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/0.10.10")]
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[-strong|-all], -fno-stack-protector, -f[no-]stack-clash-protection
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            // The other half of what a hardened build asks for, and it is a question about the
398            // frame rather than about the function, so it is a switch rather than a level.
399            "-fstack-clash-protection" => opts.stack_clash = true,
400            "-fno-stack-clash-protection" => opts.stack_clash = false,
401            // GCC drops its own include directory along with the system ones, because its
402            // headers are half of a pair with the library's and half a pair is worse than
403            // none. A build that passes this is supplying the whole set itself.
404            "-nostdinc" => nostdinc = true,
405            "-o" => {
406                output = Some(args.get(i).ok_or_else(|| err("-o requires an argument"))?.clone());
407                i += 1;
408            }
409            // The flags that take a directory only in the separated form. GCC spells them
410            // this way and nothing writes `-iquotedir`, so accepting the joined form would
411            // mean guessing at a path that starts with the flag's own letters.
412            // Apple's spelling of `--sysroot`, and the one its own build systems pass. The
413            // two mean the same thing here: the configured directories are under there rather
414            // than under the root.
415            "-isysroot" => {
416                let dir = args.get(i).ok_or_else(|| err("-isysroot requires an argument"))?;
417                i += 1;
418                sysroot = Some(PathBuf::from(dir));
419            }
420            "-iquote" | "-isystem" | "-idirafter" => {
421                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
422                i += 1;
423                match arg {
424                    "-iquote" => opts.search.push_quote(dir.clone()),
425                    "-isystem" => opts.search.push_system(dir.clone()),
426                    _ => opts.search.push_after(dir.clone()),
427                }
428            }
429            "-iprefix" => {
430                iprefix = args.get(i).ok_or_else(|| err("-iprefix requires an argument"))?.clone();
431                i += 1;
432            }
433            // Where GCC puts these is not where its manual says it puts them, and this is the
434            // measured answer rather than the documented one: `-iwithprefix` lands in the
435            // `-isystem` slot and not the `-idirafter` slot, and `-iwithprefixbefore` lands in
436            // the `-I` slot. A cross build that uses them is relying on the behaviour, since
437            // that is the compiler it was developed against.
438            "-iwithprefix" | "-iwithprefixbefore" => {
439                let dir = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
440                i += 1;
441                let dir = format!("{iprefix}{dir}");
442                if arg == "-iwithprefix" {
443                    opts.search.push_system(dir);
444                } else {
445                    opts.search.push_bracket(dir);
446                }
447            }
448            "-include" | "-imacros" => {
449                let name = args.get(i).ok_or_else(|| err(format!("{arg} requires an argument")))?;
450                i += 1;
451                opts.preincludes
452                    .push(Preinclude { name: name.clone(), macros_only: arg == "-imacros" });
453            }
454            // The flag `-iquote` was introduced to replace, still passed by build systems old
455            // enough to predate the replacement. It is not a directory: it says that every `-I`
456            // so far is for quoted includes only, and that a quoted include stops looking next
457            // to the file that wrote it.
458            "-I-" => opts.search.split_quote_chain(),
459            "-x" => {
460                let lang = args.get(i).ok_or_else(|| err("-x requires an argument"))?;
461                i += 1;
462                forced = if lang == "none" {
463                    None
464                } else {
465                    Some(InputKind::from_x_arg(lang).map_err(|e| err(format!("{e}")))?)
466                };
467            }
468            // Not a GCC flag. spec/03-architecture.md section 3.5 compiles several
469            // translation units in one process rather than making the build system fork, and
470            // section 3.8's determinism check compares `-j1` against `-j16`, so the knob has
471            // to exist and has to be spelled the way `make` spells it.
472            // `-DFOO`, `-D FOO` and the same for `-U` and `-I`. Both forms are in wide use
473            // and a build system may produce either, so both are read here rather than
474            // being normalised by whatever generated the command line.
475            _ if arg.starts_with("-D") => {
476                let value = joined_or_next(arg, 2, args, &mut i)?;
477                opts.defines.push(value);
478            }
479            _ if arg.starts_with("-U") => {
480                let value = joined_or_next(arg, 2, args, &mut i)?;
481                opts.undefines.push(value);
482            }
483            _ if arg.starts_with("-I") => {
484                let dir = joined_or_next(arg, 2, args, &mut i)?;
485                opts.search.push_bracket(dir);
486            }
487            _ if arg.starts_with("-std=") => {
488                let name = &arg["-std=".len()..];
489                let (std, gnu) = Std::from_flag(name)
490                    .ok_or_else(|| err(format!("unknown dialect `{name}`, see --help")))?;
491                opts.std = std;
492                opts.gnu_extensions = gnu;
493            }
494            // Section 4.5. The claim decides which half of glibc's `sys/cdefs.h` we are
495            // handed, so a differential run that does not set it is comparing two compilers
496            // that believe they are different compilers.
497            // GCC packs these into one flag, so `-dDI` is two of them. Letters in the family
498            // that we have not written yet are accepted and ignored, because a dump is a
499            // debugging aid and a build that asks for one should still compile. A letter
500            // outside the family falls through to the unknown option error, which is what
501            // keeps `-dumpversion` from being read as a dump of nothing.
502            _ if Dumps::is_family(arg) => {
503                opts.dumps.add(&arg[2..]);
504            }
505            // One name at a time, which is what a build that means its own `memcpy` and the
506            // library's everything else writes. The name is not checked against a list, because
507            // the flag is about what the program means by a name and a program is allowed to mean
508            // something by a name this compiler has never heard of.
509            _ if arg.starts_with("-fno-builtin-") => {
510                opts.no_builtin.push(arg["-fno-builtin-".len()..].to_owned());
511            }
512            _ if arg.starts_with("-fgnuc-version=") => {
513                let v = &arg["-fgnuc-version=".len()..];
514                opts.gnuc = v.parse().map_err(err)?;
515            }
516            // spec/13-gnu-compat.md section 13.3 promises this flag an error that says why rather
517            // than the unknown option one, because a build reaching for it is asking for a feature
518            // and deserves to be told it is not coming rather than told the spelling is wrong.
519            // The negative form is what this compiler does anyway, so it is taken and dropped.
520            "-fnested-functions" => {
521                return Err(err(
522                    "nested functions are not supported: a call to one goes through a trampoline \
523                     written on the stack, which no target that enforces an unexecutable stack \
524                     allows",
525                ));
526            }
527            "-fno-nested-functions" => {}
528            // Which of the two links the output is for, which is a real difference and not a
529            // description of what happens anyway. Everything here is position independent either
530            // way, and what these decide is whether a name may be one another object defines or
531            // replaces, because a link that produces an executable puts every name in the same
532            // program and a link that produces a shared library does not.
533            //
534            // It matters that they are accepted at all, whatever they then do. Every autoconf and
535            // cmake build puts `-fPIC` on the compile line, so a compiler that rejects it cannot
536            // be the `CC` of a project that has a configure script, whatever else it can do. That
537            // is how this was found: building SQLite's test fixture stopped on it.
538            "-fPIC" | "-fpic" => opts.pic = Pic::Library,
539            // Not a synonym of the pair above, which is what they were treated as until #756. The
540            // library is the expensive answer and gcc makes it the one that has to be asked for,
541            // so this is also what nothing at all means.
542            "-fPIE" | "-fpie" => opts.pic = Pic::Executable,
543            // A different question from the pair above, and the one every distribution build of a
544            // shared library answers. `-fPIC` decides how an address is reached, and this decides
545            // whether the optimizer may believe a body it can see, because an exported name is one
546            // the dynamic linker may find another definition of first. On by default, which is
547            // gcc's arrangement and is the honest answer, and off is a promise the build makes and
548            // nothing checks.
549            "-fsemantic-interposition" => opts.interposition = true,
550            "-fno-semantic-interposition" => opts.interposition = false,
551            // Two requests rather than one, and the same table answers both, so what decides is
552            // whether either of them is standing. gcc arranges it the same way: the asynchronous
553            // one is the default here and it implies the other, and a line that asks for a table
554            // and against an asynchronous one gets a table.
555            "-fasynchronous-unwind-tables" => opts.async_unwind_tables = true,
556            "-fno-asynchronous-unwind-tables" => opts.async_unwind_tables = false,
557            "-funwind-tables" => opts.unwind_tables = true,
558            "-fno-unwind-tables" => opts.unwind_tables = false,
559            // The other direction is a request, not a description, and it is one this compiler
560            // cannot grant, so it gets the treatment section 13.3 asks for rather than the unknown
561            // option error. Answering it by carrying on would be answering a different question:
562            // the code would still be position independent, which is correct everywhere an
563            // ordinary program runs and is wrong in a kernel, where the flag is written precisely
564            // because there is no loader to fill a global offset table in.
565            "-fno-pic" | "-fno-pie" => {
566                return Err(err(
567                    "position dependent code is not supported: an address that may be in another \
568                     object is loaded out of the global offset table, and nothing here emits the \
569                     absolute form this asks for. Use -no-pie if what you meant was how to link",
570                ));
571            }
572            // A section per function and a section per variable, which is what makes
573            // `--gc-sections` able to drop anything: a linker can leave out a section nothing
574            // reaches and cannot leave out half of one. Both directions are taken, and the off
575            // one is the default rather than a refusal, since a build that writes it is asking
576            // for what happens anyway.
577            "-ffunction-sections" => opts.function_sections = true,
578            "-fno-function-sections" => opts.function_sections = false,
579            "-fdata-sections" => opts.data_sections = true,
580            "-fno-data-sections" => opts.data_sections = false,
581            // Another description of what this compiler does. A file scope declaration with no
582            // initializer is written into `.bss` as an ordinary defined symbol, not offered to the
583            // linker as a common one for it to merge, which is what `-fno-common` asks for and what
584            // gcc has done by default since 10. Nothing in the front end produces `Linkage::Common`
585            // at all.
586            "-fno-common" => {}
587            // And the request, which is the one that cannot be granted. It is a real difference and
588            // not a preference: two files each writing `int g;` link under `-fcommon` and are a
589            // duplicate definition without it, which is the whole reason the flag survives.
590            "-fcommon" => {
591                return Err(err(
592                    "a tentative definition is written into .bss as its own symbol here, and \
593                     nothing emits the common symbol this asks the linker to merge. Give the \
594                     variable a definition in one file and declare it extern in the others",
595                ));
596            }
597            // Both directions of this one are taken, which is the exception to the rule above, and
598            // the reason is which way being wrong costs something.
599            //
600            // Nothing here derives anything from the type an object is accessed through. The IR has
601            // somewhere to put a type based aliasing node and lowering fills it with nothing on
602            // every access, so no pass has one to read and the alias analysis falls back to what it
603            // can see. That makes `-fno-strict-aliasing` a description, the way `-fPIC` is.
604            //
605            // `-fstrict-aliasing` is a request to assume more than that, and this compiler assumes
606            // less. Answering a request for a weaker guarantee by giving a stronger one is safe in
607            // a way the `-fno-pic` case is not: every program that is correct under the assumption
608            // is correct without it, only slower. And `-O2` implies it, so a build that spells it
609            // out is a build that would stop on a compiler that refused it, for no gain at all.
610            "-fstrict-aliasing" | "-fno-strict-aliasing" => {}
611            // About temporary files rather than about code. There is nothing between the phases of
612            // one compilation here to write to a file in the first place.
613            "-pipe" => {}
614            // Nothing here writes colour, so all of these are the same answer, and it is the answer
615            // that costs nothing: the diagnostics come out plain either way and no build depends on
616            // an escape sequence being there. Taken rather than refused because cmake writes
617            // `-fdiagnostics-color=always` on every compile line when the generator is ninja, which
618            // makes this the second most common flag after `-fPIC` to stop a build over a question
619            // about how the text looks.
620            "-fdiagnostics-color" | "-fno-diagnostics-color" => {}
621            _ if arg.starts_with("-fdiagnostics-color=") => {}
622            // The link flags. None of them changes the compilation, which is why they are
623            // collected apart from `opts` and why `-lm` on a `-c` line is a note rather than an
624            // error: it is a thing said to a linker that is not going to run.
625            "-static" => link.is_static = true,
626            "-shared" => link.shared = true,
627            "-pie" => link.pie = Some(true),
628            "-no-pie" | "-nopie" => link.pie = Some(false),
629            "-nostdlib" => link.no_stdlib = true,
630            "-nostartfiles" => link.no_startfiles = true,
631            "-nodefaultlibs" => link.no_defaultlibs = true,
632            "-fno-builtins-lib" => link.no_builtins_lib = true,
633            "-fbuiltins-lib" => link.no_builtins_lib = false,
634            "-rdynamic" | "-export-dynamic" => link.export_dynamic = true,
635            "-s" => link.strip = true,
636            "-Xlinker" => {
637                let next = args.get(i).ok_or_else(|| err("-Xlinker requires an argument"))?;
638                i += 1;
639                link.passthrough.push(next.clone());
640            }
641            _ if arg.starts_with("-Wl,") => {
642                // Commas separate arguments rather than being part of one, which is what makes
643                // `-Wl,-rpath,/opt/lib` two words to the linker and one word here.
644                link.passthrough.extend(arg["-Wl,".len()..].split(',').map(str::to_owned));
645            }
646            _ if arg.starts_with("-fuse-ld=") => {
647                link.use_ld = Some(arg["-fuse-ld=".len()..].to_owned());
648            }
649            _ if arg.starts_with("-l") && arg.len() > 2 => {
650                inputs.push(Input::library(&arg[2..]));
651            }
652            "-l" => {
653                let next = args.get(i).ok_or_else(|| err("-l requires an argument"))?;
654                i += 1;
655                inputs.push(Input::library(next));
656            }
657            _ if arg.starts_with("-L") => {
658                link.search.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
659            }
660            _ if arg.starts_with("-B") => {
661                link.prefixes.push(PathBuf::from(joined_or_next(arg, 2, args, &mut i)?));
662            }
663            _ if arg.starts_with("-j") => {
664                jobs = Jobs::parse(&arg[2..]).map_err(err)?;
665            }
666            _ if arg.starts_with("--sysroot=") => {
667                sysroot = Some(PathBuf::from(&arg["--sysroot=".len()..]));
668            }
669            _ if arg.starts_with("--target=") => {
670                let t = &arg["--target=".len()..];
671                opts.target = t.parse().map_err(|e| err(format!("{e}")))?;
672            }
673            _ if arg.starts_with("--emit=") => {
674                let k = &arg["--emit=".len()..];
675                opts.emit = k
676                    .parse()
677                    .map_err(|()| err(format!("unknown --emit kind `{k}`, see --help")))?;
678            }
679            // A bare `-O` is `-O1`, which is what GCC has and what a hand written makefile tends
680            // to write. `-Og` is GCC's level for a build somebody is going to step through, and
681            // it is `-O1` with the transformations that move code around left out; this compiler
682            // has no such level yet, so it is the nearest one and `--print-pipeline` says what
683            // that came to rather than the flag pretending otherwise.
684            "-O" | "-Og" => opts.opt_level = rucc_session::OptLevel::O1,
685            // The union of `-O3` and `-ffast-math`, and the second half of that changes what
686            // floating point arithmetic means. Refused rather than taken as `-O3`, because a
687            // build that asks for fast math and is quietly given ordinary arithmetic gets a
688            // slower program than it asked for and a build that is given fast math it did not
689            // ask for gets a wrong one.
690            "-Ofast" => {
691                return Err(err(
692                    "-Ofast is -O3 with fast math, and fast math is not implemented, see \
693                     spec/04-driver-and-cli.md section 4.6",
694                ));
695            }
696            _ if arg.starts_with("-O") => {
697                opts.opt_level = arg[2..]
698                    .parse()
699                    .map_err(|()| err(format!("unknown optimization level `{arg}`")))?;
700            }
701            // What every name gets when nothing in the source said, which the attribute in the
702            // source overrides rather than the other way round. Before the optimizer's `-f`
703            // family below for the reason the tier below it is.
