cc/target/
parser.rs

Help
1use std::env;
2
3use crate::{target::TargetInfo, utilities::OnceLock, Error, ErrorKind};
4
5#[derive(Debug)]
6struct TargetInfoParserInner {
7    full_arch: Box<str>,
8    arch: Box<str>,
9    vendor: Box<str>,
10    os: Box<str>,
11    env: Box<str>,
12    abi: Box<str>,
13}
14
15impl TargetInfoParserInner {
16    fn from_cargo_environment_variables() -> Result<Self, Error> {
17        // `TARGET` must be present.
18        //
19        // No need to emit `rerun-if-env-changed` for this,
20        // as it is controlled by Cargo itself.
21        #[allow(clippy::disallowed_methods)]
22        let target_name = env::var("TARGET").map_err(|err| {
23            Error::new(
24                ErrorKind::EnvVarNotFound,
25                format!("failed reading TARGET: {err}"),
26            )
27        })?;
28
29        // Parse the full architecture name from the target name.
30        let (full_arch, _rest) = target_name.split_once('-').ok_or(Error::new(
31            ErrorKind::InvalidTarget,
32            format!("target `{target_name}` only had a single component (at least two required)"),
33        ))?;
34
35        let cargo_env = |name, fallback: Option<&str>| -> Result<Box<str>, Error> {
36            // No need to emit `rerun-if-env-changed` for these,
37            // as they are controlled by Cargo itself.
38            #[allow(clippy::disallowed_methods)]
39            match env::var(name) {
40                Ok(var) => Ok(var.into_boxed_str()),
41                Err(err) => match fallback {
42                    Some(fallback) => Ok(fallback.into()),
43                    None => Err(Error::new(
44                        ErrorKind::EnvVarNotFound,
45                        format!("did not find fallback information for target `{target_name}`, and failed reading {name}: {err}"),
46                    )),
47                },
48            }
49        };
50
51        // Prefer to use `CARGO_ENV_*` if set, since these contain the most
52        // correct information relative to the current `rustc`, and makes it
53        // possible to support custom target JSON specs unknown to `rustc`.
54        //
55        // NOTE: If the user is using an older `rustc`, that data may be older
56        // than our pre-generated data, but we still prefer Cargo's view of
57        // the world, since at least `cc` won't differ from `rustc` in that
58        // case.
59        //
60        // These may not be set in case the user depended on being able to
61        // just set `TARGET` outside of build scripts; in those cases, fall
62        // back back to data from the known set of target names instead.
63        //
64        // See discussion in #1225 for further details.
65        let fallback_target = TargetInfo::from_rustc_target(&target_name).ok();
66        let ft = fallback_target.as_ref();
67        let arch = cargo_env("CARGO_CFG_TARGET_ARCH", ft.map(|t| t.arch))?;
68        let vendor = cargo_env("CARGO_CFG_TARGET_VENDOR", ft.map(|t| t.vendor))?;
69        let os = cargo_env("CARGO_CFG_TARGET_OS", ft.map(|t| t.os))?;
70        let env = cargo_env("CARGO_CFG_TARGET_ENV", ft.map(|t| t.env))?;
71        // `target_abi` was stabilized in Rust 1.78, which is higher than our
72        // MSRV, so it may not always be available; In that case, fall back to
73        // `""`, which is _probably_ correct for unknown target names.
74        let abi = cargo_env("CARGO_CFG_TARGET_ABI", ft.map(|t| t.abi))
75            .unwrap_or_else(|_| String::default().into_boxed_str());
76
77        Ok(Self {
78            full_arch: full_arch.to_string().into_boxed_str(),
79            arch,
80            vendor,
81            os,
82            env,
83            abi,
84        })
85    }
86}
87
88/// Parser for [`TargetInfo`], contains cached information.
89#[derive(Default, Debug)]
90pub(crate) struct TargetInfoParser(OnceLock<Result<TargetInfoParserInner, Error>>);
91
92impl TargetInfoParser {
93    pub fn parse_from_cargo_environment_variables(&self) -> Result<TargetInfo<'_>, Error> {
94        match self
95            .0
96            .get_or_init(TargetInfoParserInner::from_cargo_environment_variables)
97        {
98            Ok(TargetInfoParserInner {
99                full_arch,
100                arch,
101                vendor,
102                os,
103                env,
104                abi,
105            }) => Ok(TargetInfo {
106                full_arch,
107                arch,
108                vendor,
109                os,
110                env,
111                abi,
112            }),
113            Err(e) => Err(e.clone()),
114        }
115    }
116}
117
118/// Parse the full architecture in the target name into the simpler
119/// `cfg(target_arch = "...")` that `rustc` exposes.
