MFC r302517:

Fix a copy/paste bug introduced during X86_64 Linuxulator work.
FreeBSD support NX bit on X86_64 processors out of the box, for i386 emulation
use READ_IMPLIES_EXEC flag, introduced in r302515.

While here move common part of mmap() and mprotect() code to the files in compat/linux
to reduce code dupcliation between Linuxulator's.

MFC r302518, r302626:

Add linux_mmap.c to the appropriate conf/files.

2 files changed, 4 insertions, 172 deletions

diff --git a/sys/amd64/linux/linux.h b/sys/amd64/linux/linux.h
index 639499a..b30568f 100644
--- a/sys/amd64/linux/linux.h
+++ b/sys/amd64/linux/linux.h
@@ -139,13 +139,6 @@ struct l_rlimit {
 	l_ulong		rlim_max;
 };
 
-/* mmap options */
-#define	LINUX_MAP_SHARED	0x0001
-#define	LINUX_MAP_PRIVATE	0x0002
-#define	LINUX_MAP_FIXED		0x0010
-#define	LINUX_MAP_ANON		0x0020
-#define	LINUX_MAP_GROWSDOWN	0x0100
-
 /*
  * stat family of syscalls
  */
diff --git a/sys/amd64/linux/linux_machdep.c b/sys/amd64/linux/linux_machdep.c
index 9664dac..80c48aa 100644
--- a/sys/amd64/linux/linux_machdep.c
+++ b/sys/amd64/linux/linux_machdep.c
@@ -83,6 +83,7 @@ __FBSDID("$FreeBSD$");
 #include <compat/linux/linux_ipc.h>
 #include <compat/linux/linux_file.h>
 #include <compat/linux/linux_misc.h>
+#include <compat/linux/linux_mmap.h>
 #include <compat/linux/linux_signal.h>
 #include <compat/linux/linux_util.h>
 #include <compat/linux/linux_emul.h>
@@ -122,181 +123,19 @@ linux_set_upcall_kse(struct thread *td, register_t stack)
 	return (0);
 }
 
-#define STACK_SIZE  (2 * 1024 * 1024)
-#define GUARD_SIZE  (4 * PAGE_SIZE)
-
 int
 linux_mmap2(struct thread *td, struct linux_mmap2_args *args)
 {
-	struct proc *p = td->td_proc;
-	struct mmap_args /* {
-		caddr_t addr;
-		size_t len;
-		int prot;
-		int flags;
-		int fd;
-		long pad;
-		off_t pos;
-	} */ bsd_args;
-	int error;
-	struct file *fp;
-	cap_rights_t rights;
-
-	LINUX_CTR6(mmap2, "0x%lx, %ld, %ld, 0x%08lx, %ld, 0x%lx",
-	    args->addr, args->len, args->prot,
-	    args->flags, args->fd, args->pgoff);
-
-	error = 0;
-	bsd_args.flags = 0;
-	fp = NULL;
-
-	/*
-	 * Linux mmap(2):
-	 * You must specify exactly one of MAP_SHARED and MAP_PRIVATE
-	 */
-	if (! ((args->flags & LINUX_MAP_SHARED) ^
-	    (args->flags & LINUX_MAP_PRIVATE)))
-		return (EINVAL);
-
-	if (args->flags & LINUX_MAP_SHARED)
-		bsd_args.flags |= MAP_SHARED;
-	if (args->flags & LINUX_MAP_PRIVATE)
-		bsd_args.flags |= MAP_PRIVATE;
-	if (args->flags & LINUX_MAP_FIXED)
-		bsd_args.flags |= MAP_FIXED;
-	if (args->flags & LINUX_MAP_ANON)
-		bsd_args.flags |= MAP_ANON;
-	else
-		bsd_args.flags |= MAP_NOSYNC;
-	if (args->flags & LINUX_MAP_GROWSDOWN)
-		bsd_args.flags |= MAP_STACK;
 
