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/0.10.9")]
30
31pub mod compile;
32pub mod deps;
33pub mod library;
34pub mod link;
35mod map;
36pub mod phase;
37pub mod preprocess;
38pub mod schedule;
39
40use std::fmt::Write as _;
41use std::io::Write as _;
42use std::path::PathBuf;
43
44use rucc_codegen::coverage::{self, Fired};
45use rucc_codegen::pressure::Pressure;
46use rucc_pp::Dependency;
47use rucc_session::{
48    Dumps, EmitKind, Options, Pic, Preinclude, Protector, SaveTemps, Session, Std, runtime,
49};
50use rucc_target::Triple;
51
52use crate::link::LinkOptions;
53
54pub use crate::compile::{Artifact, Compiled, Temps, compile, compile_ir};
55pub use crate::phase::{Input, InputKind, Job, LinkJob, Output, Phase, Plan};
56pub use crate::preprocess::{OsFileSystem, Preprocessed, preprocess};
57pub use crate::schedule::Jobs;
58
59/// The compiler's version, taken from the workspace manifest.
60pub const VERSION: &str = env!("CARGO_PKG_VERSION");
61
62/// What the command line asked for.
63#[derive(Debug, Clone, PartialEq, Eq)]
64pub enum Action {
65    /// Print usage and exit successfully.
66    Help,
67    /// Print the version and exit successfully.
68    Version,
69    /// Print one line and exit successfully, which is what the `-dump` and `-print` family do.
70    ///
71    /// A build system asks these before it compiles anything, and what it does with the answer
72    /// is paste it into a path or into another command line, so each one is a single line with
73    /// no decoration around it.
74    Print(String),
75    /// Print the resolved configuration and exit successfully.
76    PrintConfig(Box<Options>),
77    /// Print the passes the level will run and exit successfully.
78    PrintPipeline(Box<Options>),
79    /// Print the phase plan and the link line and exit successfully, which is `-###`.
80    PrintPlan {
81        /// The resolved options, which is what says what the link line is for.
82        opts: Box<Options>,
83        /// What to do to each input, and in what order.
84        plan: Box<Plan>,
85        /// What the command line said about linking.
86        link: Box<LinkOptions>,
87    },
88    /// Compile the given inputs.
89    Compile {
90        /// The resolved options.
91        opts: Box<Options>,
92        /// What to do to each input, and in what order.
93        plan: Box<Plan>,
94        /// What the command line said about linking.
95        link: Box<LinkOptions>,
96        /// How many translation units to compile at once.
97        jobs: Jobs,
98        /// Whether `-v` asked for the plan to be printed while it runs.
99        verbose: bool,
100    },
101}
102
103/// Why a command line was rejected.
104#[derive(Debug, Clone, PartialEq, Eq)]
105pub struct CliError {
106    /// The message, lowercase and without a trailing period, in the same shape as any other
107    /// diagnostic.
108    pub message: String,
109}
110
111impl std::fmt::Display for CliError {
112    fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
113        f.write_str(&self.message)
114    }
115}
116
117impl std::error::Error for CliError {}
118
119fn err(message: impl Into<String>) -> CliError {
120    CliError { message: message.into() }
121}
122
123/// A question the command line asked instead of asking for a compilation.
124///
125/// These are answered after the loop rather than where they are read, because every one of them
126/// is about the target or about the library search and the last word on both is the end of the
127/// command line.
128enum Query {
129    /// `-dumpmachine`, the triple.
130    Machine,
131    /// `-dumpversion` and `-dumpfullversion`, which are the same three numbers here.
132    Version,
133    /// `-print-multiarch`, the directory name a distribution files this target under.
134    Multiarch,
135    /// `-print-search-dirs`, in the three lines GCC prints.
136    SearchDirs,
137    /// `-print-file-name=<name>`, the full path of a library file.
138    FileName(String),
139    /// `-print-prog-name=<name>`, the full path of a program.
140    ProgName(String),
141    /// `-print-libgcc-file-name`, which is `-print-file-name=libgcc.a` under another spelling.
142    Libgcc,
143}
144
145/// Usage text.
146///
147/// Deliberately short. `spec/04-driver-and-cli.md` puts the full flag reference in the
148/// manual page, because a `--help` nobody can read in one screen is a `--help` nobody reads.
149pub const USAGE: &str = "\
150rucc, an optimizing C compiler
151
152usage: rucc [options] file...
153
154options:
155  -c                     compile and assemble, do not link
156  -S                     compile only, emit assembly
157  -E                     preprocess only
158  -o <file>              write output to <file>, or to standard output for -
159  -D <name>[=<value>], -U <name>      define a macro, or undefine one after every -D
160  -I <dir>               add <dir> to the include search path
161  -iquote -isystem -idirafter <dir>   the other chains, -nostdinc drops ours
162  -I-, -iprefix <p>, -iwithprefix[before] <dir>   the older spellings of those
163  -include <file>, -imacros <file>    read <file> first, the second for its macros only
164  --sysroot=<dir>        look for the library's headers under <dir>, -isysroot too
165  -P, -dM                with -E: leave out the markers, or dump the macros
166  -M -MM -MD -MMD        write a make rule for the source, the last two compile as well
167  -MF <file> -MT <t> -MQ <t> -MP   where the rule goes, what it builds, targets with no recipe
168  -std=<dialect>         c89 through c23, and the gnu spellings
169  -fgnuc-version=<v>     the GCC release to claim, default 7.0.0
170  -x <lang>              treat later inputs as <lang>, or none to stop
171  -O<level>              optimize: 0, 1, 2, 3, s, z
172  -fsafety=<tier>        check memory safety: off, detect, enforce, kernel
173  -f<pass> -fno-<pass> -fdump-ir=<what> -fopt-info[-<kind>][=FILE]
174  -fpass-fuel=<pass>=<n>, -fpass-fuel-global=<n>   stop a pass, or all of them, after n
175  -fdisable-<pass>[=<funcs>], -fenable-<pass>[=<funcs>]   run a pass on some functions only
176  -g -g0 -gdwarf-5, -fno-omit-frame-pointer, -mno-red-zone   debug info, frame pointer, red zone
177  -fstack-protector, -fstack-protector-strong, -fstack-protector-all, -fno-stack-protector
178  -ffunction-sections -fdata-sections   a section per function or variable, for --gc-sections
179  -fvisibility=<what>    default, hidden, internal or protected, when nothing in the source said
180  -l<name>, -L <dir>, -B <dir>   link a library, where to look for one, where our own tools are
181  -fPIC -fpic -fPIE -fpie, -fno-common, -f[no-]strict-aliasing, -pipe   what it does anyway
182  -static -shared -pie -no-pie -nostdlib -nostartfiles -nodefaultlibs -rdynamic -s   how to link
183  -Wl,<arg>, -Xlinker <arg>, -fuse-ld=<name>   hand an argument to the linker, or pick one
184  -Werror -pedantic -pedantic-errors -w   how much to say, and whether it is fatal
185  -m64 -march= -mtune= -mcpu= -mabi= -mcmodel=   what machine to generate for
186  -pthread               build for more than one thread, and link the library for it
187  -dumpmachine -dumpversion -print-multiarch -print-search-dirs   what this compiler is
188  -print-file-name=<name> -print-prog-name=<name>   where a file or a program is
189  -j[n]                  compile n translation units at once, default all
190  -v, -###               print each phase as it runs, or without running any
191  -save-temps[=cwd|obj], -time   keep the .i and the .s, say how long each step took
192  --target=<triple>      generate code for <triple>
193  --emit=<kind>          exe, obj, asm, preprocessed, tast, ir, mir-final,
194                         safety-summary, type-granules
195  --print-config, --print-pipeline    print the configuration or the pipeline, and exit
196  --version              print the version and exit
197  -h, --help             print this message and exit
198
199See spec/04-driver-and-cli.md for the full flag reference.
200";
201
202/// The argument of a flag that may be joined to it or may be the next word.
203///
204/// `-DFOO` and `-D FOO` are the same thing, and `at` is where the flag's own letters end.
205fn joined_or_next(
206    arg: &str,
207    at: usize,
208    args: &[String],
209    i: &mut usize,
210) -> Result<String, CliError> {
211    if arg.len() > at {
212        return Ok(arg[at..].to_owned());
213    }
214    let next = args.get(*i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
215    *i += 1;
216    Ok(next.clone())
217}
218
219/// Parses a command line, without the program name.
220///
221/// # Errors
222///
223/// Returns the message to print when the arguments do not name a compilation this compiler
224/// can attempt.
225pub fn parse_args(args: &[String]) -> Result<Action, CliError> {
226    let host = Triple::host()
227        .ok_or_else(|| err("this host is not a supported target and no --target was given"))?;
228    let mut opts = Options::new(host);
229    let mut inputs: Vec<Input> = Vec::new();
230    let mut print_config = false;
231    let mut print_pipeline = false;
232    let mut print_plan = false;
233    let mut verbose = false;
234    let mut jobs = Jobs::default();
235    let mut nostdinc = false;
236    let mut sysroot: Option<PathBuf> = None;
237    let mut output = None;
238    let mut link = LinkOptions::default();
239    let mut query: Option<Query> = None;
240    let mut threads = false;
241    // `-x` applies to inputs that come after it and stays in effect until the next one, which
242    // is why it is tracked across the loop rather than attached to a single argument.
243    let mut forced: Option<InputKind> = None;
244    // What `-iprefix` last said, stuck on the front of every later `-iwithprefix`. It applies to
245    // the flags after it and not the ones before, so a command line may set it more than once.
246    // GCC's default is its own installed header directory with the last component taken off,
247    // which is a path a cross compiler's build system knows and passes; there is no equivalent
248    // here, so with no `-iprefix` the prefix is nothing and `-iwithprefix` names a directory
249    // outright.
250    let mut iprefix = String::new();
251
252    let mut i = 0;
253    while i < args.len() {
254        let arg = args[i].as_str();
255        i += 1;
256        match arg {
257            "-h" | "--help" => return Ok(Action::Help),
258            "--version" => return Ok(Action::Version),
259            "--print-config" => print_config = true,
260            "--print-pipeline" => print_pipeline = true,
261            "-###" => print_plan = true,
262            "-v" => verbose = true,
263            // The files a compilation goes through, kept rather than thrown away. The bare
264            // spelling means `=obj` and not `=cwd`, which is not what the manual says and is what
265            // gcc 16 does; `SaveTemps::Object` carries the measurement.
266            "-save-temps" => opts.save_temps = SaveTemps::Object,
267            _ if arg.starts_with("-save-temps=") => {
268                opts.save_temps = arg["-save-temps=".len()..].parse().map_err(err)?;
269            }
270            // How long each step took. A misspelling of this is worth rejecting rather than
271            // ignoring, since a run that says nothing looks like a compilation that took no time.
272            "-time" => opts.time = true,
273            "-c" => opts.emit = EmitKind::Object,
274            "-S" => opts.emit = EmitKind::Asm,
275            "-E" => opts.emit = EmitKind::Preprocessed,
276            "-g" => opts.debug_info = true,
277            // GCC's own levels of how much debug information to write. Zero is none and every
278            // other number is some, and this compiler has one amount, so the numbers above zero
279            // all mean the same thing here. `-ggdb` is the same flag asking for whatever the
280            // debugger on the machine prefers, which is what we emit anyway.
281            "-g0" => opts.debug_info = false,
282            "-g1" | "-g2" | "-g3" | "-ggdb" | "-ggdb1" | "-ggdb2" | "-ggdb3" => {
283                opts.debug_info = true;
284            }
285            // The version of DWARF to write. We write DWARF 5 and nothing else, so a build that
286            // asks for another version is told rather than handed a file it cannot read.
287            "-gdwarf" | "-gdwarf-5" => opts.debug_info = true,
288            _ if arg.starts_with("-gdwarf-") => {
289                return Err(err(format!(
290                    "{arg}: this compiler writes DWARF 5 and no other version, see \
291                     spec/11-debug-info.md"
292                )));
293            }
294            "-Werror" => opts.warnings_are_errors = true,
295            // Nothing that is not fatal is said at all. Read at the one place a diagnostic goes
296            // through rather than here, so that a warning `-w` dropped is not counted either.
297            "-w" => opts.warnings = false,
298            "-pedantic-errors" => {
299                opts.pedantic = true;
300                opts.warnings_are_errors = true;
301            }
302            "-P" => opts.line_markers = false,
303            // The dependency family, which section 4.4 calls required because every build system
304            // that generates its own makefiles asks for it. The two that end in `D` write a file
305            // beside the object and let the compilation happen, and the two that do not write to
306            // standard output and stop after it. Nothing here turns the system headers back on
307            // once a flag has turned them off, which is GCC's behaviour and is why `-MM -M` is
308            // `-MM`: the flag asking for fewer of them is the one with something to say.
309            "-M" => {
310                opts.deps.emit = true;
311                opts.deps.instead_of_compiling = true;
312            }
313            "-MM" => {
314                opts.deps.emit = true;
315                opts.deps.instead_of_compiling = true;
316                opts.deps.system_headers = false;
317            }
318            "-MD" => opts.deps.emit = true,
319            "-MMD" => {
320                opts.deps.emit = true;
321                opts.deps.system_headers = false;
322            }
323            "-MP" => opts.deps.phony = true,
324            // These three take a word and only in the separated form, which is how GCC spells
325            // them and how every build system writes them.
326            "-MF" | "-MT" | "-MQ" => {
327                let value =
328                    args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
329                i += 1;
330                match arg {
331                    "-MF" => opts.deps.file = Some(value.clone()),
332                    // The whole of the difference between the two. `-MT` is for a build that has
333                    // already escaped what it is passing, and `-MQ` is for one that has a name
334                    // and wants it to arrive as that name.
335                    "-MT" => opts.deps.targets.push(value.clone()),
336                    _ => opts.deps.targets.push(deps::escaped(value)),
337                }
338            }
339            // The questions a build system asks before it compiles anything. Answered after the
340            // loop, because each one is about the target or the library search and the command
341            // line has not finished saying what those are.
342            "-dumpmachine" => query = Some(Query::Machine),
343            "-dumpversion" | "-dumpfullversion" => query = Some(Query::Version),
344            "-print-multiarch" => query = Some(Query::Multiarch),
345            "-print-search-dirs" => query = Some(Query::SearchDirs),
346            "-print-libgcc-file-name" => query = Some(Query::Libgcc),
347            _ if arg.starts_with("-print-file-name=") => {
348                query = Some(Query::FileName(arg["-print-file-name=".len()..].to_owned()));
349            }
350            _ if arg.starts_with("-print-prog-name=") => {
351                query = Some(Query::ProgName(arg["-print-prog-name=".len()..].to_owned()));
352            }
353            // A program built to run in more than one thread. On every platform this compiler
354            // targets that is a macro the library's headers read and one more library on the
355            // link line, and the library is added after the loop so that it lands after the
356            // objects that refer to it.
357            "-pthread" | "-pthreads" => {
358                opts.defines.push("_REENTRANT".to_owned());
359                threads = true;
360            }
361            "-ansi" => {
362                opts.std = Std::C89;
363                opts.gnu_extensions = false;
364            }
365            // `-Wpedantic` is the same flag under the name the `-W` family gives it, which is
366            // the spelling a build system that groups its warning flags tends to write.
367            "-pedantic" | "-Wpedantic" => opts.pedantic = true,
368            // Both directions, because a build that needs this for one directory turns it back
369            // off for the next one rather than leaving it on for the whole tree.
370            "-fpermissive" => opts.permissive = true,
371            "-fno-permissive" => opts.permissive = false,
372            "-ffreestanding" => opts.hosted = false,
373            "-fhosted" => opts.hosted = true,
374            "-fno-builtin" => opts.builtins = false,
375            "-fbuiltin" => opts.builtins = true,
376            // The C89 dialects are under GNU's reading whatever this says, so turning it off
377            // there is turning off something the dialect asked for, which is accepted and does
378            // nothing. gcc refuses that command line, and there is nothing it could have meant.
379            "-fgnu89-inline" => opts.gnu89_inline = true,
380            "-fno-gnu89-inline" => opts.gnu89_inline = false,
381            // Both directions of each, because a build system that wants one of these usually
382            // writes it beside the flag that turns it back off for one directory.
383            "-fno-omit-frame-pointer" => opts.frame_pointer = true,
384            "-fomit-frame-pointer" => opts.frame_pointer = false,
385            "-mno-red-zone" => opts.red_zone = false,
386            "-mred-zone" => opts.red_zone = true,
387            // Four flags rather than one with an argument, which is how gcc spells them and how
388            // every build line writes them. Last one wins, because a package build puts
389            // `-fstack-protector-strong` in its global flags and a directory that cannot have one
390            // turns it back off on the line after.
391            "-fno-stack-protector" | "-fno-stack-protector-all" | "-fno-stack-protector-strong" => {
392                opts.protector = Protector::None;
393            }
394            "-fstack-protector" => opts.protector = Protector::Buffers,
395            "-fstack-protector-strong" => opts.protector = Protector::Strong,
396            "-fstack-protector-all" => opts.protector = Protector::All,
397            // GCC drops its own include directory along with the system ones, because its
398            // headers are half of a pair with the library's and half a pair is worse than
399            // none. A build that passes this is supplying the whole set itself.
400            "-nostdinc" => nostdinc = true,
401            "-o" => {
402                output = Some(args.get(i).ok_or_else(|| err("-o requires an argument"))?.clone());
403                i += 1;
404            }
405            // The flags that take a directory only in the separated form. GCC spells them
406            // this way and nothing writes `-iquotedir`, so accepting the joined form would
407            // mean guessing at a path that starts with the flag's own letters.
