.contains("extern"), "the way out is worth saying: {}", e.message);
1888    }
1889
1890    #[test]
1891    fn asking_for_position_dependent_code_is_told_why_it_is_not_coming() {
1892        for flag in ["-fno-pic", "-fno-pie"] {
1893            let e = parse_args(&args(&[flag, "a.c"])).unwrap_err();
1894            assert!(e.message.contains("global offset table"), "{flag}: {}", e.message);
1895            // The one it may have meant, since the two are a letter apart and one of them is
1896            // about linking and is taken.
1897            assert!(e.message.contains("-no-pie"), "{flag}: {}", e.message);
1898        }
1899    }
1900
1901    #[test]
1902    fn an_unsupported_target_names_itself() {
1903        let e = parse_args(&args(&["--target=sparc64-linux-gnu", "a.c"])).unwrap_err();
1904        assert!(e.message.contains("sparc64"), "{}", e.message);
1905    }
1906
1907    #[test]
1908    fn no_inputs_is_an_error_but_print_config_needs_none() {
1909        assert!(parse_args(&args(&[])).is_err());
1910        assert!(matches!(parse_args(&args(&["--print-config"])), Ok(Action::PrintConfig(_))));
1911    }
1912
1913    #[test]
1914    fn print_config_reports_the_target_it_was_given_not_the_host() {
1915        let a = parse_args(&args(&["--print-config", "--target=riscv64-linux-musl"])).unwrap();
1916        let Action::PrintConfig(opts) = a else { panic!("expected a configuration dump") };
1917        let text = print_config(&opts);
1918        assert!(text.contains("target: riscv64-unknown-linux-musl"), "{text}");
1919        assert!(text.contains("char-signed: false"), "{text}");
1920        assert!(text.contains("object-format: elf"), "{text}");
1921        assert!(text.contains("va-list: void-pointer"), "{text}");
1922        // RISC-V has a register file and this compiler has not written it down yet, and the
1923        // dump says which of those two it is rather than leaving the line out.
1924        assert!(text.contains("registers: none"), "{text}");
1925    }
1926
1927    #[test]
1928    fn print_config_has_one_key_per_line_and_a_fixed_order() {
1929        let opts = Options::new("x86_64-unknown-linux-gnu".parse().unwrap());
1930        let text = print_config(&opts);
1931        let keys: Vec<&str> =
1932            text.lines().map(|l| l.split(':').next().unwrap_or_default()).collect();
1933        assert_eq!(keys[0], "version");
1934        assert_eq!(keys[1], "target");
1935        assert_eq!(keys.len(), 19);
1936        assert!(text.ends_with('\n'));
1937    }
1938
1939    #[test]
1940    fn the_safety_tier_is_read_off_the_command_line_and_a_wrong_one_is_refused() {
1941        let (opts, _) = compile(&["a.c"]);
1942        assert_eq!(opts.safety, rucc_session::Safety::Off);
1943
1944        for (flag, tier) in [
1945            ("-fsafety=detect", rucc_session::Safety::Detect),
1946            ("-fsafety=enforce", rucc_session::Safety::Enforce),
1947            ("-fsafety=kernel", rucc_session::Safety::Kernel),
1948            ("-fsafety=off", rucc_session::Safety::Off),
1949        ] {
1950            let (opts, _) = compile(&[flag, "a.c"]);
1951            assert_eq!(opts.safety, tier, "{flag}");
1952        }
1953
1954        // The last one wins, the way every other repeated flag on this command line does.
1955        let (opts, _) = compile(&["-fsafety=enforce", "-fsafety=off", "a.c"]);
1956        assert_eq!(opts.safety, rucc_session::Safety::Off);
1957
1958        // A misspelled tier is refused rather than ignored. Silently compiling without the
1959        // monitor a build asked for is the one failure mode this feature cannot have.
1960        let e = parse_args(&args(&["-fsafety=on", "a.c"])).unwrap_err();
1961        assert!(e.message.contains("is not a safety tier"), "{}", e.message);
1962        assert!(parse_args(&args(&["-fsafety", "a.c"])).is_err());
1963    }
1964
1965    #[test]
1966    fn print_pipeline_answers_with_the_passes_the_level_asked_for() {
1967        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
1968        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1969        let text = print_pipeline(&opts);
1970        assert!(text.starts_with("level: -O2\n"), "{text}");
1971        assert!(text.contains("fold"), "{text}");
1972
1973        let a = parse_args(&args(&["--print-pipeline"])).unwrap();
1974        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1975        // One pass runs at `-O0` and it is the one that removes code nothing reaches, which is
1976        // not an optimization. See issue 359.
1977        assert!(print_pipeline(&opts).contains("1: simplify-cfg,"), "{}", print_pipeline(&opts));
1978
1979        let a = parse_args(&args(&["--print-pipeline", "-fno-simplify-cfg"])).unwrap();
1980        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1981        // And with that one turned off there is nothing left, which the dump says rather than
1982        // printing an empty list.
1983        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
1984    }
1985
1986    #[test]
1987    fn print_pipeline_takes_the_toggles_into_account() {
1988        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fno-fold"])).unwrap();
1989        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
1990        let text = print_pipeline(&opts);
1991        // The one that was named is gone and the rest of the level is not, which is the whole
1992        // of what a toggle promises.
1993        assert!(!text.contains("fold"), "{text}");
1994        assert!(text.contains("dce"), "{text}");
1995
1996        // Every pass the compiler has, named off. Built from the registry rather than written
1997        // out, so a pass added later is turned off here too and this keeps testing the thing it
1998        // is about, which is that the toggles can empty a level.
1999        let mut off = vec!["--print-pipeline".to_owned(), "-O2".to_owned()];
2000        off.extend(rucc_opt::PASSES.iter().map(|p| format!("-fno-{}", p.name())));
2001        let spelled: Vec<&str> = off.iter().map(String::as_str).collect();
2002        let a = parse_args(&args(&spelled)).unwrap();
2003        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2004        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2005    }
2006
2007    #[test]
2008    fn print_pipeline_says_when_a_budget_will_stop_the_run_short() {
2009        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2010        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2011        assert!(!print_pipeline(&opts).contains("global fuel"));
2012
2013        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fpass-fuel-global=4"])).unwrap();
2014        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2015        let text = print_pipeline(&opts);
2016        // Because the listing is the answer to what this compilation will do, and a run that
2017        // stops after four rewrites is not doing what the level says it does.