120fn parse_arch(full_arch: &str) -> Option<&str> {
121    // NOTE: Some of these don't necessarily match an existing target in
122    // `rustc`. They're parsed anyhow to be as forward-compatible as possible,
123    // while still being correct.
124    //
125    // See also:
126    // https://docs.rs/cfg-expr/0.18.0/cfg_expr/targets/index.html
127    // https://docs.rs/target-lexicon/0.13.2/target_lexicon/enum.Architecture.html
128    // https://gcc.gnu.org/onlinedocs/gcc/Submodel-Options.html
129    // `clang -print-targets`
130    Some(match full_arch {
131        arch if arch.starts_with("mipsisa32r6") => "mips32r6", // mipsisa32r6 | mipsisa32r6el
132        arch if arch.starts_with("mipsisa64r6") => "mips64r6", // mipsisa64r6 | mipsisa64r6el
133
134        arch if arch.starts_with("mips64") => "mips64", // mips64 | mips64el
135        arch if arch.starts_with("mips") => "mips",     // mips | mipsel
136
137        arch if arch.starts_with("loongarch64") => "loongarch64",
138        arch if arch.starts_with("loongarch32") => "loongarch32",
139
140        arch if arch.starts_with("powerpc64") => "powerpc64", // powerpc64 | powerpc64le
141        arch if arch.starts_with("powerpc") => "powerpc",
142        arch if arch.starts_with("ppc64") => "powerpc64",
143        arch if arch.starts_with("ppc") => "powerpc",
144
145        arch if arch.starts_with("x86_64") => "x86_64", // x86_64 | x86_64h
146        arch if arch.starts_with("i") && arch.ends_with("86") => "x86", // i386 | i586 | i686
147
148        "arm64ec" => "arm64ec", // https://github.com/rust-lang/rust/issues/131172
149        arch if arch.starts_with("aarch64") => "aarch64", // arm64e | arm64_32
150        arch if arch.starts_with("arm64") => "aarch64", // aarch64 | aarch64_be
151
152        arch if arch.starts_with("arm") => "arm", // arm | armv7s | armeb | ...
153        arch if arch.starts_with("thumb") => "arm", // thumbv4t | thumbv7a | thumbv8m | ...
154
155        arch if arch.starts_with("riscv64") => "riscv64",
156        arch if arch.starts_with("riscv32") => "riscv32",
157
158        arch if arch.starts_with("wasm64") => "wasm64",
159        arch if arch.starts_with("wasm32") => "wasm32", // wasm32 | wasm32v1
160        "asmjs" => "wasm32",
161
162        arch if arch.starts_with("nvptx64") => "nvptx64",
163        arch if arch.starts_with("nvptx") => "nvptx",
164
165        arch if arch.starts_with("bpf") => "bpf", // bpfeb | bpfel
166
167        // https://github.com/bytecodealliance/wasmtime/tree/v30.0.1/pulley
168        arch if arch.starts_with("pulley64") => "pulley64",
169        arch if arch.starts_with("pulley32") => "pulley32",
170
171        // https://github.com/Clever-ISA/Clever-ISA
172        arch if arch.starts_with("clever") => "clever",
173
174        "sparc" | "sparcv7" | "sparcv8" => "sparc",
175        "sparc64" | "sparcv9" => "sparc64",
176
177        "amdgcn" => "amdgpu",
178        "avr" => "avr",
179        "csky" => "csky",
180        "hexagon" => "hexagon",
181        "m68k" => "m68k",
182        "msp430" => "msp430",
183        "r600" => "r600",
184        "s390x" => "s390x",
185        "xtensa" => "xtensa",
186
187        _ => return None,
188    })
189}
190
191/// Parse environment and ABI from the last component of the target name.