408            // Apple's spelling of `--sysroot`, and the one its own build systems pass. The
409            // two mean the same thing here: the configured directories are under there rather
410            // than under the root.
411            "-isysroot" => {
412                let dir = args.get(i).ok_or_else(|| err("-isysroot requires an argument"))?;
413                i += 1;
414                sysroot = Some(PathBuf::from(dir));
415            }
416            "-iquote" | "-isystem" | "-idirafter" => {
417                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
418                i += 1;
419                match arg {
420                    "-iquote" => opts.search.push_quote(dir.clone()),
421                    "-isystem" => opts.search.push_system(dir.clone()),
422                    _ => opts.search.push_after(dir.clone()),
423                }
424            }
425            "-iprefix" => {
426                iprefix = args.get(i).ok_or_else(|| err("-iprefix requires an argument"))?.clone();
427                i += 1;
428            }
429            // Where GCC puts these is not where its manual says it puts them, and this is the
430            // measured answer rather than the documented one: `-iwithprefix` lands in the
431            // `-isystem` slot and not the `-idirafter` slot, and `-iwithprefixbefore` lands in
432            // the `-I` slot. A cross build that uses them is relying on the behaviour, since
433            // that is the compiler it was developed against.
434            "-iwithprefix" | "-iwithprefixbefore" => {
435                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
436                i += 1;
437                let dir = format!("{iprefix}{dir}");
438                if arg == "-iwithprefix" {
439                    opts.search.push_system(dir);
440                } else {
441                    opts.search.push_bracket(dir);
442                }
443            }
444            "-include" | "-imacros" => {
445                let name = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
446                i += 1;
447                opts.preincludes
448                    .push(Preinclude { name: name.clone(), macros_only: arg == "-imacros" });
449            }
450            // The flag `-iquote` was introduced to replace, still passed by build systems old
451            // enough to predate the replacement. It is not a directory: it says that every `-I`
452            // so far is for quoted includes only, and that a quoted include stops looking next
453            // to the file that wrote it.
454            "-I-" => opts.search.split_quote_chain(),
455            "-x" => {
456                let lang = args.get(i).ok_or_else(|| err("-x requires an argument"))?;
457                i += 1;
458                forced = if lang == "none" {
459                    None
460                } else {
461                    Some(InputKind::from_x_arg(lang).map_err(|e| err(format!("{e}")))?)
462                };
463            }
464            // Not a GCC flag. spec/03-architecture.md section 3.5 compiles several
465            // translation units in one process rather than making the build system fork, and
466            // section 3.8's determinism check compares `-j1` against `-j16`, so the knob has
467            // to exist and has to be spelled the way `make` spells it.
468            // `-DFOO`, `-D FOO` and the same for `-U` and `-I`. Both forms are in wide use
469            // and a build system may produce either, so both are read here rather than
470            // being normalised by whatever generated the command line.
471            _ if arg.starts_with("-D") => {
472                let value = joined_or_next(arg, 2, args, &mut i)?;
473                opts.defines.push(value);
474            }
475            _ if arg.starts_with("-U") => {
476                let value = joined_or_next(arg, 2, args, &mut i)?;
477                opts.undefines.push(value);
478            }
479            _ if arg.starts_with("-I") => {
480                let dir = joined_or_next(arg, 2, args, &mut i)?;
481                opts.search.push_bracket(dir);
482            }
483            _ if arg.starts_with("-std=") => {
484                let name = &arg["-std=".len()..];
485                let (std, gnu) = Std::from_flag(name)
486                    .ok_or_else(|| err(format!("unknown dialect `{name}`, see --help")))?;
487                opts.std = std;
488                opts.gnu_extensions = gnu;
489            }
490            // Section 4.5. The claim decides which half of glibc's `sys/cdefs.h` we are
491            // handed, so a differential run that does not set it is comparing two compilers
492            // that believe they are different compilers.
493            // GCC packs these into one flag, so `-dDI` is two of them. Letters in the family
494            // that we have not written yet are accepted and ignored, because a dump is a
495            // debugging aid and a build that asks for one should still compile. A letter
496            // outside the family falls through to the unknown option error, which is what
497            // keeps `-dumpversion` from being read as a dump of nothing.
498            _ if Dumps::is_family(arg) => {
499                opts.dumps.add(&arg[2..]);
500            }
501            // One name at a time, which is what a build that means its own `memcpy` and the
502            // library's everything else writes. The name is not checked against a list, because
503            // the flag is about what the program means by a name and a program is allowed to mean
504            // something by a name this compiler has never heard of.
505            _ if arg.starts_with("-fno-builtin-") => {
506                opts.no_builtin.push(arg["-fno-builtin-".len()..].to_owned());
507            }
508            _ if arg.starts_with("-fgnuc-version=") => {
509                let v = &arg["-fgnuc-version=".len()..];
510                opts.gnuc = v.parse().map_err(err)?;
511            }
512            // spec/13-gnu-compat.md section 13.3 promises this flag an error that says why rather
513            // than the unknown option one, because a build reaching for it is asking for a feature
514            // and deserves to be told it is not coming rather than told the spelling is wrong.
515            // The negative form is what this compiler does anyway, so it is taken and dropped.
516            "-fnested-functions" => {
517                return Err(err(
518                    "nested functions are not supported: a call to one goes through a trampoline \
519                     written on the stack, which no target that enforces an unexecutable stack \
520                     allows",
521                ));
522            }
523            "-fno-nested-functions" => {}
524            // Which of the two links the output is for, which is a real difference and not a
525            // description of what happens anyway. Everything here is position independent either
526            // way, and what these decide is whether a name may be one another object defines or
527            // replaces, because a link that produces an executable puts every name in the same
528            // program and a link that produces a shared library does not.
529            //
530            // It matters that they are accepted at all, whatever they then do. Every autoconf and
531            // cmake build puts `-fPIC` on the compile line, so a compiler that rejects it cannot
532            // be the `CC` of a project that has a configure script, whatever else it can do. That
533            // is how this was found: building SQLite's test fixture stopped on it.
534            "-fPIC" | "-fpic" => opts.pic = Pic::Library,
535            // Not a synonym of the pair above, which is what they were treated as until #756. The
536            // library is the expensive answer and gcc makes it the one that has to be asked for,
537            // so this is also what nothing at all means.
538            "-fPIE" | "-fpie" => opts.pic = Pic::Executable,
539            // A different question from the pair above, and the one every distribution build of a
540            // shared library answers. `-fPIC` decides how an address is reached, and this decides
541            // whether the optimizer may believe a body it can see, because an exported name is one
542            // the dynamic linker may find another definition of first. On by default, which is
543            // gcc's arrangement and is the honest answer, and off is a promise the build makes and
544            // nothing checks.
545            "-fsemantic-interposition" => opts.interposition = true,
546            "-fno-semantic-interposition" => opts.interposition = false,
547            // Two requests rather than one, and the same table answers both, so what decides is
548            // whether either of them is standing. gcc arranges it the same way: the asynchronous
549            // one is the default here and it implies the other, and a line that asks for a table
550            // and against an asynchronous one gets a table.
551            "-fasynchronous-unwind-tables" => opts.async_unwind_tables = true,
552            "-fno-asynchronous-unwind-tables" => opts.async_unwind_tables = false,
553            "-funwind-tables" => opts.unwind_tables = true,
554            "-fno-unwind-tables" => opts.unwind_tables = false,
555            // The other direction is a request, not a description, and it is one this compiler
556            // cannot grant, so it gets the treatment section 13.3 asks for rather than the unknown
557            // option error. Answering it by carrying on would be answering a different question:
558            // the code would still be position independent, which is correct everywhere an
559            // ordinary program runs and is wrong in a kernel, where the flag is written precisely
560            // because there is no loader to fill a global offset table in.
561            "-fno-pic" | "-fno-pie" => {
562                return Err(err(
563                    "position dependent code is not supported: an address that may be in another \
564                     object is loaded out of the global offset table, and nothing here emits the \
565                     absolute form this asks for. Use -no-pie if what you meant was how to link",
566                ));
567            }
568            // A section per function and a section per variable, which is what makes
569            // `--gc-sections` able to drop anything: a linker can leave out a section nothing
570            // reaches and cannot leave out half of one. Both directions are taken, and the off
571            // one is the default rather than a refusal, since a build that writes it is asking
572            // for what happens anyway.
573            "-ffunction-sections" => opts.function_sections = true,
574            "-fno-function-sections" => opts.function_sections = false,
575            "-fdata-sections" => opts.data_sections = true,
576            "-fno-data-sections" => opts.data_sections = false,
577            // Another description of what this compiler does. A file scope declaration with no
578            // initializer is written into `.bss` as an ordinary defined symbol, not offered to the
579            // linker as a common one for it to merge, which is what `-fno-common` asks for and what
580            // gcc has done by default since 10. Nothing in the front end produces `Linkage::Common`
581            // at all.
582            "-fno-common" => {}
583            // And the request, which is the one that cannot be granted. It is a real difference and
584            // not a preference: two files each writing `int g;` link under `-fcommon` and are a
585            // duplicate definition without it, which is the whole reason the flag survives.
586            "-fcommon" => {
587                return Err(err(
588                    "a tentative definition is written into .bss as its own symbol here, and \
589                     nothing emits the common symbol this asks the linker to merge. Give the \
590                     variable a definition in one file and declare it extern in the others",
591                ));
592            }
593            // Both directions of this one are taken, which is the exception to the rule above, and
594            // the reason is which way being wrong costs something.
595            //
596            // Nothing here derives anything from the type an object is accessed through. The IR has
597            // somewhere to put a type based aliasing node and lowering fills it with nothing on
598            // every access, so no pass has one to read and the alias analysis falls back to what it
599            // can see. That makes `-fno-strict-aliasing` a description, the way `-fPIC` is.
600            //
601            // `-fstrict-aliasing` is a request to assume more than that, and this compiler assumes
602            // less. Answering a request for a weaker guarantee by giving a stronger one is safe in
603            // a way the `-fno-pic` case is not: every program that is correct under the assumption
604            // is correct without it, only slower. And `-O2` implies it, so a build that spells it
605            // out is a build that would stop on a compiler that refused it, for no gain at all.
606            "-fstrict-aliasing" | "-fno-strict-aliasing" => {}
607            // About temporary files rather than about code. There is nothing between the phases of
608            // one compilation here to write to a file in the first place.
609            "-pipe" => {}
610            // Nothing here writes colour, so all of these are the same answer, and it is the answer
611            // that costs nothing: the diagnostics come out plain either way and no build depends on
612            // an escape sequence being there. Taken rather than refused because cmake writes
613            // `-fdiagnostics-color=always` on every compile line when the generator is ninja, which
614            // makes this the second most common flag after `-fPIC` to stop a build over a question
615            // about how the text looks.
616            "-fdiagnostics-color" | "-fno-diagnostics-color" => {}
617            _ if arg.starts_with("-fdiagnostics-color=") => {}
618            // The link flags. None of them changes the compilation, which is why they are
619            // collected apart from `opts` and why `-lm` on a `-c` line is a note rather than an
620            // error: it is a thing said to a linker that is not going to run.
621            "-static" => link.is_static = true,
622            "-shared" => link.shared = true,
623            "-pie" => link.pie = Some(true),
624            "-no-pie" | "-nopie" => link.pie = Some(false),
625            "-nostdlib" => link.no_stdlib = true,
626            "-nostartfiles" => link.no_startfiles = true,
627            "-nodefaultlibs" => link.no_defaultlibs = true,
628            "-fno-builtins-lib" => link.no_builtins_lib = true,
629            "-fbuiltins-lib" => link.no_builtins_lib = false,
630            "-rdynamic" | "-export-dynamic" => link.export_dynamic = true,
631            "-s" => link.strip = true,
632            "-Xlinker" => {
633                let next = args.get(i).ok_or_else(|| err("-Xlinker requires an argument"))?;
634                i += 1;
635                link.passthrough.push(next.clone());
636            }
637            _ if arg.starts_with("-Wl,") => {
638                // Commas separate arguments rather than being part of one, which is what makes
639                // `-Wl,-rpath,/opt/lib` two words to the linker and one word here.
640                link.passthrough.extend(arg["-Wl,".len()..].split(',').map(str::to_owned));
641            }
642            _ if arg.starts_with("-fuse-ld=") => {
643                link.use_ld = Some(arg["-fuse-ld=".len()..].to_owned());
644            }
645            _ if arg.starts_with("-l") && arg.len() > 2 => {
646                inputs.push(Input::library(&arg[2..]));
647            }
648            "-l" => {
649                let next = args.get(i).ok_or_else(|| err("-l requires an argument"))?;
650                i += 1;
651                inputs.push(Input::library(next));
652            }
653            _ if arg.starts_with("-L") => {
654                link.search.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
655            }
656            _ if arg.starts_with("-B") => {
657                link.prefixes.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
658            }
659            _ if arg.starts_with("-j") => {
660                jobs = Jobs::parse(&arg[2..]).map_err(err)?;
661            }
662            _ if arg.starts_with("--sysroot=") => {
663                sysroot = Some(PathBuf::from(&arg["--sysroot=".len()..]));
664            }
665            _ if arg.starts_with("--target=") => {
666                let t = &arg["--target=".len()..];
667                opts.target = t.parse().map_err(|e| err(format!("{e}")))?;
668            }
669            _ if arg.starts_with("--emit=") => {
670                let k = &arg["--emit=".len()..];
671                opts.emit = k
672                    .parse()
673                    .map_err(|()| err(format!("unknown --emit kind `{k}`, see --help")))?;
674            }
675            // A bare `-O` is `-O1`, which is what GCC has and what a hand written makefile tends
676            // to write. `-Og` is GCC's level for a build somebody is going to step through, and
677            // it is `-O1` with the transformations that move code around left out; this compiler
678            // has no such level yet, so it is the nearest one and `--print-pipeline` says what
679            // that came to rather than the flag pretending otherwise.
680            "-O" | "-Og" => opts.opt_level = rucc_session::OptLevel::O1,
681            // The union of `-O3` and `-ffast-math`, and the second half of that changes what
682            // floating point arithmetic means. Refused rather than taken as `-O3`, because a
683            // build that asks for fast math and is quietly given ordinary arithmetic gets a
684            // slower program than it asked for and a build that is given fast math it did not
685            // ask for gets a wrong one.
686            "-Ofast" => {
687                return Err(err(
688                    "-Ofast is -O3 with fast math, and fast math is not implemented, see \
689                     spec/04-driver-and-cli.md section 4.6",
690                ));
691            }
692            _ if arg.starts_with("-O") => {
693                opts.opt_level = arg[2..]
694                    .parse()
695                    .map_err(|()| err(format!("unknown optimization level `{arg}`")))?;
696            }
697            // What every name gets when nothing in the source said, which the attribute in the
698            // source overrides rather than the other way round. Before the optimizer's `-f`
699            // family below for the reason the tier below it is.
700            _ if arg.starts_with("-fvisibility=") => {
701                let seen = &arg["-fvisibility=".len()..];
702                opts.visibility = seen.parse().map_err(|()| {
703                    err(format!(
704                        "`{seen}` is not a visibility, which is default, hidden, internal or \
705                         protected"
706                    ))
707                })?;
708            }
709            // The memory safety monitor, from section 15.4 of
710            // `spec/safe-memory/15-integration.md`. Before the optimizer's `-f` family below,
711            // because a pass that took the name `safety=detect` would otherwise be handed the
712            // flag, and the tier is not a pass.
713            _ if arg.starts_with("-fsafety=") => {
714                let tier = &arg["-fsafety=".len()..];
715                opts.safety = tier.parse().map_err(|()| {
716                    err(format!(
717                        "`{tier}` is not a safety tier, which is off, detect, enforce or kernel"
718                    ))
719                })?;
720            }
721            // The optimizer's own flags, from section 9.10 of `spec/09-optimizer.md`. These come
722            // after every `-f` the rest of the compiler answers to, so a pass can never take a
723            // name that already means something else on the command line.
724            _ if arg.starts_with("-fpass-fuel=") => {
725                let (name, count) = arg["-fpass-fuel=".len()..]
726                    .split_once('=')
727                    .ok_or_else(|| err("-fpass-fuel= is spelled <pass>=<count>"))?;
728                if rucc_opt::pass::find(name).is_none() {
729                    return Err(err(format!(
730                        "`{name}` is not a pass this compiler has, see --print-pipeline"
731                    )));
732                }
733                let count: u32 = count
734                    .parse()
735                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
736                opts.pass_fuel.push((name.to_owned(), count));
737            }
738            _ if arg.starts_with("-fpass-fuel-global=") => {
739                let count = &arg["-fpass-fuel-global=".len()..];
740                let count: u32 = count
741                    .parse()
742                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
743                opts.pass_fuel_global = Some(count);
744            }
745            // Everything from `-fopt-info` to the end of the argument, which is optional
746            // keywords joined by hyphens and an optional `=<file>`. Checked here rather than
747            // where the remarks are printed, because by then the compilation somebody wanted
748            // to hear about is over.
