kernel: execve syscall

kernel-src/arch/x86-64/syscall.asm

@@ -11,6 +11,7 @@ extern syscall_getpid
 extern syscall_fstat
 extern syscall_fstatat
 extern syscall_fork
+extern syscall_execve
 syscalltab:
 dq syscall_print
 dq syscall_mmap
@@ -24,7 +25,8 @@ dq syscall_getpid
 dq syscall_fstat
 dq syscall_fstatat
 dq syscall_fork
-syscallcount equ 12
+dq syscall_execve
+syscallcount equ 13
 section .text
 global arch_syscall_entry
 ; on entry:

kernel-src/sys/syscalls/execve.c

@@ -0,0 +1,148 @@
+#include <kernel/syscalls.h>
+#include <kernel/alloc.h>
+#include <string.h>
+#include <errno.h>
+#include <kernel/vmm.h>
+#include <arch/cpu.h>
+#include <kernel/elf.h>
+#include <kernel/scheduler.h>
+#include <logging.h>
+
+static void freevec(char **v) {
+	char **it = v;
+	while (*it)
+		free(*it++);
+
+	free(v);
+}
+
+syscallret_t syscall_execve(context_t *context, char *upath, char *uargv[], char *uenvp[]) {
+	syscallret_t ret = {
+		.ret = -1
+	};
+
+	// TODO user str ops
+	char *path = alloc(strlen(upath) + 1);
+	if (path == NULL) {
+		ret.errno = ENOMEM;
+		return ret;
+	}
+	strcpy(path, upath);
+
+	size_t argsize = 0;
+	size_t envsize = 0;
+	char **argv = NULL;
+	char **envp = NULL;
+	vmmcontext_t *vmmctx = NULL;
+	vmmcontext_t *oldctx = _cpu()->vmmctx;
+	vnode_t *node = NULL;
+	vnode_t *refnode = NULL;
+	char *interp = NULL;
+
+	while (uargv[argsize++]);
+	while (uenvp[envsize++]);
+
+	argv = alloc((argsize + 1) * sizeof(char *));
+	envp = alloc((envsize + 1) * sizeof(char *));
+
+	if (argv == NULL || envp == NULL) {
+		ret.errno = ENOMEM;
+		goto error;
+	}
+
+	for (int i = 0; i < argsize - 1; ++i) {
+		argv[i] = alloc(strlen(uargv[i]) + 1);
+		if (argv[i] == NULL) {
+			ret.errno = ENOMEM;
+			goto error;
+		}
+		strcpy(argv[i], uargv[i]);
+	}
+
+	for (int i = 0; i < envsize - 1; ++i) {
+		envp[i] = alloc(strlen(uenvp[i]) + 1);
+		if (envp[i] == NULL) {
+			ret.errno = ENOMEM;
+			goto error;
+		}
+		strcpy(envp[i], uenvp[i]);
+	}
+
+	vmmctx = vmm_newcontext();
+	if (vmmctx == NULL) {
+		ret.errno = ENOMEM;
+		goto error;
+	}
+
+	refnode = *path == '/' ? sched_getroot() : sched_getcwd();
+
+	ret.errno = vfs_lookup(&node, vfsroot, "/usr/bin/bash", NULL, 0);
+	if (ret.errno)
+		goto error;
+
+	// TODO permission and type checks and all
+
+	vmm_switchcontext(vmmctx);
+
+	auxv64list_t auxv64;
+	void *entry;
+	ret.errno = elf_load(node, NULL, &entry, &interp, &auxv64);
+	if (ret.errno)
+		goto error;
+
+	if (interp) {
+		vnode_t *interpnode;
+		ret.errno = vfs_open(vfsroot, interp, 0, &interpnode);
+		if (ret.errno)
+			goto error;
+
+		auxv64list_t interpauxv;
+		char *interpinterp = NULL;
+		ret.errno = elf_load(interpnode, INTERP_BASE, &entry, &interpinterp, &interpauxv);
+		if (ret.errno)
+			goto error;
+
+		__assert(interpinterp == NULL);
+	}
+
+	void *stack = elf_preparestack(STACK_TOP, &auxv64, argv, envp);
+	if (stack == NULL) {
+		ret.errno = ENOMEM;
+		goto error;
+	}
+
+	// TODO kill other threads and close CLOEXEC fds
+
+	error:
+	free(path);
+	if (argv)
+		freevec(argv);
+
+	if (envp)
+		freevec(envp);
+
+	if (vmmctx && ret.errno) {
+		vmm_switchcontext(oldctx);
+		vmm_destroycontext(vmmctx, 0);
+	}
+
+	if (refnode)
+		VOP_RELEASE(refnode);
+
+	if (node)
+		VOP_RELEASE(node);
+
+	if (interp)
+		free(interp);
+
+	if (ret.errno == 0) {
+		// TODO sync mmap files (unmap all lower memory first before destroying context?)
+		vmm_destroycontext(oldctx, 0);
+		CTX_SP(context) = (uint64_t)stack;
+		CTX_IP(context) = (uint64_t)entry;
+	}
+
+	printf("exec ret: %lu %lu\n", ret.ret, ret.errno);
+
+	return ret;
+}

patches/mlibc/jinx-working-patch.patch

@@ -133,10 +133,10 @@ index 0000000..e5b087a
 +
 diff --git mlibc-workdir/sysdeps/astral/generic/generic.cpp mlibc-workdir/sysdeps/astral/generic/generic.cpp
 new file mode 100644
-index 0000000..3961e7f
+index 0000000..e108341
 --- /dev/null
 +++ mlibc-workdir/sysdeps/astral/generic/generic.cpp
-@@ -0,0 +1,185 @@
+@@ -0,0 +1,190 @@
 +#include <bits/ensure.h>
 +#include <mlibc/debug.hpp>
 +#include <mlibc/all-sysdeps.hpp>
@@ -200,6 +200,11 @@ index 0000000..3961e7f
 +		return 0;
 +	}
 +
++	int sys_execve(const char *path, char *const argv[], char *const envp[]) {
++		long ret = 0;
++		return syscall(SYSCALL_EXECVE, &ret, (uint64_t)path, (uint64_t)argv, (uint64_t)envp);
++	}
++
 +	int sys_fork(pid_t *pid) {
 +		long ret = 0;
 +		long error = syscall(SYSCALL_FORK, &ret);
@@ -341,10 +346,10 @@ index 0000000..990eca0
 +#endif
 diff --git mlibc-workdir/sysdeps/astral/include/astral/syscall.h mlibc-workdir/sysdeps/astral/include/astral/syscall.h
 new file mode 100644
-index 0000000..16c462d
+index 0000000..e23be0d
 --- /dev/null
 +++ mlibc-workdir/sysdeps/astral/include/astral/syscall.h
-@@ -0,0 +1,35 @@
+@@ -0,0 +1,36 @@
 +#ifndef _SYSCALL_H_INCLUDE
 +#define _SYSCALL_H_INCLUDE
 +
@@ -360,6 +365,7 @@ index 0000000..16c462d
 +#define SYSCALL_FSTAT 9
 +#define SYSCALL_FSTATAT 10
 +#define SYSCALL_FORK 11
++#define SYSCALL_EXECVE 12
 +
 +#include <stddef.h>
 +#include <stdint.h>