704            _ if arg.starts_with("-fvisibility=") => {
705                let seen = &arg["-fvisibility=".len()..];
706                opts.visibility = seen.parse().map_err(|()| {
707                    err(format!(
708                        "`{seen}` is not a visibility, which is default, hidden, internal or \
709                         protected"
710                    ))
711                })?;
712            }
713            // The memory safety monitor, from section 15.4 of
714            // `spec/safe-memory/15-integration.md`. Before the optimizer's `-f` family below,
715            // because a pass that took the name `safety=detect` would otherwise be handed the
716            // flag, and the tier is not a pass.
717            _ if arg.starts_with("-fsafety=") => {
718                let tier = &arg["-fsafety=".len()..];
719                opts.safety = tier.parse().map_err(|()| {
720                    err(format!(
721                        "`{tier}` is not a safety tier, which is off, detect, enforce or kernel"
722                    ))
723                })?;
724            }
725            // The optimizer's own flags, from section 9.10 of `spec/09-optimizer.md`. These come
726            // after every `-f` the rest of the compiler answers to, so a pass can never take a
727            // name that already means something else on the command line.
728            _ if arg.starts_with("-fpass-fuel=") => {
729                let (name, count) = arg["-fpass-fuel=".len()..]
730                    .split_once('=')
731                    .ok_or_else(|| err("-fpass-fuel= is spelled <pass>=<count>"))?;
732                if rucc_opt::pass::find(name).is_none() {
733                    return Err(err(format!(
734                        "`{name}` is not a pass this compiler has, see --print-pipeline"
735                    )));
736                }
737                let count: u32 = count
738                    .parse()
739                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
740                opts.pass_fuel.push((name.to_owned(), count));
741            }
742            _ if arg.starts_with("-fpass-fuel-global=") => {
743                let count = &arg["-fpass-fuel-global=".len()..];
744                let count: u32 = count
745                    .parse()
746                    .map_err(|_| err(format!("`{count}` is not a number of transformations")))?;
747                opts.pass_fuel_global = Some(count);
748            }
749            // Everything from `-fopt-info` to the end of the argument, which is optional
750            // keywords joined by hyphens and an optional `=<file>`. Checked here rather than
751            // where the remarks are printed, because by then the compilation somebody wanted
752            // to hear about is over.
753            _ if arg == "-fopt-info"
754                || arg.starts_with("-fopt-info=")
755                || arg.starts_with("-fopt-info-") =>
756            {
757                let rest = &arg["-fopt-info".len()..];
758                let (kinds, file) = match rest.split_once('=') {
759                    Some((kinds, file)) => (kinds, Some(file)),
760                    None => (rest, None),
761                };
762                let kinds = kinds.strip_prefix('-').unwrap_or(kinds);
763                rucc_opt::Wants::none().add(kinds).map_err(err)?;
764                opts.opt_info.push(kinds.to_owned());
765                if let Some(file) = file {
766                    if file.is_empty() {
767                        return Err(err("-fopt-info= was given no file to write to"));
768                    }
769                    opts.opt_info_file = Some(file.to_owned());
770                }
771            }
772            _ if arg.starts_with("-fdump-ir=") => {
773                // Checked here rather than where the dumps are taken, because the compilation
774                // that would have been dumped is over by then.
775                let spec = &arg["-fdump-ir=".len()..];
776                rucc_opt::Dumps::default().add(spec).map_err(err)?;
777                opts.dump_ir.push(spec.to_owned());
778            }
779            // Before the bare `-f<pass>` below, because a pass called `enable-something` would
780            // otherwise take the flag away from the gate. Checked here rather than where the
781            // pipeline reads it, for the reason that applies to all of these: a misspelled pass
782            // name that quietly gated nothing looks exactly like a pass that is not the guilty
783            // one, and a bisection would carry on past the thing it was looking for.
784            _ if arg.starts_with("-fdisable-") || arg.starts_with("-fenable-") => {
785                let on = arg.starts_with("-fenable-");
786                let spec = &arg[if on { "-fenable-".len() } else { "-fdisable-".len() }..];
787                rucc_opt::Gates::default().add(on, spec).map_err(err)?;
788                opts.pass_gates.push((on, spec.to_owned()));
789            }
790            _ if arg.strip_prefix("-fno-").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
791                opts.passes.push((arg["-fno-".len()..].to_owned(), false));
792            }
793            _ if arg.strip_prefix("-f").is_some_and(|n| rucc_opt::pass::find(n).is_some()) => {
794                opts.passes.push((arg["-f".len()..].to_owned(), true));
795            }
796            // The unstable options, spelled the way rustc spells them and carrying the same
797            // promise, which is none: one of these may change or go away in any release. They are
798            // measurements and debugging aids rather than things a build asks for, which is why
799            // none of them is in the usage text and all of them are in section 4.11 of
800            // `spec/04-driver-and-cli.md`.
801            "-Zverify-each" => opts.verify_each = true,
802            _ if arg.starts_with("-Zrule-coverage=") => {
803                let file = &arg["-Zrule-coverage=".len()..];
804                if file.is_empty() {
805                    return Err(err("-Zrule-coverage= needs a file to write to"));
806                }
807                opts.rule_coverage = Some(file.to_owned());
808            }
809            _ if arg.starts_with("-Zregister-pressure=") => {
810                let file = &arg["-Zregister-pressure=".len()..];
811                if file.is_empty() {
812                    return Err(err("-Zregister-pressure= needs a file to write to"));
813                }
814                opts.register_pressure = Some(file.to_owned());
815            }
816            _ if arg.starts_with("-Z") => {
817                return Err(err(format!(
818                    "`{arg}` is not an unstable option this compiler has, see \
819                     spec/04-driver-and-cli.md section 4.11 for the ones it does"
820                )));
821            }
822            // The word size, which is a statement about the target and is taken as one. A build
823            // that says the size the target already has is saying nothing, and one that says the
824            // other size is asking for a target this compiler does not have, which it is told
825            // rather than being given the wrong one.
826            "-m64" | "-m32" | "-mx32" => {
827                let want: u32 = match arg {
828                    "-m64" => 64,
829                    _ => 32,
830                };
831                let have = rucc_target::TargetInfo::new(opts.target).pointer_width;
832                if have != want {
833                    return Err(err(format!(
834                        "{arg} asks for a {want} bit target and {} is {have} bit, use \
835                         --target= to name the one you mean",
836                        opts.target
837                    )));
838                }
839            }
840            // Which processor in the family to generate for. This compiler emits the base
841            // instruction set of the architecture and nothing above it, so a program built with
842            // any of these runs on the machine that was named; it is a program that could have
843            // been faster rather than a program that is wrong, which is what makes these safe to
844            // take and ignore where a flag that changed the meaning of the code would not be.
845            _ if arg.starts_with("-march=")
846                || arg.starts_with("-mtune=")
847                || arg.starts_with("-mcpu=") => {}
848            // The calling convention, which is not safe to ignore. Taken when it names the one
849            // the target already uses and refused otherwise.
850            _ if arg.starts_with("-mabi=") => {
851                let want = &arg["-mabi=".len()..];
852                let have = match opts.target.arch {
853                    rucc_target::Arch::X86_64 => "sysv",
854                    rucc_target::Arch::Aarch64 => "lp64",
855                    rucc_target::Arch::Riscv64 => "lp64d",
856                };
857                if want != have {
858                    return Err(err(format!(
859                        "{arg}: {} uses the {have} convention and this compiler has no other",
860                        opts.target
861                    )));
862                }
863            }
864            // How far apart the pieces of the program may be. The small model is what we emit and
865            // it is every hosted program's default; the kernel model is a different one and a
866            // build that asks for it and does not get it links and then does not run.
867            "-mcmodel=small" => {}
868            _ if arg.starts_with("-mcmodel=") => {
869                return Err(err(format!(
870                    "{arg}: this compiler emits the small code model and no other, see \
871                     spec/12-targets.md"
872                )));
873            }
874            // GCC's own scripting language for how the driver builds a command line.
875            // `spec/04-driver-and-cli.md` section 4.4 settles that we will not have it, so a
876            // build reaching for it is told which flags do the same job.
877            _ if arg.starts_with("-specs=") => {
878                return Err(err(
879                    "-specs= is not supported: the parts of it builds rely on are -B, -L, \
880                     -nostdlib, -nostartfiles and -Wl,, see spec/04-driver-and-cli.md \
881                     section 4.4",
882                ));
883            }
884            // Arguments meant for a separate assembler or preprocessor, which this compiler does
885            // not have: both are inside it and neither reads a command line. Refused rather than
886            // dropped, because every one of these says something about the output and a build
887            // that asked for `-Wa,--noexecstack` and was silently given an executable stack got
888            // the opposite of what it asked for.
889            _ if arg.starts_with("-Wa,") || arg.starts_with("-Wp,") => {
890                return Err(err(format!(
891                    "`{arg}` is an argument for a separate assembler or preprocessor, and both \
892                     are inside this compiler rather than programs it runs"
893                )));
894            }
895            "-Xassembler" | "-Xpreprocessor" => {
896                return Err(err(format!(
897                    "{arg} hands an argument to a separate assembler or preprocessor, and both \
898                     are inside this compiler rather than programs it runs"
899                )));
900            }
901            // Everything else in the `-W` family. `spec/04-driver-and-cli.md` section 4.1 has
902            // this one as a rule about build systems rather than about warnings: autoconf finds
903            // out whether a warning flag exists by passing it and looking at the exit status, so
904            // a compiler that refuses one it has not heard of fails a configure script written
905            // for a GCC newer than itself. The names are not checked against a list because this
906            // compiler has no warning groups for a list to be of, which #485 is about.
907            _ if arg.starts_with("-W") => {}
908            // Flags that name something this compiler does not do and would not do differently
909            // if it did. `-fno-ident` is about a comment in the output that we do not write
910            // either way, and the others are about a way of ordering the compilation that has
911            // been GCC's only way for twenty years. Section 4.1 asks for the list to be short
912            // and for adding to it to be deliberate, which is why it is written out here.
913            "-fno-ident"
914            | "-fident"
915            | "-funit-at-a-time"
916            | "-fno-unit-at-a-time"
917            | "-shared-libgcc"
918            | "-static-libgcc" => {}
919            _ if arg.starts_with('-') && arg.len() > 1 => {
920                // Silently ignoring an unknown flag is how a build ends up not doing what
921                // its author asked. spec/13-gnu-compat.md section 13.4 makes this an error
922                // for the flags that change code generation, and the safe default until the
923                // flag table is populated is to reject everything we do not know.
924                return Err(err(format!("unknown option `{arg}`")));
925            }
926            _ => inputs.push(Input { path: arg.to_owned(), forced, library: false }),
927        }
928    }
929
930    // Last, so that it lands after every `-isystem` the command line gave. That is GCC's
931    // order: a directory the user names outranks the compiler's own, and the compiler's own
932    // outranks the library's. It is pushed after the loop rather than before it because
933    // `SearchPath` appends within a group and the position is what the order is.
934    // The same directory the headers were looked for under, because a sysroot is a statement
935    // about a whole installation and not about half of one.
936    link.sysroot = sysroot.clone();
937    // After the loop rather than where `-pthread` was read, so that it lands after the objects
938    // that refer to it. A static link takes the definitions it needs from a library when it
939    // reaches it and not afterwards, so a library before the objects is a library that answers
940    // nothing.
941    if threads {
942        inputs.push(Input::library("pthread"));
943    }
944    if let Some(query) = query {
945        return Ok(Action::Print(answer(&query, &opts, &link)));
946    }
947    // `-M` and `-MM` produce the rule and nothing else, so the run stops after phase 4 whatever
948    // else the command line asked for. Read here rather than where the flag was, because a `-c`
949    // written after it has to lose and the loop cannot know that until it has ended. The output
950    // file is where the rule goes rather than where an object would have gone, and the last
951    // phase being the preprocessor is what makes that true without a second rule for it.
952    if opts.deps.instead_of_compiling {
953        opts.emit = EmitKind::Preprocessed;
954    }
955    if !nostdinc {
956        opts.search.push_system(runtime::DIR);
957        // And the library's after ours, which is the other half of the same order. They go on
958        // here rather than at the point `--target=` or `--sysroot=` was read because either
959        // one changes the answer and the last word on both is the end of the loop.
960        for dir in library::system_dirs(opts.target, sysroot.as_deref()) {
961            opts.search.push_system(dir);
962        }
963    }
964    // Once, here, rather than as each directory is pushed. A `-I` that names a system
965    // directory has to lose to the system entry and the system entry is added last, so the
966    // question cannot be answered until the whole path is known.
967    opts.search.remove_duplicates();
968
969    // The target has to be resolved before the configuration is printed, so this check comes
970    // after the loop rather than at the point `--print-config` was seen.
971    if print_config {
972        return Ok(Action::PrintConfig(Box::new(opts)));
973    }
974    if print_pipeline {
975        return Ok(Action::PrintPipeline(Box::new(opts)));
976    }
977    let plan = Plan::new(&opts, &inputs, output.as_deref()).map_err(|e| err(e.message))?;
978    if print_plan {
979        return Ok(Action::PrintPlan {
980            opts: Box::new(opts),
981            plan: Box::new(plan),
982            link: Box::new(link),
983        });
984    }
985    Ok(Action::Compile {
986        opts: Box::new(opts),
987        plan: Box::new(plan),
988        link: Box::new(link),
989        jobs,
990        verbose,
991    })
992}
993
994/// What one of the `-dump` and `-print` flags prints.
995///
996/// GCC prints the name back unchanged when it cannot find the file a `-print` flag asked about,
997/// which is what makes the answer safe to paste into a link line whether or not the file is
998/// there, and this does the same.
999fn answer(query: &Query, opts: &Options, link: &LinkOptions) -> String {
1000    let found = |name: &str| {
1001        link::find_in_search(link, opts.target, name)
1002            .map_or_else(|| name.to_owned(), |path| path.display().to_string())
1003    };
1004    match query {
1005        Query::Machine => opts.target.to_string(),
1006        Query::Version => VERSION.to_owned(),
1007        Query::Multiarch => link::multiarch(opts.target),
1008        // The three lines GCC prints, in its order and with its punctuation, because what reads
1009        // them is a script written against that shape. There is no installation directory to
1010        // report: this compiler is one binary that works wherever it is copied, and the headers
1011        // it ships are inside it, so `install` is where the binary is and nothing is under it.
1012        Query::SearchDirs => {
1013            let here = std::env::current_exe()
1014                .ok()
1015                .and_then(|p| p.parent().map(std::path::Path::to_path_buf))
1016                .unwrap_or_default();
1017            let list = |dirs: &[PathBuf]| {
1018                dirs.iter().map(|d| d.display().to_string()).collect::<Vec<_>>().join(":")
1019            };
1020            let libraries = link::search_dirs(link, opts.target);
1021            format!(
1022                "install: {}\nprograms: ={}\nlibraries: ={}",
1023                here.display(),
1024                list(&link.prefixes),
1025                list(&libraries)
1026            )
1027        }
1028        Query::FileName(name) => found(name),
1029        // The name GCC gives the library of routines a compiler's output calls that the C
1030        // library does not have. Ours is built in and there is no file, so the answer is the
1031        // name itself, which is what GCC prints when it cannot find one either.
1032        Query::Libgcc => found("libgcc.a"),
1033        // A program rather than a library: the linker and the archiver are the ones a build asks
1034        // about, and this compiler finds them on the path or under `-B` rather than shipping
1035        // them, so the name back is the honest answer unless a `-B` prefix holds one.
1036        Query::ProgName(name) => link
1037            .prefixes
1038            .iter()
1039            .map(|dir| dir.join(name))
1040            .find(|path| path.is_file())
1041            .map_or_else(|| name.clone(), |path| path.display().to_string()),
1042    }
1043}
1044
1045/// Renders the passes this level will run, in order, with what each one does.