749            _ if arg == "-fopt-info"
750                || arg.starts_with("-fopt-info=")
751                || arg.starts_with("-fopt-info-") =>
752            {
753                let rest = &arg["-fopt-info".len()..];
754                let (kinds, file) = match rest.split_once('=') {
755                    Some((kinds, file)) => (kinds, Some(file)),
756                    None => (rest, None),
757                };
758                let kinds = kinds.strip_prefix('-').unwrap_or(kinds);
759                rucc_opt::Wants::none().add(kinds).map_err(err)?;
760                opts.opt_info.push(kinds.to_owned());
761                if let Some(file) = file {
762                    if file.is_empty() {
763                        return Err(err("-fopt-info= was given no file to write to"));
764                    }
765                    opts.opt_info_file = Some(file.to_owned());
766                }
767            }
768            _ if arg.starts_with("-fdump-ir=") => {
769                // Checked here rather than where the dumps are taken, because the compilation
770                // that would have been dumped is over by then.
771                let spec = &arg["-fdump-ir=".len()..];
772                rucc_opt::Dumps::default().add(spec).map_err(err)?;
773                opts.dump_ir.push(spec.to_owned());
774            }
775            // Before the bare `-f<pass>` below, because a pass called `enable-something` would
776            // otherwise take the flag away from the gate. Checked here rather than where the
777            // pipeline reads it, for the reason that applies to all of these: a misspelled pass
778            // name that quietly gated nothing looks exactly like a pass that is not the guilty
779            // one, and a bisection would carry on past the thing it was looking for.
780            _ if arg.starts_with("-fdisable-") || arg.starts_with("-fenable-") => {
781                let on = arg.starts_with("-fenable-");
782                let spec = &arg[if on { "-fenable-".len() } else { "-fdisable-".len() }..];
783                rucc_opt::Gates::default().add(on, spec).map_err(err)?;
784                opts.pass_gates.push((on, spec.to_owned()));
785            }
786            _ if arg.strip_prefix("-fno-").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
787                opts.passes.push((arg["-fno-".len()..].to_owned(), false));
788            }
789            _ if arg.strip_prefix("-f").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
790                opts.passes.push((arg["-f".len()..].to_owned(), true));
791            }
792            // The unstable options, spelled the way rustc spells them and carrying the same
793            // promise, which is none: one of these may change or go away in any release. They are
794            // measurements and debugging aids rather than things a build asks for, which is why
795            // none of them is in the usage text and all of them are in section 4.11 of
796            // `spec/04-driver-and-cli.md`.
797            "-Zverify-each" => opts.verify_each = true,
798            _ if arg.starts_with("-Zrule-coverage=") => {
799                let file = &arg["-Zrule-coverage=".len()..];
800                if file.is_empty() {
801                    return Err(err("-Zrule-coverage= needs a file to write to"));
802                }
803                opts.rule_coverage = Some(file.to_owned());
804            }
805            _ if arg.starts_with("-Zregister-pressure=") => {
806                let file = &arg["-Zregister-pressure=".len()..];
807                if file.is_empty() {
808                    return Err(err("-Zregister-pressure= needs a file to write to"));
809                }
810                opts.register_pressure = Some(file.to_owned());
811            }
812            _ if arg.starts_with("-Z") => {
813                return Err(err(format!(
814                    "`{arg}` is not an unstable option this compiler has, see \
815                     spec/04-driver-and-cli.md section 4.11 for the ones it does"
816                )));
817            }
818            // The word size, which is a statement about the target and is taken as one. A build
819            // that says the size the target already has is saying nothing, and one that says the
820            // other size is asking for a target this compiler does not have, which it is told
821            // rather than being given the wrong one.
822            "-m64" | "-m32" | "-mx32" => {
823                let want: u32 = match arg {
824                    "-m64" => 64,
825                    _ => 32,
826                };
827                let have = rucc_target::TargetInfo::new(opts.target).pointer_width;
828                if have != want {
829                    return Err(err(format!(
830                        "{arg} asks for a {want} bit target and {} is {have} bit, use \
831                         --target= to name the one you mean",
832                        opts.target
833                    )));
834                }
835            }
836            // Which processor in the family to generate for. This compiler emits the base
837            // instruction set of the architecture and nothing above it, so a program built with
838            // any of these runs on the machine that was named; it is a program that could have
839            // been faster rather than a program that is wrong, which is what makes these safe to
840            // take and ignore where a flag that changed the meaning of the code would not be.
841            _ if arg.starts_with("-march=")
842                || arg.starts_with("-mtune=")
843                || arg.starts_with("-mcpu=") => {}
844            // The calling convention, which is not safe to ignore. Taken when it names the one
845            // the target already uses and refused otherwise.
846            _ if arg.starts_with("-mabi=") => {
847                let want = &arg["-mabi=".len()..];
848                let have = match opts.target.arch {
849                    rucc_target::Arch::X86_64 => "sysv",
850                    rucc_target::Arch::Aarch64 => "lp64",
851                    rucc_target::Arch::Riscv64 => "lp64d",
852                };
853                if want != have {
854                    return Err(err(format!(
855                        "{arg}: {} uses the {have} convention and this compiler has no other",
856                        opts.target
857                    )));
858                }
859            }
860            // How far apart the pieces of the program may be. The small model is what we emit and
861            // it is every hosted program's default; the kernel model is a different one and a
862            // build that asks for it and does not get it links and then does not run.
863            "-mcmodel=small" => {}
864            _ if arg.starts_with("-mcmodel=") => {
865                return Err(err(format!(
866                    "{arg}: this compiler emits the small code model and no other, see \
867                     spec/12-targets.md"
868                )));
869            }
870            // GCC's own scripting language for how the driver builds a command line.
871            // `spec/04-driver-and-cli.md` section 4.4 settles that we will not have it, so a
872            // build reaching for it is told which flags do the same job.
873            _ if arg.starts_with("-specs=") => {
874                return Err(err(
875                    "-specs= is not supported: the parts of it builds rely on are -B, -L, \
876                     -nostdlib, -nostartfiles and -Wl,, see spec/04-driver-and-cli.md \
877                     section 4.4",
878                ));
879            }
880            // Arguments meant for a separate assembler or preprocessor, which this compiler does
881            // not have: both are inside it and neither reads a command line. Refused rather than
882            // dropped, because every one of these says something about the output and a build
883            // that asked for `-Wa,--noexecstack` and was silently given an executable stack got
884            // the opposite of what it asked for.
885            _ if arg.starts_with("-Wa,") || arg.starts_with("-Wp,") => {
886                return Err(err(format!(
887                    "`{arg}` is an argument for a separate assembler or preprocessor, and both \
888                     are inside this compiler rather than programs it runs"
889                )));
890            }
891            "-Xassembler" | "-Xpreprocessor" => {
892                return Err(err(format!(
893                    "{arg} hands an argument to a separate assembler or preprocessor, and both \
894                     are inside this compiler rather than programs it runs"
895                )));
896            }
897            // Everything else in the `-W` family. `spec/04-driver-and-cli.md` section 4.1 has
898            // this one as a rule about build systems rather than about warnings: autoconf finds
899            // out whether a warning flag exists by passing it and looking at the exit status, so
900            // a compiler that refuses one it has not heard of fails a configure script written
901            // for a GCC newer than itself. The names are not checked against a list because this
902            // compiler has no warning groups for a list to be of, which #485 is about.
903            _ if arg.starts_with("-W") => {}
904            // Flags that name something this compiler does not do and would not do differently
905            // if it did. `-fno-ident` is about a comment in the output that we do not write
906            // either way, and the others are about a way of ordering the compilation that has
907            // been GCC's only way for twenty years. Section 4.1 asks for the list to be short
908            // and for adding to it to be deliberate, which is why it is written out here.
909            "-fno-ident"
910            | "-fident"
911            | "-funit-at-a-time"
912            | "-fno-unit-at-a-time"
913            | "-shared-libgcc"
914            | "-static-libgcc" => {}
915            _ if arg.starts_with('-') && arg.len() > 1 => {
916                // Silently ignoring an unknown flag is how a build ends up not doing what
917                // its author asked. spec/13-gnu-compat.md section 13.4 makes this an error
918                // for the flags that change code generation, and the safe default until the
919                // flag table is populated is to reject everything we do not know.
920                return Err(err(format!("unknown option `{arg}`")));
921            }
922            _ => inputs.push(Input { path: arg.to_owned(), forced, library: false }),
923        }
924    }
925
926    // Last, so that it lands after every `-isystem` the command line gave. That is GCC's
927    // order: a directory the user names outranks the compiler's own, and the compiler's own
928    // outranks the library's. It is pushed after the loop rather than before it because
929    // `SearchPath` appends within a group and the position is what the order is.
930    // The same directory the headers were looked for under, because a sysroot is a statement
931    // about a whole installation and not about half of one.
932    link.sysroot = sysroot.clone();
933    // After the loop rather than where `-pthread` was read, so that it lands after the objects
934    // that refer to it. A static link takes the definitions it needs from a library when it
935    // reaches it and not afterwards, so a library before the objects is a library that answers
936    // nothing.
937    if threads {
938        inputs.push(Input::library("pthread"));
939    }
940    if let Some(query) = query {
941        return Ok(Action::Print(answer(&query, &opts, &link)));
942    }
943    // `-M` and `-MM` produce the rule and nothing else, so the run stops after phase 4 whatever
944    // else the command line asked for. Read here rather than where the flag was, because a `-c`
945    // written after it has to lose and the loop cannot know that until it has ended. The output
946    // file is where the rule goes rather than where an object would have gone, and the last
947    // phase being the preprocessor is what makes that true without a second rule for it.
948    if opts.deps.instead_of_compiling {
949        opts.emit = EmitKind::Preprocessed;
950    }
951    if !nostdinc {
952        opts.search.push_system(runtime::DIR);
953        // And the library's after ours, which is the other half of the same order. They go on
954        // here rather than at the point `--target=` or `--sysroot=` was read because either
955        // one changes the answer and the last word on both is the end of the loop.
956        for dir in library::system_dirs(opts.target, sysroot.as_deref()) {
957            opts.search.push_system(dir);
958        }
959    }
960    // Once, here, rather than as each directory is pushed. A `-I` that names a system
961    // directory has to lose to the system entry and the system entry is added last, so the
962    // question cannot be answered until the whole path is known.
963    opts.search.remove_duplicates();
964
965    // The target has to be resolved before the configuration is printed, so this check comes
966    // after the loop rather than at the point `--print-config` was seen.
967    if print_config {
968        return Ok(Action::PrintConfig(Box::new(opts)));
969    }
970    if print_pipeline {
971        return Ok(Action::PrintPipeline(Box::new(opts)));
972    }
973    let plan = Plan::new(&opts, &inputs, output.as_deref()).map_err(|e| err(e.message))?;
974    if print_plan {
975        return Ok(Action::PrintPlan {
976            opts: Box::new(opts),
977            plan: Box::new(plan),
978            link: Box::new(link),
979        });
980    }
981    Ok(Action::Compile {
982        opts: Box::new(opts),
983        plan: Box::new(plan),
984        link: Box::new(link),
985        jobs,
986        verbose,
987    })
988}
989
990/// What one of the `-dump` and `-print` flags prints.
991///
992/// GCC prints the name back unchanged when it cannot find the file a `-print` flag asked about,
993/// which is what makes the answer safe to paste into a link line whether or not the file is
994/// there, and this does the same.
995fn answer(query: &Query, opts: &Options, link: &LinkOptions) -> String {
996    let found = |name: &str| {
997        link::find_in_search(link, opts.target, name)
998            .map_or_else(|| name.to_owned(), |path| path.display().to_string())
999    };
1000    match query {
1001        Query::Machine => opts.target.to_string(),
1002        Query::Version => VERSION.to_owned(),
1003        Query::Multiarch => link::multiarch(opts.target),
1004        // The three lines GCC prints, in its order and with its punctuation, because what reads
1005        // them is a script written against that shape. There is no installation directory to
1006        // report: this compiler is one binary that works wherever it is copied, and the headers
1007        // it ships are inside it, so `install` is where the binary is and nothing is under it.
1008        Query::SearchDirs => {
1009            let here = std::env::current_exe()
1010                .ok()
1011                .and_then(|p| p.parent().map(std::path::Path::to_path_buf))
1012                .unwrap_or_default();
1013            let list = |dirs: &[PathBuf]| {
1014                dirs.iter().map(|d| d.display().to_string()).collect::<Vec<_>>().join(":")
1015            };
1016            let libraries = link::search_dirs(link, opts.target);
1017            format!(
1018                "install: {}\nprograms: ={}\nlibraries: ={}",
1019                here.display(),
1020                list(&link.prefixes),
1021                list(&libraries)
1022            )
1023        }
1024        Query::FileName(name) => found(name),
1025        // The name GCC gives the library of routines a compiler's output calls that the C
1026        // library does not have. Ours is built in and there is no file, so the answer is the
1027        // name itself, which is what GCC prints when it cannot find one either.
1028        Query::Libgcc => found("libgcc.a"),
1029        // A program rather than a library: the linker and the archiver are the ones a build asks
1030        // about, and this compiler finds them on the path or under `-B` rather than shipping
1031        // them, so the name back is the honest answer unless a `-B` prefix holds one.
1032        Query::ProgName(name) => link
1033            .prefixes
1034            .iter()
1035            .map(|dir| dir.join(name))
1036            .find(|path| path.is_file())
1037            .map_or_else(|| name.clone(), |path| path.display().to_string()),
1038    }
1039}
1040
1041/// Renders the passes this level will run, in order, with what each one does.
1042///
1043/// The level is the whole of the answer unless a `-f` flag edited it, which is section 9.1 of
1044/// `spec/09-optimizer.md`: a level is a list somebody wrote down rather than something that
1045/// emerges from which flags happen to be set, and this is how that list is read.
1046#[must_use]
1047pub fn print_pipeline(opts: &Options) -> String {
1048    let mut settings = rucc_opt::Options::for_level(opts.opt_level);
1049    settings.toggles.clone_from(&opts.passes);
1050    settings.global_fuel = opts.pass_fuel_global;
1051    for (on, spec) in &opts.pass_gates {
1052        // Every spelling was checked while the arguments were parsed, so there is nothing here
1053        // this can refuse, and a listing is not the place to report it if there were.
1054        let _ = settings.gates.add(*on, spec);
1055    }
1056    rucc_opt::pipeline::print(&settings)
1057}
1058
1059/// Renders the resolved configuration.
1060///
1061/// One `key: value` per line, sorted by nothing in particular but fixed in order, because
1062/// this output is diffed across hosts in CI and a reordering would read as a change.
1063#[must_use]
1064pub fn print_config(opts: &Options) -> String {
1065    let sess = Session::new(opts.clone());
1066    let t = &sess.target;
1067    let mut out = String::new();
1068    let _ = writeln!(out, "version: {VERSION}");
1069    // The three field triple the driver was given rather than the ten field tuple it widens to,
1070    // because this output is what a build system reads to find out what it asked for. The tuple is
1071    // the compiler's model of the machine and this line is a receipt for a command line.
1072    let _ = writeln!(out, "target: {}", opts.target);
1073    let _ = writeln!(out, "arch: {}", opts.target.arch.as_str());
1074    let _ = writeln!(out, "os: {}", opts.target.os.as_str());
1075    let _ = writeln!(out, "env: {}", opts.target.env.as_str());
1076    let _ = writeln!(out, "object-format: {}", t.object_format.as_str());
1077    let _ = writeln!(out, "pointer-width: {}", t.pointer_width);
1078    let _ = writeln!(out, "long-width: {}", t.long_width);
1079    let _ = writeln!(out, "long-double-width: {}", t.long_double_width);
1080    let _ = writeln!(out, "endian: {}", if t.little_endian { "little" } else { "big" });
1081    let _ = writeln!(out, "char-signed: {}", t.char_is_signed);
1082    let _ = writeln!(out, "va-list: {}", t.va_list.map_or("none", |list| list.as_str()));
1083    // The register file as a count per class, which is enough to tell a target whose registers
1084    // are described from one whose are not without printing sixteen names nobody asked for.
1085    let regs: Vec<String> = t
1086        .regs
1087        .classes()
1088        .map(|(class, info)| format!("{} {}", info.name, t.regs.len(class)))
1089        .collect();
1090    let _ = writeln!(
1091        out,
1092        "registers: {}",
1093        if regs.is_empty() { "none".to_string() } else { regs.join(", ") }
1094    );
1095    let _ = writeln!(out, "opt-level: {}", sess.opts.opt_level);
1096    let _ = writeln!(out, "safety: {}", sess.opts.safety);
1097    let _ = writeln!(out, "emit: {}", sess.opts.emit.as_str());
1098    let _ = writeln!(out, "debug-info: {}", sess.opts.debug_info);
1099    let _ = writeln!(out, "frame-pointer: {}", sess.opts.frame_pointer);
1100    let _ = writeln!(out, "red-zone: {}", sess.opts.red_zone);
1101    let _ = writeln!(out, "stack-protector: {}", sess.opts.protector);
1102    // Last because it is the one key with more than one line under it, and the only one
1103    // whose value is a property of the machine rather than of the command line.
1104    for dir in sess.opts.search.dirs() {
1105        let system = if dir.is_system { " (system)" } else { "" };
1106        let _ = writeln!(out, "include: {}{system}", dir.path.display());
1107    }
1108    out
1109}
1110
1111/// The output name the make target is taken from, which is the `-o` argument or nothing.
1112///
1113/// A run that stops at the preprocessor has not named an object, whatever its `-o` says: under
1114/// `-E` that argument is the preprocessed text and under `-M` it is the rule itself, and neither
1115/// is a file `make` would rebuild by running this rule. GCC agrees and falls back to the source
1116/// name in both, which is why a `-MD -E -o out.i` writes `out.d` holding a rule for `a.o`. From
1117/// `-S` on the argument does name what the rule builds, and it is used as written.
