driver: only depend on ld for linking (todo)

5 files changed, 155 insertions, 47 deletions

diff --git a/README.md b/README.md
index 3665d6c..ec9979b 100644
--- a/README.md
+++ b/README.md
@@ -15,10 +15,7 @@ and backends like [QBE](https://c9x.me/compile/) and [LLVM](https://llvm.org/).
 ## Requirements
 
 `antcc` is written in standard C11 and can be built with any conforming
-compiler toolchain.  The `Makefile` requires GNU Make. At runtime, an existing
-C compiler is currently required for calling the linker with the appropiate
-libc runtime paths (eventually, the driver should only depend on the linker by
-determining those linker paths and flags at `configure` time).
+compiler toolchain.  The `Makefile` requires GNU Make.
 
 ## Building
 
diff --git a/configure b/configure
index 54199fc..43c7ad0 100755
--- a/configure
+++ b/configure
@@ -14,15 +14,17 @@ Options:
   --help                    Print this help mesage
   --host=target-triple      $host
   --cc=cc                   C compiler [cc]
+  --ld=ld                   Linker [ld]
   --prefix=PREFIX           Installation path [/usr/local]
-  --sysroot=/path           System root [/usr]
+  --sysroot=/path           System root []
   --includedir="/a/b",...   System include dirs
 EOF
 }
 
-sysroot=/usr
+sysroot=
 prefix=/usr/local
 host=
+linker=
 
 for arg ; do
 	case "$arg" in
@@ -31,6 +33,7 @@ for arg ; do
     --includedir=*) host_include_dirs=${arg#*=} ;;
 	CC=*) CC=${arg#*=} ;;
 	--cc=*) CC=${arg#*=} ;;
+	--ld=*) linker=${arg#*=} ;;
 	CFLAGS=*) CFLAGS=${arg#*=} ;;
 	--cflags=*) CFLAGS=${arg#*=} ;;
     --prefix=*) prefix=$(sh -c "echo ${arg#*=}") ;;
@@ -40,7 +43,16 @@ for arg ; do
 done
 
 : ${CC:=cc}
-: ${host:=$($CC -dumpmachine)}
+if ! test -n "$host"; then
+    host=$($CC -dumpmachine)
+else
+    # try target-triple-ld (cross compiling)
+    if ! test -n "$linker" && command -v "$host"-ld >/dev/null; then
+        linker="$host"-ld
+        sysroot=$(($host-cc -print-sysroot || $host-gcc -print-sysroot) 2>/dev/null)
+    fi
+fi
+: ${linker:=ld}
 test -n "$host" || die "cannot determine host"
 
 host_arch=
@@ -54,26 +66,74 @@ x86_64-*) host_arch=x86_64 ;;
 amd64-*) host_arch=x86_64 ;;
 aarch64-*) host_arch=aarch64 ;;
 arm64-*) host_arch=aarch64 ;;
-*) host_arch=unk ;;
+*) die "unsupported target '$host'"
 esac
 
