internal/rosa: remove xz utils hack

This replaces all xz invocations with the native decompressor. This also significantly cleans up the Gentoo stage3 bootstrap path.

Signed-off-by: Ophestra <cat@gensokyo.uk>

6 files changed, 131 insertions, 176 deletions

diff --git a/internal/rosa/busybox.go b/internal/rosa/busybox.go
deleted file mode 100644
index ae900a04..00000000
--- a/internal/rosa/busybox.go
+++ /dev/null
@@ -1,124 +0,0 @@
-package rosa
-
-import (
-	"fmt"
-	"io"
-	"net/http"
-	"os"
-	"time"
-
-	"hakurei.app/fhs"
-	"hakurei.app/internal/pkg"
-)
-
-// busyboxBin is a busybox binary distribution installed under bin/busybox.
-type busyboxBin struct {
-	// Underlying busybox binary.
-	bin pkg.FileArtifact
-}
-
-// Kind returns the hardcoded [pkg.Kind] value.
-func (a busyboxBin) Kind() pkg.Kind { return kindBusyboxBin }
-
-// Params is a noop.
-func (a busyboxBin) Params(*pkg.IContext) {}
-
-// IsExclusive returns false: Cure performs a trivial filesystem write.
-func (busyboxBin) IsExclusive() bool { return false }
-
-// Inputs returns the underlying busybox [pkg.FileArtifact].
-func (a busyboxBin) Inputs() []pkg.Artifact {
-	return []pkg.Artifact{a.bin}
-}
-
-func init() {
-	pkg.Register(kindBusyboxBin, func(r *pkg.IRReader) pkg.Artifact {
-		a := busyboxBin{r.Next().(pkg.FileArtifact)}
-		if _, ok := r.Finalise(); ok {
-			panic(pkg.ErrUnexpectedChecksum)
-		}
-		return a
-	})
-}
-
-// String returns the reporting name of the underlying file prefixed with expand.
-func (a busyboxBin) String() string {
-	return "expand-" + a.bin.(fmt.Stringer).String()
-}
-
-// Cure installs the underlying busybox [pkg.File] to bin/busybox.
-func (a busyboxBin) Cure(t *pkg.TContext) (err error) {
-	var r io.ReadCloser
-	if r, err = t.Open(a.bin); err != nil {
-		return
-	}
-	defer func() {
-		closeErr := r.Close()
-		if err == nil {
-			err = closeErr
-		}
-	}()
-
-	binDir := t.GetWorkDir().Append("bin")
-	if err = os.MkdirAll(binDir.String(), 0700); err != nil {
-		return
-	}
-
-	var w *os.File
-	if w, err = os.OpenFile(
-		binDir.Append("busybox").String(),
-		os.O_WRONLY|os.O_CREATE|os.O_EXCL,
-		0500,
-	); err != nil {
-		return
-	}
-	defer func() {
-		closeErr := w.Close()
-		if err == nil {
-			err = closeErr
-		}
-	}()
-
-	_, err = io.Copy(w, r)
-	return
-}
-
-// newBusyboxBin returns a [pkg.Artifact] containing a busybox installation from
-// the https://busybox.net/downloads/binaries/ binary release.
-func (s *S) newBusyboxBin() pkg.Artifact {
-	var version, url, checksum string
-	switch s.arch {
-	case "amd64":
-		version = "1.35.0"
-		url = "https://busybox.net/downloads/binaries/" +
-			version + "-" + s.linuxArch() + "-linux-musl/busybox"
-		checksum = "L7OBIsPu9enNHn7FqpBT1kOg_mCLNmetSeNMA3i4Y60Z5jTgnlX3qX3zcQtLx5AB"
-	case "arm64":
-		version = "1.31.0"
-		url = "https://busybox.net/downloads/binaries/" +
-			version + "-defconfig-multiarch-musl/busybox-armv8l"
-		checksum = "npJjBO7iwhjW6Kx2aXeSxf8kXhVgTCDChOZTTsI8ZfFfa3tbsklxRiidZQdrVERg"
-
-	default:
-		panic("unsupported target " + s.arch)
-	}
-
-	return pkg.NewExec(
-		"busybox-bin-"+version, s.arch, nil, pkg.ExecTimeoutMax, false, false,
-		fhs.AbsRoot, []string{
-			"PATH=/system/bin",
-		},
-		AbsSystem.Append("bin", "busybox"),
-		[]string{"hush", "-c", "" +