1118fn deps_target_output<'a>(opts: &Options, plan: &'a Plan) -> Option<&'a str> {
1119    if opts.emit == EmitKind::Preprocessed { None } else { plan.output.as_deref() }
1120}
1121
1122/// Writes to a path the command line named rather than one the plan derived, where `-` is
1123/// standard output.
1124fn write_named(path: &str, bytes: &[u8]) -> Result<(), String> {
1125    if path == "-" {
1126        return write_out(&Output::Stdout, bytes);
1127    }
1128    write_out(&Output::File(path.to_owned()), bytes)
1129}
1130
1131/// Writes the make rule for one input, and reports whether it got there.
1132///
1133/// A rule with no file of its own goes where the compilation it replaced would have written,
1134/// which is what makes the usual makefile recipe work: `rucc -M $< -o $@` leaves the rule in
1135/// `$@`, and the same line with the `-o` left off puts it on standard output.
1136fn write_deps(
1137    opts: &Options,
1138    plan: &Plan,
1139    job: &Job,
1140    found: &[Dependency],
1141    stderr: &mut impl std::io::Write,
1142) -> bool {
1143    let targets = if opts.deps.targets.is_empty() {
1144        vec![deps::default_target(&job.input, deps_target_output(opts, plan))]
1145    } else {
1146        opts.deps.targets.clone()
1147    };
1148    let rule = deps::rule(&opts.deps, &targets, &job.input, found);
1149    // The file, on the other hand, is named after the `-o` in every mode that still has one to
1150    // spend, which is every mode except the two that spend it on the rule.
1151    let wrote = match deps::default_file(&opts.deps, &job.input, plan.output.as_deref()) {
1152        // A `-MF` on a run that had nowhere else to put the rule leaves the file the `-o`
1153        // named empty rather than absent, because a makefile that named it as a target of its
1154        // own is a makefile that will look for it.
1155        Some(path) => write_named(&path, rule.as_bytes()).and_then(|()| {
1156            if opts.deps.instead_of_compiling { write_out(&job.output, b"") } else { Ok(()) }
1157        }),
1158        None => write_out(&job.output, rule.as_bytes()),
1159    };
1160    if let Err(e) = wrote {
1161        let _ = writeln!(stderr, "rucc: error: {e}");
1162        return false;
1163    }
1164    true
1165}
1166
1167/// Runs phase 4 over every input that has one, and writes what came out.
1168///
1169/// One input that fails does not stop the others. A build that reports every file it could
1170/// not preprocess in one run is worth more than one that stops at the first, and the exit
1171/// status is still a failure either way.
1172fn preprocess_all(opts: &Options, plan: &Plan) -> i32 {
1173    let fs = OsFileSystem::new();
1174    let mut stderr = std::io::stderr().lock();
1175    let mut failed = false;
1176    for job in &plan.jobs {
1177        if !job.phases.first().is_some_and(|p| *p == Phase::Preprocess) {
1178            // An input that is already preprocessed, or an object file. GCC passes these
1179            // through untouched, and the plan has already said so in its notes.
1180            continue;
1181        }
1182        let started = std::time::Instant::now();
1183        let result = preprocess(opts, &job.input, &fs);
1184        if opts.time {
1185            say_time(&job.input, started.elapsed(), &mut stderr);
1186        }
1187        for message in &result.messages {
1188            let _ = writeln!(stderr, "{message}");
1189        }
1190        if result.failed() {
1191            failed = true;
1192            continue;
1193        }
1194        if opts.deps.emit {
1195            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1196            // `-M` and `-MM` asked for the rule instead of the text, so there is nothing else
1197            // to write. The other two asked for both and fall through to the text below.
1198            if opts.deps.instead_of_compiling {
1199                continue;
1200            }
1201        }
1202        if let Err(e) = write_out(&job.output, result.text.as_bytes()) {
1203            let _ = writeln!(stderr, "rucc: error: {e}");
1204            failed = true;
1205        }
1206    }
1207    i32::from(failed)
1208}
1209
1210/// Runs the front end over every input that has a compile phase, and writes what came out.
1211///
1212/// The same rule as [`preprocess_all`]: one input that fails does not stop the others, and the
1213/// exit status is a failure either way. An input that is already assembly or an object has no
1214/// compile phase and is passed over here, which the plan has already said in its notes.
1215fn compile_all(opts: &Options, plan: &Plan) -> i32 {
1216    let fs = OsFileSystem::new();
1217    let mut stderr = std::io::stderr().lock();
1218    let mut failed = false;
1219    let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1220    failed |= !ok;
1221    let mut fired = Fired::new();
1222    let mut pressure = Pressure::new();
1223    for job in &plan.jobs {
1224        if !job.phases.contains(&Phase::Compile) {
1225            continue;
1226        }
1227        // An input of IR is read back rather than compiled, since the C it came from is not
1228        // here any more. Everything after this is the same, so the two paths meet again at the
1229        // messages and the file the result is written to.
1230        let started = std::time::Instant::now();
1231        let result = if job.kind == InputKind::Ir {
1232            compile_ir(opts, &job.input, &fs)
1233        } else {
1234            compile(opts, &job.input, &fs)
1235        };
1236        if opts.time {
1237            say_time(&job.input, started.elapsed(), &mut stderr);
1238        }
1239        fired.merge(&result.fired);
1240        pressure.merge(&result.pressure);
1241        failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1242        failed |= !remarks.write(&result.remarks, &mut stderr);
1243        for message in &result.messages {
1244            let _ = writeln!(stderr, "{message}");
1245        }
1246        // Before the failure below, because a compilation that stopped in the back end is exactly
1247        // the one whose preprocessed source somebody wants to look at.
1248        failed |= !write_temps(job, &result.temps, &mut stderr);
1249        if result.failed() {
1250            failed = true;
1251            continue;
1252        }
1253        // `-MD` and `-MMD` write the rule beside the object and let the compilation happen, so
1254        // this is the one path where both files come out of the same run. An input of IR has no
1255        // dependencies to report and produces an empty list, which produces a rule naming only
1256        // itself, and that is the honest answer rather than a missing file.
1257        if opts.deps.emit {
1258            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1259        }
1260        if let Err(e) = write_out(&job.output, result.artifact.bytes()) {
1261            let _ = writeln!(stderr, "rucc: error: {e}");
1262            failed = true;
1263        }
1264    }
1265    failed |= !write_coverage(opts, &fired, &mut stderr);
1266    failed |= !write_pressure(opts, &pressure, &mut stderr);
1267    i32::from(failed)
1268}
1269
1270/// A directory for the object files only the link step ever sees, removed when it goes away.
1271///
1272/// `-c` writes its object where the user can see it and linking does not, which is the whole of
1273/// the difference: a `rucc a.c b.c` leaves an executable behind and nothing else, the same as
1274/// every other compiler. Removing them on drop rather than at the end of a function is so that a
1275/// link that failed leaves nothing behind either.
1276struct Scratch {
1277    /// Where the objects go.
1278    dir: PathBuf,
1279}
1280
1281impl Scratch {
1282    /// Makes one, under whatever the platform calls its temporary directory.
1283    ///
1284    /// The name carries the process id so that two compilers running at once do not share a
1285    /// directory, which they would otherwise do the moment two of them compiled a file of the
1286    /// same name.
1287    fn new() -> Result<Scratch, String> {
1288        let dir = std::env::temp_dir().join(format!("rucc-{}", std::process::id()));
1289        std::fs::create_dir_all(&dir).map_err(|e| format!("{}: {e}", dir.display()))?;
1290        Ok(Scratch { dir })
1291    }
1292}
1293
1294impl Drop for Scratch {
1295    fn drop(&mut self) {
1296        let _ = std::fs::remove_dir_all(&self.dir);
1297    }
1298}
1299
1300/// The link line the plan describes, for `-###`.
1301///
1302/// The names in it are the hints the plan carries rather than the temporaries a real compilation
1303/// would choose, because `-###` prints the line without having compiled anything and so has
1304/// nothing to point at. That also makes the printed line readable rather than naming a directory
1305/// that only exists while a compilation is running.
1306fn link_line(opts: &Options, link: &LinkOptions, job: &LinkJob) -> Result<String, link::Error> {
1307    let linker = link::find(opts.target, link)?;
1308    let args = link::line(opts.target, link, &job.inputs, &job.output)?;
1309    Ok(link::render(&linker, &args))
1310}
1311
1312/// Compiles everything, then links it.
1313///
1314/// The objects go in a directory that is removed afterwards, which is why this is not
1315/// [`compile_all`] followed by a link: the plan says an object feeding the linker is temporary
1316/// and does not say where, because where is a question that only has an answer once something is
1317/// running.
1318fn link_all(opts: &Options, plan: &Plan, link: &LinkOptions, verbose: bool) -> i32 {
1319    let Some(job) = &plan.link else {
1320        // Every path into here comes from a plan whose last phase is the link, and such a plan
1321        // has a link job. Saying so is cheaper than an unwrap that would have to be explained.
1322        let mut stderr = std::io::stderr().lock();
1323        let _ = writeln!(stderr, "rucc: error: there is nothing to link");
1324        return 1;
1325    };
1326    // Before anything is compiled, because a linker that is not on the machine is worth knowing
1327    // about in the second it takes to look rather than after the compilation.
1328    let linker = match link::find(opts.target, link) {
1329        Ok(linker) => linker,
1330        Err(why) => return complain(why),
1331    };
1332
1333    let scratch = match Scratch::new() {
1334        Ok(scratch) => scratch,
1335        Err(why) => return complain(format!("could not make a place for the object files: {why}")),
1336    };
1337
1338    let fs = OsFileSystem::new();
1339    let mut failed = false;
1340    // One per job, in job order, which is what lets the link line below be rebuilt with the real
1341    // paths in it: every job contributes exactly one file to the line and does so in this order.
1342    let mut produced: Vec<String> = Vec::with_capacity(plan.jobs.len());
1343    let mut fired = Fired::new();
1344    let mut pressure = Pressure::new();
1345    {
1346        let mut stderr = std::io::stderr().lock();
1347        let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1348        failed |= !ok;
1349        for (at, job) in plan.jobs.iter().enumerate() {
1350            let out = match &job.output {
1351                Output::Temporary(hint) => {
1352                    // The index because two inputs in different directories can have the same
1353                    // name, and the two objects of `rucc a/x.c b/x.c` must not be one file.
1354                    scratch.dir.join(format!("{at}-{hint}")).display().to_string()
1355                }
1356                Output::File(path) => path.clone(),
1357                // A job feeding the linker never writes to standard output, since the plan gives
1358                // it a temporary. This is here so that the match is total rather than a panic.
1359                Output::Stdout => continue,
1360            };
1361            produced.push(out.clone());
1362            if !job.phases.contains(&Phase::Compile) {
1363                continue;
1364            }
1365            let started = std::time::Instant::now();
1366            let result = if job.kind == InputKind::Ir {
1367                compile_ir(opts, &job.input, &fs)
1368            } else {
1369                compile(opts, &job.input, &fs)
1370            };
1371            if opts.time {
1372                say_time(&job.input, started.elapsed(), &mut stderr);
1373            }
1374            fired.merge(&result.fired);
1375            pressure.merge(&result.pressure);
1376            failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1377            failed |= !remarks.write(&result.remarks, &mut stderr);
1378            for message in &result.messages {
1379                let _ = writeln!(stderr, "{message}");
1380            }
1381            failed |= !write_temps(job, &result.temps, &mut stderr);
1382            if result.failed() {
1383                failed = true;
1384                continue;
1385            }
1386            // A `-MD` on a command line that links writes the rule next to the executable and
1387            // names the executable as its target, since that is the file this source builds
1388            // here. The object it went through is in a temporary directory and is gone by the
1389            // time `make` reads any of this.
1390            if opts.deps.emit {
1391                failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1392            }
1393            if !matches!(result.artifact, Artifact::Object(_)) {
1394                // Worth saying rather than writing whatever it is and letting the linker read it.
1395                // An empty file is a valid empty linker script, so a link handed one gets as far
1396                // as reporting every symbol of this file undefined, which is a page of messages
1397                // about something that went wrong here.
1398                let _ = writeln!(
1399                    stderr,
1400                    "rucc: internal error: {}: no object file was produced for the link",
1401                    job.input
1402                );
1403                failed = true;
1404                continue;
1405            }
1406            if let Err(e) = std::fs::write(&out, result.artifact.bytes()) {
1407                let _ = writeln!(stderr, "rucc: error: {out}: {e}");
1408                failed = true;
1409            }
1410        }
1411        failed |= !write_coverage(opts, &fired, &mut stderr);
1412        failed |= !write_pressure(opts, &pressure, &mut stderr);
1413    }
1414    if failed {
1415        // Nothing is linked from a compilation that did not finish. A linker run over the objects
1416        // that did compile would report every function of the file that did not as undefined,
1417        // which is a page of messages about a mistake already reported once.
1418        return 1;
1419    }
1420
1421    // The items in command line order with the temporaries filled in. A library contributes no
1422    // job and passes through, and every file item takes the next job's real output, which is
1423    // what keeps a library that was written between two objects between them here.
1424    let mut outputs = produced.into_iter();
1425    let mut items = Vec::with_capacity(job.inputs.len());
1426    for item in &job.inputs {
1427        match item {
1428            link::Item::Library(name) => items.push(link::Item::Library(name.clone())),
1429            link::Item::File(_) => match outputs.next() {
1430                Some(path) => items.push(link::Item::File(path)),
1431                None => return complain("the plan asks the linker for a file nothing produced"),
1432            },
1433        }
1434    }
1435
1436    let args = match link::line(opts.target, link, &items, &job.output) {
1437        Ok(args) => args,
1438        Err(why) => return complain(why),
1439    };
1440    if verbose {
1441        let mut stderr = std::io::stderr().lock();
1442        let _ = writeln!(stderr, "{}", link::render(&linker, &args));
1443    }
1444    let started = std::time::Instant::now();
1445    let ran = link::run(&linker, &args);
1446    if opts.time {
1447        // The one step of a compilation that really is another program, so this line is the same
1448        // measurement gcc's is and names the linker the way gcc names `collect2`.
1449        let mut stderr = std::io::stderr().lock();
1450        say_time(&linker.name, started.elapsed(), &mut stderr);
1451    }
1452    match ran {
1453        Ok(()) => 0,
1454        // The linker has already said what was wrong on its own error output, and repeating that
1455        // linking failed would only push its message further up the screen.
1456        Err(link::Error::Refused { .. }) => 1,
1457        Err(why) => complain(why),
1458    }
1459}
1460
1461/// Prints one driver level message and gives back the exit status that goes with it.
1462fn complain(why: impl std::fmt::Display) -> i32 {
1463    let mut stderr = std::io::stderr().lock();
1464    let _ = writeln!(stderr, "rucc: error: {why}");
1465    1
1466}
1467
1468/// Writes what `-Zrule-coverage=FILE` asked for, and says whether it could.
1469///
1470/// Once for the whole command line rather than once per input, because the question is which
1471/// lowering rules this run of the compiler reached and a file per input would leave the reader
1472/// unioning files to find out something one process already knew.
1473///
1474/// A file that could not be written is a failure and not a warning. What asks for this is a
1475/// measurement run, and a measurement that quietly did not happen is worse than one that stopped.
1476fn write_coverage(opts: &Options, fired: &Fired, stderr: &mut impl std::io::Write) -> bool {
1477    let Some(path) = &opts.rule_coverage else { return true };
1478    let Some(table) = coverage::table(opts.target.arch) else {
1479        let _ = writeln!(
1480            stderr,
1481            "rucc: error: there are no lowering rules for {} yet, so there is no coverage of them \
1482             to report",
1483            opts.target
1484        );
1485        return false;
1486    };
1487    match std::fs::write(path, fired.listing(table)) {
1488        Ok(()) => true,
1489        Err(e) => {
1490            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1491            false
1492        }
1493    }
1494}
1495
1496/// Writes what `-Zregister-pressure=FILE` asked for, and says whether it could.
1497///
1498/// Once for the whole command line, for the reason [`write_coverage`] gives, and a file that could
1499/// not be written is a failure for the reason it gives too. There is no equivalent of the missing
1500/// rule table here, since every target this compiles for has an allocator, and a run that reached
1501/// no back end at all writes an empty listing rather than nothing: a measurement of a build that
1502/// produced no code is still an answer and it is the honest one.
1503fn write_pressure(opts: &Options, pressure: &Pressure, stderr: &mut impl std::io::Write) -> bool {
1504    let Some(path) = &opts.register_pressure else { return true };
1505    match std::fs::write(path, pressure.listing()) {
1506        Ok(()) => true,
1507        Err(e) => {
1508            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1509            false
1510        }
1511    }
1512}
1513
1514/// Where the `-fopt-info` remarks go, and how much of the run has already gone there.
1515///
1516/// Standard error by default, and one file for the whole run when `-fopt-info=<file>` named one.
1517/// A file rather than the diagnostic stream is what a harness wants: the corpus in
1518/// `tamnd/rucc-corpus` matches a rejection against what the compiler said on standard error, and
1519/// a few thousand remarks mixed into that would bury it.
1520struct Remarks {
1521    /// The file, if there is one.
1522    file: Option<String>,
1523    /// Whether anything has been written to it yet, which decides between truncating and
1524    /// appending. One file holds the whole run rather than the last input in it.
1525    started: bool,
1526}
1527
1528impl Remarks {
1529    /// Prepares the destination, emptying the file if there is one.