1046///
1047/// The level is the whole of the answer unless a `-f` flag edited it, which is section 9.1 of
1048/// `spec/09-optimizer.md`: a level is a list somebody wrote down rather than something that
1049/// emerges from which flags happen to be set, and this is how that list is read.
1050#[must_use]
1051pub fn print_pipeline(opts: &Options) -> String {
1052    let mut settings = rucc_opt::Options::for_level(opts.opt_level);
1053    settings.toggles.clone_from(&opts.passes);
1054    settings.global_fuel = opts.pass_fuel_global;
1055    for (on, spec) in &opts.pass_gates {
1056        // Every spelling was checked while the arguments were parsed, so there is nothing here
1057        // this can refuse, and a listing is not the place to report it if there were.
1058        let _ = settings.gates.add(*on, spec);
1059    }
1060    rucc_opt::pipeline::print(&settings)
1061}
1062
1063/// Renders the resolved configuration.
1064///
1065/// One `key: value` per line, sorted by nothing in particular but fixed in order, because
1066/// this output is diffed across hosts in CI and a reordering would read as a change.
1067#[must_use]
1068pub fn print_config(opts: &Options) -> String {
1069    let sess = Session::new(opts.clone());
1070    let t = &sess.target;
1071    let mut out = String::new();
1072    let _ = writeln!(out, "version: {VERSION}");
1073    // The three field triple the driver was given rather than the ten field tuple it widens to,
1074    // because this output is what a build system reads to find out what it asked for. The tuple is
1075    // the compiler's model of the machine and this line is a receipt for a command line.
1076    let _ = writeln!(out, "target: {}", opts.target);
1077    let _ = writeln!(out, "arch: {}", opts.target.arch.as_str());
1078    let _ = writeln!(out, "os: {}", opts.target.os.as_str());
1079    let _ = writeln!(out, "env: {}", opts.target.env.as_str());
1080    let _ = writeln!(out, "object-format: {}", t.object_format.as_str());
1081    let _ = writeln!(out, "pointer-width: {}", t.pointer_width);
1082    let _ = writeln!(out, "long-width: {}", t.long_width);
1083    let _ = writeln!(out, "long-double-width: {}", t.long_double_width);
1084    let _ = writeln!(out, "endian: {}", if t.little_endian { "little" } else { "big" });
1085    let _ = writeln!(out, "char-signed: {}", t.char_is_signed);
1086    let _ = writeln!(out, "va-list: {}", t.va_list.map_or("none", |list| list.as_str()));
1087    // The register file as a count per class, which is enough to tell a target whose registers
1088    // are described from one whose are not without printing sixteen names nobody asked for.
1089    let regs: Vec<String> = t
1090        .regs
1091        .classes()
1092        .map(|(class, info)| format!("{} {}", info.name, t.regs.len(class)))
1093        .collect();
1094    let _ = writeln!(
1095        out,
1096        "registers: {}",
1097        if regs.is_empty() { "none".to_string() } else { regs.join(", ") }
1098    );
1099    let _ = writeln!(out, "opt-level: {}", sess.opts.opt_level);
1100    let _ = writeln!(out, "safety: {}", sess.opts.safety);
1101    let _ = writeln!(out, "emit: {}", sess.opts.emit.as_str());
1102    let _ = writeln!(out, "debug-info: {}", sess.opts.debug_info);
1103    let _ = writeln!(out, "frame-pointer: {}", sess.opts.frame_pointer);
1104    let _ = writeln!(out, "red-zone: {}", sess.opts.red_zone);
1105    let _ = writeln!(out, "stack-protector: {}", sess.opts.protector);
1106    let _ = writeln!(out, "stack-clash-protection: {}", sess.opts.stack_clash);
1107    // Last because it is the one key with more than one line under it, and the only one
1108    // whose value is a property of the machine rather than of the command line.
1109    for dir in sess.opts.search.dirs() {
1110        let system = if dir.is_system { " (system)" } else { "" };
1111        let _ = writeln!(out, "include: {}{system}", dir.path.display());
1112    }
1113    out
1114}
1115
1116/// The output name the make target is taken from, which is the `-o` argument or nothing.
1117///
1118/// A run that stops at the preprocessor has not named an object, whatever its `-o` says: under
1119/// `-E` that argument is the preprocessed text and under `-M` it is the rule itself, and neither
1120/// is a file `make` would rebuild by running this rule. GCC agrees and falls back to the source
1121/// name in both, which is why a `-MD -E -o out.i` writes `out.d` holding a rule for `a.o`. From
1122/// `-S` on the argument does name what the rule builds, and it is used as written.
1123fn deps_target_output<'a>(opts: &Options, plan: &'a Plan) -> Option<&'a str> {
1124    if opts.emit == EmitKind::Preprocessed { None } else { plan.output.as_deref() }
1125}
1126
1127/// Writes to a path the command line named rather than one the plan derived, where `-` is
1128/// standard output.
1129fn write_named(path: &str, bytes: &[u8]) -> Result<(), String> {
1130    if path == "-" {
1131        return write_out(&Output::Stdout, bytes);
1132    }
1133    write_out(&Output::File(path.to_owned()), bytes)
1134}
1135
1136/// Writes the make rule for one input, and reports whether it got there.
1137///
1138/// A rule with no file of its own goes where the compilation it replaced would have written,
1139/// which is what makes the usual makefile recipe work: `rucc -M $< -o $@` leaves the rule in
1140/// `$@`, and the same line with the `-o` left off puts it on standard output.
1141fn write_deps(
1142    opts: &Options,
1143    plan: &Plan,
1144    job: &Job,
1145    found: &[Dependency],
1146    stderr: &mut impl std::io::Write,
1147) -> bool {
1148    let targets = if opts.deps.targets.is_empty() {
1149        vec![deps::default_target(&job.input, deps_target_output(opts, plan))]
1150    } else {
1151        opts.deps.targets.clone()
1152    };
1153    let rule = deps::rule(&opts.deps, &targets, &job.input, found);
1154    // The file, on the other hand, is named after the `-o` in every mode that still has one to
1155    // spend, which is every mode except the two that spend it on the rule.
1156    let wrote = match deps::default_file(&opts.deps, &job.input, plan.output.as_deref()) {
1157        // A `-MF` on a run that had nowhere else to put the rule leaves the file the `-o`
1158        // named empty rather than absent, because a makefile that named it as a target of its
1159        // own is a makefile that will look for it.
1160        Some(path) => write_named(&path, rule.as_bytes()).and_then(|()| {
1161            if opts.deps.instead_of_compiling { write_out(&job.output, b"") } else { Ok(()) }
1162        }),
1163        None => write_out(&job.output, rule.as_bytes()),
1164    };
1165    if let Err(e) = wrote {
1166        let _ = writeln!(stderr, "rucc: error: {e}");
1167        return false;
1168    }
1169    true
1170}
1171
1172/// Runs phase 4 over every input that has one, and writes what came out.
1173///
1174/// One input that fails does not stop the others. A build that reports every file it could
1175/// not preprocess in one run is worth more than one that stops at the first, and the exit
1176/// status is still a failure either way.
1177fn preprocess_all(opts: &Options, plan: &Plan) -> i32 {
1178    let fs = OsFileSystem::new();
1179    let mut stderr = std::io::stderr().lock();
1180    let mut failed = false;
1181    for job in &plan.jobs {
1182        if !job.phases.first().is_some_and(|p| *p == Phase::Preprocess) {
1183            // An input that is already preprocessed, or an object file. GCC passes these
1184            // through untouched, and the plan has already said so in its notes.
1185            continue;
1186        }
1187        let started = std::time::Instant::now();
1188        let result = preprocess(opts, &job.input, &fs);
1189        if opts.time {
1190            say_time(&job.input, started.elapsed(), &mut stderr);
1191        }
1192        for message in &result.messages {
1193            let _ = writeln!(stderr, "{message}");
1194        }
1195        if result.failed() {
1196            failed = true;
1197            continue;
1198        }
1199        if opts.deps.emit {
1200            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1201            // `-M` and `-MM` asked for the rule instead of the text, so there is nothing else
1202            // to write. The other two asked for both and fall through to the text below.
1203            if opts.deps.instead_of_compiling {
1204                continue;
1205            }
1206        }
1207        if let Err(e) = write_out(&job.output, result.text.as_bytes()) {
1208            let _ = writeln!(stderr, "rucc: error: {e}");
1209            failed = true;
1210        }
1211    }
1212    i32::from(failed)
1213}
1214
1215/// Runs the front end over every input that has a compile phase, and writes what came out.
1216///
1217/// The same rule as [`preprocess_all`]: one input that fails does not stop the others, and the
1218/// exit status is a failure either way. An input that is already assembly or an object has no
1219/// compile phase and is passed over here, which the plan has already said in its notes.
1220fn compile_all(opts: &Options, plan: &Plan) -> i32 {
1221    let fs = OsFileSystem::new();
1222    let mut stderr = std::io::stderr().lock();
1223    let mut failed = false;
1224    let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1225    failed |= !ok;
1226    let mut fired = Fired::new();
1227    let mut pressure = Pressure::new();
1228    for job in &plan.jobs {
1229        if !job.phases.contains(&Phase::Compile) {
1230            continue;
1231        }
1232        // An input of IR is read back rather than compiled, since the C it came from is not
1233        // here any more. Everything after this is the same, so the two paths meet again at the
1234        // messages and the file the result is written to.
1235        let started = std::time::Instant::now();
1236        let result = if job.kind == InputKind::Ir {
1237            compile_ir(opts, &job.input, &fs)
1238        } else {
1239            compile(opts, &job.input, &fs)
1240        };
1241        if opts.time {
1242            say_time(&job.input, started.elapsed(), &mut stderr);
1243        }
1244        fired.merge(&result.fired);
1245        pressure.merge(&result.pressure);
1246        failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1247        failed |= !remarks.write(&result.remarks, &mut stderr);
1248        for message in &result.messages {
1249            let _ = writeln!(stderr, "{message}");
1250        }
1251        // Before the failure below, because a compilation that stopped in the back end is exactly
1252        // the one whose preprocessed source somebody wants to look at.
1253        failed |= !write_temps(job, &result.temps, &mut stderr);
1254        if result.failed() {
1255            failed = true;
1256            continue;
1257        }
1258        // `-MD` and `-MMD` write the rule beside the object and let the compilation happen, so
1259        // this is the one path where both files come out of the same run. An input of IR has no
1260        // dependencies to report and produces an empty list, which produces a rule naming only
1261        // itself, and that is the honest answer rather than a missing file.
1262        if opts.deps.emit {
1263            failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1264        }
1265        if let Err(e) = write_out(&job.output, result.artifact.bytes()) {
1266            let _ = writeln!(stderr, "rucc: error: {e}");
1267            failed = true;
1268        }
1269    }
1270    failed |= !write_coverage(opts, &fired, &mut stderr);
1271    failed |= !write_pressure(opts, &pressure, &mut stderr);
1272    i32::from(failed)
1273}
1274
1275/// A directory for the object files only the link step ever sees, removed when it goes away.
1276///
1277/// `-c` writes its object where the user can see it and linking does not, which is the whole of
1278/// the difference: a `rucc a.c b.c` leaves an executable behind and nothing else, the same as
1279/// every other compiler. Removing them on drop rather than at the end of a function is so that a
1280/// link that failed leaves nothing behind either.
1281struct Scratch {
1282    /// Where the objects go.
1283    dir: PathBuf,
1284}
1285
1286impl Scratch {
1287    /// Makes one, under whatever the platform calls its temporary directory.
1288    ///
1289    /// The name carries the process id so that two compilers running at once do not share a
1290    /// directory, which they would otherwise do the moment two of them compiled a file of the
1291    /// same name.
1292    fn new() -> Result<Scratch, String> {
1293        let dir = std::env::temp_dir().join(format!("rucc-{}", std::process::id()));
1294        std::fs::create_dir_all(&dir).map_err(|e| format!("{}: {e}", dir.display()))?;
1295        Ok(Scratch { dir })
1296    }
1297}
1298
1299impl Drop for Scratch {
1300    fn drop(&mut self) {
1301        let _ = std::fs::remove_dir_all(&self.dir);
1302    }
1303}
1304
1305/// The link line the plan describes, for `-###`.
1306///
1307/// The names in it are the hints the plan carries rather than the temporaries a real compilation
1308/// would choose, because `-###` prints the line without having compiled anything and so has
1309/// nothing to point at. That also makes the printed line readable rather than naming a directory
1310/// that only exists while a compilation is running.
1311fn link_line(opts: &Options, link: &LinkOptions, job: &LinkJob) -> Result<String, link::Error> {
1312    let linker = link::find(opts.target, link)?;
1313    let args = link::line(opts.target, link, &job.inputs, &job.output)?;
1314    Ok(link::render(&linker, &args))
1315}
1316
1317/// Compiles everything, then links it.
1318///
1319/// The objects go in a directory that is removed afterwards, which is why this is not
1320/// [`compile_all`] followed by a link: the plan says an object feeding the linker is temporary
1321/// and does not say where, because where is a question that only has an answer once something is
1322/// running.
1323fn link_all(opts: &Options, plan: &Plan, link: &LinkOptions, verbose: bool) -> i32 {
1324    let Some(job) = &plan.link else {
1325        // Every path into here comes from a plan whose last phase is the link, and such a plan
1326        // has a link job. Saying so is cheaper than an unwrap that would have to be explained.
1327        let mut stderr = std::io::stderr().lock();
1328        let _ = writeln!(stderr, "rucc: error: there is nothing to link");
1329        return 1;
1330    };
1331    // Before anything is compiled, because a linker that is not on the machine is worth knowing
1332    // about in the second it takes to look rather than after the compilation.
1333    let linker = match link::find(opts.target, link) {
1334        Ok(linker) => linker,
1335        Err(why) => return complain(why),
1336    };
1337
1338    let scratch = match Scratch::new() {
1339        Ok(scratch) => scratch,
1340        Err(why) => return complain(format!("could not make a place for the object files: {why}")),
1341    };
1342
1343    let fs = OsFileSystem::new();
1344    let mut failed = false;
1345    // One per job, in job order, which is what lets the link line below be rebuilt with the real
1346    // paths in it: every job contributes exactly one file to the line and does so in this order.
1347    let mut produced: Vec<String> = Vec::with_capacity(plan.jobs.len());
1348    let mut fired = Fired::new();
1349    let mut pressure = Pressure::new();
1350    {
1351        let mut stderr = std::io::stderr().lock();
1352        let (mut remarks, ok) = Remarks::new(opts.opt_info_file.as_ref(), &mut stderr);
1353        failed |= !ok;
1354        for (at, job) in plan.jobs.iter().enumerate() {
1355            let out = match &job.output {
1356                Output::Temporary(hint) => {
1357                    // The index because two inputs in different directories can have the same
1358                    // name, and the two objects of `rucc a/x.c b/x.c` must not be one file.
1359                    scratch.dir.join(format!("{at}-{hint}")).display().to_string()
1360                }
1361                Output::File(path) => path.clone(),
1362                // A job feeding the linker never writes to standard output, since the plan gives
1363                // it a temporary. This is here so that the match is total rather than a panic.
1364                Output::Stdout => continue,
1365            };
1366            produced.push(out.clone());
1367            if !job.phases.contains(&Phase::Compile) {
1368                continue;
1369            }
1370            let started = std::time::Instant::now();
1371            let result = if job.kind == InputKind::Ir {
1372                compile_ir(opts, &job.input, &fs)
1373            } else {
1374                compile(opts, &job.input, &fs)
1375            };
1376            if opts.time {
1377                say_time(&job.input, started.elapsed(), &mut stderr);
1378            }
1379            fired.merge(&result.fired);
1380            pressure.merge(&result.pressure);
1381            failed |= !write_dumps(&job.input, &result.dumps, &mut stderr);
1382            failed |= !remarks.write(&result.remarks, &mut stderr);
1383            for message in &result.messages {
1384                let _ = writeln!(stderr, "{message}");
1385            }
1386            failed |= !write_temps(job, &result.temps, &mut stderr);
1387            if result.failed() {
1388                failed = true;
1389                continue;
1390            }
1391            // A `-MD` on a command line that links writes the rule next to the executable and
1392            // names the executable as its target, since that is the file this source builds
1393            // here. The object it went through is in a temporary directory and is gone by the
1394            // time `make` reads any of this.
