Kernel+LibC: Implement the openat() syscall

POSIX's openat() is very similar to open(), except you also provide a
file descriptor referring to a directory from which relative paths
should be resolved.

Passing it the magical fd number AT_FDCWD means "resolve from current
directory" (which is indeed also what open() normally does.)

This fixes libarchive's bsdtar, since it was trying to do something
extremely wrong in the absence of openat() support. The issue has
recently been fixed upstream in libarchive:

https://github.com/libarchive/libarchive/issues/1239

However, we should have openat() support anyway, so I went ahead and
implemented it. :^)

Fixes #748.

Kernel/Process.cpp

@@ -1351,15 +1351,12 @@ int Process::sys$open(const Syscall::SC_open_params* params)
 {
     if (!validate_read_typed(params))
         return -EFAULT;
-    auto* path = params->path;
-    auto path_length = params->path_length;
-    auto options = params->options;
-    auto mode = params->mode;
+    auto [path, path_length, options, mode] = *params;
+    if (!validate_read(path, path_length))
+        return -EFAULT;
 #ifdef DEBUG_IO
     dbgprintf("%s(%u) sys$open(\"%s\")\n", name().characters(), pid(), path);
 #endif
-    if (!validate_read(path, path_length))
-        return -EFAULT;
     int fd = alloc_fd();
     if (fd < 0)
         return fd;
@@ -1375,6 +1372,46 @@ int Process::sys$open(const Syscall::SC_open_params* params)
     return fd;
 }
 
+int Process::sys$openat(const Syscall::SC_openat_params* params)
+{
+    if (!validate_read_typed(params))
+        return -EFAULT;
+    auto [dirfd, path, path_length, options, mode] = *params;
+    if (!validate_read(path, path_length))
+        return -EFAULT;
+#ifdef DEBUG_IO
+    dbgprintf("%s(%u) sys$openat(%d, \"%s\")\n", dirfd, name().characters(), pid(), path);
+#endif
+    int fd = alloc_fd();
+    if (fd < 0)
+        return fd;
+
+    RefPtr<Custody> base;
+    if (dirfd == AT_FDCWD) {
+        base = current_directory();
+    } else {
+        auto* base_description = file_description(dirfd);
+        if (!base_description)
+            return -EBADF;
+        if (!base_description->is_directory())
+            return -ENOTDIR;
+        if (!base_description->custody())
+            return -EINVAL;
+        base = base_description->custody();
+    }
+
+    auto result = VFS::the().open(path, options, mode & ~umask(), *base);
+    if (result.is_error())
+        return result.error();
+    auto description = result.value();
+    if (options & O_DIRECTORY && !description->is_directory())
+        return -ENOTDIR; // FIXME: This should be handled by VFS::open.
+    description->set_file_flags(options);
+    u32 fd_flags = (options & O_CLOEXEC) ? FD_CLOEXEC : 0;
+    m_fds[fd].set(move(description), fd_flags);
+    return fd;
+}
+
 int Process::alloc_fd(int first_candidate_fd)
 {
     int fd = -EMFILE;

Kernel/Process.h

@@ -124,6 +124,7 @@ public:
     pid_t sys$getppid();
     mode_t sys$umask(mode_t);
     int sys$open(const Syscall::SC_open_params*);
+    int sys$openat(const Syscall::SC_openat_params*);
     int sys$close(int fd);
     ssize_t sys$read(int fd, u8*, ssize_t);
     ssize_t sys$write(int fd, const u8*, ssize_t);

Kernel/Syscall.h

@@ -134,7 +134,8 @@ struct timespec;
     __ENUMERATE_SYSCALL(fchdir)                 \
     __ENUMERATE_SYSCALL(getrandom)              \
     __ENUMERATE_SYSCALL(clock_gettime)          \
-    __ENUMERATE_SYSCALL(clock_nanosleep)
+    __ENUMERATE_SYSCALL(clock_nanosleep)        \
+    __ENUMERATE_SYSCALL(openat)
 
 namespace Syscall {
 
@@ -179,6 +180,14 @@ struct SC_open_params {
     u16 mode;
 };
 
+struct SC_openat_params {
+    int dirfd;
+    const char* path;
+    int path_length;
+    int options;
+    u16 mode;
+};
+
 struct SC_select_params {
     int nfds;
     fd_set* readfds;

Kernel/UnixTypes.h

@@ -448,3 +448,6 @@ struct ifreq {
 #define ifr_llprio ifr_ifru.ifru_metric       // link layer priority
 #define ifr_hwaddr ifr_ifru.ifru_hwaddr       // MAC address
 };
+
+#define AT_FDCWD -100
+