1530    ///
1531    /// Emptied here rather than at the first remark, because a run where no pass had anything to
1532    /// say should leave an empty file and not yesterday's. An absent file and an empty one are
1533    /// different facts and something reading this will act on the difference.
1534    fn new(file: Option<&String>, stderr: &mut impl std::io::Write) -> (Self, bool) {
1535        let mut ok = true;
1536        if let Some(path) = file {
1537            if let Err(e) = std::fs::write(path, "") {
1538                let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1539                ok = false;
1540            }
1541        }
1542        (Self { file: file.cloned(), started: false }, ok)
1543    }
1544
1545    /// Writes one input's remarks, and says whether that worked.
1546    ///
1547    /// A file that cannot be written is a failure and not a warning, for the reason
1548    /// [`write_dumps`] gives: remarks that quietly did not arrive look exactly like a compilation
1549    /// where nothing happened.
1550    fn write(&mut self, text: &str, stderr: &mut impl std::io::Write) -> bool {
1551        if text.is_empty() {
1552            return true;
1553        }
1554        let Some(path) = &self.file else {
1555            let _ = write!(stderr, "{text}");
1556            return true;
1557        };
1558        let opened = std::fs::OpenOptions::new()
1559            .write(true)
1560            .append(self.started)
1561            .truncate(!self.started)
1562            .create(true)
1563            .open(path);
1564        self.started = true;
1565        let result =
1566            opened.and_then(|mut file| std::io::Write::write_all(&mut file, text.as_bytes()));
1567        if let Err(e) = result {
1568            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1569            return false;
1570        }
1571        true
1572    }
1573}
1574
1575/// Writes what `-fdump-ir=` asked to see, one file per dump.
1576///
1577/// The name is the input file with the dump's own name and `.ir` after it, so a directory listing
1578/// after a run is the passes in the order they ran, per input. They go in the working directory
1579/// rather than beside the output, because a dump is something a person asked for at a prompt and
1580/// the working directory is where that person is.
1581///
1582/// A file that could not be written is a failure and not a warning, for the reason
1583/// [`write_coverage`] gives: what asked for this is somebody debugging a pass, and a dump that
1584/// quietly did not happen looks exactly like a pass that did not run.
1585fn write_dumps(input: &str, dumps: &[rucc_opt::Dump], stderr: &mut impl std::io::Write) -> bool {
1586    let stem = std::path::Path::new(input)
1587        .file_name()
1588        .map_or_else(|| input.to_owned(), |name| name.to_string_lossy().into_owned());
1589    let mut ok = true;
1590    for dump in dumps {
1591        let path = format!("{stem}.{}.ir", dump.name);
1592        if let Err(e) = std::fs::write(&path, &dump.text) {
1593            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1594            ok = false;
1595        }
1596    }
1597    ok
1598}
1599
1600/// Writes the files `-save-temps` kept, which is nothing at all unless it was given.
1601///
1602/// A file that could not be written is a failure rather than a warning, for the reason
1603/// [`write_dumps`] gives: somebody asked for these by name, and one that quietly did not happen
1604/// looks like a compilation that never went through that step.
1605fn write_temps(job: &Job, temps: &Temps, stderr: &mut impl std::io::Write) -> bool {
1606    let mut ok = true;
1607    let kept = [(job.saved_text(), &temps.preprocessed), (job.saved_asm(), &temps.assembly)];
1608    for (path, text) in kept {
1609        // A step the compilation did not reach has nothing to keep, and a job that is not keeping
1610        // that step has nowhere to put it. Either way there is no file here.
1611        let (Some(path), Some(text)) = (path, text) else { continue };
1612        if let Err(e) = std::fs::write(&path, text) {
1613            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1614            ok = false;
1615        }
1616    }
1617    ok
1618}
1619
1620/// One line of `-time`, which is what a step was called and how long it took.
1621///
1622/// GCC's two numbers are the user and the system time of a subprocess it ran. This compiler runs
1623/// no subprocess for anything but the link, so what is measured here is the wall clock of the
1624/// step and the second column is always zero. The shape of the line is kept because a person
1625/// reading it next to gcc's should not have to work out which column is which.
1626fn say_time(name: &str, took: std::time::Duration, stderr: &mut impl std::io::Write) {
1627    let _ = writeln!(stderr, "# {name} {:.2} {:.2}", took.as_secs_f64(), 0.0);
1628}
1629
1630/// Writes one job's result where the plan said it goes.
1631///
1632/// # Errors
1633///
1634/// Returns the message to print, which names the file when there is one, because "permission
1635/// denied" on its own does not say which file was refused.
1636fn write_out(output: &Output, bytes: &[u8]) -> Result<(), String> {
1637    match output {
1638        Output::Stdout => {
1639            let mut stdout = std::io::stdout().lock();
1640            stdout.write_all(bytes).map_err(|e| format!("writing to standard output: {e}"))
1641        }
1642        Output::File(path) | Output::Temporary(path) => {
1643            std::fs::write(path, bytes).map_err(|e| format!("{path}: {e}"))
1644        }
1645    }
1646}
1647
1648/// Runs the driver and returns the process exit code.
1649///
1650/// `args` excludes the program name. Output goes to `stdout` and errors to `stderr`, which
1651/// is the one place in the compiler that is true.
1652pub fn run(args: &[String]) -> i32 {
1653    match parse_args(args) {
1654        Ok(Action::Help) => {
1655            print!("{USAGE}");
1656            0
1657        }
1658        Ok(Action::Version) => {
1659            println!("rucc {VERSION}");
1660            0
1661        }
1662        Ok(Action::Print(line)) => {
1663            println!("{line}");
1664            0
1665        }
1666        Ok(Action::PrintConfig(opts)) => {
1667            print!("{}", print_config(&opts));
1668            0
1669        }
1670        Ok(Action::PrintPipeline(opts)) => {
1671            print!("{}", print_pipeline(&opts));
1672            0
1673        }
1674        Ok(Action::PrintPlan { opts, plan, link }) => {
1675            print!("{}", plan.render());
1676            // The line as it would be typed, which is the half of `-###` that section 4.3 says
1677            // arrives with the link. It is printed even when the linker is not on this machine,
1678            // because what a build wants from `-###` is what the compiler would do.
1679            if let Some(job) = &plan.link {
1680                match link_line(&opts, &link, job) {
1681                    Ok(line) => println!("{line}"),
1682                    Err(why) => {
1683                        let mut stderr = std::io::stderr().lock();
1684                        let _ = writeln!(stderr, "rucc: error: {why}");
1685                        return 1;
1686                    }
1687                }
1688            }
1689            0
1690        }
1691        Ok(Action::Compile { opts, plan, link, jobs, verbose }) => {
1692            {
1693                let mut stderr = std::io::stderr().lock();
1694                if verbose {
1695                    let _ = write!(stderr, "{}", plan.render());
1696                    let _ = writeln!(stderr, "workers: {}", jobs.count());
1697                }
1698            }
1699            if opts.emit == EmitKind::Preprocessed {
1700                return preprocess_all(&opts, &plan);
1701            }
1702            if opts.emit != EmitKind::Executable {
1703                return compile_all(&opts, &plan);
1704            }
1705            link_all(&opts, &plan, &link, verbose)
1706        }
1707        Err(e) => {
1708            let mut stderr = std::io::stderr().lock();
1709            let _ = writeln!(stderr, "rucc: error: {e}");
1710            let _ = writeln!(stderr, "rucc: note: run `rucc --help` for usage");
1711            1
1712        }
1713    }
1714}
1715
1716#[cfg(test)]
1717mod tests {
1718    use rucc_session::{GnucVersion, IncludeForm, OptLevel, Visibility};
1719
1720    use super::*;
1721
1722    fn args(s: &[&str]) -> Vec<String> {
1723        s.iter().map(|x| (*x).to_owned()).collect()
1724    }
1725
1726    #[test]
1727    fn help_and_version_win_over_everything_else() {
1728        assert_eq!(parse_args(&args(&["-c", "--help", "x.c"])).unwrap(), Action::Help);
1729        assert_eq!(parse_args(&args(&["--version"])).unwrap(), Action::Version);
1730    }
1731
1732    fn compile(s: &[&str]) -> (Box<Options>, Box<Plan>) {
1733        match parse_args(&args(s)).expect("expected a compilation") {
1734            Action::Compile { opts, plan, .. } => (opts, plan),
1735            other => panic!("expected a compilation, got {other:?}"),
1736        }
1737    }
1738
1739    fn linking(s: &[&str]) -> (Box<LinkOptions>, Box<Plan>) {
1740        match parse_args(&args(s)).expect("expected a compilation") {
1741            Action::Compile { link, plan, .. } => (link, plan),
1742            other => panic!("expected a compilation, got {other:?}"),
1743        }
1744    }
1745
1746    #[test]
1747    fn collects_inputs_and_flags() {
1748        let (opts, plan) = compile(&["-c", "-O2", "-g", "a.c", "b.c"]);
1749        let paths: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
1750        assert_eq!(paths, vec!["a.c", "b.c"]);
1751        assert_eq!(opts.opt_level, OptLevel::O2);
1752        assert_eq!(opts.emit, EmitKind::Object);
1753        assert!(opts.debug_info);
1754    }
1755
1756    /// The unstable options, which are spelled apart from everything else on purpose: what is
1757    /// under `-Z` promises nothing, and a build that reaches for one should have had to say so.
1758    #[test]
1759    fn an_unstable_option_is_taken_and_one_that_does_not_exist_is_refused() {
1760        let (opts, _) = compile(&["-c", "-Zrule-coverage=/tmp/rules.cov", "a.c"]);
1761        assert_eq!(opts.rule_coverage.as_deref(), Some("/tmp/rules.cov"));
1762
1763        let (plain, _) = compile(&["-c", "a.c"]);
1764        assert_eq!(plain.rule_coverage, None, "nothing is measured unless it was asked for");
1765
1766        assert!(parse_args(&args(&["-Zrule-coverage=", "a.c"])).is_err(), "a file with no name");
1767        let unknown = parse_args(&args(&["-Zwhat", "a.c"])).expect_err("there is no such option");
1768        assert!(unknown.message.contains("4.11"), "{}", unknown.message);
1769    }
1770
1771    /// The other measurement written to a file, which reads the same way and fails the same way.
1772    #[test]
1773    fn where_the_register_pressure_goes_is_asked_for_the_same_way() {
1774        let (opts, _) = compile(&["-c", "-O2", "-Zregister-pressure=/tmp/spills.txt", "a.c"]);
1775        assert_eq!(opts.register_pressure.as_deref(), Some("/tmp/spills.txt"));
1776
1777        let (plain, _) = compile(&["-c", "a.c"]);
1778        assert_eq!(plain.register_pressure, None, "nothing is measured unless it was asked for");
1779
1780        assert!(parse_args(&args(&["-Zregister-pressure=", "a.c"])).is_err(), "no file named");
1781    }
1782
1783    #[test]
1784    fn a_bare_dash_o_means_o1_the_way_gcc_reads_it() {
1785        let (opts, _) = compile(&["-O", "a.c"]);
1786        assert_eq!(opts.opt_level, OptLevel::O1);
1787    }
1788
1789    #[test]
1790    fn dash_x_applies_to_later_inputs_only_and_none_stops_it() {
1791        let (_, plan) = compile(&["a.o", "-x", "c", "b.txt", "-x", "none", "c.o"]);
1792        assert_eq!(plan.jobs[0].kind, InputKind::LinkerInput);
1793        assert_eq!(plan.jobs[1].kind, InputKind::C);
1794        assert_eq!(plan.jobs[2].kind, InputKind::LinkerInput);
1795    }
1796
1797    #[test]
1798    fn dash_j_reaches_the_scheduler_and_defaults_to_the_machine() {
1799        let (_, _, jobs) = match parse_args(&args(&["-j4", "a.c"])).unwrap() {
1800            Action::Compile { opts, plan, jobs, .. } => (opts, plan, jobs),
1801            other => panic!("expected a compilation, got {other:?}"),
1802        };
1803        assert_eq!(jobs.count(), 4);
1804
1805        let default = match parse_args(&args(&["a.c"])).unwrap() {
1806            Action::Compile { jobs, .. } => jobs,
1807            other => panic!("expected a compilation, got {other:?}"),
1808        };
1809        assert_eq!(default, Jobs::available());
1810        assert!(parse_args(&args(&["-j0", "a.c"])).is_err());
1811    }
1812
1813    #[test]
1814    fn triple_hash_prints_the_plan_and_runs_nothing() {
1815        let a = parse_args(&args(&["-###", "-c", "a.c"])).unwrap();
1816        let Action::PrintPlan { plan, .. } = a else { panic!("expected a plan dump") };
1817        assert!(plan.render().contains("a.c: preprocess, compile, assemble -> a.o"));
1818    }
1819
1820    #[test]
1821    fn the_flag_that_keeps_the_intermediate_files_has_three_spellings_and_two_meanings() {
1822        // The bare one is `=obj` and not `=cwd`. gcc's manual says the opposite and gcc 16 does
1823        // this, and following the compiler is what makes a build that reads either of them find
1824        // the files where they are.
1825        assert_eq!(compile(&["-c", "-save-temps", "a.c"]).0.save_temps, SaveTemps::Object);
1826        assert_eq!(compile(&["-c", "-save-temps=obj", "a.c"]).0.save_temps, SaveTemps::Object);
1827        assert_eq!(compile(&["-c", "-save-temps=cwd", "a.c"]).0.save_temps, SaveTemps::Cwd);
1828        assert_eq!(compile(&["-c", "a.c"]).0.save_temps, SaveTemps::No);
1829        // The last one on the line decides, the way it does for every other flag with an
1830        // argument, and a keyword that is neither is fatal rather than ignored: a run that kept
1831        // nothing and said nothing looks exactly like one where the files were not produced.
1832        let (opts, _) = compile(&["-c", "-save-temps", "-save-temps=cwd", "a.c"]);
1833        assert_eq!(opts.save_temps, SaveTemps::Cwd);
1834        let e = parse_args(&args(&["-c", "-save-temps=nowhere", "a.c"])).unwrap_err();
1835        assert!(e.message.contains("accepted: cwd, obj"), "{}", e.message);
1836    }
1837
1838    #[test]
1839    fn the_flag_that_times_each_step_reaches_the_options_and_changes_nothing_else() {
1840        let (opts, plan) = compile(&["-c", "-time", "a.c"]);
1841        let (plain, without) = compile(&["-c", "a.c"]);
1842        assert!(opts.time);
1843        assert!(!plain.time);
1844        // Against the same line without the flag rather than against a spelling of the object's
1845        // name, since what the object is called is the host's business and this is not about that.
1846        assert_eq!(plan.jobs[0].output, without.jobs[0].output);
1847    }
1848
1849    #[test]
1850    fn dash_x_names_what_it_accepts_when_it_does_not_know_a_language() {
1851        let e = parse_args(&args(&["-x", "fortran", "a.c"])).unwrap_err();
1852        assert!(e.message.contains("assembler-with-cpp"), "{}", e.message);
1853    }
1854
1855    #[test]
1856    fn an_unknown_flag_is_an_error_rather_than_a_shrug() {
1857        let e = parse_args(&args(&["-fno-such-thing", "a.c"])).unwrap_err();
1858        assert!(e.message.contains("unknown option"), "{}", e.message);
1859    }
1860
1861    /// `-fpermissive` and the flag that turns it back off, which a build writes beside it when
1862    /// one directory needs the older rules and the rest of the tree does not.
1863    #[test]
1864    fn permissive_reads_in_both_directions_and_the_last_one_wins() {
1865        let (opts, _) = compile(&["-c", "a.c"]);
1866        assert!(!opts.permissive, "off unless it is asked for");
1867
1868        let (opts, _) = compile(&["-c", "-fpermissive", "a.c"]);
1869        assert!(opts.permissive);
1870
1871        let (opts, _) = compile(&["-c", "-fpermissive", "-fno-permissive", "a.c"]);
1872        assert!(!opts.permissive);
1873    }
1874
1875    #[test]
1876    fn asking_for_nested_functions_is_told_why_it_is_not_coming() {
1877        let e = parse_args(&args(&["-fnested-functions", "a.c"])).unwrap_err();
1878        assert!(e.message.contains("trampoline"), "{}", e.message);
1879        assert!(parse_args(&args(&["-fno-nested-functions", "a.c"])).is_ok());
1880    }
1881
1882    #[test]
1883    fn the_flag_every_configure_script_writes_is_taken() {
1884        // All four spellings, because a build writes whichever one its macros picked and a
1885        // compiler that takes three of them is a compiler that fails on the fourth.
1886        for flag in ["-fPIC", "-fpic", "-fPIE", "-fpie"] {
1887            let (opts, _) = compile(&["-c", flag, "a.c"]);
1888            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1889        }
1890    }
1891
1892    #[test]
1893    fn a_table_is_written_unless_the_build_says_nothing_will_walk_it() {
1894        let (opts, _) = compile(&["-c", "a.c"]);
1895        assert!(opts.unwinds(), "the default is off");
1896        let (opts, _) = compile(&["-c", "-fno-asynchronous-unwind-tables", "a.c"]);
1897        assert!(!opts.unwinds(), "the build was not taken at its word");
1898        let (opts, _) = compile(&[
1899            "-c",
1900            "-fno-asynchronous-unwind-tables",
1901            "-fasynchronous-unwind-tables",
1902            "a.c",
1903        ]);
1904        assert!(opts.unwinds(), "the last flag did not win");
1905        // The weaker request, which the same table answers, so a line that asks for a table and
1906        // against an asynchronous one gets one. That is gcc's arrangement and it turns up when a
1907        // build turns the asynchronous one off globally and a directory asks for a table back.