192fn parse_envabi(last_component: &str) -> Option<(&str, &str)> {
193    let (env, abi) = match last_component {
194        // Combined environment and ABI
195
196        // gnullvm | gnueabi | gnueabihf | gnuabiv2 | gnuabi64 | gnuspe | gnux32 | gnu_ilp32
197        env_and_abi if env_and_abi.starts_with("gnu") => {
198            let abi = env_and_abi.strip_prefix("gnu").unwrap();
199            let abi = abi.strip_prefix("_").unwrap_or(abi);
200            ("gnu", abi)
201        }
202        // musl | musleabi | musleabihf | muslabi64 | muslspe
203        env_and_abi if env_and_abi.starts_with("musl") => {
204            ("musl", env_and_abi.strip_prefix("musl").unwrap())
205        }
206        // uclibc | uclibceabi | uclibceabihf
207        env_and_abi if env_and_abi.starts_with("uclibc") => {
208            ("uclibc", env_and_abi.strip_prefix("uclibc").unwrap())
209        }
210        // newlib | newlibeabihf
211        env_and_abi if env_and_abi.starts_with("newlib") => {
212            ("newlib", env_and_abi.strip_prefix("newlib").unwrap())
213        }
214
215        // Environments
216        "msvc" => ("msvc", ""),
217        "ohos" => ("ohos", ""),
218        "qnx700" => ("nto70", ""),
219        "qnx710_iosock" => ("nto71_iosock", ""),
220        "qnx710" => ("nto71", ""),
221        "qnx800" => ("nto80", ""),
222        "sgx" => ("sgx", ""),
223        "threads" => ("threads", ""),
224
225        // ABIs
226        "abi64" => ("", "abi64"),
227        "abiv2" => ("", "spe"),
228        "eabi" => ("", "eabi"),
229        "eabihf" => ("", "eabihf"),
230        "macabi" => ("", "macabi"),
231        "sim" => ("", "sim"),
232        "softfloat" => ("", "softfloat"),
233        "spe" => ("", "spe"),
234        "x32" => ("", "x32"),
235
236        // Badly named targets, ELF is already known from target OS.
237        // Probably too late to fix now though.
238        "elf" => ("", ""),
239        // Undesirable to expose to user code (yet):
240        // https://github.com/rust-lang/rust/pull/131166#issuecomment-2389541917
241        "freestanding" => ("", ""),
242
243        _ => return None,
244    };
245    Some((env, abi))
246}
247
248impl<'a> TargetInfo<'a> {
249    pub(crate) fn from_rustc_target(target: &'a str) -> Result<Self, Error> {
250        // FIXME(madsmtm): This target should be renamed, cannot be parsed
251        // with the means we do below (since `none` must not be interpreted
252        // as an env/ABI).
253        if target == "x86_64-unknown-linux-none" {
254            return Ok(Self {
255                full_arch: "x86_64",
256                arch: "x86_64",
257                vendor: "unknown",
258                os: "linux",
259                env: "",
260                abi: "",
261            });
262        }
263
264        let mut components = target.split('-');
265
266        // Insist that the target name contains at least a valid architecture.
267        let full_arch = components.next().ok_or(Error::new(
268            ErrorKind::InvalidTarget,
269            "target was empty".to_string(),
270        ))?;
271        let arch = parse_arch(full_arch).ok_or_else(|| {
272            Error::new(
273                ErrorKind::UnknownTarget,
274                format!("target `{target}` had an unknown architecture"),
275            )
276        })?;
277
278        // Newer target names have begun omitting the vendor, so the only
279        // component we know must be there is the OS name.
280        let components: Vec<_> = components.collect();
281        let (vendor, os, mut env, mut abi) = match &*components {
282            [] => {
283                return Err(Error::new(
284                    ErrorKind::InvalidTarget,
285                    format!("target `{target}` must have at least two components"),
286                ))
287            }
288            // Two components; format is `arch-os`.
289            [os] => ("unknown", *os, "", ""),
290            // The three-component case is a bit tricky to handle, it could
291            // either have the format `arch-vendor-os` or `arch-os-env+abi`.
292            [vendor_or_os, os_or_envabi] => {
293                // We differentiate between these by checking if the last
294                // component is an env/ABI; if it isn't, then it's probably
295                // an OS instead.
296                if let Some((env, abi)) = parse_envabi(os_or_envabi) {
297                    ("unknown", *vendor_or_os, env, abi)
298                } else {
299                    (*vendor_or_os, *os_or_envabi, "", "")
300                }
301            }
302            // Four components; format is `arch-vendor-os-env+abi`.