+linkflags=
+
+findgcclibdir () {
+    : ${GCCLIBDIR:=$(dirname $(("$host"-gcc -print-file-name=crtbegin.o || $CC -print-file-name=crtbegin.o) 2>/dev/null))}
+    if ! test -n $GCCLIBDIR; then
+        die "cannot determine gcclibdir"
+    fi
+}
+
 case "$host" in
-*-linux-*)
-    host_os=linux ;;
+*-linux-*gnu*)
+    host_os=linux
+    host_abi=gnu
+    case $host_arch in
+    x86_64) DYNAMIC_LINKER=$sysroot/lib64/ld-linux-x86-64.so.2 ;;
+    aarch64) DYNAMIC_LINKER=$sysroot/lib/ld-linux-aarch64.so.1
+    esac
+    ldstartfiles='"-l:crt1.o", "-l:crti.o", "-l:crtbegin.o"'
+    ldendfiles='"-lc", "-l:crtend.o", "-l:crtn.o"'
+    ldstartfiles_pie='"-l:Scrt1.o", "-l:crti.o", "-l:crtbeginS.o"'
+    ldendfiles_pie='"-lc", "-l:crtendS.o", "-l:crtn.o"'
+    findgcclibdir
+    linkflags='"-L", "'$GCCLIBDIR'",'
+    host_predefs='
+    "__SIZE_TYPE__=unsigned long int", /* work around https://sourceware.org/bugzilla/show_bug.cgi?id=33969 */
+    '
+    ;;
+*-linux-*musl*)
+    host_os=linux
+    host_abi=musl
+    case $host_arch in
+    x86_64) DYNAMIC_LINKER=$sysroot/lib/ld-musl-x86_64.so.1 ;;
+    aarch64) DYNAMIC_LINKER=$sysroot/lib/ld-linux-aarch64.so.1 ;;
+    esac
+    if test -d $sysroot/usr/lib/musl/lib; then
+        # musl cross compiler in glibc system (probably)
+        findgcclibdir
+        linkflags='"-nostdlib", "-L", "'$sysroot'/usr/lib/musl/lib", "-L", "'$GCCLIBDIR'",'
+        ldstartfiles='"'$sysroot'/usr/lib/musl/lib/crt1.o","'$sysroot'/usr/lib/musl/lib/crti.o", "-l:crtbeginS.o"'
+        ldendfiles='"-lc", "-l:crtendS.o", "'$sysroot'/usr/lib/musl/lib/crtn.o"'
+        ldstartfiles_pie='"'$sysroot'/usr/lib/musl/lib/Scrt1.o", "'$sysroot'/usr/lib/musl/lib/crti.o", "-l:crtbeginS.o"'
+        ldendfiles_pie="$ldendfiles"
+    else
+        ldstartfiles='"-l:crt1.o", "-l:crti.o"'
+        ldendfiles='"-lc", "-l:crtn.o"'
+        ldstartfiles_pie='"-l:Scrt1.o", "-l:crti.o"'
+        ldendfiles_pie="$ldendfiles"
+    fi
+    ;;
 *-openbsd*)
     host_os=openbsd
+    DYNAMIC_LINKER=$sysroot/usr/libexec/ld.so
+    ldstartfiles='"-l:crt0.o", "-l:crtbegin.o"'
+    ldendfiles='"-lc", "-l:crtend.o"'
+    ldstartfiles_pie='"-l:crt0.o", "-l:crtbeginS.o"'
+    ldendfiles_pie='"-lc", "-l:crtendS.o"'
+    linkflags='"-L'$sysroot'/usr/lib"'
     host_predefs='
     "_ANSI_LIBRARY", /* works around __only_inline functions in libc headers */
     '
     ;;
-*) host_os=unknown ;;
-esac
-
-case "$host" in
-*-gnu) host_abi=gnu ;;
-*-musl) host_abi=musl ;;
-*) host_abi=none ;;
+*)
+    die "unknown/unsupported target '$host'"
 esac
+echo dynamic linker: $DYNAMIC_LINKER
 
 tryincldir() {
     if stat "$1"/ >/dev/null 2>/dev/null; then
@@ -83,11 +143,17 @@ tryincldir() {
 
 if ! test -n "$host_include_dirs"; then
     # try to query host toolchain first
-    if hostccdirs=$($CC -E -v -xc /dev/null 2>&1 | sed -n '/#include <...> search starts here:/,/End of search list./{
+    for sfx in cc gcc; do
+        if command -v "$host"-$sfx > /dev/null; then
+            cc="$host"-$sfx
+        fi
+    done
+    : ${cc:=$CC}
+    if hostccdirs=$($cc -E -v -xc /dev/null 2>&1 | sed -n '/#include <...> search starts here:/,/End of search list./{
     s/^[[:space:]]*//
     /^\/.*/p
 }') && test -n "$hostccdirs"; then
-        echo "Using $CC's system include paths:"
+        echo "Using $cc's system include paths:"
         for d in $hostccdirs; do
             dir=$(realpath "$d")
             case "$dir" in
@@ -99,15 +165,16 @@ if ! test -n "$host_include_dirs"; then
         done
     else
         echo "Falling back to generic system include paths"
-        tryincldir "$sysroot/local/include/$host"
-        tryincldir "$sysroot/local/include"
-        tryincldir "$sysroot/include/$host"
-        tryincldir "$sysroot/include"
+        tryincldir "$sysroot/usr/local/include/$host"
+        tryincldir "$sysroot/usr/local/include"
+        tryincldir "$sysroot/usr/include/$host"
+        tryincldir "$sysroot/usr/include"
     fi
 fi
 