-			"busybox mkdir -p /work/system/bin/ && " +
-			"busybox cp /system/bin/busybox /work/system/bin/ && " +
-			"busybox --install -s /work/system/bin/"},
-		pkg.Path(AbsSystem, true, busyboxBin{pkg.NewHTTPGet(
-			&http.Client{Transport: &http.Transport{
-				// busybox website is really slow to respond
-				TLSHandshakeTimeout: 2 * time.Minute,
-			}}, url,
-			mustDecode(checksum),
-		)}),
-	)
-}
diff --git a/internal/rosa/gentoo.go b/internal/rosa/gentoo.go
new file mode 100644
index 00000000..a0a8e525
--- /dev/null
+++ b/internal/rosa/gentoo.go
@@ -0,0 +1,104 @@
+package rosa
+
+import (
+	"errors"
+	"io/fs"
+	"os"
+	"path/filepath"
+
+	"hakurei.app/internal/pkg"
+)
+
+// gentooOverlay are symlinks on top of a Gentoo LLVM stage3 tarball for
+// compatibility with the Rosa OS stage0 distribution.
+type gentooOverlay struct{ stage3 pkg.Artifact }
+
+// Kind returns the hardcoded [pkg.Kind] value.
+func (a gentooOverlay) Kind() pkg.Kind { return kindGentooOverlay }
+
+// Params is a noop.
+func (a gentooOverlay) Params(*pkg.IContext) {}
+
+// IsExclusive returns false: Cure performs a trivial filesystem write.
+func (gentooOverlay) IsExclusive() bool { return false }
+
+// Inputs returns the underlying Gentoo stage3.
+func (a gentooOverlay) Inputs() []pkg.Artifact {
+	return []pkg.Artifact{a.stage3}
+}
+
+func init() {
+	pkg.Register(kindGentooOverlay, func(r *pkg.IRReader) pkg.Artifact {
+		a := gentooOverlay{r.Next()}
+		if _, ok := r.Finalise(); ok {
+			panic(pkg.ErrUnexpectedChecksum)
+		}
+		return a
+	})
+}
+
+// String returns a hardcoded name.
+func (a gentooOverlay) String() string { return "gentoo-overlay" }
+
+// Revision returns a hardcoded value incremented on layout changes.
+func (a gentooOverlay) Revision() uint64 { return 0 }
+
+// Cure installs the overlay.
+func (a gentooOverlay) Cure(f *pkg.FContext) (err error) {
+	pathname, _ := f.GetArtifact(a.stage3)
+	var stage3 *os.Root
+	if stage3, err = os.OpenRoot(pathname.String()); err != nil {
+		return err
+	}
+	defer func() {
+		closeErr := stage3.Close()
+		if err == nil {
+			err = closeErr
+		}
+	}()
+	fsys := stage3.FS()
+
+	system := f.GetWorkDir().Append("system")
+	bin := system.Append("bin")
+	if err = os.MkdirAll(bin.String(), 0700); err != nil {
+		return
+	}
+
+	linknames := []string{
+		"../../bin/sh",
+	}
+
+	var dents []os.DirEntry
+
+	const llvm = "usr/lib/llvm"
+	dents, err = fs.ReadDir(fsys, llvm)
+	if err != nil {
+		return
+	}
+	if len(dents) != 1 {
+		return errors.New("stage3 /" + llvm + " contains more than one entry")
+	}
+
+	llvmBin := filepath.Join(llvm, dents[0].Name(), "bin")
+	dents, err = fs.ReadDir(fsys, llvmBin)
+	if err != nil {
+		return
+	}
+	for _, dent := range dents {
+		linknames = append(linknames, filepath.Join(
+			"../..",
+			llvmBin,
+			dent.Name(),
+		))
+	}
+
+	for _, linkname := range linknames {
+		if err = os.Symlink(
+			linkname,
+			bin.Append(filepath.Base(linkname)).String(),
+		); err != nil {
+			return
+		}
+	}
+	return
+}
diff --git a/internal/rosa/llvm_test.go b/internal/rosa/llvm_test.go
index af5f6530..2b1d2e7d 100644
--- a/internal/rosa/llvm_test.go
+++ b/internal/rosa/llvm_test.go
@@ -8,7 +8,7 @@ import (
 )
 
 func TestLLVMInputs(t *testing.T) {
-	const wantInputCount = 473
+	const wantInputCount = 470
 