1395            if opts.deps.emit {
1396                failed |= !write_deps(opts, plan, job, &result.deps, &mut stderr);
1397            }
1398            if !matches!(result.artifact, Artifact::Object(_)) {
1399                // Worth saying rather than writing whatever it is and letting the linker read it.
1400                // An empty file is a valid empty linker script, so a link handed one gets as far
1401                // as reporting every symbol of this file undefined, which is a page of messages
1402                // about something that went wrong here.
1403                let _ = writeln!(
1404                    stderr,
1405                    "rucc: internal error: {}: no object file was produced for the link",
1406                    job.input
1407                );
1408                failed = true;
1409                continue;
1410            }
1411            if let Err(e) = std::fs::write(&out, result.artifact.bytes()) {
1412                let _ = writeln!(stderr, "rucc: error: {out}: {e}");
1413                failed = true;
1414            }
1415        }
1416        failed |= !write_coverage(opts, &fired, &mut stderr);
1417        failed |= !write_pressure(opts, &pressure, &mut stderr);
1418    }
1419    if failed {
1420        // Nothing is linked from a compilation that did not finish. A linker run over the objects
1421        // that did compile would report every function of the file that did not as undefined,
1422        // which is a page of messages about a mistake already reported once.
1423        return 1;
1424    }
1425
1426    // The items in command line order with the temporaries filled in. A library contributes no
1427    // job and passes through, and every file item takes the next job's real output, which is
1428    // what keeps a library that was written between two objects between them here.
1429    let mut outputs = produced.into_iter();
1430    let mut items = Vec::with_capacity(job.inputs.len());
1431    for item in &job.inputs {
1432        match item {
1433            link::Item::Library(name) => items.push(link::Item::Library(name.clone())),
1434            link::Item::File(_) => match outputs.next() {
1435                Some(path) => items.push(link::Item::File(path)),
1436                None => return complain("the plan asks the linker for a file nothing produced"),
1437            },
1438        }
1439    }
1440
1441    let args = match link::line(opts.target, link, &items, &job.output) {
1442        Ok(args) => args,
1443        Err(why) => return complain(why),
1444    };
1445    if verbose {
1446        let mut stderr = std::io::stderr().lock();
1447        let _ = writeln!(stderr, "{}", link::render(&linker, &args));
1448    }
1449    let started = std::time::Instant::now();
1450    let ran = link::run(&linker, &args);
1451    if opts.time {
1452        // The one step of a compilation that really is another program, so this line is the same
1453        // measurement gcc's is and names the linker the way gcc names `collect2`.
1454        let mut stderr = std::io::stderr().lock();
1455        say_time(&linker.name, started.elapsed(), &mut stderr);
1456    }
1457    match ran {
1458        Ok(()) => 0,
1459        // The linker has already said what was wrong on its own error output, and repeating that
1460        // linking failed would only push its message further up the screen.
1461        Err(link::Error::Refused { .. }) => 1,
1462        Err(why) => complain(why),
1463    }
1464}
1465
1466/// Prints one driver level message and gives back the exit status that goes with it.
1467fn complain(why: impl std::fmt::Display) -> i32 {
1468    let mut stderr = std::io::stderr().lock();
1469    let _ = writeln!(stderr, "rucc: error: {why}");
1470    1
1471}
1472
1473/// Writes what `-Zrule-coverage=FILE` asked for, and says whether it could.
1474///
1475/// Once for the whole command line rather than once per input, because the question is which
1476/// lowering rules this run of the compiler reached and a file per input would leave the reader
1477/// unioning files to find out something one process already knew.
1478///
1479/// A file that could not be written is a failure and not a warning. What asks for this is a
1480/// measurement run, and a measurement that quietly did not happen is worse than one that stopped.
1481fn write_coverage(opts: &Options, fired: &Fired, stderr: &mut impl std::io::Write) -> bool {
1482    let Some(path) = &opts.rule_coverage else { return true };
1483    let Some(table) = coverage::table(opts.target.arch) else {
1484        let _ = writeln!(
1485            stderr,
1486            "rucc: error: there are no lowering rules for {} yet, so there is no coverage of them \
1487             to report",
1488            opts.target
1489        );
1490        return false;
1491    };
1492    match std::fs::write(path, fired.listing(table)) {
1493        Ok(()) => true,
1494        Err(e) => {
1495            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1496            false
1497        }
1498    }
1499}
1500
1501/// Writes what `-Zregister-pressure=FILE` asked for, and says whether it could.
1502///
1503/// Once for the whole command line, for the reason [`write_coverage`] gives, and a file that could
1504/// not be written is a failure for the reason it gives too. There is no equivalent of the missing
1505/// rule table here, since every target this compiles for has an allocator, and a run that reached
1506/// no back end at all writes an empty listing rather than nothing: a measurement of a build that
1507/// produced no code is still an answer and it is the honest one.
1508fn write_pressure(opts: &Options, pressure: &Pressure, stderr: &mut impl std::io::Write) -> bool {
1509    let Some(path) = &opts.register_pressure else { return true };
1510    match std::fs::write(path, pressure.listing()) {
1511        Ok(()) => true,
1512        Err(e) => {
1513            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1514            false
1515        }
1516    }
1517}
1518
1519/// Where the `-fopt-info` remarks go, and how much of the run has already gone there.
1520///
1521/// Standard error by default, and one file for the whole run when `-fopt-info=<file>` named one.
1522/// A file rather than the diagnostic stream is what a harness wants: the corpus in
1523/// `tamnd/rucc-corpus` matches a rejection against what the compiler said on standard error, and
1524/// a few thousand remarks mixed into that would bury it.
1525struct Remarks {
1526    /// The file, if there is one.
1527    file: Option<String>,
1528    /// Whether anything has been written to it yet, which decides between truncating and
1529    /// appending. One file holds the whole run rather than the last input in it.
1530    started: bool,
1531}
1532
1533impl Remarks {
1534    /// Prepares the destination, emptying the file if there is one.
1535    ///
1536    /// Emptied here rather than at the first remark, because a run where no pass had anything to
1537    /// say should leave an empty file and not yesterday's. An absent file and an empty one are
1538    /// different facts and something reading this will act on the difference.
1539    fn new(file: Option<&String>, stderr: &mut impl std::io::Write) -> (Self, bool) {
1540        let mut ok = true;
1541        if let Some(path) = file {
1542            if let Err(e) = std::fs::write(path, "") {
1543                let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1544                ok = false;
1545            }
1546        }
1547        (Self { file: file.cloned(), started: false }, ok)
1548    }
1549
1550    /// Writes one input's remarks, and says whether that worked.
1551    ///
1552    /// A file that cannot be written is a failure and not a warning, for the reason
1553    /// [`write_dumps`] gives: remarks that quietly did not arrive look exactly like a compilation
1554    /// where nothing happened.
1555    fn write(&mut self, text: &str, stderr: &mut impl std::io::Write) -> bool {
1556        if text.is_empty() {
1557            return true;
1558        }
1559        let Some(path) = &self.file else {
1560            let _ = write!(stderr, "{text}");
1561            return true;
1562        };
1563        let opened = std::fs::OpenOptions::new()
1564            .write(true)
1565            .append(self.started)
1566            .truncate(!self.started)
1567            .create(true)
1568            .open(path);
1569        self.started = true;
1570        let result =
1571            opened.and_then(|mut file| std::io::Write::write_all(&mut file, text.as_bytes()));
1572        if let Err(e) = result {
1573            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1574            return false;
1575        }
1576        true
1577    }
1578}
1579
1580/// Writes what `-fdump-ir=` asked to see, one file per dump.
1581///
1582/// The name is the input file with the dump's own name and `.ir` after it, so a directory listing
1583/// after a run is the passes in the order they ran, per input. They go in the working directory
1584/// rather than beside the output, because a dump is something a person asked for at a prompt and
1585/// the working directory is where that person is.
1586///
1587/// A file that could not be written is a failure and not a warning, for the reason
1588/// [`write_coverage`] gives: what asked for this is somebody debugging a pass, and a dump that
1589/// quietly did not happen looks exactly like a pass that did not run.
1590fn write_dumps(input: &str, dumps: &[rucc_opt::Dump], stderr: &mut impl std::io::Write) -> bool {
1591    let stem = std::path::Path::new(input)
1592        .file_name()
1593        .map_or_else(|| input.to_owned(), |name| name.to_string_lossy().into_owned());
1594    let mut ok = true;
1595    for dump in dumps {
1596        let path = format!("{stem}.{}.ir", dump.name);
1597        if let Err(e) = std::fs::write(&path, &dump.text) {
1598            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1599            ok = false;
1600        }
1601    }
1602    ok
1603}
1604
1605/// Writes the files `-save-temps` kept, which is nothing at all unless it was given.
1606///
1607/// A file that could not be written is a failure rather than a warning, for the reason
1608/// [`write_dumps`] gives: somebody asked for these by name, and one that quietly did not happen
1609/// looks like a compilation that never went through that step.
1610fn write_temps(job: &Job, temps: &Temps, stderr: &mut impl std::io::Write) -> bool {
1611    let mut ok = true;
1612    let kept = [(job.saved_text(), &temps.preprocessed), (job.saved_asm(), &temps.assembly)];
1613    for (path, text) in kept {
1614        // A step the compilation did not reach has nothing to keep, and a job that is not keeping
1615        // that step has nowhere to put it. Either way there is no file here.
1616        let (Some(path), Some(text)) = (path, text) else { continue };
1617        if let Err(e) = std::fs::write(&path, text) {
1618            let _ = writeln!(stderr, "rucc: error: {path}: {e}");
1619            ok = false;
1620        }
1621    }
1622    ok
1623}
1624
1625/// One line of `-time`, which is what a step was called and how long it took.
1626///
1627/// GCC's two numbers are the user and the system time of a subprocess it ran. This compiler runs
1628/// no subprocess for anything but the link, so what is measured here is the wall clock of the
1629/// step and the second column is always zero. The shape of the line is kept because a person
1630/// reading it next to gcc's should not have to work out which column is which.
1631fn say_time(name: &str, took: std::time::Duration, stderr: &mut impl std::io::Write) {
1632    let _ = writeln!(stderr, "# {name} {:.2} {:.2}", took.as_secs_f64(), 0.0);
1633}
1634
1635/// Writes one job's result where the plan said it goes.
1636///
1637/// # Errors
1638///
1639/// Returns the message to print, which names the file when there is one, because "permission
1640/// denied" on its own does not say which file was refused.
1641fn write_out(output: &Output, bytes: &[u8]) -> Result<(), String> {
1642    match output {
1643        Output::Stdout => {
1644            let mut stdout = std::io::stdout().lock();
1645            stdout.write_all(bytes).map_err(|e| format!("writing to standard output: {e}"))
1646        }
1647        Output::File(path) | Output::Temporary(path) => {
1648            std::fs::write(path, bytes).map_err(|e| format!("{path}: {e}"))
1649        }
1650    }
1651}
1652
1653/// Runs the driver and returns the process exit code.
1654///
1655/// `args` excludes the program name. Output goes to `stdout` and errors to `stderr`, which
1656/// is the one place in the compiler that is true.
1657pub fn run(args: &[String]) -> i32 {
1658    match parse_args(args) {
1659        Ok(Action::Help) => {
1660            print!("{USAGE}");
1661            0
1662        }
1663        Ok(Action::Version) => {
1664            println!("rucc {VERSION}");
1665            0
1666        }
1667        Ok(Action::Print(line)) => {
1668            println!("{line}");
1669            0
1670        }
1671        Ok(Action::PrintConfig(opts)) => {
1672            print!("{}", print_config(&opts));
1673            0
1674        }
1675        Ok(Action::PrintPipeline(opts)) => {
1676            print!("{}", print_pipeline(&opts));
1677            0
1678        }
1679        Ok(Action::PrintPlan { opts, plan, link }) => {
1680            print!("{}", plan.render());
1681            // The line as it would be typed, which is the half of `-###` that section 4.3 says
1682            // arrives with the link. It is printed even when the linker is not on this machine,
1683            // because what a build wants from `-###` is what the compiler would do.
1684            if let Some(job) = &plan.link {
1685                match link_line(&opts, &link, job) {
1686                    Ok(line) => println!("{line}"),
1687                    Err(why) => {
1688                        let mut stderr = std::io::stderr().lock();
1689                        let _ = writeln!(stderr, "rucc: error: {why}");
1690                        return 1;
1691                    }
1692                }
1693            }
1694            0
1695        }
1696        Ok(Action::Compile { opts, plan, link, jobs, verbose }) => {
1697            {
1698                let mut stderr = std::io::stderr().lock();
1699                if verbose {
1700                    let _ = write!(stderr, "{}", plan.render());
1701                    let _ = writeln!(stderr, "workers: {}", jobs.count());
1702                }
1703            }
1704            if opts.emit == EmitKind::Preprocessed {
1705                return preprocess_all(&opts, &plan);
1706            }
1707            if opts.emit != EmitKind::Executable {
1708                return compile_all(&opts, &plan);
1709            }
1710            link_all(&opts, &plan, &link, verbose)
1711        }
1712        Err(e) => {
1713            let mut stderr = std::io::stderr().lock();
1714            let _ = writeln!(stderr, "rucc: error: {e}");
1715            let _ = writeln!(stderr, "rucc: note: run `rucc --help` for usage");
1716            1
1717        }
1718    }
1719}
1720
1721#[cfg(test)]
1722mod tests {
1723    use rucc_session::{GnucVersion, IncludeForm, OptLevel, Visibility};
1724
1725    use super::*;
1726
1727    fn args(s: &[&str]) -> Vec<String> {
1728        s.iter().map(|x| (*x).to_owned()).collect()
1729    }
1730
1731    #[test]
1732    fn help_and_version_win_over_everything_else() {
1733        assert_eq!(parse_args(&args(&["-c", "--help", "x.c"])).unwrap(), Action::Help);
1734        assert_eq!(parse_args(&args(&["--version"])).unwrap(), Action::Version);
1735    }
1736
1737    fn compile(s: &[&str]) -> (Box<Options>, Box<Plan>) {
1738        match parse_args(&args(s)).expect("expected a compilation") {
1739            Action::Compile { opts, plan, .. } => (opts, plan),
1740            other => panic!("expected a compilation, got {other:?}"),
1741        }
1742    }
1743
1744    fn linking(s: &[&str]) -> (Box<LinkOptions>, Box<Plan>) {
1745        match parse_args(&args(s)).expect("expected a compilation") {
1746            Action::Compile { link, plan, .. } => (link, plan),
1747            other => panic!("expected a compilation, got {other:?}"),
1748        }
1749    }
1750
1751    #[test]
1752    fn collects_inputs_and_flags() {
1753        let (opts, plan) = compile(&["-c", "-O2", "-g", "a.c", "b.c"]);
1754        let paths: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
1755        assert_eq!(paths, vec!["a.c", "b.c"]);
1756        assert_eq!(opts.opt_level, OptLevel::O2);
1757        assert_eq!(opts.emit, EmitKind::Object);
1758        assert!(opts.debug_info);
1759    }
1760
1761    /// The unstable options, which are spelled apart from everything else on purpose: what is
1762    /// under `-Z` promises nothing, and a build that reaches for one should have had to say so.