 set -e
 
+echo using linker: "$linker"
 echo using arch: "$host_arch"
 echo using os: "$host_os"
 echo using abi: "$host_abi"
@@ -119,9 +186,15 @@ echo "/** GENERATED WITH $0 $@ **/
 #define HOST_OS     OS$host_os
 #define HOST_ABI    ABI$host_abi
 #define HOST_INCLUDE_DIRS $host_include_dirs
+#define HOST_LD \"$linker\"
 #define HOST_CC \"$CC\"
 static const char *const host_predefs[] = {$host_predefs 0
 };
+static const char *const host_linkcmd[] = {\"--dynamic-linker=$DYNAMIC_LINKER\", $linkflags};
+static const char *const host_ldstartfiles[] = {$ldstartfiles};
+static const char *const host_ldendfiles[] = {$ldendfiles};
+static const char *const host_ldstartfiles_pie[] = {$ldstartfiles_pie};
+static const char *const host_ldendfiles_pie[] = {$ldendfiles_pie};
 "  > src/hostconfig.h
 
 echo "# GENERATED WITH $0 $@
diff --git a/src/a_main.c b/src/a_main.c
index 33f0919..b947d41 100644
--- a/src/a_main.c
+++ b/src/a_main.c
@@ -113,6 +113,7 @@ typedef struct Task {
    vec_of(InFile) inf;
    char **runargs;
    vec_of(const char *) linkargs;
+   bool link_with_cc;
    bool verbose, run, syntaxonly;
 } Task;
 static InFile infilebuf[16];
@@ -134,6 +135,7 @@ optparse(char **args)
             ft ? ft : ftdetect(arg-1),
             arg[-1] != '-' ? arg-1 : "/dev/stdin"
          }));
+         vpush(&task.linkargs, NULL);
          ft = IFTauto;
          if (task.run) {
             task.runargs = args+1;
@@ -203,8 +205,24 @@ optparse(char **args)
          const char *s = *++args;
          if (!s) fatal(NULL, "missing target name");
          task.targ = s;
-      } else if (*arg == 'l' || *arg == 'L' || *arg == 'B' || !strcmp(arg, "shared") || !strcmp(arg, "pthread")) {
+      } else if (*arg == 'l' || *arg == 'L' || *arg == 'B' || !strcmp(arg, "shared") || !strcmp(arg, "pthread") || !strcmp(arg, "static")) {
+         /* XXX having some issues with linker commands for -shared */
+         if (!strcmp(arg, "shared"))
+            task.link_with_cc = 1;
          vpush(&task.linkargs, arg-1);
+      } else if (!memcmp(arg, "Wl,", 3)) {
+         if (task.link_with_cc) {
+            vpush(&task.linkargs, arg-1);
+         } else {
+            for (char *larg = (char*)arg+3, *next; larg && *larg; larg = next) {
+               next = strchr(larg, ',');
+               if (next) {
+                  *next = '\0';
+                  ++next;
+               }
+               vpush(&task.linkargs, larg);
+            }
+         }
       } else if (!strcmp(arg, "v") || !strcmp(arg, "-verbose")) {
          task.verbose = 1;
       } else if (!strcmp(arg, "c")) {
@@ -429,6 +447,7 @@ static void
 findlinkcmd(CmdArgs *cmd)
 {
    if (task.targ && iscrosscc()) {
+      task.link_with_cc = 1;
       /* try to find a cross compiling toolchain, e.g. aarch64-linux-gnu-gcc */
       static const char *ccs[] = {"cc", "gcc", "clang"};
       char cross[1024];
@@ -451,8 +470,16 @@ findlinkcmd(CmdArgs *cmd)
       } else {
          fatal(NULL, "cannot link to cross-compilation target: no appropiate toolchain installed");
       }
-   } else {
+   } else if (task.link_with_cc) {
       vpush(cmd, HOST_CC);
+   } else {
+      vpush(cmd, HOST_LD);
+      vpushn(cmd, host_linkcmd, countof(host_linkcmd));
+      if (ccopt.pie) {
+         vpushn(cmd, host_ldstartfiles_pie, countof(host_ldstartfiles_pie));
+      } else {
+         vpushn(cmd, host_ldstartfiles, countof(host_ldstartfiles));
+      }
    }
 }
 