2018        assert!(text.contains("global fuel: 4"), "{text}");
2019    }
2020
2021    /// A pass is turned on and off by its own name, and the order the flags were given in is
2022    /// kept, because the last spelling of a name is the one that decides.
2023    #[test]
2024    fn a_pass_is_named_by_dash_f_and_unnamed_by_dash_f_no() {
2025        let (opts, _) = compile(&["-c", "-O0", "-ffold", "-fno-fold", "-ffold", "a.c"]);
2026        assert_eq!(
2027            opts.passes,
2028            [("fold".to_owned(), true), ("fold".to_owned(), false), ("fold".to_owned(), true)]
2029        );
2030
2031        let e = parse_args(&args(&["-fno-such-pass", "a.c"])).unwrap_err();
2032        assert!(e.message.contains("unknown option"), "{}", e.message);
2033    }
2034
2035    #[test]
2036    fn pass_fuel_names_a_pass_and_a_count_and_refuses_anything_else() {
2037        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel=fold=3", "a.c"]);
2038        assert_eq!(opts.pass_fuel, [("fold".to_owned(), 3)]);
2039
2040        let e = parse_args(&args(&["-fpass-fuel=fold", "a.c"])).unwrap_err();
2041        assert!(e.message.contains("<pass>=<count>"), "{}", e.message);
2042        let e = parse_args(&args(&["-fpass-fuel=nosuch=3", "a.c"])).unwrap_err();
2043        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2044        let e = parse_args(&args(&["-fpass-fuel=fold=lots", "a.c"])).unwrap_err();
2045        assert!(e.message.contains("not a number"), "{}", e.message);
2046    }
2047
2048    #[test]
2049    fn global_pass_fuel_is_a_count_on_its_own_and_defaults_to_no_limit() {
2050        let (opts, _) = compile(&["-c", "-O2", "a.c"]);
2051        assert_eq!(opts.pass_fuel_global, None);
2052
2053        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel-global=12", "a.c"]);
2054        assert_eq!(opts.pass_fuel_global, Some(12));
2055        // And it is not the per pass flag with a longer name, so neither spelling swallows the
2056        // other.
2057        assert!(opts.pass_fuel.is_empty());
2058
2059        let e = parse_args(&args(&["-fpass-fuel-global=lots", "a.c"])).unwrap_err();
2060        assert!(e.message.contains("not a number"), "{}", e.message);
2061    }
2062
2063    #[test]
2064    fn a_gate_names_a_pass_and_optionally_the_functions_it_covers() {
2065        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold", "-fenable-fold=2-4,main", "a.c"]);
2066        assert_eq!(
2067            opts.pass_gates,
2068            [(false, "fold".to_owned()), (true, "fold=2-4,main".to_owned())],
2069            "the order is what decides, so it has to survive the parse"
2070        );
2071
2072        let e = parse_args(&args(&["-fdisable-nosuch", "a.c"])).unwrap_err();
2073        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2074        let e = parse_args(&args(&["-fenable-fold=9-2", "a.c"])).unwrap_err();
2075        assert!(e.message.contains("ends before it starts"), "{}", e.message);
2076        let e = parse_args(&args(&["-fdisable-fold=", "a.c"])).unwrap_err();
2077        assert!(e.message.contains("is empty"), "{}", e.message);
2078    }
2079
2080    #[test]
2081    fn the_pipeline_listing_says_which_passes_a_gate_touched() {
2082        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold=main", "a.c"]);
2083        let text = print_pipeline(&opts);
2084        assert!(text.contains("fold, "), "{text}");
2085        assert!(text.contains("[off for main]"), "{text}");
2086    }
2087
2088    /// The spelling is checked while the arguments are read, because a dump that names a pass
2089    /// this compiler does not have is a typo, and a typo found after the compilation has run is
2090    /// found too late to be any use.
2091    #[test]
2092    fn a_dump_is_checked_when_it_is_asked_for_rather_than_when_it_is_taken() {
2093        let (opts, _) = compile(&["-c", "-O2", "-fdump-ir=all", "-fdump-ir=after-fold", "a.c"]);
2094        assert_eq!(opts.dump_ir, ["all", "after-fold"]);
2095
2096        let e = parse_args(&args(&["-fdump-ir=after-nosuch", "a.c"])).unwrap_err();
2097        assert!(e.message.contains("nosuch"), "{}", e.message);
2098        assert!(parse_args(&args(&["-fdump-ir=sideways-fold", "a.c"])).is_err());
2099    }
2100
2101    /// Every spelling `-fopt-info` takes, and the one it does not.
2102    ///
2103    /// The keywords are checked here for the same reason a dump's pass name is: a person who
2104    /// misspelled one gets no output, and no output is also what a compilation where nothing
2105    /// happened looks like. Telling those two apart is the entire reason to reach for this flag.
2106    #[test]
2107    fn opt_info_takes_kinds_and_a_file_and_refuses_a_kind_it_does_not_have() {
2108        let (opts, _) = compile(&["-c", "-O2", "-fopt-info", "a.c"]);
2109        assert_eq!(opts.opt_info, [""], "a bare flag asks for the rewrites");
2110        assert_eq!(opts.opt_info_file, None, "and goes to standard error");
2111
2112        let (opts, _) = compile(&["-c", "-O2", "-fopt-info-missed-note", "a.c"]);
2113        assert_eq!(opts.opt_info, ["missed-note"]);
2114
2115        // Two flags add up rather than the second replacing the first, and the file is the last
2116        // one that named a file, which is how GCC treats both.