1763    #[test]
1764    fn an_unstable_option_is_taken_and_one_that_does_not_exist_is_refused() {
1765        let (opts, _) = compile(&["-c", "-Zrule-coverage=/tmp/rules.cov", "a.c"]);
1766        assert_eq!(opts.rule_coverage.as_deref(), Some("/tmp/rules.cov"));
1767
1768        let (plain, _) = compile(&["-c", "a.c"]);
1769        assert_eq!(plain.rule_coverage, None, "nothing is measured unless it was asked for");
1770
1771        assert!(parse_args(&args(&["-Zrule-coverage=", "a.c"])).is_err(), "a file with no name");
1772        let unknown = parse_args(&args(&["-Zwhat", "a.c"])).expect_err("there is no such option");
1773        assert!(unknown.message.contains("4.11"), "{}", unknown.message);
1774    }
1775
1776    /// The other measurement written to a file, which reads the same way and fails the same way.
1777    #[test]
1778    fn where_the_register_pressure_goes_is_asked_for_the_same_way() {
1779        let (opts, _) = compile(&["-c", "-O2", "-Zregister-pressure=/tmp/spills.txt", "a.c"]);
1780        assert_eq!(opts.register_pressure.as_deref(), Some("/tmp/spills.txt"));
1781
1782        let (plain, _) = compile(&["-c", "a.c"]);
1783        assert_eq!(plain.register_pressure, None, "nothing is measured unless it was asked for");
1784
1785        assert!(parse_args(&args(&["-Zregister-pressure=", "a.c"])).is_err(), "no file named");
1786    }
1787
1788    #[test]
1789    fn a_bare_dash_o_means_o1_the_way_gcc_reads_it() {
1790        let (opts, _) = compile(&["-O", "a.c"]);
1791        assert_eq!(opts.opt_level, OptLevel::O1);
1792    }
1793
1794    #[test]
1795    fn dash_x_applies_to_later_inputs_only_and_none_stops_it() {
1796        let (_, plan) = compile(&["a.o", "-x", "c", "b.txt", "-x", "none", "c.o"]);
1797        assert_eq!(plan.jobs[0].kind, InputKind::LinkerInput);
1798        assert_eq!(plan.jobs[1].kind, InputKind::C);
1799        assert_eq!(plan.jobs[2].kind, InputKind::LinkerInput);
1800    }
1801
1802    #[test]
1803    fn dash_j_reaches_the_scheduler_and_defaults_to_the_machine() {
1804        let (_, _, jobs) = match parse_args(&args(&["-j4", "a.c"])).unwrap() {
1805            Action::Compile { opts, plan, jobs, .. } => (opts, plan, jobs),
1806            other => panic!("expected a compilation, got {other:?}"),
1807        };
1808        assert_eq!(jobs.count(), 4);
1809
1810        let default = match parse_args(&args(&["a.c"])).unwrap() {
1811            Action::Compile { jobs, .. } => jobs,
1812            other => panic!("expected a compilation, got {other:?}"),
1813        };
1814        assert_eq!(default, Jobs::available());
1815        assert!(parse_args(&args(&["-j0", "a.c"])).is_err());
1816    }
1817
1818    #[test]
1819    fn triple_hash_prints_the_plan_and_runs_nothing() {
1820        let a = parse_args(&args(&["-###", "-c", "a.c"])).unwrap();
1821        let Action::PrintPlan { plan, .. } = a else { panic!("expected a plan dump") };
1822        assert!(plan.render().contains("a.c: preprocess, compile, assemble -> a.o"));
1823    }
1824
1825    #[test]
1826    fn the_flag_that_keeps_the_intermediate_files_has_three_spellings_and_two_meanings() {
1827        // The bare one is `=obj` and not `=cwd`. gcc's manual says the opposite and gcc 16 does
1828        // this, and following the compiler is what makes a build that reads either of them find
1829        // the files where they are.
1830        assert_eq!(compile(&["-c", "-save-temps", "a.c"]).0.save_temps, SaveTemps::Object);
1831        assert_eq!(compile(&["-c", "-save-temps=obj", "a.c"]).0.save_temps, SaveTemps::Object);
1832        assert_eq!(compile(&["-c", "-save-temps=cwd", "a.c"]).0.save_temps, SaveTemps::Cwd);
1833        assert_eq!(compile(&["-c", "a.c"]).0.save_temps, SaveTemps::No);
1834        // The last one on the line decides, the way it does for every other flag with an
1835        // argument, and a keyword that is neither is fatal rather than ignored: a run that kept
1836        // nothing and said nothing looks exactly like one where the files were not produced.
1837        let (opts, _) = compile(&["-c", "-save-temps", "-save-temps=cwd", "a.c"]);
1838        assert_eq!(opts.save_temps, SaveTemps::Cwd);
1839        let e = parse_args(&args(&["-c", "-save-temps=nowhere", "a.c"])).unwrap_err();
1840        assert!(e.message.contains("accepted: cwd, obj"), "{}", e.message);
1841    }
1842
1843    #[test]
1844    fn the_flag_that_times_each_step_reaches_the_options_and_changes_nothing_else() {
1845        let (opts, plan) = compile(&["-c", "-time", "a.c"]);
1846        let (plain, without) = compile(&["-c", "a.c"]);
1847        assert!(opts.time);
1848        assert!(!plain.time);
1849        // Against the same line without the flag rather than against a spelling of the object's
1850        // name, since what the object is called is the host's business and this is not about that.
1851        assert_eq!(plan.jobs[0].output, without.jobs[0].output);
1852    }
1853
1854    #[test]
1855    fn dash_x_names_what_it_accepts_when_it_does_not_know_a_language() {
1856        let e = parse_args(&args(&["-x", "fortran", "a.c"])).unwrap_err();
1857        assert!(e.message.contains("assembler-with-cpp"), "{}", e.message);
1858    }
1859
1860    #[test]
1861    fn an_unknown_flag_is_an_error_rather_than_a_shrug() {
1862        let e = parse_args(&args(&["-fno-such-thing", "a.c"])).unwrap_err();
1863        assert!(e.message.contains("unknown option"), "{}", e.message);
1864    }
1865
1866    /// `-fpermissive` and the flag that turns it back off, which a build writes beside it when
1867    /// one directory needs the older rules and the rest of the tree does not.
1868    #[test]
1869    fn permissive_reads_in_both_directions_and_the_last_one_wins() {
1870        let (opts, _) = compile(&["-c", "a.c"]);
1871        assert!(!opts.permissive, "off unless it is asked for");
1872
1873        let (opts, _) = compile(&["-c", "-fpermissive", "a.c"]);
1874        assert!(opts.permissive);
1875
1876        let (opts, _) = compile(&["-c", "-fpermissive", "-fno-permissive", "a.c"]);
1877        assert!(!opts.permissive);
1878    }
1879
1880    #[test]
1881    fn asking_for_nested_functions_is_told_why_it_is_not_coming() {
1882        let e = parse_args(&args(&["-fnested-functions", "a.c"])).unwrap_err();
1883        assert!(e.message.contains("trampoline"), "{}", e.message);
1884        assert!(parse_args(&args(&["-fno-nested-functions", "a.c"])).is_ok());
1885    }
1886
1887    #[test]
1888    fn the_flag_every_configure_script_writes_is_taken() {
1889        // All four spellings, because a build writes whichever one its macros picked and a
1890        // compiler that takes three of them is a compiler that fails on the fourth.
1891        for flag in ["-fPIC", "-fpic", "-fPIE", "-fpie"] {
1892            let (opts, _) = compile(&["-c", flag, "a.c"]);
1893            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1894        }
1895    }
1896
1897    #[test]
1898    fn a_table_is_written_unless_the_build_says_nothing_will_walk_it() {
1899        let (opts, _) = compile(&["-c", "a.c"]);
1900        assert!(opts.unwinds(), "the default is off");
1901        let (opts, _) = compile(&["-c", "-fno-asynchronous-unwind-tables", "a.c"]);
1902        assert!(!opts.unwinds(), "the build was not taken at its word");
1903        let (opts, _) = compile(&[
1904            "-c",
1905            "-fno-asynchronous-unwind-tables",
1906            "-fasynchronous-unwind-tables",
1907            "a.c",
1908        ]);
1909        assert!(opts.unwinds(), "the last flag did not win");
1910        // The weaker request, which the same table answers, so a line that asks for a table and
1911        // against an asynchronous one gets one. That is gcc's arrangement and it turns up when a
1912        // build turns the asynchronous one off globally and a directory asks for a table back.
1913        let (opts, _) =
1914            compile(&["-c", "-fno-asynchronous-unwind-tables", "-funwind-tables", "a.c"]);
1915        assert!(opts.unwinds(), "the weaker request was dropped");
1916        let (opts, _) = compile(&["-c", "-fno-unwind-tables", "a.c"]);
1917        assert!(opts.unwinds(), "the weaker negative turned off the stronger request");
1918        let (opts, _) =
1919            compile(&["-c", "-fno-unwind-tables", "-fno-asynchronous-unwind-tables", "a.c"]);
1920        assert!(!opts.unwinds(), "both were turned off and one stayed on");
1921    }
1922
1923    #[test]
1924    fn the_flags_that_describe_what_this_compiler_already_does_are_taken() {
1925        // Every one of these is on a real build line somewhere and every one of them was an
1926        // unknown option. What they have in common is that the answer rucc gives is the answer
1927        // they ask for, so there is nothing to implement and nothing to refuse.
1928        for flag in [
1929            "-fno-common",
1930            "-fstrict-aliasing",
1931            "-fno-strict-aliasing",
1932            "-pipe",
1933            "-fdiagnostics-color",
1934            "-fno-diagnostics-color",
1935            "-fdiagnostics-color=always",
1936            "-fdiagnostics-color=never",
1937            "-fdiagnostics-color=auto",
1938        ] {
1939            let (opts, _) = compile(&["-c", flag, "a.c"]);
1940            assert_eq!(opts.emit, EmitKind::Object, "{flag}");
1941        }
1942    }
1943
1944    #[test]
1945    fn asking_the_linker_to_merge_tentative_definitions_is_told_why_it_is_not_coming() {
1946        // The one of that family that is a request rather than a description, and it is a real
1947        // difference: two files each writing `int g;` link under it and do not without it.
1948        let e = parse_args(&args(&["-fcommon", "a.c"])).unwrap_err();
1949        assert!(e.message.contains(".bss"), "{}", e.message);
1950        assert!(e.message.contains("extern"), "the way out is worth saying: {}", e.message);
1951    }
1952
1953    #[test]
1954    fn asking_for_position_dependent_code_is_told_why_it_is_not_coming() {
1955        for flag in ["-fno-pic", "-fno-pie"] {
1956            let e = parse_args(&args(&[flag, "a.c"])).unwrap_err();
1957            assert!(e.message.contains("global offset table"), "{flag}: {}", e.message);
1958            // The one it may have meant, since the two are a letter apart and one of them is
1959            // about linking and is taken.
1960            assert!(e.message.contains("-no-pie"), "{flag}: {}", e.message);
1961        }
1962    }
1963
1964    #[test]
1965    fn an_unsupported_target_names_itself() {
1966        let e = parse_args(&args(&["--target=sparc64-linux-gnu", "a.c"])).unwrap_err();
1967        assert!(e.message.contains("sparc64"), "{}", e.message);
1968    }
1969
1970    #[test]
1971    fn no_inputs_is_an_error_but_print_config_needs_none() {
1972        assert!(parse_args(&args(&[])).is_err());
1973        assert!(matches!(parse_args(&args(&["--print-config"])), Ok(Action::PrintConfig(_))));
1974    }
1975
1976    #[test]
1977    fn print_config_reports_the_target_it_was_given_not_the_host() {
1978        let a = parse_args(&args(&["--print-config", "--target=riscv64-linux-musl"])).unwrap();
1979        let Action::PrintConfig(opts) = a else { panic!("expected a configuration dump") };
1980        let text = print_config(&opts);
1981        assert!(text.contains("target: riscv64-unknown-linux-musl"), "{text}");
1982        assert!(text.contains("char-signed: false"), "{text}");
1983        assert!(text.contains("object-format: elf"), "{text}");
1984        assert!(text.contains("va-list: void-pointer"), "{text}");
1985        // RISC-V has a register file and this compiler has not written it down yet, and the
1986        // dump says which of those two it is rather than leaving the line out.
1987        assert!(text.contains("registers: none"), "{text}");
1988    }
1989
1990    #[test]
1991    fn print_config_has_one_key_per_line_and_a_fixed_order() {
1992        let opts = Options::new("x86_64-unknown-linux-gnu".parse().unwrap());
1993        let text = print_config(&opts);
1994        let keys: Vec<&str> =
1995            text.lines().map(|l| l.split(':').next().unwrap_or_default()).collect();
1996        assert_eq!(keys[0], "version");
1997        assert_eq!(keys[1], "target");
1998        assert_eq!(keys.len(), 21);
1999        assert!(text.ends_with('\n'));
2000    }
2001
2002    #[test]
2003    fn the_safety_tier_is_read_off_the_command_line_and_a_wrong_one_is_refused() {
2004        let (opts, _) = compile(&["a.c"]);
2005        assert_eq!(opts.safety, rucc_session::Safety::Off);
2006
2007        for (flag, tier) in [
2008            ("-fsafety=detect", rucc_session::Safety::Detect),
2009            ("-fsafety=enforce", rucc_session::Safety::Enforce),
2010            ("-fsafety=kernel", rucc_session::Safety::Kernel),
2011            ("-fsafety=off", rucc_session::Safety::Off),
2012        ] {
2013            let (opts, _) = compile(&[flag, "a.c"]);
2014            assert_eq!(opts.safety, tier, "{flag}");
2015        }
2016
2017        // The last one wins, the way every other repeated flag on this command line does.
2018        let (opts, _) = compile(&["-fsafety=enforce", "-fsafety=off", "a.c"]);
2019        assert_eq!(opts.safety, rucc_session::Safety::Off);
2020
2021        // A misspelled tier is refused rather than ignored. Silently compiling without the
2022        // monitor a build asked for is the one failure mode this feature cannot have.
2023        let e = parse_args(&args(&["-fsafety=on", "a.c"])).unwrap_err();
2024        assert!(e.message.contains("is not a safety tier"), "{}", e.message);
2025        assert!(parse_args(&args(&["-fsafety", "a.c"])).is_err());
2026    }
2027
2028    #[test]
2029    fn print_pipeline_answers_with_the_passes_the_level_asked_for() {
2030        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2031        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2032        let text = print_pipeline(&opts);
2033        assert!(text.starts_with("level: -O2\n"), "{text}");
2034        assert!(text.contains("fold"), "{text}");
2035
2036        let a = parse_args(&args(&["--print-pipeline"])).unwrap();
2037        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2038        // One pass runs at `-O0` and it is the one that removes code nothing reaches, which is
2039        // not an optimization. See issue 359.
2040        assert!(print_pipeline(&opts).contains("1: simplify-cfg,"), "{}", print_pipeline(&opts));
2041
2042        let a = parse_args(&args(&["--print-pipeline", "-fno-simplify-cfg"])).unwrap();
2043        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2044        // And with that one turned off there is nothing left, which the dump says rather than
2045        // printing an empty list.
2046        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2047    }
2048
2049    #[test]
2050    fn print_pipeline_takes_the_toggles_into_account() {
2051        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fno-fold"])).unwrap();
2052        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2053        let text = print_pipeline(&opts);
2054        // The one that was named is gone and the rest of the level is not, which is the whole
2055        // of what a toggle promises.
2056        assert!(!text.contains("fold"), "{text}");
2057        assert!(text.contains("dce"), "{text}");
2058
2059        // Every pass the compiler has, named off. Built from the registry rather than written
2060        // out, so a pass added later is turned off here too and this keeps testing the thing it
2061        // is about, which is that the toggles can empty a level.