 	_, llvm := rosa.Native().Std().MustLoad(rosa.H("llvm"))
 	var n int
diff --git a/internal/rosa/package/gnu.az b/internal/rosa/package/gnu.az
index d1da6b8a..7cfcc726 100644
--- a/internal/rosa/package/gnu.az
+++ b/internal/rosa/package/gnu.az
@@ -292,11 +292,14 @@ package findutils {
 	anitya = 812;
 
 	version# = "4.11.0";
-	source = remoteFile {
+	source = remoteTar {
 		url = "https://ftpmirror.gnu.org/gnu/findutils/findutils-"+version+".tar.xz";
+		compress = xz;
 		checksum = "X-qsv9aUoBL583q-4iZcBI3_7Ufd3fSIqFmNLXWdEpUihlRfMpGGyAUt8SAvDKae";
 	};
 
+	writable = true;
+	chmod = true;
 	early = `
 echo '#!/bin/sh' > gnulib-tests/test-c32ispunct.sh
 echo 'int main(){return 0;}' > gnulib-tests/test-fstatat.c
@@ -307,7 +310,6 @@ echo 'int main(){return 0;}' > tests/xargs/test-sigusr.c
 
 	inputs = [
 		diffutils,
-		xz,
 		sed,
 	];
 }
diff --git a/internal/rosa/rosa.go b/internal/rosa/rosa.go
index de4229ea..456f82e7 100644
--- a/internal/rosa/rosa.go
+++ b/internal/rosa/rosa.go
@@ -26,8 +26,8 @@ const (
 	// kindEtc is the kind of [pkg.Artifact] of cureEtc.
 	kindEtc = iota + pkg.KindCustomOffset
 
-	// kindBusyboxBin is the kind of [pkg.Artifact] of busyboxBin.
-	kindBusyboxBin
+	// kindGentooOverlay is the kind of [pkg.Artifact] of gentooOverlay.
+	kindGentooOverlay
 )
 
 // mustDecode is like [pkg.MustDecode], but replaces the zero value and prints
@@ -97,15 +97,9 @@ func (s *S) earlyLDFLAGS(static bool) string {
 }
 
 const (
-	// _stageBusybox denotes a busybox installation from the busyboxBin
-	// binary distribution. This is defined as a [Stage] to make use of the
-	// toolchain abstractions to preprocess stageGentoo and is not a real,
-	// functioning toolchain. It does not contain any compilers.
-	_stageBusybox Stage = iota
-
 	// stageGentoo denotes the toolchain in a Gentoo stage3 tarball. Special
 	// care must be taken to compile correctly against this stage.
-	stageGentoo
+	stageGentoo Stage = iota
 
 	// stageIntermediateGentoo is like stageIntermediate, but compiled against
 	// stageGentoo.
@@ -239,11 +233,6 @@ func (t Toolchain) New(
 