1908        let (opts, _) =
1909            compile(&["-c", "-fno-asynchronous-unwind-tables", "-funwind-tables", "a.c"]);
1910        assert!(opts.unwinds(), "the weaker request was dropped");
1911        let (opts, _) = compile(&["-c", "-fno-unwind-tables", "a.c"]);
1912        assert!(opts.unwinds(), "the weaker negative turned off the stronger request");
1913        let (opts, _) =
1914            compile(&["-c", "-fno-unwind-tables", "-fno-asynchronous-unwind-tables", "a.c"]);
1915        assert!(!opts.unwinds(), "both were turned off and one stayed on");
1916    }
1917
1918    #[test]
1919    fn the_flags_that_describe_what_this_compiler_already_does_are_taken() {
1920        // Every one of these is on a real build line somewhere and every one of them was an
1921        // unknown option. What they have in common is that the answer rucc gives is the answer
1922        // they ask for, so there is nothing to implement and nothing to refuse.
1923        for flag in [
1924            "-fno-common",
1925            "-fstrict-aliasing",
1926            "-fno-strict-aliasing",
1927            "-pipe",
1928            "-fdiagnostics-color",
1929            "-fno-diagnostics-color",
1930            "-fdiagnostics-color=always",
1931            "-fdiagnostics-color=never",
1932            "-fdiagnostics-color=auto",
1933        ] {
1934            let (opts, _) = compile(&["-c", flag, "a.c"]);
1935            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1936        }
1937    }
1938
1939    #[test]
1940    fn asking_the_linker_to_merge_tentative_definitions_is_told_why_it_is_not_coming() {
1941        // The one of that family that is a request rather than a description, and it is a real
1942        // difference: two files each writing `int g;` link under it and do not without it.
1943        let e = parse_args(&args(&["-fcommon", "a.c"])).unwrap_err();
1944        assert!(e.message.contains(".bss"), "{}", e.message);
1945        assert!(e.message.contains("extern"), "the way out is worth saying: {}", e.message);
1946    }
1947
1948    #[test]
1949    fn asking_for_position_dependent_code_is_told_why_it_is_not_coming() {
1950        for flag in ["-fno-pic", "-fno-pie"] {
1951            let e = parse_args(&args(&[flag, "a.c"])).unwrap_err();
1952            assert!(e.message.contains("global offset table"), "{flag}: {}", e.message);
1953            // The one it may have meant, since the two are a letter apart and one of them is
1954            // about linking and is taken.
1955            assert!(e.message.contains("-no-pie"), "{flag}: {}", e.message);
1956        }
1957    }
1958
1959    #[test]
1960    fn an_unsupported_target_names_itself() {
1961        let e = parse_args(&args(&["--target=sparc64-linux-gnu", "a.c"])).unwrap_err();
1962        assert!(e.message.contains("sparc64"), "{}", e.message);
1963    }
1964
1965    #[test]
1966    fn no_inputs_is_an_error_but_print_config_needs_none() {
1967        assert!(parse_args(&args(&[])).is_err());
1968        assert!(matches!(parse_args(&args(&["--print-config"])), Ok(Action::PrintConfig(_))));
1969    }
1970
1971    #[test]
1972    fn print_config_reports_the_target_it_was_given_not_the_host() {
1973        let a = parse_args(&args(&["--print-config", "--target=riscv64-linux-musl"])).unwrap();
1974        let Action::PrintConfig(opts) = a else { panic!("expected a configuration dump") };
1975        let text = print_config(&opts);
1976        assert!(text.contains("target: riscv64-unknown-linux-musl"), "{text}");
1977        assert!(text.contains("char-signed: false"), "{text}");
1978        assert!(text.contains("object-format: elf"), "{text}");
1979        assert!(text.contains("va-list: void-pointer"), "{text}");
1980        // RISC-V has a register file and this compiler has not written it down yet, and the
1981        // dump says which of those two it is rather than leaving the line out.
1982        assert!(text.contains("registers: none"), "{text}");
1983    }
1984
1985    #[test]
1986    fn print_config_has_one_key_per_line_and_a_fixed_order() {
1987        let opts = Options::new("x86_64-unknown-linux-gnu".parse().unwrap());
1988        let text = print_config(&opts);
1989        let keys: Vec<&str> =
1990            text.lines().map(|l| l.split(':').next().unwrap_or_default()).collect();
1991        assert_eq!(keys[0], "version");
1992        assert_eq!(keys[1], "target");
1993        assert_eq!(keys.len(), 20);
1994        assert!(text.ends_with('\n'));
1995    }
1996
1997    #[test]
1998    fn the_safety_tier_is_read_off_the_command_line_and_a_wrong_one_is_refused() {
1999        let (opts, _) = compile(&["a.c"]);
2000        assert_eq!(opts.safety, rucc_session::Safety::Off);
2001
2002        for (flag, tier) in [
2003            ("-fsafety=detect", rucc_session::Safety::Detect),
2004            ("-fsafety=enforce", rucc_session::Safety::Enforce),
2005            ("-fsafety=kernel", rucc_session::Safety::Kernel),
2006            ("-fsafety=off", rucc_session::Safety::Off),
2007        ] {
2008            let (opts, _) = compile(&[flag, "a.c"]);
2009            assert_eq!(opts.safety, tier, "{flag}");
2010        }
2011
2012        // The last one wins, the way every other repeated flag on this command line does.
2013        let (opts, _) = compile(&["-fsafety=enforce", "-fsafety=off", "a.c"]);
2014        assert_eq!(opts.safety, rucc_session::Safety::Off);
2015
2016        // A misspelled tier is refused rather than ignored. Silently compiling without the
2017        // monitor a build asked for is the one failure mode this feature cannot have.
2018        let e = parse_args(&args(&["-fsafety=on", "a.c"])).unwrap_err();
2019        assert!(e.message.contains("is not a safety tier"), "{}", e.message);
2020        assert!(parse_args(&args(&["-fsafety", "a.c"])).is_err());
2021    }
2022
2023    #[test]
2024    fn print_pipeline_answers_with_the_passes_the_level_asked_for() {
2025        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2026        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2027        let text = print_pipeline(&opts);
2028        assert!(text.starts_with("level: -O2\n"), "{text}");
2029        assert!(text.contains("fold"), "{text}");
2030
2031        let a = parse_args(&args(&["--print-pipeline"])).unwrap();
2032        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2033        // One pass runs at `-O0` and it is the one that removes code nothing reaches, which is
2034        // not an optimization. See issue 359.
2035        assert!(print_pipeline(&opts).contains("1: simplify-cfg,"), "{}", print_pipeline(&opts));
2036
2037        let a = parse_args(&args(&["--print-pipeline", "-fno-simplify-cfg"])).unwrap();
2038        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2039        // And with that one turned off there is nothing left, which the dump says rather than
2040        // printing an empty list.
2041        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2042    }
2043
2044    #[test]
2045    fn print_pipeline_takes_the_toggles_into_account() {
2046        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fno-fold"])).unwrap();
2047        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2048        let text = print_pipeline(&opts);
2049        // The one that was named is gone and the rest of the level is not, which is the whole
2050        // of what a toggle promises.
2051        assert!(!text.contains("fold"), "{text}");
2052        assert!(text.contains("dce"), "{text}");
2053
2054        // Every pass the compiler has, named off. Built from the registry rather than written
2055        // out, so a pass added later is turned off here too and this keeps testing the thing it
2056        // is about, which is that the toggles can empty a level.
2057        let mut off = vec!["--print-pipeline".to_owned(), "-O2".to_owned()];
2058        off.extend(rucc_opt::PASSES.iter().map(|p| format!("-fno-{}", p.name())));
2059        let spelled: Vec<&str> = off.iter().map(String::as_str).collect();
2060        let a = parse_args(&args(&spelled)).unwrap();
2061        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2062        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2063    }
2064
2065    #[test]
2066    fn print_pipeline_says_when_a_budget_will_stop_the_run_short() {
2067        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2068        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2069        assert!(!print_pipeline(&opts).contains("global fuel"));
2070
2071        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fpass-fuel-global=4"])).unwrap();
2072        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2073        let text = print_pipeline(&opts);
2074        // Because the listing is the answer to what this compilation will do, and a run that
2075        // stops after four rewrites is not doing what the level says it does.
2076        assert!(text.contains("global fuel: 4"), "{text}");
2077    }
2078
2079    /// A pass is turned on and off by its own name, and the order the flags were given in is
2080    /// kept, because the last spelling of a name is the one that decides.
2081    #[test]
2082    fn a_pass_is_named_by_dash_f_and_unnamed_by_dash_f_no() {
2083        let (opts, _) = compile(&["-c", "-O0", "-ffold", "-fno-fold", "-ffold", "a.c"]);
2084        assert_eq!(
2085            opts.passes,
2086            [("fold".to_owned(), true), ("fold".to_owned(), false), ("fold".to_owned(), true)]
2087        );
2088
2089        let e = parse_args(&args(&["-fno-such-pass", "a.c"])).unwrap_err();
2090        assert!(e.message.contains("unknown option"), "{}", e.message);
2091    }
2092
2093    #[test]
2094    fn pass_fuel_names_a_pass_and_a_count_and_refuses_anything_else() {
2095        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel=fold=3", "a.c"]);
2096        assert_eq!(opts.pass_fuel, [("fold".to_owned(), 3)]);
2097
2098        let e = parse_args(&args(&["-fpass-fuel=fold", "a.c"])).unwrap_err();
2099        assert!(e.message.contains("<pass>=<count>"), "{}", e.message);
2100        let e = parse_args(&args(&["-fpass-fuel=nosuch=3", "a.c"])).unwrap_err();
2101        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2102        let e = parse_args(&args(&["-fpass-fuel=fold=lots", "a.c"])).unwrap_err();
2103        assert!(e.message.contains("not a number"), "{}", e.message);
2104    }
2105
2106    #[test]
2107    fn global_pass_fuel_is_a_count_on_its_own_and_defaults_to_no_limit() {
2108        let (opts, _) = compile(&["-c", "-O2", "a.c"]);
2109        assert_eq!(opts.pass_fuel_global, None);
2110
2111        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel-global=12", "a.c"]);
2112        assert_eq!(opts.pass_fuel_global, Some(12));
2113        // And it is not the per pass flag with a longer name, so neither spelling swallows the
2114        // other.
2115        assert!(opts.pass_fuel.is_empty());
2116
2117        let e = parse_args(&args(&["-fpass-fuel-global=lots", "a.c"])).unwrap_err();
2118        assert!(e.message.contains("not a number"), "{}", e.message);
2119    }
2120
2121    #[test]
2122    fn a_gate_names_a_pass_and_optionally_the_functions_it_covers() {
2123        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold", "-fenable-fold=2-4,main", "a.c"]);
2124        assert_eq!(
2125            opts.pass_gates,
2126            [(false, "fold".to_owned()), (true, "fold=2-4,main".to_owned())],
2127            "the order is what decides, so it has to survive the parse"
2128        );
2129
2130        let e = parse_args(&args(&["-fdisable-nosuch", "a.c"])).unwrap_err();
2131        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2132        let e = parse_args(&args(&["-fenable-fold=9-2", "a.c"])).unwrap_err();
2133        assert!(e.message.contains("ends before it starts"), "{}", e.message);
2134        let e = parse_args(&args(&["-fdisable-fold=", "a.c"])).unwrap_err();
2135        assert!(e.message.contains("is empty"), "{}", e.message);
2136    }
2137
2138    #[test]
2139    fn the_pipeline_listing_says_which_passes_a_gate_touched() {
2140        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold=main", "a.c"]);
2141        let text = print_pipeline(&opts);
2142        assert!(text.contains("fold, "), "{text}");
2143        assert!(text.contains("[off for main]"), "{text}");
2144    }
2145
2146    /// The spelling is checked while the arguments are read, because a dump that names a pass
2147    /// this compiler does not have is a typo, and a typo found after the compilation has run is
2148    /// found too late to be any use.
2149    #[test]
2150    fn a_dump_is_checked_when_it_is_asked_for_rather_than_when_it_is_taken() {
2151        let (opts, _) = compile(&["-c", "-O2", "-fdump-ir=all", "-fdump-ir=after-fold", "a.c"]);
2152        assert_eq!(opts.dump_ir, ["all", "after-fold"]);
2153
2154        let e = parse_args(&args(&["-fdump-ir=after-nosuch", "a.c"])).unwrap_err();
2155        assert!(e.message.contains("nosuch"), "{}", e.message);
2156        assert!(parse_args(&args(&["-fdump-ir=sideways-fold", "a.c"])).is_err());
2157    }
2158
2159    /// Every spelling `-fopt-info` takes, and the one it does not.
2160    ///
2161    /// The keywords are checked here for the same reason a dump's pass name is: a person who
2162    /// misspelled one gets no output, and no output is also what a compilation where nothing
2163    /// happened looks like. Telling those two apart is the entire reason to reach for this flag.
2164    #[test]
2165    fn opt_info_takes_kinds_and_a_file_and_refuses_a_kind_it_does_not_have() {
2166        let (opts, _) = compile(&["-c", "-O2", "-fopt-info", "a.c"]);
2167        assert_eq!(opts.opt_info, [""], "a bare flag asks for the rewrites");
2168        assert_eq!(opts.opt_info_file, None, "and goes to standard error");
2169
2170        let (opts, _) = compile(&["-c", "-O2", "-fopt-info-missed-note", "a.c"]);
2171        assert_eq!(opts.opt_info, ["missed-note"]);
2172
2173        // Two flags add up rather than the second replacing the first, and the file is the last
2174        // one that named a file, which is how GCC treats both.
2175        let (opts, _) =
2176            compile(&["-c", "-O2", "-fopt-info-missed=one.txt", "-fopt-info-all=two.txt", "a.c"]);
2177        assert_eq!(opts.opt_info, ["missed", "all"]);
2178        assert_eq!(opts.opt_info_file.as_deref(), Some("two.txt"));
2179
2180        let e = parse_args(&args(&["-fopt-info-vectorized", "a.c"])).unwrap_err();
2181        assert!(e.message.contains("vectorized"), "{}", e.message);
2182        assert!(e.message.contains("`missed`"), "{}", e.message);
2183        let e = parse_args(&args(&["-fopt-info-missed=", "a.c"])).unwrap_err();
2184        assert!(e.message.contains("no file"), "{}", e.message);
2185    }
2186
2187    #[test]
2188    fn verify_each_is_unstable_and_off_unless_it_was_asked_for() {
2189        let (opts, _) = compile(&["-c", "-Zverify-each", "a.c"]);
2190        assert!(opts.verify_each);
2191        assert!(!USAGE.contains("verify-each"), "an unstable option stays out of the usage text");
2192    }
2193
2194    #[test]
2195    fn dash_o_needs_an_argument() {
2196        let e = parse_args(&args(&["a.c", "-o"])).unwrap_err();
2197        assert_eq!(e.message, "-o requires an argument");
2198    }
2199
2200    #[test]
2201    fn dash_d_and_dash_u_are_read_joined_or_separated_and_keep_their_order() {
2202        let (opts, _) = compile(&["-DFOO=1", "-D", "BAR", "-UBAZ", "-U", "QUX", "a.c"]);
2203        assert_eq!(opts.defines, ["FOO=1", "BAR"]);
2204        assert_eq!(opts.undefines, ["BAZ", "QUX"]);
2205    }
2206
2207    #[test]
2208    fn the_include_flags_land_on_the_chain_each_one_names() {
2209        // A sysroot with nothing under it, so that the library's own directories are the
2210        // same on every machine this test runs on, which is none of them.
2211        let (opts, _) = compile(&[
2212            "-Ii",
2213            "-iquote",
2214            "q",
2215            "-isystem",
2216            "sys",
2217            "-idirafter",
2218            "after",
2219            "--sysroot=/nowhere-at-all",
2220            "a.c",
2221        ]);
2222        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2223        // The compiler's own headers sit after every `-isystem` and before `-idirafter`,
2224        // which is where GCC puts its own: a directory the user named outranks ours.
2225        assert_eq!(dirs, ["q", "i", "sys", runtime::DIR, "after"]);
2226        assert!(!opts.search.dirs()[1].is_system);
2227        assert!(opts.search.dirs()[2].is_system);
2228    }
2229
2230    #[test]
2231    fn the_librarys_headers_come_after_the_compilers_own_and_go_away_with_them() {
2232        // Which machine this runs on decides what is on the path, so the test is about the
2233        // order rather than about the names: ours is on it, the library's follow it, and
2234        // `-nostdinc` is the one flag that takes both halves of the pair off at once.
2235        let (opts, _) = compile(&["a.c"]);
2236        let dirs = opts.search.dirs();
2237        let ours = dirs.iter().position(|d| d.path.to_str() == Some(runtime::DIR));
2238        assert_eq!(ours, Some(0), "{dirs:?}");
2239        assert!(dirs[1..].iter().all(|d| d.is_system), "{dirs:?}");
2240        let (bare, _) = compile(&["-nostdinc", "a.c"]);
2241        assert!(bare.search.dirs().is_empty(), "{:?}", bare.search.dirs());
2242    }
2243
2244    #[test]
2245    fn a_sysroot_moves_the_librarys_directories_and_nothing_else() {
2246        let (opts, _) = compile(&["-isystem", "sys", "--sysroot=/nowhere-at-all", "a.c"]);
2247        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2248        assert_eq!(dirs, ["sys", runtime::DIR]);
2249    }
2250
2251    #[test]
2252    fn dash_i_dash_moves_the_bracket_directories_into_the_quoted_chain() {
2253        let (opts, _) =
2254            compile(&["-Iinc1", "-iquote", "inc2", "-I-", "-Iinc3", "-nostdinc", "a.c"]);
2255        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2256        assert_eq!(dirs, ["inc1", "inc2", "inc3"]);
2257        // An angled include sees only what came after the flag.