2117        let (opts, _) =
2118            compile(&["-c", "-O2", "-fopt-info-missed=one.txt", "-fopt-info-all=two.txt", "a.c"]);
2119        assert_eq!(opts.opt_info, ["missed", "all"]);
2120        assert_eq!(opts.opt_info_file.as_deref(), Some("two.txt"));
2121
2122        let e = parse_args(&args(&["-fopt-info-vectorized", "a.c"])).unwrap_err();
2123        assert!(e.message.contains("vectorized"), "{}", e.message);
2124        assert!(e.message.contains("`missed`"), "{}", e.message);
2125        let e = parse_args(&args(&["-fopt-info-missed=", "a.c"])).unwrap_err();
2126        assert!(e.message.contains("no file"), "{}", e.message);
2127    }
2128
2129    #[test]
2130    fn verify_each_is_unstable_and_off_unless_it_was_asked_for() {
2131        let (opts, _) = compile(&["-c", "-Zverify-each", "a.c"]);
2132        assert!(opts.verify_each);
2133        assert!(!USAGE.contains("verify-each"), "an unstable option stays out of the usage text");
2134    }
2135
2136    #[test]
2137    fn dash_o_needs_an_argument() {
2138        let e = parse_args(&args(&["a.c", "-o"])).unwrap_err();
2139        assert_eq!(e.message, "-o requires an argument");
2140    }
2141
2142    #[test]
2143    fn dash_d_and_dash_u_are_read_joined_or_separated_and_keep_their_order() {
2144        let (opts, _) = compile(&["-DFOO=1", "-D", "BAR", "-UBAZ", "-U", "QUX", "a.c"]);
2145        assert_eq!(opts.defines, ["FOO=1", "BAR"]);
2146        assert_eq!(opts.undefines, ["BAZ", "QUX"]);
2147    }
2148
2149    #[test]
2150    fn the_include_flags_land_on_the_chain_each_one_names() {
2151        // A sysroot with nothing under it, so that the library's own directories are the
2152        // same on every machine this test runs on, which is none of them.
2153        let (opts, _) = compile(&[
2154            "-Ii",
2155            "-iquote",
2156            "q",
2157            "-isystem",
2158            "sys",
2159            "-idirafter",
2160            "after",
2161            "--sysroot=/nowhere-at-all",
2162            "a.c",
2163        ]);
2164        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2165        // The compiler's own headers sit after every `-isystem` and before `-idirafter`,
2166        // which is where GCC puts its own: a directory the user named outranks ours.
2167        assert_eq!(dirs, ["q", "i", "sys", runtime::DIR, "after"]);
2168        assert!(!opts.search.dirs()[1].is_system);
2169        assert!(opts.search.dirs()[2].is_system);
2170    }
2171
2172    #[test]
2173    fn the_librarys_headers_come_after_the_compilers_own_and_go_away_with_them() {
2174        // Which machine this runs on decides what is on the path, so the test is about the
2175        // order rather than about the names: ours is on it, the library's follow it, and
2176        // `-nostdinc` is the one flag that takes both halves of the pair off at once.
2177        let (opts, _) = compile(&["a.c"]);
2178        let dirs = opts.search.dirs();
2179        let ours = dirs.iter().position(|d| d.path.to_str() == Some(runtime::DIR));
2180        assert_eq!(ours, Some(0), "{dirs:?}");
2181        assert!(dirs[1..].iter().all(|d| d.is_system), "{dirs:?}");
2182        let (bare, _) = compile(&["-nostdinc", "a.c"]);
2183        assert!(bare.search.dirs().is_empty(), "{:?}", bare.search.dirs());
2184    }
2185
2186    #[test]
2187    fn a_sysroot_moves_the_librarys_directories_and_nothing_else() {
2188        let (opts, _) = compile(&["-isystem", "sys", "--sysroot=/nowhere-at-all", "a.c"]);
2189        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2190        assert_eq!(dirs, ["sys", runtime::DIR]);
2191    }
2192
2193    #[test]
2194    fn dash_i_dash_moves_the_bracket_directories_into_the_quoted_chain() {
2195        let (opts, _) =
2196            compile(&["-Iinc1", "-iquote", "inc2", "-I-", "-Iinc3", "-nostdinc", "a.c"]);
2197        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2198        assert_eq!(dirs, ["inc1", "inc2", "inc3"]);
2199        // An angled include sees only what came after the flag.
2200        assert_eq!(opts.search.start(IncludeForm::Angled), 2);
2201        assert!(!opts.search.searches_current_dir());
2202    }
2203
2204    #[test]
2205    fn the_prefix_flags_stick_what_iprefix_said_on_the_front_of_what_follows_it() {
2206        let (opts, _) = compile(&[
2207            "-iprefix",
2208            "/tools/",
2209            "-iwithprefix",
2210            "late",
2211            "-iwithprefixbefore",
2212            "early",
2213            "-iprefix",
2214            "/other/",
2215            "-iwithprefix",
2216            "last",
2217            "-nostdinc",
2218            "a.c",
2219        ]);
2220        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2221        // `-iwithprefixbefore` is an `-I` and the other two are `-isystem`, which is where GCC
2222        // puts them rather than where its manual says it does.
2223        assert_eq!(dirs, ["/tools/early", "/tools/late", "/other/last"]);
2224        assert!(!opts.search.dirs()[0].is_system);
2225        assert!(opts.search.dirs()[1].is_system);
2226    }
2227
2228    #[test]
2229    fn the_files_named_on_the_command_line_keep_their_order_and_which_flag_named_them() {
2230        let (opts, _) =
2231            compile(&["-include", "one.h", "-imacros", "two.h", "-include", "3.h", "a.c"]);
2232        let names: Vec<&str> = opts.preincludes.iter().map(|p| p.name.as_str()).collect();
2233        assert_eq!(names, ["one.h", "two.h", "3.h"]);
2234        assert_eq!(opts.preincludes.iter().filter(|p| p.macros_only).count(), 1);
2235    }
2236
2237    #[test]
2238    fn nostdinc_takes_the_compilers_own_headers_off_the_path() {
2239        let (opts, _) = compile(&["-Ii", "-nostdinc", "a.c"]);
2240        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2241        assert_eq!(dirs, ["i"]);
2242    }
2243
2244    #[test]
2245    fn the_dialect_flags_set_the_language_and_the_extensions_separately() {
2246        let (opts, _) = compile(&["-std=gnu11", "a.c"]);
2247        assert_eq!(opts.std, Std::C11);
2248        assert!(opts.gnu_extensions);
2249
2250        let (opts, _) = compile(&["-std=iso9899:1999", "a.c"]);
2251        assert_eq!(opts.std, Std::C99);
2252        assert!(!opts.gnu_extensions);
2253
2254        let (opts, _) = compile(&["-ansi", "a.c"]);
2255        assert_eq!(opts.std, Std::C89);
2256        assert!(!opts.gnu_extensions);
2257
2258        let e = parse_args(&args(&["-std=c94jr", "a.c"])).unwrap_err();
2259        assert!(e.message.contains("unknown dialect"), "{}", e.message);
2260    }
2261
2262    #[test]
2263    fn the_dump_letters_are_a_family_and_everything_else_beginning_with_d_is_not() {
2264        let (opts, _) = compile(&["-dM", "a.c"]);
2265        assert!(opts.dumps.macros);
2266
2267        // Packed, the way GCC takes them, and a letter in the family we have not written yet
2268        // is accepted and does nothing rather than failing a build.