303            [vendor, os, envabi] => {
304                let (env, abi) = parse_envabi(envabi).ok_or_else(|| {
305                    Error::new(
306                        ErrorKind::UnknownTarget,
307                        format!("unknown environment/ABI `{envabi}` in target `{target}`"),
308                    )
309                })?;
310                (*vendor, *os, env, abi)
311            }
312            _ => {
313                return Err(Error::new(
314                    ErrorKind::InvalidTarget,
315                    format!("too many components in target `{target}`"),
316                ))
317            }
318        };
319
320        // Part of the architecture name is carried over into the ABI.
321        match full_arch {
322            // https://github.com/rust-lang/compiler-team/issues/830
323            arch if arch.starts_with("riscv32e") => {
324                abi = "ilp32e";
325            }
326            _ => {}
327        }
328
329        // Various environment/ABIs are determined based on OS name.
330        match os {
331            "3ds" | "rtems" | "espidf" => env = "newlib",
332            "vxworks" => env = "gnu",
333            "redox" => env = "relibc",
334            "aix" => abi = "vec-extabi",
335            _ => {}
336        }
337
338        // Extra overrides for badly named targets.
339        match target {
340            // Actually simulator targets.
341            "i386-apple-ios" | "x86_64-apple-ios" | "x86_64-apple-tvos" => {
342                abi = "sim";
343            }
344            // Name should've contained `muslabi64`.
345            "mips64-openwrt-linux-musl" => {
346                abi = "abi64";
347            }
348            // Specifies abi even though not in name.
349            "armv6-unknown-freebsd" | "armv6k-nintendo-3ds" | "armv7-unknown-freebsd" => {
350                abi = "eabihf";
351            }
352            // Specifies abi even though not in name.
353            "armv7-unknown-linux-ohos" | "armv7-unknown-trusty" => {
354                abi = "eabi";
355            }
356            _ => {}
357        }
358
359        let os = match os {
360            // Horizon is the common/internal OS name for 3DS and the Switch.
361            "3ds" | "switch" => "horizon",
362            // FIXME(madsmtm): macOS targets are badly named.
363            "darwin" => "macos",
364
365            // WASI targets contain the preview version in them too. Should've
366            // been `wasi-p1`/`wasi-p2`, but that's probably too late now.
367            os if os.starts_with("wasi") => {
368                env = os.strip_prefix("wasi").unwrap();
369                "wasi"
370            }
371            // FIXME(madsmtm): Badly named targets `*-linux-androideabi`,
372            // should be `*-android-eabi`.
373            "androideabi" => {
374                abi = "eabi";
375                "android"
376            }
377
378            os => os,
379        };
380
381        let vendor = match vendor {
382            // esp, esp32, esp32s2 etc.
383            vendor if vendor.starts_with("esp") => "espressif",
384            // FIXME(madsmtm): Badly named targets `*-linux-android*`,
385            // "linux" makes no sense as the vendor name.
386            "linux" if os == "android" || os == "androideabi" => "unknown",
387            // FIXME(madsmtm): Fix in `rustc` after
388            // https://github.com/rust-lang/compiler-team/issues/850.
389            "wali" => "unknown",
390            // Some Linux distributions set their name as the target vendor,
391            // so we have to assume that it can be an arbitary string.
392            vendor => vendor,
393        };
394
395        // Intentionally also marked as an ABI:
396        // https://github.com/rust-lang/rust/pull/86922
397        if vendor == "fortanix" {
398            abi = "fortanix";
399        }
400        if vendor == "uwp" {
401            abi = "uwp";
402        }
403
404        Ok(Self {
405            full_arch,
406            arch,
407            vendor,
408            os,
409            env,
410            abi,
411        })
412    }
413}
414
415#[cfg(test)]
416#[allow(unexpected_cfgs)]
417mod tests {
418    use std::process::Command;
419
420    use super::TargetInfo;
421    use crate::ErrorKind;
422
423    // Test tier 1 targets.
424    #[test]
425    fn tier1() {
426        let targets = [
427            "aarch64-unknown-linux-gnu",
428            "aarch64-apple-darwin",
429            "i686-pc-windows-gnu",
430            "i686-pc-windows-msvc",
431            "i686-unknown-linux-gnu",
432            "x86_64-apple-darwin",
433            "x86_64-pc-windows-gnu",
434            "x86_64-pc-windows-msvc",
435            "x86_64-unknown-linux-gnu",
436        ];
437
438        for target in targets {
439            // Check that they parse.