2062        let mut off = vec!["--print-pipeline".to_owned(), "-O2".to_owned()];
2063        off.extend(rucc_opt::PASSES.iter().map(|p| format!("-fno-{}", p.name())));
2064        let spelled: Vec<&str> = off.iter().map(String::as_str).collect();
2065        let a = parse_args(&args(&spelled)).unwrap();
2066        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2067        assert!(print_pipeline(&opts).contains("no passes"), "{}", print_pipeline(&opts));
2068    }
2069
2070    #[test]
2071    fn print_pipeline_says_when_a_budget_will_stop_the_run_short() {
2072        let a = parse_args(&args(&["--print-pipeline", "-O2"])).unwrap();
2073        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2074        assert!(!print_pipeline(&opts).contains("global fuel"));
2075
2076        let a = parse_args(&args(&["--print-pipeline", "-O2", "-fpass-fuel-global=4"])).unwrap();
2077        let Action::PrintPipeline(opts) = a else { panic!("expected a pipeline dump") };
2078        let text = print_pipeline(&opts);
2079        // Because the listing is the answer to what this compilation will do, and a run that
2080        // stops after four rewrites is not doing what the level says it does.
2081        assert!(text.contains("global fuel: 4"), "{text}");
2082    }
2083
2084    /// A pass is turned on and off by its own name, and the order the flags were given in is
2085    /// kept, because the last spelling of a name is the one that decides.
2086    #[test]
2087    fn a_pass_is_named_by_dash_f_and_unnamed_by_dash_f_no() {
2088        let (opts, _) = compile(&["-c", "-O0", "-ffold", "-fno-fold", "-ffold", "a.c"]);
2089        assert_eq!(
2090            opts.passes,
2091            [("fold".to_owned(), true), ("fold".to_owned(), false), ("fold".to_owned(), true)]
2092        );
2093
2094        let e = parse_args(&args(&["-fno-such-pass", "a.c"])).unwrap_err();
2095        assert!(e.message.contains("unknown option"), "{}", e.message);
2096    }
2097
2098    #[test]
2099    fn pass_fuel_names_a_pass_and_a_count_and_refuses_anything_else() {
2100        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel=fold=3", "a.c"]);
2101        assert_eq!(opts.pass_fuel, [("fold".to_owned(), 3)]);
2102
2103        let e = parse_args(&args(&["-fpass-fuel=fold", "a.c"])).unwrap_err();
2104        assert!(e.message.contains("<pass>=<count>"), "{}", e.message);
2105        let e = parse_args(&args(&["-fpass-fuel=nosuch=3", "a.c"])).unwrap_err();
2106        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2107        let e = parse_args(&args(&["-fpass-fuel=fold=lots", "a.c"])).unwrap_err();
2108        assert!(e.message.contains("not a number"), "{}", e.message);
2109    }
2110
2111    #[test]
2112    fn global_pass_fuel_is_a_count_on_its_own_and_defaults_to_no_limit() {
2113        let (opts, _) = compile(&["-c", "-O2", "a.c"]);
2114        assert_eq!(opts.pass_fuel_global, None);
2115
2116        let (opts, _) = compile(&["-c", "-O2", "-fpass-fuel-global=12", "a.c"]);
2117        assert_eq!(opts.pass_fuel_global, Some(12));
2118        // And it is not the per pass flag with a longer name, so neither spelling swallows the
2119        // other.
2120        assert!(opts.pass_fuel.is_empty());
2121
2122        let e = parse_args(&args(&["-fpass-fuel-global=lots", "a.c"])).unwrap_err();
2123        assert!(e.message.contains("not a number"), "{}", e.message);
2124    }
2125
2126    #[test]
2127    fn a_gate_names_a_pass_and_optionally_the_functions_it_covers() {
2128        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold", "-fenable-fold=2-4,main", "a.c"]);
2129        assert_eq!(
2130            opts.pass_gates,
2131            [(false, "fold".to_owned()), (true, "fold=2-4,main".to_owned())],
2132            "the order is what decides, so it has to survive the parse"
2133        );
2134
2135        let e = parse_args(&args(&["-fdisable-nosuch", "a.c"])).unwrap_err();
2136        assert!(e.message.contains("--print-pipeline"), "{}", e.message);
2137        let e = parse_args(&args(&["-fenable-fold=9-2", "a.c"])).unwrap_err();
2138        assert!(e.message.contains("ends before it starts"), "{}", e.message);
2139        let e = parse_args(&args(&["-fdisable-fold=", "a.c"])).unwrap_err();
2140        assert!(e.message.contains("is empty"), "{}", e.message);
2141    }
2142
2143    #[test]
2144    fn the_pipeline_listing_says_which_passes_a_gate_touched() {
2145        let (opts, _) = compile(&["-c", "-O2", "-fdisable-fold=main", "a.c"]);
2146        let text = print_pipeline(&opts);
2147        assert!(text.contains("fold, "), "{text}");
2148        assert!(text.contains("[off for main]"), "{text}");
2149    }
2150
2151    /// The spelling is checked while the arguments are read, because a dump that names a pass
2152    /// this compiler does not have is a typo, and a typo found after the compilation has run is
2153    /// found too late to be any use.
2154    #[test]
2155    fn a_dump_is_checked_when_it_is_asked_for_rather_than_when_it_is_taken() {
2156        let (opts, _) = compile(&["-c", "-O2", "-fdump-ir=all", "-fdump-ir=after-fold", "a.c"]);
2157        assert_eq!(opts.dump_ir, ["all", "after-fold"]);
2158
2159        let e = parse_args(&args(&["-fdump-ir=after-nosuch", "a.c"])).unwrap_err();
2160        assert!(e.message.contains("nosuch"), "{}", e.message);
2161        assert!(parse_args(&args(&["-fdump-ir=sideways-fold", "a.c"])).is_err());
2162    }
2163
2164    /// Every spelling `-fopt-info` takes, and the one it does not.
2165    ///
2166    /// The keywords are checked here for the same reason a dump's pass name is: a person who
2167    /// misspelled one gets no output, and no output is also what a compilation where nothing
2168    /// happened looks like. Telling those two apart is the entire reason to reach for this flag.
2169    #[test]
2170    fn opt_info_takes_kinds_and_a_file_and_refuses_a_kind_it_does_not_have() {
2171        let (opts, _) = compile(&["-c", "-O2", "-fopt-info", "a.c"]);
2172        assert_eq!(opts.opt_info, [""], "a bare flag asks for the rewrites");
2173        assert_eq!(opts.opt_info_file, None, "and goes to standard error");
2174
2175        let (opts, _) = compile(&["-c", "-O2", "-fopt-info-missed-note", "a.c"]);
2176        assert_eq!(opts.opt_info, ["missed-note"]);
2177
2178        // Two flags add up rather than the second replacing the first, and the file is the last
2179        // one that named a file, which is how GCC treats both.
2180        let (opts, _) =
2181            compile(&["-c", "-O2", "-fopt-info-missed=one.txt", "-fopt-info-all=two.txt", "a.c"]);
2182        assert_eq!(opts.opt_info, ["missed", "all"]);
2183        assert_eq!(opts.opt_info_file.as_deref(), Some("two.txt"));
2184
2185        let e = parse_args(&args(&["-fopt-info-vectorized", "a.c"])).unwrap_err();
2186        assert!(e.message.contains("vectorized"), "{}", e.message);
2187        assert!(e.message.contains("`missed`"), "{}", e.message);
2188        let e = parse_args(&args(&["-fopt-info-missed=", "a.c"])).unwrap_err();
2189        assert!(e.message.contains("no file"), "{}", e.message);
2190    }
2191
2192    #[test]
2193    fn verify_each_is_unstable_and_off_unless_it_was_asked_for() {
2194        let (opts, _) = compile(&["-c", "-Zverify-each", "a.c"]);
2195        assert!(opts.verify_each);
2196        assert!(!USAGE.contains("verify-each"), "an unstable option stays out of the usage text");
2197    }
2198
2199    #[test]
2200    fn dash_o_needs_an_argument() {
2201        let e = parse_args(&args(&["a.c", "-o"])).unwrap_err();
2202        assert_eq!(e.message, "-o requires an argument");
2203    }
2204
2205    #[test]
2206    fn dash_d_and_dash_u_are_read_joined_or_separated_and_keep_their_order() {
2207        let (opts, _) = compile(&["-DFOO=1", "-D", "BAR", "-UBAZ", "-U", "QUX", "a.c"]);
2208        assert_eq!(opts.defines, ["FOO=1", "BAR"]);
2209        assert_eq!(opts.undefines, ["BAZ", "QUX"]);
2210    }
2211
2212    #[test]
2213    fn the_include_flags_land_on_the_chain_each_one_names() {
2214        // A sysroot with nothing under it, so that the library's own directories are the
2215        // same on every machine this test runs on, which is none of them.
2216        let (opts, _) = compile(&[
2217            "-Ii",
2218            "-iquote",
2219            "q",
2220            "-isystem",
2221            "sys",
2222            "-idirafter",
2223            "after",
2224            "--sysroot=/nowhere-at-all",
2225            "a.c",
2226        ]);
2227        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2228        // The compiler's own headers sit after every `-isystem` and before `-idirafter`,
2229        // which is where GCC puts its own: a directory the user named outranks ours.
2230        assert_eq!(dirs, ["q", "i", "sys", runtime::DIR, "after"]);
2231        assert!(!opts.search.dirs()[1].is_system);
2232        assert!(opts.search.dirs()[2].is_system);
2233    }
2234
2235    #[test]
2236    fn the_librarys_headers_come_after_the_compilers_own_and_go_away_with_them() {
2237        // Which machine this runs on decides what is on the path, so the test is about the
2238        // order rather than about the names: ours is on it, the library's follow it, and
2239        // `-nostdinc` is the one flag that takes both halves of the pair off at once.
2240        let (opts, _) = compile(&["a.c"]);
2241        let dirs = opts.search.dirs();
2242        let ours = dirs.iter().position(|d| d.path.to_str() == Some(runtime::DIR));
2243        assert_eq!(ours, Some(0), "{dirs:?}");
2244        assert!(dirs[1..].iter().all(|d| d.is_system), "{dirs:?}");
2245        let (bare, _) = compile(&["-nostdinc", "a.c"]);
2246        assert!(bare.search.dirs().is_empty(), "{:?}", bare.search.dirs());
2247    }
2248
2249    #[test]
2250    fn a_sysroot_moves_the_librarys_directories_and_nothing_else() {
2251        let (opts, _) = compile(&["-isystem", "sys", "--sysroot=/nowhere-at-all", "a.c"]);
2252        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2253        assert_eq!(dirs, ["sys", runtime::DIR]);
2254    }
2255
2256    #[test]
2257    fn dash_i_dash_moves_the_bracket_directories_into_the_quoted_chain() {
2258        let (opts, _) =
2259            compile(&["-Iinc1", "-iquote", "inc2", "-I-", "-Iinc3", "-nostdinc", "a.c"]);
2260        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2261        assert_eq!(dirs, ["inc1", "inc2", "inc3"]);
2262        // An angled include sees only what came after the flag.
2263        assert_eq!(opts.search.start(IncludeForm::Angled), 2);
2264        assert!(!opts.search.searches_current_dir());
2265    }
2266
2267    #[test]
2268    fn the_prefix_flags_stick_what_iprefix_said_on_the_front_of_what_follows_it() {
2269        let (opts, _) = compile(&[
2270            "-iprefix",
2271            "/tools/",
2272            "-iwithprefix",
2273            "late",
2274            "-iwithprefixbefore",
2275            "early",
2276            "-iprefix",
2277            "/other/",
2278            "-iwithprefix",
2279            "last",
2280            "-nostdinc",
2281            "a.c",
2282        ]);
2283        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2284        // `-iwithprefixbefore` is an `-I` and the other two are `-isystem`, which is where GCC
2285        // puts them rather than where its manual says it does.
2286        assert_eq!(dirs, ["/tools/early", "/tools/late", "/other/last"]);
2287        assert!(!opts.search.dirs()[0].is_system);
2288        assert!(opts.search.dirs()[1].is_system);
2289    }
2290
2291    #[test]
2292    fn the_files_named_on_the_command_line_keep_their_order_and_which_flag_named_them() {
2293        let (opts, _) =
2294            compile(&["-include", "one.h", "-imacros", "two.h", "-include", "3.h", "a.c"]);
2295        let names: Vec<&str> = opts.preincludes.iter().map(|p| p.name.as_str()).collect();
2296        assert_eq!(names, ["one.h", "two.h", "3.h"]);
2297        assert_eq!(opts.preincludes.iter().filter(|p| p.macros_only).count(), 1);
2298    }
2299
2300    #[test]
2301    fn nostdinc_takes_the_compilers_own_headers_off_the_path() {
2302        let (opts, _) = compile(&["-Ii", "-nostdinc", "a.c"]);
2303        let dirs: Vec<&str> = opts.search.dirs().iter().filter_map(|d| d.path.to_str()).collect();
2304        assert_eq!(dirs, ["i"]);
2305    }
2306
2307    #[test]
2308    fn the_dialect_flags_set_the_language_and_the_extensions_separately() {
2309        let (opts, _) = compile(&["-std=gnu11", "a.c"]);
2310        assert_eq!(opts.std, Std::C11);
2311        assert!(opts.gnu_extensions);
2312
2313        let (opts, _) = compile(&["-std=iso9899:1999", "a.c"]);
2314        assert_eq!(opts.std, Std::C99);
2315        assert!(!opts.gnu_extensions);
2316
2317        let (opts, _) = compile(&["-ansi", "a.c"]);
2318        assert_eq!(opts.std, Std::C89);
2319        assert!(!opts.gnu_extensions);
2320
2321        let e = parse_args(&args(&["-std=c94jr", "a.c"])).unwrap_err();
2322        assert!(e.message.contains("unknown dialect"), "{}", e.message);
2323    }
2324
2325    #[test]
2326    fn the_dump_letters_are_a_family_and_everything_else_beginning_with_d_is_not() {
2327        let (opts, _) = compile(&["-dM", "a.c"]);
2328        assert!(opts.dumps.macros);
2329
2330        // Packed, the way GCC takes them, and a letter in the family we have not written yet
2331        // is accepted and does nothing rather than failing a build.
2332        let (opts, _) = compile(&["-dDM", "a.c"]);
2333        assert!(opts.dumps.macros);
2334        let (opts, _) = compile(&["-dD", "a.c"]);
2335        assert!(!opts.dumps.macros);
2336
2337        let (opts, _) = compile(&["a.c"]);
2338        assert!(!opts.dumps.any());
2339
2340        // `-dumpversion` is a different flag that happens to start the same way, and it is read
2341        // as itself rather than as a dump of nothing.
2342        assert_eq!(printed(&["-dumpversion", "a.c"]), VERSION);
2343    }
2344
2345    #[test]
2346    fn the_gcc_version_claimed_is_a_flag_and_the_short_spellings_are_the_ones_people_write() {
2347        let (opts, _) = compile(&["a.c"]);
2348        assert_eq!(
2349            opts.gnuc,
2350            GnucVersion { major: 7, minor: 0, patch: 0 },
2351            "the lowest claim a modern glibc gives its own declarations to"
2352        );
2353
2354        let (opts, _) = compile(&["-fgnuc-version=15.1.0", "a.c"]);
2355        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 1, patch: 0 });
2356
2357        // A missing component is zero. `gcc -dumpversion` says `15` on a release with no
2358        // patchlevel and a harness that pastes that back has to be understood.