2269        let (opts, _) = compile(&["-dDM", "a.c"]);
2270        assert!(opts.dumps.macros);
2271        let (opts, _) = compile(&["-dD", "a.c"]);
2272        assert!(!opts.dumps.macros);
2273
2274        let (opts, _) = compile(&["a.c"]);
2275        assert!(!opts.dumps.any());
2276
2277        // `-dumpversion` is a different flag that happens to start the same way, and it is read
2278        // as itself rather than as a dump of nothing.
2279        assert_eq!(printed(&["-dumpversion", "a.c"]), VERSION);
2280    }
2281
2282    #[test]
2283    fn the_gcc_version_claimed_is_a_flag_and_the_short_spellings_are_the_ones_people_write() {
2284        let (opts, _) = compile(&["a.c"]);
2285        assert_eq!(
2286            opts.gnuc,
2287            GnucVersion { major: 7, minor: 0, patch: 0 },
2288            "the lowest claim a modern glibc gives its own declarations to"
2289        );
2290
2291        let (opts, _) = compile(&["-fgnuc-version=15.1.0", "a.c"]);
2292        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 1, patch: 0 });
2293
2294        // A missing component is zero. `gcc -dumpversion` says `15` on a release with no
2295        // patchlevel and a harness that pastes that back has to be understood.
2296        let (opts, _) = compile(&["-fgnuc-version=15", "a.c"]);
2297        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 0, patch: 0 });
2298
2299        let (opts, _) = compile(&["-fgnuc-version=13.2", "a.c"]);
2300        assert_eq!(opts.gnuc, GnucVersion { major: 13, minor: 2, patch: 0 });
2301
2302        let e = parse_args(&args(&["-fgnuc-version=15.x", "a.c"])).unwrap_err();
2303        assert!(e.message.contains("minor that is not a number"), "{}", e.message);
2304
2305        let e = parse_args(&args(&["-fgnuc-version=1.2.3.4", "a.c"])).unwrap_err();
2306        assert!(e.message.contains("more than three"), "{}", e.message);
2307    }
2308
2309    #[test]
2310    fn pedantic_has_two_spellings_and_is_not_the_same_knob_as_the_dialect() {
2311        let (opts, _) = compile(&["-std=c17", "-pedantic", "a.c"]);
2312        assert!(opts.pedantic);
2313        assert_eq!(opts.std, Std::C17);
2314
2315        // The `-W` family's name for it, which is what a build that groups its warning flags
2316        // tends to write.
2317        let (opts, _) = compile(&["-Wpedantic", "a.c"]);
2318        assert!(opts.pedantic);
2319
2320        let (opts, _) = compile(&["-std=c17", "a.c"]);
2321        assert!(!opts.pedantic, "a dialect on its own does not diagnose an extension");
2322    }
2323
2324    #[test]
2325    fn dash_p_and_dash_ffreestanding_reach_the_options() {
2326        let (opts, _) = compile(&["-E", "-P", "-ffreestanding", "a.c"]);
2327        assert!(!opts.line_markers);
2328        assert!(!opts.hosted);
2329        assert_eq!(opts.emit, EmitKind::Preprocessed);
2330    }
2331
2332    /// The two ways a build says it means its own function by a name the C library also has.
2333    ///
2334    /// `-fno-builtin` is all of them and `-fno-builtin-<name>` is one, and the second is what a
2335    /// build writes when it means its own `memcpy` and the library's everything else. The name is
2336    /// kept as it was written and not checked against anything, because a program is allowed to
2337    /// mean something by a name this compiler has never heard of.
2338    #[test]
2339    fn the_builtin_flags_are_read_in_both_directions_and_one_name_at_a_time() {
2340        let (opts, _) = compile(&["-c", "a.c"]);
2341        assert!(opts.builtins, "a library name means the library function by default");
2342        assert!(opts.no_builtin.is_empty());
2343
2344        let (opts, _) = compile(&["-c", "-fno-builtin", "a.c"]);
2345        assert!(!opts.builtins);
2346
2347        let (opts, _) = compile(&["-c", "-fno-builtin", "-fbuiltin", "a.c"]);
2348        assert!(opts.builtins, "the last mention decides");
2349
2350        let (opts, _) = compile(&["-c", "-fno-builtin-memcpy", "-fno-builtin-nonesuch", "a.c"]);
2351        assert!(opts.builtins, "one name is not the family");
2352        assert_eq!(opts.no_builtin, vec!["memcpy".to_owned(), "nonesuch".to_owned()]);
2353    }
2354
2355    /// `-fvisibility=`, which is on every cmake project that cares about which names it exports
2356    /// and which was refused as an unknown option until now.
2357    ///
2358    /// Four spellings and three answers. `internal` is hidden plus a promise about never taking
2359    /// the address across a component boundary, and nothing derives anything from that promise
2360    /// here, so it comes out as the weaker of the two rather than as a refusal that stops a build
2361    /// over a distinction this compiler does not make.
2362    #[test]
2363    fn visibility_takes_the_four_spellings_gcc_takes_and_refuses_the_rest() {
2364        let (opts, _) = compile(&["-c", "a.c"]);
2365        assert_eq!(opts.visibility, Visibility::Default, "exported unless something says not");
2366
2367        for (written, wanted) in [
2368            ("default", Visibility::Default),
2369            ("hidden", Visibility::Hidden),
2370            ("internal", Visibility::Hidden),
2371            ("protected", Visibility::Protected),
2372        ] {
2373            let (opts, _) = compile(&["-c", &format!("-fvisibility={written}"), "a.c"]);
2374            assert_eq!(opts.visibility, wanted, "{written}");
2375        }
2376
2377        // The last mention decides, which is what every other flag of this shape does and what a
2378        // build that turns something off for one directory relies on.