 	var support []pkg.Artifact
 	switch t.stage {
-	case _stageBusybox:
-		name += "-early"
-		support = slices.Concat([]pkg.Artifact{t.newBusyboxBin()}, extra)
-		env = fixupEnviron(env, nil, "/system/bin")
-
 	case stageGentoo, stageEarly:
 		name += "-boot"
 		support = append(support, extra...)
@@ -251,22 +240,17 @@ func (t Toolchain) New(
 		if t.stage == stageEarly {
 			_, a := t.MustLoad(_stage0Dist)
 			support = append(support, a)
+		} else if t.gentooStage3 == nil {
+			panic(os.ErrInvalid)
 		} else {
-			support = append(support, t.S.New(_stageBusybox).New("gentoo", 0, nil, nil, nil, `
-tar -C /work -xf /usr/src/stage3.tar.xz
-rm -rf /work/dev/ /work/proc/
-ln -vs ../usr/bin /work/bin
-mkdir -vp /work/system/bin
-(cd /work/system/bin && ln -vs \
-	../../bin/sh \
-	../../usr/lib/llvm/*/bin/* \
-	.)
-`, pkg.Path(AbsUsrSrc.Append("stage3.tar.xz"), false,
-				pkg.NewHTTPGet(
-					nil, t.gentooStage3,
-					t.gentooStage3Checksum,
-				),
-			)))
+			support = append(support,
+				t.gentooStage3,
+				gentooOverlay{t.gentooStage3},
+			)
+			env = append(env,
+				"CC=clang",
+				"CXX=clang++",
+			)
 		}
 		env = fixupEnviron(env, []string{
 			EnvTriple + "=" + t.triple(),
@@ -516,20 +500,6 @@ func (t Toolchain) NewPackage(
 		sourceSuffix string
 	)
 
-	if _, ok := source.(pkg.FileArtifact); ok {
-		if attr.Writable || attr.Chmod ||
-			wantsChmod || wantsWrite ||
-			len(attr.Patches) > 0 {
-			panic("source processing requested on a xz-compressed tarball")
-		}
-
-		sourceSuffix = ".tar.xz"
-		scriptEarly += `
-tar -C /usr/src/ -xf '/usr/src/` + name + `.tar.xz'
-mv '/usr/src/` + rn + `' '/usr/src/` + name + `'
-`
-	}
-
 	dir, create := helper.wantsDir()
 	helperScriptEarly := helper.scriptEarly()
 	if attr.EnterSource ||
diff --git a/internal/rosa/state.go b/internal/rosa/state.go
index 36fe9451..516efd81 100644
--- a/internal/rosa/state.go
+++ b/internal/rosa/state.go
@@ -213,10 +213,8 @@ type S struct {
 	// Cached [pkg.Artifact].
 	c [_stageEnd]sync.Map
 
-	// URL of a Gentoo stage3 tarball.
-	gentooStage3 string
-	// Expected checksum of gentooStage3.
-	gentooStage3Checksum pkg.Checksum
+	// Unpacked Gentoo LLVM stage3.
+	gentooStage3 pkg.Artifact
 }
 
 // Clone returns a copy of s.
@@ -485,6 +483,7 @@ func (s *S) getFrame() azalea.Frame {
 			k("gzip"):         pkg.Gzip + compressOffset,
 			k("bzip2"):        pkg.Bzip2 + compressOffset,
 			k("zstd"):         pkg.Zstd + compressOffset,
+			k("xz"):           pkg.XZ + compressOffset,
 		}
 
 		s.frame.Val = map[unique.Handle[azalea.Ident]]any{
@@ -803,6 +802,7 @@ func (s *S) getFrame() azalea.Frame {
 				k("gzip"):  uint32(pkg.Gzip),
 				k("bzip2"): uint32(pkg.Bzip2),
 				k("zstd"):  uint32(pkg.Zstd),
+				k("xz"):    uint32(pkg.XZ),
 			}},
 
 			// high-level helpers
@@ -1372,12 +1372,15 @@ func (s *S) SetSource(fsys fs.FS) error {
 // SetGentooStage3 sets the Gentoo stage3 tarball url and checksum. It panics
 // if given zero values or if these values have already been set.
 func (s *S) SetGentooStage3(url string, checksum pkg.Checksum) {
-	if s.gentooStage3 != "" {
+	if s.gentooStage3 != nil {
 		panic(errors.New("attempting to set Gentoo stage3 url twice"))
 	}
 	if url == "" {
 		panic(errors.New("attempting to set Gentoo stage3 url to the zero value"))
 	}
-	s.gentooStage3, s.gentooStage3Checksum = url, checksum
+	s.gentooStage3 = pkg.NewTar(pkg.NewDecompress(
+		pkg.NewHTTPGet(nil, url, checksum),
+		pkg.XZ,
+	))
 	s.DropCaches(s.Arch(), s.Flags())
 }