2359        let (opts, _) = compile(&["-fgnuc-version=15", "a.c"]);
2360        assert_eq!(opts.gnuc, GnucVersion { major: 15, minor: 0, patch: 0 });
2361
2362        let (opts, _) = compile(&["-fgnuc-version=13.2", "a.c"]);
2363        assert_eq!(opts.gnuc, GnucVersion { major: 13, minor: 2, patch: 0 });
2364
2365        let e = parse_args(&args(&["-fgnuc-version=15.x", "a.c"])).unwrap_err();
2366        assert!(e.message.contains("minor that is not a number"), "{}", e.message);
2367
2368        let e = parse_args(&args(&["-fgnuc-version=1.2.3.4", "a.c"])).unwrap_err();
2369        assert!(e.message.contains("more than three"), "{}", e.message);
2370    }
2371
2372    #[test]
2373    fn pedantic_has_two_spellings_and_is_not_the_same_knob_as_the_dialect() {
2374        let (opts, _) = compile(&["-std=c17", "-pedantic", "a.c"]);
2375        assert!(opts.pedantic);
2376        assert_eq!(opts.std, Std::C17);
2377
2378        // The `-W` family's name for it, which is what a build that groups its warning flags
2379        // tends to write.
2380        let (opts, _) = compile(&["-Wpedantic", "a.c"]);
2381        assert!(opts.pedantic);
2382
2383        let (opts, _) = compile(&["-std=c17", "a.c"]);
2384        assert!(!opts.pedantic, "a dialect on its own does not diagnose an extension");
2385    }
2386
2387    #[test]
2388    fn dash_p_and_dash_ffreestanding_reach_the_options() {
2389        let (opts, _) = compile(&["-E", "-P", "-ffreestanding", "a.c"]);
2390        assert!(!opts.line_markers);
2391        assert!(!opts.hosted);
2392        assert_eq!(opts.emit, EmitKind::Preprocessed);
2393    }
2394
2395    /// The two ways a build says it means its own function by a name the C library also has.
2396    ///
2397    /// `-fno-builtin` is all of them and `-fno-builtin-<name>` is one, and the second is what a
2398    /// build writes when it means its own `memcpy` and the library's everything else. The name is
2399    /// kept as it was written and not checked against anything, because a program is allowed to
2400    /// mean something by a name this compiler has never heard of.
2401    #[test]
2402    fn the_builtin_flags_are_read_in_both_directions_and_one_name_at_a_time() {
2403        let (opts, _) = compile(&["-c", "a.c"]);
2404        assert!(opts.builtins, "a library name means the library function by default");
2405        assert!(opts.no_builtin.is_empty());
2406
2407        let (opts, _) = compile(&["-c", "-fno-builtin", "a.c"]);
2408        assert!(!opts.builtins);
2409
2410        let (opts, _) = compile(&["-c", "-fno-builtin", "-fbuiltin", "a.c"]);
2411        assert!(opts.builtins, "the last mention decides");
2412
2413        let (opts, _) = compile(&["-c", "-fno-builtin-memcpy", "-fno-builtin-nonesuch", "a.c"]);
2414        assert!(opts.builtins, "one name is not the family");
2415        assert_eq!(opts.no_builtin, vec!["memcpy".to_owned(), "nonesuch".to_owned()]);
2416    }
2417
2418    /// `-fvisibility=`, which is on every cmake project that cares about which names it exports
2419    /// and which was refused as an unknown option until now.
2420    ///
2421    /// Four spellings and three answers. `internal` is hidden plus a promise about never taking
2422    /// the address across a component boundary, and nothing derives anything from that promise
2423    /// here, so it comes out as the weaker of the two rather than as a refusal that stops a build
2424    /// over a distinction this compiler does not make.
2425    #[test]
2426    fn visibility_takes_the_four_spellings_gcc_takes_and_refuses_the_rest() {
2427        let (opts, _) = compile(&["-c", "a.c"]);
2428        assert_eq!(opts.visibility, Visibility::Default, "exported unless something says not");
2429
2430        for (written, wanted) in [
2431            ("default", Visibility::Default),
2432            ("hidden", Visibility::Hidden),
2433            ("internal", Visibility::Hidden),
2434            ("protected", Visibility::Protected),
2435        ] {
2436            let (opts, _) = compile(&["-c", &format!("-fvisibility={written}"), "a.c"]);
2437            assert_eq!(opts.visibility, wanted, "{written}");
2438        }
2439
2440        // The last mention decides, which is what every other flag of this shape does and what a
2441        // build that turns something off for one directory relies on.
2442        let (opts, _) = compile(&["-c", "-fvisibility=hidden", "-fvisibility=default", "a.c"]);
2443        assert_eq!(opts.visibility, Visibility::Default, "the last mention decides");
2444
2445        // A spelling gcc does not take is refused rather than read as the default, because a
2446        // build that meant hidden and got exported is a library with the wrong interface and
2447        // nothing said about it anywhere.
2448        let failed = parse_args(&args(&["-fvisibility=none", "a.c"])).expect_err("refused");
2449        assert!(failed.to_string().contains("is not a visibility"), "{failed}");
2450    }
2451
2452    /// `-ffunction-sections` and `-fdata-sections`, which are what make `--gc-sections` able to
2453    /// drop anything: a linker can leave out a section nothing reaches and cannot leave out half of
2454    /// one. A kernel and an embedded image are both linked that way.
2455    ///
2456    /// Two flags rather than one because gcc has two, and a build that asks for one of them and not
2457    /// the other is a build that measured something: splitting the code is nearly free at link time
2458    /// and splitting the data can defeat the linker's ordering of what is next to what.
2459    #[test]
2460    fn a_section_per_function_and_a_section_per_variable_are_asked_for_one_at_a_time() {
2461        let (opts, _) = compile(&["-c", "a.c"]);
2462        assert!(!opts.function_sections, "one text section unless something says otherwise");
2463        assert!(!opts.data_sections);
2464
2465        let (opts, _) = compile(&["-c", "-ffunction-sections", "a.c"]);
2466        assert!(opts.function_sections);
2467        assert!(!opts.data_sections, "one flag is not the other");
2468
2469        let (opts, _) = compile(&["-c", "-fdata-sections", "a.c"]);
2470        assert!(opts.data_sections);
2471        assert!(!opts.function_sections);
2472
2473        // Both directions taken, and the off one is what happens anyway rather than a refusal,
2474        // since a build that writes it is asking for the default.
2475        let (opts, _) = compile(&[
2476            "-c",
2477            "-ffunction-sections",
2478            "-fno-function-sections",
2479            "-fdata-sections",
2480            "-fno-data-sections",
2481            "a.c",
2482        ]);
2483        assert!(!opts.function_sections, "the last mention decides");
2484        assert!(!opts.data_sections, "the last mention decides");
2485    }
2486
2487    /// `-fgnu89-inline`, which is off by default and is not implied by anything on the command
2488    /// line, since the dialect asks for GNU's reading further in rather than through this.
2489    #[test]
2490    fn gnu89_inline_is_off_until_it_is_asked_for_and_the_last_mention_decides() {
2491        let (opts, _) = compile(&["-c", "a.c"]);
2492        assert!(!opts.gnu89_inline, "C's reading of inline by default");
2493
2494        let (opts, _) = compile(&["-c", "-fgnu89-inline", "a.c"]);
2495        assert!(opts.gnu89_inline);
2496
2497        let (opts, _) = compile(&["-c", "-fgnu89-inline", "-fno-gnu89-inline", "a.c"]);
2498        assert!(!opts.gnu89_inline, "the last mention decides");
2499
2500        // The C89 dialects are under GNU's reading whether this was written or not, so the flag
2501        // stays off there and the dialect is what the checker and the macro set both ask. That is
2502        // also why `-std=c89 -fno-gnu89-inline` needs no diagnostic: it asks for the reading the
2503        // dialect already has. gcc refuses that command line, which is measured in the issue.
2504        let (opts, _) = compile(&["-c", "-std=c89", "a.c"]);
2505        assert!(!opts.gnu89_inline);
2506    }
2507
2508    /// Both spellings of both frame flags, since a build that wants one usually writes the
2509    /// other beside it for the one file that has to be compiled the ordinary way.
2510    #[test]
2511    fn the_two_frame_flags_are_read_in_both_directions() {
2512        let (opts, _) = compile(&["-c", "a.c"]);
2513        assert!(!opts.frame_pointer, "gcc omits it above -O0 and so does this");
2514        assert!(opts.red_zone, "the psABI has one and nothing said not to use it");
2515
2516        let (opts, _) = compile(&["-c", "-fno-omit-frame-pointer", "-mno-red-zone", "a.c"]);
2517        assert!(opts.frame_pointer);
2518        assert!(!opts.red_zone);
2519
2520        let (opts, _) = compile(&[
2521            "-c",
2522            "-fno-omit-frame-pointer",
2523            "-fomit-frame-pointer",
2524            "-mno-red-zone",
2525            "-mred-zone",
2526            "a.c",
2527        ]);
2528        assert!(!opts.frame_pointer, "the last one wins, as it does in gcc");
2529        assert!(opts.red_zone);
2530    }
2531
2532    /// Four flags rather than one with an argument, which is how gcc spells them, and the negative
2533    /// spelled three ways because a build that turns one off writes whichever it turned on.
2534    #[test]
2535    fn the_stack_protector_is_four_flags_and_the_last_one_wins() {
2536        let (opts, _) = compile(&["-c", "a.c"]);
2537        assert_eq!(opts.protector, Protector::None, "gcc protects nothing unless it was asked");
2538
2539        for (flag, want) in [
2540            ("-fstack-protector", Protector::Buffers),
2541            ("-fstack-protector-strong", Protector::Strong),
2542            ("-fstack-protector-all", Protector::All),
2543        ] {
2544            let (opts, _) = compile(&["-c", flag, "a.c"]);
2545            assert_eq!(opts.protector, want, "{flag}");
2546        }
2547
2548        // What a package build does: the strong one in the global flags and one directory that
2549        // cannot have a protector turning it off on the line after.
2550        for off in ["-fno-stack-protector", "-fno-stack-protector-strong"] {
2551            let (opts, _) = compile(&["-c", "-fstack-protector-strong", off, "a.c"]);
2552            assert_eq!(opts.protector, Protector::None, "{off}");
2553        }
2554        let (opts, _) = compile(&["-c", "-fno-stack-protector", "-fstack-protector-all", "a.c"]);
2555        assert_eq!(opts.protector, Protector::All, "the last one wins either way round");
2556    }
2557
2558    /// A switch rather than a level, because how a frame is taken is one question and which
2559    /// functions get a canary is another, and gcc spells it that way for the same reason.
2560    #[test]
2561    fn taking_a_frame_a_page_at_a_time_is_off_until_it_is_asked_for() {
2562        let (opts, _) = compile(&["-c", "a.c"]);
2563        assert!(!opts.stack_clash, "gcc takes a frame in one subtraction unless it was asked");
2564
2565        let (opts, _) = compile(&["-c", "-fstack-clash-protection", "a.c"]);
2566        assert!(opts.stack_clash);
2567
2568        // The same shape a package build uses for the protector: on in the global flags and off
2569        // for the one directory that cannot have it.
2570        let (opts, _) =
2571            compile(&["-c", "-fstack-clash-protection", "-fno-stack-clash-protection", "a.c"]);
2572        assert!(!opts.stack_clash);
2573        let (opts, _) =
2574            compile(&["-c", "-fno-stack-clash-protection", "-fstack-clash-protection", "a.c"]);
2575        assert!(opts.stack_clash, "the last one wins either way round");
2576
2577        // The two are independent, since one is about the frame and the other about the function.
2578        let (opts, _) =
2579            compile(&["-c", "-fstack-clash-protection", "-fstack-protector-strong", "a.c"]);
2580        assert!(opts.stack_clash);
2581        assert_eq!(opts.protector, Protector::Strong);
2582    }
2583
2584    #[test]
2585    fn the_link_flags_are_collected_apart_from_the_compilation() {
2586        let (link, _) = linking(&[
2587            "-static",
2588            "-nostartfiles",
2589            "-rdynamic",
2590            "-s",
2591            "-fuse-ld=mold",
2592            "-L/opt/lib",
2593            "-B",
2594            "/opt/tools",
2595            "a.c",
2596        ]);
2597        assert!(link.is_static);
2598        assert!(link.no_startfiles);
2599        assert!(link.export_dynamic);
2600        assert!(link.strip);
2601        assert_eq!(link.use_ld.as_deref(), Some("mold"));
2602        assert_eq!(link.search, vec![PathBuf::from("/opt/lib")]);
2603        assert_eq!(link.prefixes, vec![PathBuf::from("/opt/tools")]);
2604    }
2605
2606    #[test]
2607    fn a_comma_in_dash_wl_separates_two_arguments() {
2608        let (link, _) = linking(&["-Wl,-rpath,/opt/lib", "-Xlinker", "--as-needed", "a.c"]);
2609        assert_eq!(link.passthrough, vec!["-rpath", "/opt/lib", "--as-needed"]);
2610    }
2611
2612    #[test]
2613    fn a_library_keeps_its_place_between_the_objects() {
2614        // Link order is semantic: `-lm` written between two files resolves for the one before
2615        // it and not for the one after, so a library cannot be collected into a list of its own.
2616        // The target is named because the suffix of an object is the target's and this asserts
2617        // on the names: the same command line on a Windows host plans two `.obj` files.
2618        let (_, plan) = linking(&["--target=x86_64-unknown-linux-gnu", "a.c", "-lm", "b.c"]);
2619        let link = plan.link.expect("expected a link step");
2620        assert_eq!(
2621            link.inputs,
2622            vec![
2623                link::Item::File("a.o".into()),
2624                link::Item::Library("m".into()),
2625                link::Item::File("b.o".into()),
2626            ]
2627        );
2628        // And it is not a job, because there is nothing to compile in a library.
2629        assert_eq!(plan.jobs.len(), 2);
2630    }
2631
2632    #[test]
2633    fn a_library_on_a_dash_c_line_is_a_note_rather_than_an_error() {
2634        let (_, plan) = linking(&["-c", "-lm", "a.c"]);
2635        assert!(plan.link.is_none());
2636        assert!(plan.notes.iter().any(|n| n.contains("-lm")), "{:?}", plan.notes);
2637    }
2638
2639    #[test]
2640    fn the_sysroot_reaches_the_linker_as_well_as_the_headers() {
2641        let (link, _) = linking(&["--sysroot=/opt/root", "a.c"]);
2642        assert_eq!(link.sysroot, Some(PathBuf::from("/opt/root")));
2643    }
2644
2645    fn printed(s: &[&str]) -> String {
2646        match parse_args(&args(s)).expect("expected an answer") {
2647            Action::Print(line) => line,
2648            other => panic!("expected an answer, got {other:?}"),
2649        }
2650    }
2651
2652    fn refused(s: &[&str]) -> String {
2653        parse_args(&args(s)).expect_err("expected a refusal").message
2654    }
2655
2656    #[test]
2657    fn a_warning_flag_this_compiler_has_not_heard_of_is_taken_rather_than_refused() {
2658        // The rule in section 4.1, and the reason for it is autoconf: a configure script finds
2659        // out whether a warning flag exists by passing it and looking at the exit status, so a
2660        // compiler that refuses one it does not know fails a script written for a newer GCC.
2661        let (opts, _) = compile(&["-Wall", "-Wextra", "-Wno-format-truncation", "-c", "a.c"]);
2662        assert!(!opts.warnings_are_errors);
2663        assert!(opts.warnings);
2664        // The two spellings that do mean something are still read.
2665        let (opts, _) = compile(&["-Werror", "-c", "a.c"]);
2666        assert!(opts.warnings_are_errors);
2667        let (opts, _) = compile(&["-w", "-c", "a.c"]);
2668        assert!(!opts.warnings);
2669        let (opts, _) = compile(&["-pedantic-errors", "-c", "a.c"]);
2670        assert!(opts.pedantic && opts.warnings_are_errors);
2671    }
2672
2673    #[test]
2674    fn an_argument_for_a_separate_tool_is_refused_rather_than_dropped() {
2675        // Every one of these says something about the output, so the wrong answer is silence.