2379        let (opts, _) = compile(&["-c", "-fvisibility=hidden", "-fvisibility=default", "a.c"]);
2380        assert_eq!(opts.visibility, Visibility::Default, "the last mention decides");
2381
2382        // A spelling gcc does not take is refused rather than read as the default, because a
2383        // build that meant hidden and got exported is a library with the wrong interface and
2384        // nothing said about it anywhere.
2385        let failed = parse_args(&args(&["-fvisibility=none", "a.c"])).expect_err("refused");
2386        assert!(failed.to_string().contains("is not a visibility"), "{failed}");
2387    }
2388
2389    /// `-ffunction-sections` and `-fdata-sections`, which are what make `--gc-sections` able to
2390    /// drop anything: a linker can leave out a section nothing reaches and cannot leave out half of
2391    /// one. A kernel and an embedded image are both linked that way.
2392    ///
2393    /// Two flags rather than one because gcc has two, and a build that asks for one of them and not
2394    /// the other is a build that measured something: splitting the code is nearly free at link time
2395    /// and splitting the data can defeat the linker's ordering of what is next to what.
2396    #[test]
2397    fn a_section_per_function_and_a_section_per_variable_are_asked_for_one_at_a_time() {
2398        let (opts, _) = compile(&["-c", "a.c"]);
2399        assert!(!opts.function_sections, "one text section unless something says otherwise");
2400        assert!(!opts.data_sections);
2401
2402        let (opts, _) = compile(&["-c", "-ffunction-sections", "a.c"]);
2403        assert!(opts.function_sections);
2404        assert!(!opts.data_sections, "one flag is not the other");
2405
2406        let (opts, _) = compile(&["-c", "-fdata-sections", "a.c"]);
2407        assert!(opts.data_sections);
2408        assert!(!opts.function_sections);
2409
2410        // Both directions taken, and the off one is what happens anyway rather than a refusal,
2411        // since a build that writes it is asking for the default.
2412        let (opts, _) = compile(&[
2413            "-c",
2414            "-ffunction-sections",
2415            "-fno-function-sections",
2416            "-fdata-sections",
2417            "-fno-data-sections",
2418            "a.c",
2419        ]);
2420        assert!(!opts.function_sections, "the last mention decides");
2421        assert!(!opts.data_sections, "the last mention decides");
2422    }
2423
2424    /// `-fgnu89-inline`, which is off by default and is not implied by anything on the command
2425    /// line, since the dialect asks for GNU's reading further in rather than through this.
2426    #[test]
2427    fn gnu89_inline_is_off_until_it_is_asked_for_and_the_last_mention_decides() {
2428        let (opts, _) = compile(&["-c", "a.c"]);
2429        assert!(!opts.gnu89_inline, "C's reading of inline by default");
2430
2431        let (opts, _) = compile(&["-c", "-fgnu89-inline", "a.c"]);
2432        assert!(opts.gnu89_inline);
2433
2434        let (opts, _) = compile(&["-c", "-fgnu89-inline", "-fno-gnu89-inline", "a.c"]);
2435        assert!(!opts.gnu89_inline, "the last mention decides");
2436
2437        // The C89 dialects are under GNU's reading whether this was written or not, so the flag
2438        // stays off there and the dialect is what the checker and the macro set both ask. That is
2439        // also why `-std=c89 -fno-gnu89-inline` needs no diagnostic: it asks for the reading the
2440        // dialect already has. gcc refuses that command line, which is measured in the issue.
2441        let (opts, _) = compile(&["-c", "-std=c89", "a.c"]);
2442        assert!(!opts.gnu89_inline);
2443    }
2444
2445    /// Both spellings of both frame flags, since a build that wants one usually writes the
2446    /// other beside it for the one file that has to be compiled the ordinary way.
2447    #[test]
2448    fn the_two_frame_flags_are_read_in_both_directions() {
2449        let (opts, _) = compile(&["-c", "a.c"]);
2450        assert!(!opts.frame_pointer, "gcc omits it above -O0 and so does this");
2451        assert!(opts.red_zone, "the psABI has one and nothing said not to use it");
2452
2453        let (opts, _) = compile(&["-c", "-fno-omit-frame-pointer", "-mno-red-zone", "a.c"]);
2454        assert!(opts.frame_pointer);
2455        assert!(!opts.red_zone);
2456
2457        let (opts, _) = compile(&[
2458            "-c",
2459            "-fno-omit-frame-pointer",
2460            "-fomit-frame-pointer",
2461            "-mno-red-zone",
2462            "-mred-zone",
2463            "a.c",
2464        ]);
2465        assert!(!opts.frame_pointer, "the last one wins, as it does in gcc");
2466        assert!(opts.red_zone);
2467    }
2468
2469    #[test]
2470    fn the_link_flags_are_collected_apart_from_the_compilation() {
2471        let (link, _) = linking(&[
2472            "-static",
2473            "-nostartfiles",
2474            "-rdynamic",
2475            "-s",
2476            "-fuse-ld=mold",
2477            "-L/opt/lib",
2478            "-B",
2479            "/opt/tools",
2480            "a.c",
2481        ]);
2482        assert!(link.is_static);
2483        assert!(link.no_startfiles);
2484        assert!(link.export_dynamic);
2485        assert!(link.strip);
2486        assert_eq!(link.use_ld.as_deref(), Some("mold"));
2487        assert_eq!(link.search, vec![PathBuf::from("/opt/lib")]);
2488        assert_eq!(link.prefixes, vec![PathBuf::from("/opt/tools")]);
2489    }
2490
2491    #[test]
2492    fn a_comma_in_dash_wl_separates_two_arguments() {
2493        let (link, _) = linking(&["-Wl,-rpath,/opt/lib", "-Xlinker", "--as-needed", "a.c"]);
2494        assert_eq!(link.passthrough, vec!["-rpath", "/opt/lib", "--as-needed"]);
2495    }
2496
2497    #[test]
2498    fn a_library_keeps_its_place_between_the_objects() {
2499        // Link order is semantic: `-lm` written between two files resolves for the one before
2500        // it and not for the one after, so a library cannot be collected into a list of its own.
