fix faccessat to support AT_EACCESS flag

this is another case of the kernel syscall failing to support flags
where it needs to, leading to horrible workarounds in userspace. this
time the workaround requires changing uid/gid, and that's not safe to
do in the current process. in the worst case, kernel resource limits
might prevent recovering the original values, and then there would be
no way to safely return. so, use the safe but horribly inefficient
alternative: forking. clone is used instead of fork to suppress
signals from the child.

fortunately this worst-case code is only needed when effective and
real ids mismatch, which mainly happens in suid programs.

src/unistd/faccessat.c

@@ -1,7 +1,52 @@
 #include <unistd.h>
+#include <fcntl.h>
 #include "syscall.h"
+#include "pthread_impl.h"
+
+struct ctx {
+	int fd;
+	const char *filename;
+	int amode;
+	int p;
+};
+
+static int checker(void *p)
+{
+	struct ctx *c = p;
+	int ret;
+	if (__syscall(SYS_setgid, __syscall(SYS_getegid))
+	    || __syscall(SYS_setuid, __syscall(SYS_geteuid)))
+		__syscall(SYS_exit, 1);
+	ret = __syscall(SYS_faccessat, c->fd, c->filename, c->amode, 0);
+	__syscall(SYS_write, c->p, &ret, sizeof ret);
+	__syscall(SYS_exit, 0);
+}
 
 int faccessat(int fd, const char *filename, int amode, int flag)
 {
-	return syscall(SYS_faccessat, fd, filename, amode, flag);
+	if (!flag || (flag==AT_EACCESS && getuid()==geteuid() && getgid()==getegid()))
+		return syscall(SYS_faccessat, fd, filename, amode, flag);
+
+	if (flag != AT_EACCESS)
+		return __syscall_ret(-EINVAL);
+
+	char stack[1024];
+	sigset_t set;
+	int ret, p[2];
+
+	if (pipe(p)) return __syscall_ret(-EBUSY);
+	struct ctx c = { .fd = fd, .filename = filename, .amode = amode, .p = p[1] };
+
+	__block_app_sigs(&set);
+	
+	ret = __clone(checker, stack+sizeof stack, 0, &c);
+	__syscall(SYS_close, p[1]);
+
+	if (ret<0 || __syscall(SYS_read, p[0], &ret, sizeof ret) != sizeof(ret))
+		ret = -EBUSY;
+	__syscall(SYS_close, p[0]);
+
+	__restore_sigs(&set);
+
+	return __syscall_ret(ret);
 }