2676        assert!(refused(&["-Wa,--noexecstack", "-c", "a.c"]).contains("separate assembler"));
2677        assert!(refused(&["-Wp,-DX", "-c", "a.c"]).contains("separate assembler"));
2678        assert!(refused(&["-specs=/x", "a.c"]).contains("-specs= is not supported"));
2679        assert!(refused(&["-mcmodel=kernel", "-c", "a.c"]).contains("small code model"));
2680        assert!(refused(&["-gdwarf-4", "-c", "a.c"]).contains("DWARF 5"));
2681        assert!(refused(&["-Ofast", "-c", "a.c"]).contains("fast math"));
2682        // The word size the target does not have, which is a target this compiler was not asked
2683        // for rather than a flag it does not know.
2684        let no32 = refused(&["--target=x86_64-unknown-linux-gnu", "-m32", "-c", "a.c"]);
2685        assert!(no32.contains("32 bit target"), "{no32}");
2686    }
2687
2688    #[test]
2689    fn the_levels_gcc_spells_differently_are_the_levels_they_mean() {
2690        assert_eq!(compile(&["-O", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2691        assert_eq!(compile(&["-Og", "-c", "a.c"]).0.opt_level, OptLevel::O1);
2692        assert_eq!(compile(&["-O2", "-c", "a.c"]).0.opt_level, OptLevel::O2);
2693    }
2694
2695    #[test]
2696    fn the_machine_flags_that_name_what_we_already_do_are_taken_and_the_rest_are_not() {
2697        let line = ["--target=x86_64-unknown-linux-gnu", "-m64", "-march=x86-64-v3"];
2698        let (opts, _) =
2699            compile(&[&line[..], &["-mtune=native", "-mabi=sysv", "-c", "a.c"]].concat());
2700        assert_eq!(opts.target.to_string(), "x86_64-unknown-linux-gnu");
2701        let wrong = refused(&["--target=x86_64-unknown-linux-gnu", "-mabi=ms", "-c", "a.c"]);
2702        assert!(wrong.contains("sysv convention"), "{wrong}");
2703    }
2704
2705    #[test]
2706    fn the_thread_flag_is_a_macro_and_a_library_and_the_library_goes_last() {
2707        let (opts, plan) = compile(&["-pthread", "-c", "a.c"]);
2708        assert!(opts.defines.iter().any(|d| d == "_REENTRANT"));
2709        // After the input, because a static link takes what it needs from a library when it
2710        // reaches it and not afterwards.
2711        let names: Vec<&str> = plan.jobs.iter().map(|j| j.input.as_str()).collect();
2712        assert_eq!(names, vec!["a.c"]);
2713    }
2714
2715    #[test]
2716    fn the_questions_a_build_system_asks_before_it_compiles_anything() {
2717        let target = "--target=x86_64-unknown-linux-gnu";
2718        assert_eq!(printed(&[target, "-dumpmachine"]), "x86_64-unknown-linux-gnu");
2719        assert_eq!(printed(&[target, "-dumpversion"]), VERSION);
2720        assert_eq!(printed(&[target, "-dumpfullversion"]), VERSION);
2721        assert_eq!(printed(&[target, "-print-multiarch"]), "x86_64-linux-gnu");
2722        // A name nothing holds comes back unchanged, which is GCC's rule and is what makes the
2723        // answer safe to paste into a link line whether or not the file is there.
2724        assert_eq!(printed(&[target, "-print-file-name=no-such-library.a"]), "no-such-library.a");
2725        assert_eq!(printed(&[target, "-print-prog-name=ld"]), "ld");
2726        let dirs = printed(&[target, "-print-search-dirs"]);
2727        assert!(dirs.starts_with("install: "), "{dirs}");
2728        assert!(dirs.contains("\nlibraries: ="), "{dirs}");
2729    }
2730
2731    #[test]
2732    fn the_two_dependency_flags_that_stop_after_the_rule_stop_after_the_rule() {
2733        let (opts, _) = compile(&["-M", "a.c"]);
2734        assert!(opts.deps.emit && opts.deps.instead_of_compiling);
2735        assert!(opts.deps.system_headers, "plain -M lists them");
2736        assert_eq!(opts.emit, EmitKind::Preprocessed);
2737
2738        // Even where a later flag asked for something else, because the family is a mode and
2739        // the mode is what the run is for.
2740        let (opts, _) = compile(&["-M", "-c", "a.c"]);
2741        assert_eq!(opts.emit, EmitKind::Preprocessed);
2742
2743        let (opts, _) = compile(&["-MM", "a.c"]);
2744        assert!(!opts.deps.system_headers);
2745    }
2746
2747    #[test]
2748    fn the_two_that_end_in_d_leave_the_compilation_alone() {
2749        let (opts, _) = compile(&["-MD", "-c", "a.c"]);
2750        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2751        assert!(opts.deps.system_headers);
2752        assert_eq!(opts.emit, EmitKind::Object);
2753
2754        let (opts, _) = compile(&["-MMD", "-c", "a.c"]);
2755        assert!(opts.deps.emit && !opts.deps.instead_of_compiling);
2756        assert!(!opts.deps.system_headers);
2757    }
2758
2759    #[test]
2760    fn nothing_puts_the_system_headers_back_once_a_flag_has_taken_them_out() {
2761        // GCC's rule, and not an oversight in it. The flag asking for fewer of them is read as
2762        // the answer, because the other one never asked the question.
2763        let (opts, _) = compile(&["-MM", "-M", "a.c"]);
2764        assert!(!opts.deps.system_headers);
2765        let (opts, _) = compile(&["-MD", "-MMD", "-c", "a.c"]);
2766        assert!(!opts.deps.system_headers);
2767        let (opts, _) = compile(&["-MMD", "-MD", "-c", "a.c"]);
2768        assert!(!opts.deps.system_headers);
2769    }
2770
2771    #[test]
2772    fn a_target_arrives_escaped_from_one_flag_and_untouched_from_the_other() {
2773        let (opts, _) = compile(&["-MM", "-MT", "a b.o", "-MQ", "a b.o", "a.c"]);
2774        assert_eq!(opts.deps.targets, vec!["a b.o".to_owned(), "a\\ b.o".to_owned()]);
2775    }
2776
2777    #[test]
2778    fn the_rest_of_the_family_is_a_file_and_a_switch() {
2779        let (opts, _) = compile(&["-MM", "-MF", "dep.d", "-MP", "a.c"]);
2780        assert_eq!(opts.deps.file.as_deref(), Some("dep.d"));
2781        assert!(opts.deps.phony);
2782
2783        for flag in ["-MF", "-MT", "-MQ"] {
2784            let e = parse_args(&args(&[flag])).unwrap_err();
2785            assert!(e.message.contains("requires an argument"), "{}", e.message);
2786        }
2787    }
2788
2789    /// A directory of sources for one test, removed when the test is done with it.
2790    struct TempTree(PathBuf);
2791
2792    impl Drop for TempTree {
2793        fn drop(&mut self) {
2794            let _ = std::fs::remove_dir_all(&self.0);
2795        }
2796    }
2797
2798    impl TempTree {
2799        fn new(name: &str, files: &[(&str, &str)]) -> TempTree {
2800            let dir = std::env::temp_dir().join(format!("rucc-deps-{}-{name}", std::process::id()));
2801            let _ = std::fs::remove_dir_all(&dir);
2802            std::fs::create_dir_all(&dir).expect("temporary directory should be writable");
2803            for (path, text) in files {
2804                let at = dir.join(path);
2805                if let Some(parent) = at.parent() {
2806                    std::fs::create_dir_all(parent).expect("creating a subdirectory should work");
2807                }
2808                std::fs::write(&at, text).expect("writing a temporary file should work");
2809            }
2810            TempTree(dir)
2811        }
2812
2813        fn path(&self, name: &str) -> String {
2814            self.0.join(name).to_string_lossy().into_owned()
2815        }
2816    }
2817
2818    #[test]
2819    fn the_rule_names_what_the_includes_found_and_names_each_of_them_once() {
2820        // End to end, because the list comes from the preprocessor and the format comes from
2821        // somewhere else, and a test of either half on its own would pass with the two of them
2822        // wired up backwards.
2823        let tree = TempTree::new(
2824            "found",
2825            &[
2826                ("a.c", "#include \"one.h\"\n#include \"two.h\"\nint main(void) { return X; }\n"),
2827                ("one.h", "#define X 0\n"),
2828                ("two.h", "#include \"one.h\"\n"),
2829            ],
2830        );
2831        let out = tree.path("dep.d");
2832        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2833        assert_eq!(code, 0);
2834
2835        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2836        let names: Vec<&str> = text.split_whitespace().collect();
2837        // The target, the source, and each header once however many times it was reached.
2838        assert_eq!(names.first(), Some(&"a.o:"), "{text}");
2839        assert_eq!(names.iter().filter(|n| n.ends_with("one.h")).count(), 1, "{text}");
2840        assert_eq!(names.iter().filter(|n| n.ends_with("two.h")).count(), 1, "{text}");
2841        // And the `-o` went to the file the rule replaced, which is left empty rather than
2842        // absent because a makefile that named it as a target will look for it.
2843        assert_eq!(std::fs::read(tree.path("a.i")).expect("the output should exist"), b"");
2844    }
2845
2846    #[test]
2847    fn a_header_that_is_only_reached_under_a_guard_is_still_a_dependency() {
2848        // The multiple-include optimization means the second reach never opens the file. It is
2849        // still a file this translation unit was built from, so it is still in the rule.
2850        let tree = TempTree::new(
2851            "guarded",
2852            &[
2853                ("a.c", "#include \"g.h\"\n#include \"g.h\"\nint main(void) { return 0; }\n"),
2854                ("g.h", "#ifndef G\n#define G\n#endif\n"),
2855            ],
2856        );
2857        let out = tree.path("dep.d");
2858        let code = run(&args(&["-MM", "-MF", &out, "-o", &tree.path("a.i"), &tree.path("a.c")]));
2859        assert_eq!(code, 0);
2860        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2861        assert_eq!(text.split_whitespace().filter(|n| n.ends_with("g.h")).count(), 1, "{text}");
2862    }
2863
2864    #[test]
2865    fn every_imacros_file_is_read_before_every_include_file_whatever_order_they_were_written() {
2866        // Measured against GCC rather than read: the two flags the other way round produce the
2867        // same output byte for byte, so the command line order between the two families does not
2868        // decide anything and the order within one does. The `-include` file here can only see
2869        // the definition if the `-imacros` file that was written after it ran first.
2870        let tree = TempTree::new(
2871            "preinclude",
2872            &[
2873                ("a.c", "int main(void) { return 0; }\n"),
2874                ("i.h", "#ifdef FROM_MACROS\nint saw_it;\n#else\nint missed_it;\n#endif\n"),
2875                ("m.h", "#define FROM_MACROS 1\nint macros_text;\n"),
2876            ],
2877        );
2878        let out = tree.path("a.i");
2879        let code = run(&args(&[
2880            "-E",
2881            "-include",
2882            &tree.path("i.h"),
2883            "-imacros",
2884            &tree.path("m.h"),
2885            "-o",
2886            &out,
2887            &tree.path("a.c"),
2888        ]));
2889        assert_eq!(code, 0);
2890        let text = std::fs::read_to_string(&out).expect("the output should have been written");
2891        assert!(text.contains("saw_it"), "{text}");
2892        // And the text of the `-imacros` file is thrown away, which is the whole difference
2893        // between the two flags.
2894        assert!(!text.contains("macros_text"), "{text}");
2895    }
2896
2897    #[test]
2898    fn a_file_the_command_line_named_is_a_prerequisite_the_same_as_one_a_directive_named() {
2899        let tree = TempTree::new(
2900            "preinclude-deps",
2901            &[
2902                ("a.c", "int main(void) { return 0; }\n"),
2903                ("i.h", "int from_include;\n"),
2904                ("m.h", "#define M 1\n"),
2905            ],
2906        );
2907        let out = tree.path("dep.d");
2908        let code = run(&args(&[
2909            "-MM",
2910            "-MF",
2911            &out,
2912            "-include",
2913            &tree.path("i.h"),
2914            "-imacros",
2915            &tree.path("m.h"),
2916            "-o",
2917            &tree.path("a.i"),
2918            &tree.path("a.c"),
2919        ]));
2920        assert_eq!(code, 0);
2921        let text = std::fs::read_to_string(&out).expect("the rule should have been written");
2922        assert!(text.contains("i.h"), "{text}");
2923        assert!(text.contains("m.h"), "{text}");
2924    }
2925
2926    #[test]
2927    fn a_command_line_include_that_is_nowhere_on_the_path_is_an_error_and_not_a_warning() {
2928        // Including the directory of the source file, which is not on the path for these: the
2929        // command line was not written there, so a name in it is relative to where the compiler
2930        // was run rather than to where the source sits.
2931        let tree = TempTree::new(
2932            "preinclude-missing",
2933            &[("sub/a.c", "int main(void) { return 0; }\n"), ("sub/beside.h", "int x;\n")],
2934        );
2935        let code = run(&args(&["-E", "-include", "beside.h", "-o", "-", &tree.path("sub/a.c")]));
2936        assert_eq!(code, 1);
2937    }
2938
2939    #[test]
2940    fn a_command_line_that_links_names_the_executable_and_not_the_object_it_went_through() {
2941        // The object a link goes through is in a temporary directory and is gone before `make`
2942        // reads any of this, so the rule that named it would be a rule for a file that is never
2943        // there. The target and the file are both the `-o`, which is the executable.
2944        let (opts, plan) = compile(&["-MD", "sub/a.c", "-o", "prog"]);
2945        assert_eq!(plan.output.as_deref(), Some("prog"));
2946        assert_eq!(deps::default_target("sub/a.c", deps_target_output(&opts, &plan)), "prog");
2947        assert_eq!(
2948            deps::default_file(&opts.deps, "sub/a.c", plan.output.as_deref()).as_deref(),
2949            Some("prog.d")
2950        );
2951    }
2952
2953    #[test]
2954    fn the_plan_keeps_the_output_name_because_the_rule_is_written_from_it() {
2955        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c", "-o", "obj/x.o"]);
2956        assert_eq!(plan.output.as_deref(), Some("obj/x.o"));
2957        let (_, plan) = compile(&["-MMD", "-c", "sub/a.c"]);
2958        assert_eq!(plan.output, None);
2959    }
2960
2961    #[test]
2962    fn usage_fits_on_a_screen() {
2963        // Not a style preference. A help text that scrolls is one nobody reads, and this is
2964        // the cheapest way to keep it honest as flags accumulate. The number goes up only when
2965        // a family of flags arrives that has nowhere to share a line, which the two pass gates
2966        // were and which the two fuel flags and `-fsafety=` now are, and it goes up by exactly
2967        // the lines that family took. The four it went up by last are the flags a build system
2968        // passes without being asked to: how much to say, what machine to generate for, threads,
2969        // and the questions `configure` asks before it compiles anything. The one it went up by
2970        // last is the second line of `--emit`, whose kinds are a family that has now outgrown
2971        // one line and has nowhere else to go. The two it went up by last are the dependency
2972        // family, which is eight flags that share nothing with anything above them. The one it
2973        // went up by last is the four spellings of position independent code, which every
2974        // configure script writes and which could only have shared the link line, and that line
2975        // is already four characters short of the limit. The two it went up by last are the rest
2976        // of the include family, which is six more flags that change where a header is looked for
2977        // and two that name a header outright. The one it went up by last is the pair that keeps
2978        // the intermediate files and times the steps, which belong next to the two flags above
2979        // them that are also about watching a compilation rather than changing one. The two it
2980        // went up by last are the section flags and the visibility flag, which are what a build
2981        // that cares about the size of what it ships and about which names it exports writes, and
2982        // the second of them was already taken and only missing from here. The one it went up by
2983        // last is the stack protector, which is four spellings of one question and which every
2984        // distribution puts on every command line it issues, so a build that reads this list
2985        // looking for it and does not find it has to go and read the specification instead.
2986        assert!(USAGE.lines().count() < 51, "usage text has grown past one screen");
2987    }
2988}