2501        // The target is named because the suffix of an object is the target's and this asserts
2502        // on the names: the same command line on a Windows host plans two `.obj` files.
2503        let (_, plan) = linking(&["--target=x86_64-unknown-linux-gnu", "a.c", "-lm", "b.c"]);
2504        let link = plan.link.expect("expected a link step");
2505        assert_eq!(
2506            link.inputs,
2507            vec![
2508                link::Item::File("a.o".into()),
2509                link::Item::Library("m".into()),
2510                link::Item::File("b.o".into()),
2511            ]
2512        );
2513        // And it is not a job, because there is nothing to compile in a library.
2514        assert_eq!(plan.jobs.len(), 2);
2515    }
2516
2517    #[test]
2518    fn a_library_on_a_dash_c_line_is_a_note_rather_than_an_error() {
2519        let (_, plan) = linking(&["-c", "-lm", "a.c"]);
2520        assert!(plan.link.is_none());
2521        assert!(plan.notes.iter().any(|n| n.contains("-lm")), "{:?}", plan.notes);
2522    }
2523
2524    #[test]
2525    fn the_sysroot_reaches_the_linker_as_well_as_the_headers() {
2526        let (link, _) = linking(&["--sysroot=/opt/root", "a.c"]);
2527        assert_eq!(link.sysroot, Some(PathBuf::from("/opt/root")));
2528    }
2529
2530    fn printed(s: &[&str]) -> String {
2531        match parse_args(&args(s)).expect("expected an answer") {
2532            Action::Print(line) => line,
2533            other => panic!("expected an answer, got {other:?}"),
2534        }
2535    }
2536
2537    fn refused(s: &[&str]) -> String {
2538        parse_args(&args(s)).expect_err("expected a refusal").message
2539    }
2540
2541    #[test]
2542    fn a_warning_flag_this_compiler_has_not_heard_of_is_taken_rather_than_refused() {
2543        // The rule in section 4.1, and the reason for it is autoconf: a configure script finds
2544        // out whether a warning flag exists by passing it and looking at the exit status, so a
2545        // compiler that refuses one it does not know fails a script written for a newer GCC.
2546        let (opts, _) = compile(&["-Wall", "-Wextra", "-Wno-format-truncation", "-c", "a.c"]);
2547        assert!(!opts.warnings_are_errors);
2548        assert!(opts.warnings);
2549        // The two spellings that do mean something are still read.
2550        let (opts, _) = compile(&["-Werror", "-c", "a.c"]);
2551        assert!(opts.warnings_are_errors);
2552        let (opts, _) = compile(&["-w", "-c", "a.c"]);
2553        assert!(!opts.warnings);
2554        let (opts, _) = compile(&["-pedantic-errors", "-c", "a.c"]);
2555        assert!(opts.pedantic && opts.warnings_are_errors);
2556    }
2557
2558    #[test]
2559    fn an_argument_for_a_separate_tool_is_refused_rather_than_dropped() {
2560        // Every one of these says something about the output, so the wrong answer is silence.
2561        assert!(refused(&["-Wa,--noexecstack", "-c", "a.c"]).contains("separate assembler"));
2562        assert!(refused(&["-Wp,-DX", "-c", "a.c"]).contains("separate assembler"));
2563        assert!(refused(&["-specs=/x", "a.c"]).contains("-specs= is not supported"));
2564        assert!(refused(&["-mcmodel=kernel", "-c", "a.c"]).contains("small code model"));
2565        assert!(refused(&["-gdwarf-4", "-c", "a.c"]).contains("DWARF 5"));
2566        assert!(refused(&["-Ofast", "-c", "a.c"]).contains("fast math"));
2567        // The word size the target does not have, which is a target this compiler was not asked
2568        // for rather than a flag it does not know.
2569        let no32 = refused(&["--target=x86_64-unknown-linux-gnu", "-m32", "-c", "a.c"]);
2570        assert!(no32.contains("32 bit target"), "{no32}");
2571    }
2572
2573    #[test]
2574    fn the_levels_gcc_spells_differently_are_the_levels_they_mean() {
2575        assert_eq!(compile(&["-O", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2576        assert_eq!(compile(&["-Og", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2577        assert_eq!(compile(&["-O2", "-c", "a.c"]).0.opt_level, OptLevel::O2);
2578    }
2579
2580    #[test]
2581    fn the_machine_flags_that_name_what_we_already_do_are_taken_and_the_rest_are_not() {
2582        let line = ["--target=x86_64-unknown-linux-gnu", "-m64", "-march=x86-64-v3"];
2583        let (opts, _) =
2584            compile(&[&line[..], &["-mtune=native", "-mabi=sysv", "-c", "a.c"]].concat());
2585        assert_eq!(opts.target.to_string(), "x86_64-unknown-linux-gnu");
2586        let wrong = refused(&["--target=x86_64-unknown-linux-gnu", "-mabi=ms", "-c", "a.c"]);
2587        assert!(wrong.contains("sysv convention"), "{wrong}");
2588    }
2589
2590    #[test]
2591    fn the_thread_flag_is_a_macro_and_a_library_and_the_library_goes_last() {
2592        let (opts, plan) = compile(&["-pthread", "-c", "a.c"]);
2593        assert!(opts.defines.iter().any(|d| d == "_REENTRANT"));
2594        // After the input, because a static link takes what it needs from a library when it
2595        // reaches it and not afterwards.
2596        let names: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
2597        assert_eq!(names, vec!["a.c"]);
2598    }
2599
2600    #[test]
2601    fn the_questions_a_build_system_asks_before_it_compiles_anything() {
2602        let target = "--target=x86_64-unknown-linux-gnu";
2603        assert_eq!(printed(&[target, "-dumpmachine"]), "x86_64-unknown-linux-gnu");
2604        assert_eq!(printed(&[target, "-dumpversion"]), VERSION);
2605        assert_eq!(printed(&[target, "-dumpfullversion"]), VERSION);
2606        assert_eq!(printed(&[target, "-print-multiarch"]), "x86_64-linux-gnu");
2607        // A name nothing holds comes back unchanged, which is GCC's rule and is what makes the
2608        // answer safe to paste into a link line whether or not the file is there.