2258        assert_eq!(opts.search.start(IncludeForm::Angled), 2);
2259        assert!(!opts.search.searches_current_dir());
2260    }
2261
2262    #[test]
2263    fn the_prefix_flags_stick_what_iprefix_said_on_the_front_of_what_follows_it() {
2264        let (opts, _) = compile(&[
2265            "-iprefix",
2266            "/tools/",
2267            "-iwithprefix",
2268            "late",
2269            "-iwithprefixbefore",
2270            "early",
2271            "-iprefix",
2272            "/other/",
2273            "-iwithprefix",
2274            "last",
2275            "-nostdinc",
2276            "a.c",
2277        ]);
2278        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2279        // `-iwithprefixbefore` is an `-I` and the other two are `-isystem`, which is where GCC
2280        // puts them rather than where its manual says it does.
2281        assert_eq!(dirs, ["/tools/early", "/tools/late", "/other/last"]);
2282        assert!(!opts.search.dirs()[0].is_system);
2283        assert!(opts.search.dirs()[1].is_system);
2284    }
2285
2286    #[test]
2287    fn the_files_named_on_the_command_line_keep_their_order_and_which_flag_named_them() {
2288        let (opts, _) =
2289            compile(&["-include", "one.h", "-imacros", "two.h", "-include", "3.h", "a.c"]);
2290        let names: Vec<&str> = opts.preincludes.iter().map(|p| p.name.as_str()).collect();
2291        assert_eq!(names, ["one.h", "two.h", "3.h"]);
2292        assert_eq!(opts.preincludes.iter().filter(|p| p.macros_only).count(), 1);
2293    }
2294
2295    #[test]
2296    fn nostdinc_takes_the_compilers_own_headers_off_the_path() {
2297        let (opts, _) = compile(&["-Ii", "-nostdinc", "a.c"]);
2298        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2299        assert_eq!(dirs, ["i"]);
2300    }
2301
2302    #[test]
2303    fn the_dialect_flags_set_the_language_and_the_extensions_separately() {
2304        let (opts, _) = compile(&["-std=gnu11", "a.c"]);
2305        assert_eq!(opts.std, Std::C11);
2306        assert!(opts.gnu_extensions);
2307
2308        let (opts, _) = compile(&["-std=iso9899:1999", "a.c"]);
2309        assert_eq!(opts.std, Std::C99);
2310        assert!(!opts.gnu_extensions);
2311
2312        let (opts, _) = compile(&["-ansi", "a.c"]);
2313        assert_eq!(opts.std, Std::C89);
2314        assert!(!opts.gnu_extensions);
2315
2316        let e = parse_args(&args(&["-std=c94jr", "a.c"])).unwrap_err();
2317        assert!(e.message.contains("unknown dialect"), "{}", e.message);
2318    }
2319
2320    #[test]
2321    fn the_dump_letters_are_a_family_and_everything_else_beginning_with_d_is_not() {
2322        let (opts, _) = compile(&["-dM", "a.c"]);
2323        assert!(opts.dumps.macros);
2324
2325        // Packed, the way GCC takes them, and a letter in the family we have not written yet
2326        // is accepted and does nothing rather than failing a build.
2327        let (opts, _) = compile(&["-dDM", "a.c"]);
2328        assert!(opts.dumps.macros);
2329        let (opts, _) = compile(&["-dD", "a.c"]);
2330        assert!(!opts.dumps.macros);
2331
2332        let (opts, _) = compile(&["a.c"]);
2333        assert!(!opts.dumps.any());
2334
2335        // `-dumpversion` is a different flag that happens to start the same way, and it is read
2336        // as itself rather than as a dump of nothing.
2337        assert_eq!(printed(&["-dumpversion", "a.c"]), VERSION);
2338    }
2339
2340    #[test]
2341    fn the_gcc_version_claimed_is_a_flag_and_the_short_spellings_are_the_ones_people_write() {
2342        let (opts, _) = compile(&["a.c"]);
2343        assert_eq!(
2344            opts.gnuc,
2345            GnucVersion { major: 7, minor: 0, patch: 0 },
2346            "the lowest claim a modern glibc gives its own declarations to"
2347        );
2348
2349        let (opts, _) = compile(&["-fgnuc-version=15.1.0", "a.c"]);
2350        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 1, patch: 0 });
2351
2352        // A missing component is zero. `gcc -dumpversion` says `15` on a release with no
2353        // patchlevel and a harness that pastes that back has to be understood.
2354        let (opts, _) = compile(&["-fgnuc-version=15", "a.c"]);
2355        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 0, patch: 0 });
2356
2357        let (opts, _) = compile(&["-fgnuc-version=13.2", "a.c"]);
2358        assert_eq!(opts.gnuc, GnucVersion { major: 13, minor: 2, patch: 0 });
2359
2360        let e = parse_args(&args(&["-fgnuc-version=15.x", "a.c"])).unwrap_err();
2361        assert!(e.message.contains("minor that is not a number"), "{}", e.message);
2362
2363        let e = parse_args(&args(&["-fgnuc-version=1.2.3.4", "a.c"])).unwrap_err();
2364        assert!(e.message.contains("more than three"), "{}", e.message);
2365    }
2366
2367    #[test]
2368    fn pedantic_has_two_spellings_and_is_not_the_same_knob_as_the_dialect() {
2369        let (opts, _) = compile(&["-std=c17", "-pedantic", "a.c"]);
2370        assert!(opts.pedantic);
2371        assert_eq!(opts.std, Std::C17);
2372
2373        // The `-W` family's name for it, which is what a build that groups its warning flags
2374        // tends to write.
2375        let (opts, _) = compile(&["-Wpedantic", "a.c"]);
2376        assert!(opts.pedantic);
2377
2378        let (opts, _) = compile(&["-std=c17", "a.c"]);
2379        assert!(!opts.pedantic, "a dialect on its own does not diagnose an extension");
2380    }
2381
2382    #[test]
2383    fn dash_p_and_dash_ffreestanding_reach_the_options() {
2384        let (opts, _) = compile(&["-E", "-P", "-ffreestanding", "a.c"]);
2385        assert!(!opts.line_markers);
2386        assert!(!opts.hosted);
2387        assert_eq!(opts.emit, EmitKind::Preprocessed);
2388    }
2389
2390    /// The two ways a build says it means its own function by a name the C library also has.
2391    ///
2392    /// `-fno-builtin` is all of them and `-fno-builtin-<name>` is one, and the second is what a
2393    /// build writes when it means its own `memcpy` and the library's everything else. The name is
2394    /// kept as it was written and not checked against anything, because a program is allowed to
2395    /// mean something by a name this compiler has never heard of.
2396    #[test]
2397    fn the_builtin_flags_are_read_in_both_directions_and_one_name_at_a_time() {
2398        let (opts, _) = compile(&["-c", "a.c"]);
2399        assert!(opts.builtins, "a library name means the library function by default");
2400        assert!(opts.no_builtin.is_empty());
2401
2402        let (opts, _) = compile(&["-c", "-fno-builtin", "a.c"]);
2403        assert!(!opts.builtins);
2404
2405        let (opts, _) = compile(&["-c", "-fno-builtin", "-fbuiltin", "a.c"]);
2406        assert!(opts.builtins, "the last mention decides");
2407
2408        let (opts, _) = compile(&["-c", "-fno-builtin-memcpy", "-fno-builtin-nonesuch", "a.c"]);
2409        assert!(opts.builtins, "one name is not the family");
2410        assert_eq!(opts.no_builtin, vec!["memcpy".to_owned(), "nonesuch".to_owned()]);
2411    }
2412
2413    /// `-fvisibility=`, which is on every cmake project that cares about which names it exports
2414    /// and which was refused as an unknown option until now.
2415    ///
2416    /// Four spellings and three answers. `internal` is hidden plus a promise about never taking
2417    /// the address across a component boundary, and nothing derives anything from that promise
2418    /// here, so it comes out as the weaker of the two rather than as a refusal that stops a build
2419    /// over a distinction this compiler does not make.
2420    #[test]
2421    fn visibility_takes_the_four_spellings_gcc_takes_and_refuses_the_rest() {
2422        let (opts, _) = compile(&["-c", "a.c"]);
2423        assert_eq!(opts.visibility, Visibility::Default, "exported unless something says not");
2424
2425        for (written, wanted) in [
2426            ("default", Visibility::Default),
2427            ("hidden", Visibility::Hidden),
2428            ("internal", Visibility::Hidden),
2429            ("protected", Visibility::Protected),
2430        ] {
2431            let (opts, _) = compile(&["-c", &format!("-fvisibility={written}"), "a.c"]);
2432            assert_eq!(opts.visibility, wanted, "{written}");
2433        }
2434
2435        // The last mention decides, which is what every other flag of this shape does and what a
2436        // build that turns something off for one directory relies on.
2437        let (opts, _) = compile(&["-c", "-fvisibility=hidden", "-fvisibility=default", "a.c"]);
2438        assert_eq!(opts.visibility, Visibility::Default, "the last mention decides");
2439
2440        // A spelling gcc does not take is refused rather than read as the default, because a
2441        // build that meant hidden and got exported is a library with the wrong interface and
2442        // nothing said about it anywhere.
2443        let failed = parse_args(&args(&["-fvisibility=none", "a.c"])).expect_err("refused");
2444        assert!(failed.to_string().contains("is not a visibility"), "{failed}");
2445    }
2446
2447    /// `-ffunction-sections` and `-fdata-sections`, which are what make `--gc-sections` able to
2448    /// drop anything: a linker can leave out a section nothing reaches and cannot leave out half of
2449    /// one. A kernel and an embedded image are both linked that way.
2450    ///
2451    /// Two flags rather than one because gcc has two, and a build that asks for one of them and not
2452    /// the other is a build that measured something: splitting the code is nearly free at link time
2453    /// and splitting the data can defeat the linker's ordering of what is next to what.
2454    #[test]
2455    fn a_section_per_function_and_a_section_per_variable_are_asked_for_one_at_a_time() {
2456        let (opts, _) = compile(&["-c", "a.c"]);
2457        assert!(!opts.function_sections, "one text section unless something says otherwise");
2458        assert!(!opts.data_sections);
2459
2460        let (opts, _) = compile(&["-c", "-ffunction-sections", "a.c"]);
2461        assert!(opts.function_sections);
2462        assert!(!opts.data_sections, "one flag is not the other");
2463
2464        let (opts, _) = compile(&["-c", "-fdata-sections", "a.c"]);
2465        assert!(opts.data_sections);
2466        assert!(!opts.function_sections);
2467
2468        // Both directions taken, and the off one is what happens anyway rather than a refusal,
2469        // since a build that writes it is asking for the default.
2470        let (opts, _) = compile(&[
2471            "-c",
2472            "-ffunction-sections",
2473            "-fno-function-sections",
2474            "-fdata-sections",
2475            "-fno-data-sections",
2476            "a.c",
2477        ]);
2478        assert!(!opts.function_sections, "the last mention decides");
2479        assert!(!opts.data_sections, "the last mention decides");
2480    }
2481
2482    /// `-fgnu89-inline`, which is off by default and is not implied by anything on the command
2483    /// line, since the dialect asks for GNU's reading further in rather than through this.
2484    #[test]
2485    fn gnu89_inline_is_off_until_it_is_asked_for_and_the_last_mention_decides() {
2486        let (opts, _) = compile(&["-c", "a.c"]);
2487        assert!(!opts.gnu89_inline, "C's reading of inline by default");
2488
2489        let (opts, _) = compile(&["-c", "-fgnu89-inline", "a.c"]);
2490        assert!(opts.gnu89_inline);
2491
2492        let (opts, _) = compile(&["-c", "-fgnu89-inline", "-fno-gnu89-inline", "a.c"]);
2493        assert!(!opts.gnu89_inline, "the last mention decides");
2494
2495        // The C89 dialects are under GNU's reading whether this was written or not, so the flag
2496        // stays off there and the dialect is what the checker and the macro set both ask. That is
2497        // also why `-std=c89 -fno-gnu89-inline` needs no diagnostic: it asks for the reading the
2498        // dialect already has. gcc refuses that command line, which is measured in the issue.
2499        let (opts, _) = compile(&["-c", "-std=c89", "a.c"]);
2500        assert!(!opts.gnu89_inline);
2501    }
2502
2503    /// Both spellings of both frame flags, since a build that wants one usually writes the
2504    /// other beside it for the one file that has to be compiled the ordinary way.
2505    #[test]
2506    fn the_two_frame_flags_are_read_in_both_directions() {
2507        let (opts, _) = compile(&["-c", "a.c"]);
2508        assert!(!opts.frame_pointer, "gcc omits it above -O0 and so does this");
2509        assert!(opts.red_zone, "the psABI has one and nothing said not to use it");
2510
2511        let (opts, _) = compile(&["-c", "-fno-omit-frame-pointer", "-mno-red-zone", "a.c"]);
2512        assert!(opts.frame_pointer);
2513        assert!(!opts.red_zone);
2514
2515        let (opts, _) = compile(&[
2516            "-c",
2517            "-fno-omit-frame-pointer",
2518            "-fomit-frame-pointer",
2519            "-mno-red-zone",
2520            "-mred-zone",
2521            "a.c",
2522        ]);
2523        assert!(!opts.frame_pointer, "the last one wins, as it does in gcc");
2524        assert!(opts.red_zone);
2525    }
2526
2527    /// Four flags rather than one with an argument, which is how gcc spells them, and the negative
2528    /// spelled three ways because a build that turns one off writes whichever it turned on.
2529    #[test]
2530    fn the_stack_protector_is_four_flags_and_the_last_one_wins() {
2531        let (opts, _) = compile(&["-c", "a.c"]);
2532        assert_eq!(opts.protector, Protector::None, "gcc protects nothing unless it was asked");
2533
2534        for (flag, want) in [
2535            ("-fstack-protector", Protector::Buffers),
2536            ("-fstack-protector-strong", Protector::Strong),
2537            ("-fstack-protector-all", Protector::All),
2538        ] {
2539            let (opts, _) = compile(&["-c", flag, "a.c"]);
2540            assert_eq!(opts.protector, want, "{flag}");
2541        }
2542
2543        // What a package build does: the strong one in the global flags and one directory that
2544        // cannot have a protector turning it off on the line after.
2545        for off in ["-fno-stack-protector", "-fno-stack-protector-strong"] {
2546            let (opts, _) = compile(&["-c", "-fstack-protector-strong", off, "a.c"]);
2547            assert_eq!(opts.protector, Protector::None, "{off}");
2548        }
2549        let (opts, _) = compile(&["-c", "-fno-stack-protector", "-fstack-protector-all", "a.c"]);
2550        assert_eq!(opts.protector, Protector::All, "the last one wins either way round");
2551    }
2552
2553    #[test]
2554    fn the_link_flags_are_collected_apart_from_the_compilation() {
2555        let (link, _) = linking(&[
2556            "-static",
2557            "-nostartfiles",
2558            "-rdynamic",
2559            "-s",
2560            "-fuse-ld=mold",
2561            "-L/opt/lib",
2562            "-B",
2563            "/opt/tools",
2564            "a.c",
2565        ]);
2566        assert!(link.is_static);
2567        assert!(link.no_startfiles);
2568        assert!(link.export_dynamic);
2569        assert!(link.strip);
2570        assert_eq!(link.use_ld.as_deref(), Some("mold"));
2571        assert_eq!(link.search, vec![PathBuf::from("/opt/lib")]);
2572        assert_eq!(link.prefixes, vec![PathBuf::from("/opt/tools")]);
2573    }
2574
2575    #[test]
2576    fn a_comma_in_dash_wl_separates_two_arguments() {
2577        let (link, _) = linking(&["-Wl,-rpath,/opt/lib", "-Xlinker", "--as-needed", "a.c"]);
2578        assert_eq!(link.passthrough, vec!["-rpath", "/opt/lib", "--as-needed"]);
2579    }
2580
2581    #[test]
2582    fn a_library_keeps_its_place_between_the_objects() {
2583        // Link order is semantic: `-lm` written between two files resolves for the one before
2584        // it and not for the one after, so a library cannot be collected into a list of its own.
2585        // The target is named because the suffix of an object is the target's and this asserts
2586        // on the names: the same command line on a Windows host plans two `.obj` files.
2587        let (_, plan) = linking(&["--target=x86_64-unknown-linux-gnu", "a.c", "-lm", "b.c"]);
2588        let link = plan.link.expect("expected a link step");
2589        assert_eq!(
2590            link.inputs,
2591            vec![
2592                link::Item::File("a.o".into()),
2593                link::Item::Library("m".into()),
2594                link::Item::File("b.o".into()),
2595            ]
2596        );
2597        // And it is not a job, because there is nothing to compile in a library.