@@ -476,19 +503,27 @@ dolink(void)
    vpush(&cmd, "-o");
    vpush(&cmd, task.out);
    assert(task.inf.n > 0);
-   for (int i = 0; i < task.inf.n; ++i) {
-      const char *o;
-      switch (task.inf.p[i].ft) {
-      case IFTc: o = task.inf.p[i].temp; break;
-      case IFTobj: case IFTar: case IFTdll:
-         o = task.inf.p[i].path; break;
-      default: assert(!"link obj?");
+
+   for (int l = 0, i = 0; l < task.linkargs.n; ++l) {
+      const char *a = task.linkargs.p[l];
+      if (!a) {
+         switch (task.inf.p[i].ft) {
+         case IFTc: a = task.inf.p[i].temp; break;
+         case IFTobj: case IFTar: case IFTdll:
+            a = task.inf.p[i].path; break;
+         default: assert(!"link obj?");
+         }
+         ++i;
       }
-      vpush(&cmd, o);
+      vpush(&cmd, a);
+   }
+   if (!task.link_with_cc) {
+      if (ccopt.pie)
+         vpushn(&cmd, host_ldendfiles_pie, countof(host_ldendfiles_pie));
+      else
+         vpushn(&cmd, host_ldendfiles, countof(host_ldendfiles));
    }
-   vpushn(&cmd, task.linkargs.p, task.linkargs.n);
    if (task.verbose) {
-      efmt("> ");
       for (int i = 0; i < cmd.n; ++i)
          efmt("%s ", cmd.p[i]);
       efmt("\n");
@@ -520,7 +555,7 @@ dorun(void)
       warn(NULL, "'-run' with cross-compiled binary");
    }
    if (task.verbose) {
-      efmt("> exec %s", task.out);
+      efmt("exec %s", task.out);
       for (char **s = task.runargs; *s; ++s)
          efmt(" %s", *s);
       efmt("\n");
@@ -682,6 +717,8 @@ prihelp(void)
    );
 }
 
+static const TargTriple host_targ = { HOST_ARCH, HOST_OS, HOST_ABI };
+
 int
 main(int argc, char **argv)
 {
@@ -694,6 +731,9 @@ main(int argc, char **argv)
    ccopt.cstd = STDC11;
    ccopt.pie = 1;
    ccopt.dbgout = &bstdout;
+   if (getenv("ANTCC_VERBOSE")) {
+      task.verbose = 1;
+   }
 
    /* parse cli ags */
    if (argc == 1) {
@@ -703,9 +743,9 @@ main(int argc, char **argv)
    optparse(argv);
 
    /* global init */
-   targ_init(task.targ);
-   if (!target.arch)
+   if (!targ_init(task.targ, &host_targ) || !target.arch) {
       fatal(NULL, "unsupported target: %s", task.targ ? task.targ : HOST_TRIPLE);
+   }
 
    for (const char *const *p = host_predefs; *p; ++p)
       cpppredef(0, *p);
diff --git a/src/a_targ.c b/src/a_targ.c
index fab23e9..1513bc6 100644
--- a/src/a_targ.c
+++ b/src/a_targ.c
@@ -67,18 +67,14 @@ parsetriple(TargTriple *trg, const char *str)
    return 1;
 }
 
-#include "hostconfig.h" /* run ./configure */
-
-void
-targ_init(const char *starg)
+bool
+targ_init(const char *starg, const TargTriple *dfault)
 {
    const struct Targ *t = NULL;
    uchar *sizes = targ_primsizes, *align = targ_primalign;
 
    if (!starg) {
-      target.arch = HOST_ARCH;
-      target.os = HOST_OS;
-      target.abi = HOST_ABI;
+      target = *dfault;
    } else if (!parsetriple(&target, starg)) {
       fatal(NULL, "unrecognized target: %s", starg);
    }
@@ -91,7 +87,7 @@ targ_init(const char *starg)
          break;
       }
    }
-   if (!t) fatal(NULL, "unsupported target: %s", starg ? starg : HOST_TRIPLE);
+   if (!t) return 0;
 
    sizes[TYBOOL] = sizes[TYCHAR] = sizes[TYSCHAR] = sizes[TYUCHAR] = 1;
    sizes[TYSHORT] = sizes[TYUSHORT] = 2;
@@ -123,4 +119,6 @@ targ_init(const char *starg)
    targ_64bit = t->ptrsize == 8;
    mctarg = t->mctarg;
    targ_arch = ISx86_64;
+
+   return 1;
 }
diff --git a/src/antcc.h b/src/antcc.h
index a666a95..59ae690 100644
--- a/src/antcc.h
+++ b/src/antcc.h
@@ -156,7 +156,7 @@ typedef struct TargTriple {
 extern TargTriple target;
 enum objkind { OBJELF };
 extern const struct MCTarg *mctarg;
-void targ_init(const char *);
+bool targ_init(const char *, const TargTriple *dfault);
 
 /*********/
 /** MEM **/