2609        assert_eq!(printed(&[target, "-print-file-name=no-such-library.a"]), "no-such-library.a");
2610        assert_eq!(printed(&[target, "-print-prog-name=ld"]), "ld");
2611        let dirs = printed(&[target, "-print-search-dirs"]);
2612        assert!(dirs.starts_with("install: "), "{dirs}");
2613        assert!(dirs.contains("\nlibraries: ="), "{dirs}");
2614    }
2615
2616    #[test]
2617    fn the_two_dependency_flags_that_stop_after_the_rule_stop_after_the_rule() {
2618        let (opts, _) = compile(&["-M", "a.c"]);
2619        assert!(opts.deps.emit && opts.deps.instead_of_compiling);
2620        assert!(opts.deps.system_headers, "plain -M lists them");
2621        assert_eq!(opts.emit, EmitKind::Preprocessed);
2622
2623        // Even where a later flag asked for something else, because the family is a mode and
2624        // the mode is what the run is for.
2625        let (opts, _) = compile(&["-M", "-c", "a.c"]);
2626        assert_eq!(opts.emit, EmitKind::Preprocessed);
2627
2628        let (opts, _) = compile(&["-MM", "a.c"]);
2629        assert!(!opts.deps.system_headers);
2630    }
2631
2632    #[test]
2633    fn the_two_that_end_in_d_leave_the_compilation_alone() {
2634        let (opts, _) = compile(&["-MD", "-c", "a.c"]);
2635        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2636        assert!(opts.deps.system_headers);
2637        assert_eq!(opts.emit, EmitKind::Object);
2638
2639        let (opts, _) = compile(&["-MMD", "-c", "a.c"]);
2640        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2641        assert!(!opts.deps.system_headers);
2642    }
2643
2644    #[test]
2645    fn nothing_puts_the_system_headers_back_once_a_flag_has_taken_them_out() {
2646        // GCC's rule, and not an oversight in it. The flag asking for fewer of them is read as
2647        // the answer, because the other one never asked the question.
2648        let (opts, _) = compile(&["-MM", "-M", "a.c"]);
2649        assert!(!opts.deps.system_headers);
2650        let (opts, _) = compile(&["-MD", "-MMD", "-c", "a.c"]);
2651        assert!(!opts.deps.system_headers);
2652        let (opts, _) = compile(&["-MMD", "-MD", "-c", "a.c"]);
2653        assert!(!opts.deps.system_headers);
2654    }
2655
2656    #[test]
2657    fn a_target_arrives_escaped_from_one_flag_and_untouched_from_the_other() {
2658        let (opts, _) = compile(&["-MM", "-MT", "a b.o", "-MQ", "a b.o", "a.c"]);
2659        assert_eq!(opts.deps.targets, vec!["a b.o".to_owned(), "a\\ b.o".to_owned()]);
2660    }
2661
2662    #[test]
2663    fn the_rest_of_the_family_is_a_file_and_a_switch() {
2664        let (opts, _) = compile(&["-MM", "-MF", "dep.d", "-MP", "a.c"]);
2665        assert_eq!(opts.deps.file.as_deref(), Some("dep.d"));
2666        assert!(opts.deps.phony);
2667
2668        for flag in ["-MF", "-MT", "-MQ"] {
2669            let e = parse_args(&args(&[flag])).unwrap_err();
2670            assert!(e.message.contains("requires an argument"), "{}", e.message);
2671        }
2672    }
2673
2674    /// A directory of sources for one test, removed when the test is done with it.
2675    struct TempTree(PathBuf);
2676
2677    impl Drop for TempTree {
2678        fn drop(&mut self) {
2679            let _ = std::fs::remove_dir_all(&self.0);
2680        }
2681    }
2682
2683    impl TempTree {
2684        fn new(name: &str, files: &[(&str, &str)]) -> TempTree {
2685            let dir = std::env::temp_dir().join(format!("rucc-deps-{}-{name}", std::process::id()));
2686            let _ = std::fs::remove_dir_all(&dir);
2687            std::fs::create_dir_all(&dir).expect("temporary directory should be writable");
2688            for (path, text) in files {
2689                let at = dir.join(path);
2690                if let Some(parent) = at.parent() {
2691                    std::fs::create_dir_all(parent).expect("creating a subdirectory should work");
2692                }
2693                std::fs::write(&at, text).expect("writing a temporary file should work");
2694            }
2695            TempTree(dir)
2696        }
2697
2698        fn path(&self, name: &str) -> String {
2699            self.0.join(name).to_string_lossy().into_owned()
2700        }
2701    }
2702
2703    #[test]
2704    fn the_rule_names_what_the_includes_found_and_names_each_of_them_once() {
2705        // End to end, because the list comes from the preprocessor and the format comes from
2706        // somewhere else, and a test of either half on its own would pass with the two of them
2707        // wired up backwards.
2708        let tree = TempTree::new(
2709            "found",
2710            &[
2711                ("a.c", "#include \"one.h\"\n#include \"two.h\"\nint main(void) { return X; }\n"),
2712                ("one.h", "#define X 0\n"),
2713                ("two.h", "#include \"one.h\"\n"),
2714            ],
2715        );
2716        let out = tree.path("dep.d");
2717        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2718        assert_eq!(code, 0);
2719
2720        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2721        let names: Vec<&str> = text.split_whitespace().collect();
2722        // The target, the source, and each header once however many times it was reached.
2723        assert_eq!(names.first(), Some(&"a.o:"), "{text}");
2724        assert_eq!(names.iter().filter(|n| n.ends_with("one.h")).count(), 1, "{text}");
2725        assert_eq!(names.iter().filter(|n| n.ends_with("two.h")).count(), 1, "{text}");
2726        // And the `-o` went to the file the rule replaced, which is left empty rather than
2727        // absent because a makefile that named it as a target will look for it.
2728        assert_eq!(std::fs::read(tree.path("a.i")).expect("the output should exist"), b"");
2729    }
2730
2731    #[test]
2732    fn a_header_that_is_only_reached_under_a_guard_is_still_a_dependency() {
2733        // The multiple-include optimization means the second reach never opens the file. It is
2734        // still a file this translation unit was built from, so it is still in the rule.