2598        assert_eq!(plan.jobs.len(), 2);
2599    }
2600
2601    #[test]
2602    fn a_library_on_a_dash_c_line_is_a_note_rather_than_an_error() {
2603        let (_, plan) = linking(&["-c", "-lm", "a.c"]);
2604        assert!(plan.link.is_none());
2605        assert!(plan.notes.iter().any(|n| n.contains("-lm")), "{:?}", plan.notes);
2606    }
2607
2608    #[test]
2609    fn the_sysroot_reaches_the_linker_as_well_as_the_headers() {
2610        let (link, _) = linking(&["--sysroot=/opt/root", "a.c"]);
2611        assert_eq!(link.sysroot, Some(PathBuf::from("/opt/root")));
2612    }
2613
2614    fn printed(s: &[&str]) -> String {
2615        match parse_args(&args(s)).expect("expected an answer") {
2616            Action::Print(line) => line,
2617            other => panic!("expected an answer, got {other:?}"),
2618        }
2619    }
2620
2621    fn refused(s: &[&str]) -> String {
2622        parse_args(&args(s)).expect_err("expected a refusal").message
2623    }
2624
2625    #[test]
2626    fn a_warning_flag_this_compiler_has_not_heard_of_is_taken_rather_than_refused() {
2627        // The rule in section 4.1, and the reason for it is autoconf: a configure script finds
2628        // out whether a warning flag exists by passing it and looking at the exit status, so a
2629        // compiler that refuses one it does not know fails a script written for a newer GCC.
2630        let (opts, _) = compile(&["-Wall", "-Wextra", "-Wno-format-truncation", "-c", "a.c"]);
2631        assert!(!opts.warnings_are_errors);
2632        assert!(opts.warnings);
2633        // The two spellings that do mean something are still read.
2634        let (opts, _) = compile(&["-Werror", "-c", "a.c"]);
2635        assert!(opts.warnings_are_errors);
2636        let (opts, _) = compile(&["-w", "-c", "a.c"]);
2637        assert!(!opts.warnings);
2638        let (opts, _) = compile(&["-pedantic-errors", "-c", "a.c"]);
2639        assert!(opts.pedantic && opts.warnings_are_errors);
2640    }
2641
2642    #[test]
2643    fn an_argument_for_a_separate_tool_is_refused_rather_than_dropped() {
2644        // Every one of these says something about the output, so the wrong answer is silence.
2645        assert!(refused(&["-Wa,--noexecstack", "-c", "a.c"]).contains("separate assembler"));
2646        assert!(refused(&["-Wp,-DX", "-c", "a.c"]).contains("separate assembler"));
2647        assert!(refused(&["-specs=/x", "a.c"]).contains("-specs= is not supported"));
2648        assert!(refused(&["-mcmodel=kernel", "-c", "a.c"]).contains("small code model"));
2649        assert!(refused(&["-gdwarf-4", "-c", "a.c"]).contains("DWARF 5"));
2650        assert!(refused(&["-Ofast", "-c", "a.c"]).contains("fast math"));
2651        // The word size the target does not have, which is a target this compiler was not asked
2652        // for rather than a flag it does not know.
2653        let no32 = refused(&["--target=x86_64-unknown-linux-gnu", "-m32", "-c", "a.c"]);
2654        assert!(no32.contains("32 bit target"), "{no32}");
2655    }
2656
2657    #[test]
2658    fn the_levels_gcc_spells_differently_are_the_levels_they_mean() {
2659        assert_eq!(compile(&["-O", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2660        assert_eq!(compile(&["-Og", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2661        assert_eq!(compile(&["-O2", "-c", "a.c"]).0.opt_level, OptLevel::O2);
2662    }
2663
2664    #[test]
2665    fn the_machine_flags_that_name_what_we_already_do_are_taken_and_the_rest_are_not() {
2666        let line = ["--target=x86_64-unknown-linux-gnu", "-m64", "-march=x86-64-v3"];
2667        let (opts, _) =
2668            compile(&[&line[..], &["-mtune=native", "-mabi=sysv", "-c", "a.c"]].concat());
2669        assert_eq!(opts.target.to_string(), "x86_64-unknown-linux-gnu");
2670        let wrong = refused(&["--target=x86_64-unknown-linux-gnu", "-mabi=ms", "-c", "a.c"]);
2671        assert!(wrong.contains("sysv convention"), "{wrong}");
2672    }
2673
2674    #[test]
2675    fn the_thread_flag_is_a_macro_and_a_library_and_the_library_goes_last() {
2676        let (opts, plan) = compile(&["-pthread", "-c", "a.c"]);
2677        assert!(opts.defines.iter().any(|d| d == "_REENTRANT"));
2678        // After the input, because a static link takes what it needs from a library when it
2679        // reaches it and not afterwards.
2680        let names: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
2681        assert_eq!(names, vec!["a.c"]);
2682    }
2683
2684    #[test]
2685    fn the_questions_a_build_system_asks_before_it_compiles_anything() {
2686        let target = "--target=x86_64-unknown-linux-gnu";
2687        assert_eq!(printed(&[target, "-dumpmachine"]), "x86_64-unknown-linux-gnu");
2688        assert_eq!(printed(&[target, "-dumpversion"]), VERSION);
2689        assert_eq!(printed(&[target, "-dumpfullversion"]), VERSION);
2690        assert_eq!(printed(&[target, "-print-multiarch"]), "x86_64-linux-gnu");
2691        // A name nothing holds comes back unchanged, which is GCC's rule and is what makes the
2692        // answer safe to paste into a link line whether or not the file is there.
2693        assert_eq!(printed(&[target, "-print-file-name=no-such-library.a"]), "no-such-library.a");
2694        assert_eq!(printed(&[target, "-print-prog-name=ld"]), "ld");
2695        let dirs = printed(&[target, "-print-search-dirs"]);
2696        assert!(dirs.starts_with("install: "), "{dirs}");
2697        assert!(dirs.contains("\nlibraries: ="), "{dirs}");
2698    }
2699
2700    #[test]
2701    fn the_two_dependency_flags_that_stop_after_the_rule_stop_after_the_rule() {
2702        let (opts, _) = compile(&["-M", "a.c"]);
2703        assert!(opts.deps.emit && opts.deps.instead_of_compiling);
2704        assert!(opts.deps.system_headers, "plain -M lists them");
2705        assert_eq!(opts.emit, EmitKind::Preprocessed);
2706
2707        // Even where a later flag asked for something else, because the family is a mode and
2708        // the mode is what the run is for.
2709        let (opts, _) = compile(&["-M", "-c", "a.c"]);
2710        assert_eq!(opts.emit, EmitKind::Preprocessed);
2711
2712        let (opts, _) = compile(&["-MM", "a.c"]);
2713        assert!(!opts.deps.system_headers);
2714    }
2715
2716    #[test]
2717    fn the_two_that_end_in_d_leave_the_compilation_alone() {
2718        let (opts, _) = compile(&["-MD", "-c", "a.c"]);
2719        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2720        assert!(opts.deps.system_headers);
2721        assert_eq!(opts.emit, EmitKind::Object);
2722
2723        let (opts, _) = compile(&["-MMD", "-c", "a.c"]);
2724        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2725        assert!(!opts.deps.system_headers);
2726    }
2727
2728    #[test]
2729    fn nothing_puts_the_system_headers_back_once_a_flag_has_taken_them_out() {
2730        // GCC's rule, and not an oversight in it. The flag asking for fewer of them is read as
2731        // the answer, because the other one never asked the question.
2732        let (opts, _) = compile(&["-MM", "-M", "a.c"]);
2733        assert!(!opts.deps.system_headers);
2734        let (opts, _) = compile(&["-MD", "-MMD", "-c", "a.c"]);
2735        assert!(!opts.deps.system_headers);
2736        let (opts, _) = compile(&["-MMD", "-MD", "-c", "a.c"]);
2737        assert!(!opts.deps.system_headers);
2738    }
2739
2740    #[test]
2741    fn a_target_arrives_escaped_from_one_flag_and_untouched_from_the_other() {
2742        let (opts, _) = compile(&["-MM", "-MT", "a b.o", "-MQ", "a b.o", "a.c"]);
2743        assert_eq!(opts.deps.targets, vec!["a b.o".to_owned(), "a\\ b.o".to_owned()]);
2744    }
2745
2746    #[test]
2747    fn the_rest_of_the_family_is_a_file_and_a_switch() {
2748        let (opts, _) = compile(&["-MM", "-MF", "dep.d", "-MP", "a.c"]);
2749        assert_eq!(opts.deps.file.as_deref(), Some("dep.d"));
2750        assert!(opts.deps.phony);
2751
2752        for flag in ["-MF", "-MT", "-MQ"] {
2753            let e = parse_args(&args(&[flag])).unwrap_err();
2754            assert!(e.message.contains("requires an argument"), "{}", e.message);
2755        }
2756    }
2757
2758    /// A directory of sources for one test, removed when the test is done with it.
2759    struct TempTree(PathBuf);
2760
2761    impl Drop for TempTree {
2762        fn drop(&mut self) {
2763            let _ = std::fs::remove_dir_all(&self.0);
2764        }
2765    }
2766
2767    impl TempTree {
2768        fn new(name: &str, files: &[(&str, &str)]) -> TempTree {
2769            let dir = std::env::temp_dir().join(format!("rucc-deps-{}-{name}", std::process::id()));
2770            let _ = std::fs::remove_dir_all(&dir);
2771            std::fs::create_dir_all(&dir).expect("temporary directory should be writable");
2772            for (path, text) in files {
2773                let at = dir.join(path);
2774                if let Some(parent) = at.parent() {
2775                    std::fs::create_dir_all(parent).expect("creating a subdirectory should work");
2776                }
2777                std::fs::write(&at, text).expect("writing a temporary file should work");
2778            }
2779            TempTree(dir)
2780        }
2781
2782        fn path(&self, name: &str) -> String {
2783            self.0.join(name).to_string_lossy().into_owned()
2784        }
2785    }
2786
2787    #[test]
2788    fn the_rule_names_what_the_includes_found_and_names_each_of_them_once() {
2789        // End to end, because the list comes from the preprocessor and the format comes from
2790        // somewhere else, and a test of either half on its own would pass with the two of them
2791        // wired up backwards.
2792        let tree = TempTree::new(
2793            "found",
2794            &[
2795                ("a.c", "#include \"one.h\"\n#include \"two.h\"\nint main(void) { return X; }\n"),
2796                ("one.h", "#define X 0\n"),
2797                ("two.h", "#include \"one.h\"\n"),
2798            ],
2799        );
2800        let out = tree.path("dep.d");
2801        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2802        assert_eq!(code, 0);
2803
2804        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2805        let names: Vec<&str> = text.split_whitespace().collect();
2806        // The target, the source, and each header once however many times it was reached.
2807        assert_eq!(names.first(), Some(&"a.o:"), "{text}");
2808        assert_eq!(names.iter().filter(|n| n.ends_with("one.h")).count(), 1, "{text}");
2809        assert_eq!(names.iter().filter(|n| n.ends_with("two.h")).count(), 1, "{text}");
2810        // And the `-o` went to the file the rule replaced, which is left empty rather than
2811        // absent because a makefile that named it as a target will look for it.
2812        assert_eq!(std::fs::read(tree.path("a.i")).expect("the output should exist"), b"");
2813    }
2814
2815    #[test]
2816    fn a_header_that_is_only_reached_under_a_guard_is_still_a_dependency() {
2817        // The multiple-include optimization means the second reach never opens the file. It is
2818        // still a file this translation unit was built from, so it is still in the rule.
2819        let tree = TempTree::new(
2820            "guarded",
2821            &[
2822                ("a.c", "#include \"g.h\"\n#include \"g.h\"\nint main(void) { return 0; }\n"),
2823                ("g.h", "#ifndef G\n#define G\n#endif\n"),
2824            ],
2825        );
2826        let out = tree.path("dep.d");
2827        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2828        assert_eq!(code, 0);
2829        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2830        assert_eq!(text.split_whitespace().filter(|n| n.ends_with("g.h")).count(), 1, "{text}");
2831    }
2832
2833    #[test]
2834    fn every_imacros_file_is_read_before_every_include_file_whatever_order_they_were_written() {
2835        // Measured against GCC rather than read: the two flags the other way round produce the
2836        // same output byte for byte, so the command line order between the two families does not
2837        // decide anything and the order within one does. The `-include` file here can only see
2838        // the definition if the `-imacros` file that was written after it ran first.
2839        let tree = TempTree::new(
2840            "preinclude",
2841            &[
2842                ("a.c", "int main(void) { return 0; }\n"),
2843                ("i.h", "#ifdef FROM_MACROS\nint saw_it;\n#else\nint missed_it;\n#endif\n"),
2844                ("m.h", "#define FROM_MACROS 1\nint macros_text;\n"),
2845            ],
2846        );
2847        let out = tree.path("a.i");
2848        let code = run(&args(&[
2849            "-E",
2850            "-include",
2851            &tree.path("i.h"),
2852            "-imacros",
2853            &tree.path("m.h"),
2854            "-o",
2855            &out,
2856            &tree.path("a.c"),
2857        ]));
2858        assert_eq!(code, 0);
2859        let text = std::fs::read_to_string(&out).expect("the output should have been written");
2860        assert!(text.contains("saw_it"), "{text}");
2861        // And the text of the `-imacros` file is thrown away, which is the whole difference
2862        // between the two flags.
2863        assert!(!text.contains("macros_text"), "{text}");
2864    }
2865
2866    #[test]
2867    fn a_file_the_command_line_named_is_a_prerequisite_the_same_as_one_a_directive_named() {
2868        let tree = TempTree::new(
2869            "preinclude-deps",
2870            &[
2871                ("a.c", "int main(void) { return 0; }\n"),
2872                ("i.h", "int from_include;\n"),
2873                ("m.h", "#define M 1\n"),
2874            ],
2875        );
2876        let out = tree.path("dep.d");
2877        let code = run(&args(&[
2878            "-MM",
2879            "-MF",
2880            &out,
2881            "-include",
2882            &tree.path("i.h"),
2883            "-imacros",
2884            &tree.path("m.h"),
2885            "-o",
2886            &tree.path("a.i"),
2887            &tree.path("a.c"),
2888        ]));
2889        assert_eq!(code, 0);
2890        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2891        assert!(text.contains("i.h"), "{text}");
2892        assert!(text.contains("m.h"), "{text}");
2893    }
2894
2895    #[test]
2896    fn a_command_line_include_that_is_nowhere_on_the_path_is_an_error_and_not_a_warning() {
2897        // Including the directory of the source file, which is not on the path for these: the
2898        // command line was not written there, so a name in it is relative to where the compiler
2899        // was run rather than to where the source sits.
2900        let tree = TempTree::new(
2901            "preinclude-missing",
2902            &[("sub/a.c", "int main(void) { return 0; }\n"), ("sub/beside.h", "int x;\n")],
2903        );
2904        let code = run(&args(&["-E", "-include", "beside.h", "-o", "-", &tree.path("sub/a.c")]));
2905        assert_eq!(code, 1);
2906    }
2907
2908    #[test]
2909    fn a_command_line_that_links_names_the_executable_and_not_the_object_it_went_through() {
2910        // The object a link goes through is in a temporary directory and is gone before `make`
2911        // reads any of this, so the rule that named it would be a rule for a file that is never
2912        // there. The target and the file are both the `-o`, which is the executable.
2913        let (opts, plan) = compile(&["-MD", "sub/a.c", "-o", "prog"]);
2914        assert_eq!(plan.output.as_deref(), Some("prog"));
2915        assert_eq!(deps::default_target("sub/a.c", deps_target_output(&opts, &plan)), "prog");
2916        assert_eq!(
2917            deps::default_file(&opts.deps, "sub/a.c", plan.output.as_deref()).as_deref(),
2918            Some("prog.d")
2919        );
2920    }
2921
2922    #[test]
2923    fn the_plan_keeps_the_output_name_because_the_rule_is_written_from_it() {
2924        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c", "-o", "obj/x.o"]);
2925        assert_eq!(plan.output.as_deref(), Some("obj/x.o"));
2926        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c"]);
2927        assert_eq!(plan.output, None);
2928    }
2929
2930    #[test]
2931    fn usage_fits_on_a_screen() {
2932        // Not a style preference. A help text that scrolls is one nobody reads, and this is
2933        // the cheapest way to keep it honest as flags accumulate. The number goes up only when
2934        // a family of flags arrives that has nowhere to share a line, which the two pass gates
2935        // were and which the two fuel flags and `-fsafety=` now are, and it goes up by exactly
2936        // the lines that family took. The four it went up by last are the flags a build system
2937        // passes without being asked to: how much to say, what machine to generate for, threads,
2938        // and the questions `configure` asks before it compiles anything. The one it went up by
2939        // last is the second line of `--emit`, whose kinds are a family that has now outgrown
2940        // one line and has nowhere else to go. The two it went up by last are the dependency
2941        // family, which is eight flags that share nothing with anything above them. The one it
2942        // went up by last is the four spellings of position independent code, which every
2943        // configure script writes and which could only have shared the link line, and that line
2944        // is already four characters short of the limit. The two it went up by last are the rest
2945        // of the include family, which is six more flags that change where a header is looked for
2946        // and two that name a header outright. The one it went up by last is the pair that keeps
2947        // the intermediate files and times the steps, which belong next to the two flags above
2948        // them that are also about watching a compilation rather than changing one. The two it
2949        // went up by last are the section flags and the visibility flag, which are what a build
2950        // that cares about the size of what it ships and about which names it exports writes, and
2951        // the second of them was already taken and only missing from here. The one it went up by
2952        // last is the stack protector, which is four spellings of one question and which every
2953        // distribution puts on every command line it issues, so a build that reads this list
2954        // looking for it and does not find it has to go and read the specification instead.
2955        assert!(USAGE.lines().count() < 51, "usage text has grown past one screen");
2956    }
2957}