440            let _ = TargetInfo::from_rustc_target(target).unwrap();
441        }
442    }
443
444    // Various custom target names not (or no longer) known by `rustc`.
445    #[test]
446    fn parse_extra() {
447        let targets = [
448            "aarch64-unknown-none-gnu",
449            "aarch64-uwp-windows-gnu",
450            "arm-frc-linux-gnueabi",
451            "arm-unknown-netbsd-eabi",
452            "armv7neon-unknown-linux-gnueabihf",
453            "armv7neon-unknown-linux-musleabihf",
454            "thumbv7-unknown-linux-gnueabihf",
455            "thumbv7-unknown-linux-musleabihf",
456            "armv7-apple-ios",
457            "wasm32-wasi",
458            "x86_64-rumprun-netbsd",
459            "x86_64-unknown-linux",
460            "x86_64-alpine-linux-musl",
461            "x86_64-chimera-linux-musl",
462            "x86_64-foxkit-linux-musl",
463            "arm-poky-linux-gnueabi",
464            "x86_64-unknown-moturus",
465        ];
466
467        for target in targets {
468            // Check that they parse.
469            let _ = TargetInfo::from_rustc_target(target).unwrap();
470        }
471    }
472
473    fn target_from_rustc_cfgs<'a>(target: &'a str, cfgs: &'a str) -> TargetInfo<'a> {
474        // Cannot determine full architecture from cfgs.
475        let (full_arch, _rest) = target.split_once('-').expect("target to have arch");
476
477        let mut target = TargetInfo {
478            full_arch,
479            arch: "invalid-none-set",
480            vendor: "invalid-none-set",
481            os: "invalid-none-set",
482            env: "invalid-none-set",
483            // Not set in older Rust versions
484            abi: "",
485        };
486
487        for cfg in cfgs.lines() {
488            if let Some((name, value)) = cfg.split_once('=') {
489                // Remove whitespace, if `rustc` decided to insert any.
490                let name = name.trim();
491                let value = value.trim();
492
493                // Remove quotes around value.
494                let value = value.strip_prefix('"').unwrap_or(value);
495                let value = value.strip_suffix('"').unwrap_or(value);
496
497                match name {
498                    "target_arch" => target.arch = value,
499                    "target_vendor" => target.vendor = value,
500                    "target_os" => target.os = value,
501                    "target_env" => target.env = value,
502                    "target_abi" => target.abi = value,
503                    _ => {}
504                }
505            } else {
506                // Skip cfgs like `debug_assertions` and `unix`.
507            }
508        }
509
510        target
511    }
512
513    #[test]
514    fn unknown_env_determined_as_unknown() {
515        let err = TargetInfo::from_rustc_target("aarch64-unknown-linux-bogus").unwrap_err();
516        assert!(matches!(err.kind, ErrorKind::UnknownTarget));
517    }
518
519    // Used in .github/workflows/test-rustc-targets.yml
520    #[test]
521    #[cfg_attr(
522        not(rustc_target_test),
523        ignore = "must enable explicitly with --cfg=rustc_target_test"
524    )]
525    fn parse_rustc_targets() {
526        let rustc = std::env::var("RUSTC").unwrap_or_else(|_| "rustc".to_string());
527
528        let target_list = Command::new(&rustc)
529            .arg("--print=target-list")
530            .output()
531            .unwrap()
532            .stdout;
533        let target_list = String::from_utf8(target_list).unwrap();
534
535        let mut has_failure = false;
536        for target in target_list.lines() {
537            let cfgs = Command::new(&rustc)
538                .arg("--target")
539                .arg(target)
540                .arg("--print=cfg")
541                .output()
542                .unwrap()
543                .stdout;
544            let cfgs = String::from_utf8(cfgs).unwrap();
545
546            let expected = target_from_rustc_cfgs(target, &cfgs);
547            let actual = TargetInfo::from_rustc_target(target);
548
549            if Some(&expected) != actual.as_ref().ok() {
550                eprintln!("failed comparing {target}:");
551                eprintln!("  expected: Ok({expected:?})");
552                eprintln!("    actual: {actual:?}");
553                eprintln!();
554                has_failure = true;
555            }
556        }
557
558        if has_failure {
559            panic!("failed comparing targets");
560        }
561    }
562}
cc/target/parser.rs

cc/target/
parser.rs