2735        let tree = TempTree::new(
2736            "guarded",
2737            &[
2738                ("a.c", "#include \"g.h\"\n#include \"g.h\"\nint main(void) { return 0; }\n"),
2739                ("g.h", "#ifndef G\n#define G\n#endif\n"),
2740            ],
2741        );
2742        let out = tree.path("dep.d");
2743        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2744        assert_eq!(code, 0);
2745        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2746        assert_eq!(text.split_whitespace().filter(|n| n.ends_with("g.h")).count(), 1, "{text}");
2747    }
2748
2749    #[test]
2750    fn every_imacros_file_is_read_before_every_include_file_whatever_order_they_were_written() {
2751        // Measured against GCC rather than read: the two flags the other way round produce the
2752        // same output byte for byte, so the command line order between the two families does not
2753        // decide anything and the order within one does. The `-include` file here can only see
2754        // the definition if the `-imacros` file that was written after it ran first.
2755        let tree = TempTree::new(
2756            "preinclude",
2757            &[
2758                ("a.c", "int main(void) { return 0; }\n"),
2759                ("i.h", "#ifdef FROM_MACROS\nint saw_it;\n#else\nint missed_it;\n#endif\n"),
2760                ("m.h", "#define FROM_MACROS 1\nint macros_text;\n"),
2761            ],
2762        );
2763        let out = tree.path("a.i");
2764        let code = run(&args(&[
2765            "-E",
2766            "-include",
2767            &tree.path("i.h"),
2768            "-imacros",
2769            &tree.path("m.h"),
2770            "-o",
2771            &out,
2772            &tree.path("a.c"),
2773        ]));
2774        assert_eq!(code, 0);
2775        let text = std::fs::read_to_string(&out).expect("the output should have been written");
2776        assert!(text.contains("saw_it"), "{text}");
2777        // And the text of the `-imacros` file is thrown away, which is the whole difference
2778        // between the two flags.
2779        assert!(!text.contains("macros_text"), "{text}");
2780    }
2781
2782    #[test]
2783    fn a_file_the_command_line_named_is_a_prerequisite_the_same_as_one_a_directive_named() {
2784        let tree = TempTree::new(
2785            "preinclude-deps",
2786            &[
2787                ("a.c", "int main(void) { return 0; }\n"),
2788                ("i.h", "int from_include;\n"),
2789                ("m.h", "#define M 1\n"),
2790            ],
2791        );
2792        let out = tree.path("dep.d");
2793        let code = run(&args(&[
2794            "-MM",
2795            "-MF",
2796            &out,
2797            "-include",
2798            &tree.path("i.h"),
2799            "-imacros",
2800            &tree.path("m.h"),
2801            "-o",
2802            &tree.path("a.i"),
2803            &tree.path("a.c"),
2804        ]));
2805        assert_eq!(code, 0);
2806        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2807        assert!(text.contains("i.h"), "{text}");
2808        assert!(text.contains("m.h"), "{text}");
2809    }
2810
2811    #[test]
2812    fn a_command_line_include_that_is_nowhere_on_the_path_is_an_error_and_not_a_warning() {
2813        // Including the directory of the source file, which is not on the path for these: the
2814        // command line was not written there, so a name in it is relative to where the compiler
2815        // was run rather than to where the source sits.
2816        let tree = TempTree::new(
2817            "preinclude-missing",
2818            &[("sub/a.c", "int main(void) { return 0; }\n"), ("sub/beside.h", "int x;\n")],
2819        );
2820        let code = run(&args(&["-E", "-include", "beside.h", "-o", "-", &tree.path("sub/a.c")]));
2821        assert_eq!(code, 1);
2822    }
2823
2824    #[test]
2825    fn a_command_line_that_links_names_the_executable_and_not_the_object_it_went_through() {
2826        // The object a link goes through is in a temporary directory and is gone before `make`
2827        // reads any of this, so the rule that named it would be a rule for a file that is never
2828        // there. The target and the file are both the `-o`, which is the executable.
2829        let (opts, plan) = compile(&["-MD", "sub/a.c", "-o", "prog"]);
2830        assert_eq!(plan.output.as_deref(), Some("prog"));
2831        assert_eq!(deps::default_target("sub/a.c", deps_target_output(&opts, &plan)), "prog");
2832        assert_eq!(
2833            deps::default_file(&opts.deps, "sub/a.c", plan.output.as_deref()).as_deref(),
2834            Some("prog.d")
2835        );
2836    }
2837
2838    #[test]
2839    fn the_plan_keeps_the_output_name_because_the_rule_is_written_from_it() {
2840        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c", "-o", "obj/x.o"]);
2841        assert_eq!(plan.output.as_deref(), Some("obj/x.o"));
2842        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c"]);
2843        assert_eq!(plan.output, None);
2844    }
2845
2846    #[test]
2847    fn usage_fits_on_a_screen() {
2848        // Not a style preference. A help text that scrolls is one nobody reads, and this is
2849        // the cheapest way to keep it honest as flags accumulate. The number goes up only when
2850        // a family of flags arrives that has nowhere to share a line, which the two pass gates
2851        // were and which the two fuel flags and `-fsafety=` now are, and it goes up by exactly
2852        // the lines that family took. The four it went up by last are the flags a build system
2853        // passes without being asked to: how much to say, what machine to generate for, threads,
2854        // and the questions `configure` asks before it compiles anything. The one it went up by
2855        // last is the second line of `--emit`, whose kinds are a family that has now outgrown
2856        // one line and has nowhere else to go. The two it went up by last are the dependency
2857        // family, which is eight flags that share nothing with anything above them. The one it
2858        // went up by last is the four spellings of position independent code, which every
2859        // configure script writes and which could only have shared the link line, and that line
2860        // is already four characters short of the limit. The two it went up by last are the rest
2861        // of the include family, which is six more flags that change where a header is looked for
2862        // and two that name a header outright. The one it went up by last is the pair that keeps
2863        // the intermediate files and times the steps, which belong next to the two flags above
2864        // them that are also about watching a compilation rather than changing one. The two it
2865        // went up by last are the section flags and the visibility flag, which are what a build
2866        // that cares about the size of what it ships and about which names it exports writes, and
2867        // the second of them was already taken and only missing from here.
2868        assert!(USAGE.lines().count() < 50, "usage text has grown past one screen");
2869    }
2870}