-	/*
-	 * PROT_READ, PROT_WRITE, or PROT_EXEC implies PROT_READ and PROT_EXEC
-	 * on Linux/i386. We do this to ensure maximum compatibility.
-	 * Linux/ia64 does the same in i386 emulation mode.
-	 */
-	bsd_args.prot = args->prot;
-	if (bsd_args.prot & (PROT_READ | PROT_WRITE | PROT_EXEC))
-		bsd_args.prot |= PROT_READ | PROT_EXEC;
-
-	/* Linux does not check file descriptor when MAP_ANONYMOUS is set. */
-	bsd_args.fd = (bsd_args.flags & MAP_ANON) ? -1 : args->fd;
-	if (bsd_args.fd != -1) {
-		/*
-		 * Linux follows Solaris mmap(2) description:
-		 * The file descriptor fildes is opened with
-		 * read permission, regardless of the
-		 * protection options specified.
-		 */
-
-		error = fget(td, bsd_args.fd,
-		    cap_rights_init(&rights, CAP_MMAP), &fp);
-		if (error != 0 )
-			return (error);
-		if (fp->f_type != DTYPE_VNODE) {
-			fdrop(fp, td);
-			return (EINVAL);
-		}
-
-		/* Linux mmap() just fails for O_WRONLY files */
-		if (!(fp->f_flag & FREAD)) {
-			fdrop(fp, td);
-			return (EACCES);
-		}
-
-		fdrop(fp, td);
-	}
-
-	if (args->flags & LINUX_MAP_GROWSDOWN) {
-		/*
-		 * The Linux MAP_GROWSDOWN option does not limit auto
-		 * growth of the region.  Linux mmap with this option
-		 * takes as addr the inital BOS, and as len, the initial
-		 * region size.  It can then grow down from addr without
-		 * limit.  However, Linux threads has an implicit internal
-		 * limit to stack size of STACK_SIZE.  Its just not
-		 * enforced explicitly in Linux.  But, here we impose
-		 * a limit of (STACK_SIZE - GUARD_SIZE) on the stack
-		 * region, since we can do this with our mmap.
-		 *
-		 * Our mmap with MAP_STACK takes addr as the maximum
-		 * downsize limit on BOS, and as len the max size of
-		 * the region.  It then maps the top SGROWSIZ bytes,
-		 * and auto grows the region down, up to the limit
-		 * in addr.
-		 *
-		 * If we don't use the MAP_STACK option, the effect
-		 * of this code is to allocate a stack region of a
-		 * fixed size of (STACK_SIZE - GUARD_SIZE).
-		 */
-
-		if ((caddr_t)PTRIN(args->addr) + args->len >
-		    p->p_vmspace->vm_maxsaddr) {
-			/*
-			 * Some Linux apps will attempt to mmap
-			 * thread stacks near the top of their
-			 * address space.  If their TOS is greater
-			 * than vm_maxsaddr, vm_map_growstack()
-			 * will confuse the thread stack with the
-			 * process stack and deliver a SEGV if they
-			 * attempt to grow the thread stack past their
-			 * current stacksize rlimit.  To avoid this,
-			 * adjust vm_maxsaddr upwards to reflect
-			 * the current stacksize rlimit rather
-			 * than the maximum possible stacksize.
-			 * It would be better to adjust the
-			 * mmap'ed region, but some apps do not check
-			 * mmap's return value.
-			 */
-			PROC_LOCK(p);
-			p->p_vmspace->vm_maxsaddr = (char *)USRSTACK -
-			    lim_cur(p, RLIMIT_STACK);
-			PROC_UNLOCK(p);
-		}
-
-		/*
-		 * This gives us our maximum stack size and a new BOS.
-		 * If we're using VM_STACK, then mmap will just map
-		 * the top SGROWSIZ bytes, and let the stack grow down
-		 * to the limit at BOS.  If we're not using VM_STACK
-		 * we map the full stack, since we don't have a way
-		 * to autogrow it.
-		 */
-		if (args->len > STACK_SIZE - GUARD_SIZE) {
-			bsd_args.addr = (caddr_t)PTRIN(args->addr);
-			bsd_args.len = args->len;
-		} else {
-			bsd_args.addr = (caddr_t)PTRIN(args->addr) -
-			    (STACK_SIZE - GUARD_SIZE - args->len);
-			bsd_args.len = STACK_SIZE - GUARD_SIZE;
-		}
-	} else {
-		bsd_args.addr = (caddr_t)PTRIN(args->addr);
-		bsd_args.len  = args->len;
-	}
-	bsd_args.pos = (off_t)args->pgoff;
-
-	error = sys_mmap(td, &bsd_args);
-
-	LINUX_CTR2(mmap2, "return: %d (%p)",
-	    error, td->td_retval[0]);
-	return (error);
+	return (linux_mmap_common(td, PTROUT(args->addr), args->len, args->prot,
+		args->flags, args->fd, args->pgoff));
 }
 
 int
 linux_mprotect(struct thread *td, struct linux_mprotect_args *uap)
 {
-	struct mprotect_args bsd_args;
-
-	LINUX_CTR(mprotect);
 
-	bsd_args.addr = uap->addr;
-	bsd_args.len = uap->len;
-	bsd_args.prot = uap->prot;
-	if (bsd_args.prot & (PROT_READ | PROT_WRITE | PROT_EXEC))
-		bsd_args.prot |= PROT_READ | PROT_EXEC;
-	return (sys_mprotect(td, &bsd_args));
+	return (linux_mprotect_common(td, PTROUT(uap->addr), uap->len, uap->prot));
 }
 
 int