PSARC/2008/226 mmapobj(2) - mmap object

6502792 Same dynamic libraries should be mapped at the same virtual addresses in different processes

5 files changed, 129 insertions, 7 deletions

diff --git a/usr/src/uts/common/sys/Makefile b/usr/src/uts/common/sys/Makefile
index d0d808d459..5fe7393f56 100644
--- a/usr/src/uts/common/sys/Makefile
+++ b/usr/src/uts/common/sys/Makefile
@@ -368,6 +368,7 @@ CHKHDRS=			\
 	miiregs.h		\
 	mixer.h			\
 	mman.h			\
+	mmapobj.h		\
 	mntent.h		\
 	mntio.h			\
 	mnttab.h		\
diff --git a/usr/src/uts/common/sys/mman.h b/usr/src/uts/common/sys/mman.h
index a155d8d68b..1c8da540d1 100644
--- a/usr/src/uts/common/sys/mman.h
+++ b/usr/src/uts/common/sys/mman.h
@@ -40,8 +40,6 @@
 #ifndef	_SYS_MMAN_H
 #define	_SYS_MMAN_H
 
-#pragma ident	"%Z%%M%	%I%	%E% SMI"
-
 #include <sys/feature_tests.h>
 
 #ifdef	__cplusplus
@@ -121,6 +119,71 @@ extern "C" {
 
 #include <sys/types.h>
 
+#endif	/* !_ASM && !_KERNEL */
+
+/* External flags for mmapobj syscall (Exclusive of MAP_* flags above) */
+#define	MMOBJ_PADDING		0x10000
+#define	MMOBJ_INTERPRET		0x20000
+
+#define	MMOBJ_ALL_FLAGS		(MMOBJ_PADDING | MMOBJ_INTERPRET)
+
+/*
+ * Values for mr_flags field of mmapobj_result_t below.
+ * The bottom 16 bits are mutually exclusive and thus only one
+ * of them can be set at a time.  Use MR_GET_TYPE below to check this value.
+ * The top 16 bits are used for flags which are not mutually exclusive and
+ * thus more than one of these flags can be set for a given mmapobj_result_t.
+ *
+ * MR_PADDING being set indicates that this memory range represents the user
+ * requested padding.
+ *
+ * MR_HDR_ELF being set indicates that the ELF header of the mapped object
+ * is mapped at mr_addr + mr_offset.
+ *
+ * MR_HDR_AOUT being set indicates that the AOUT (4.x) header of the mapped
+ * object is mapped at mr_addr + mr_offset.
+ */
+
+/*
+ * External flags for mr_flags field below.
+ */
+#define	MR_PADDING	0x1
+#define	MR_HDR_ELF	0x2
+#define	MR_HDR_AOUT	0x3
+
+/*
+ * Internal flags for mr_flags field below.
+ */
+#ifdef	_KERNEL
+#define	MR_RESV	0x80000000	/* overmapped /dev/null */
+#endif	/* _KERNEL */
+
+#define	MR_TYPE_MASK 0x0000ffff
+#define	MR_GET_TYPE(val)	((val) & MR_TYPE_MASK)
+
+#if	!defined(_ASM)
+typedef struct mmapobj_result {
+	caddr_t		mr_addr;	/* mapping address */
+	size_t		mr_msize;	/* mapping size */
+	size_t		mr_fsize;	/* file size */
+	size_t		mr_offset;	/* offset into file */
+	uint_t		mr_prot;	/* the protections provided */
+	uint_t		mr_flags;	/* info on the mapping */
+} mmapobj_result_t;
+
+#if defined(_KERNEL) || defined(_SYSCALL32)
+typedef struct mmapobj_result32 {
+	caddr32_t	mr_addr;	/* mapping address */
+	size32_t	mr_msize;	/* mapping size */
+	size32_t	mr_fsize;	/* file size */
+	size32_t	mr_offset;	/* offset into file */
+	uint_t		mr_prot;	/* the protections provided */
+	uint_t		mr_flags;	/* info on the mapping */
+} mmapobj_result32_t;
+#endif	/* defined(_KERNEL) || defined(_SYSCALL32) */
+#endif	/* !defined(_ASM) */
+
+#if	!defined(_ASM) && !defined(_KERNEL)
 /*
  * large file compilation environment setup
  *
@@ -157,6 +220,7 @@ extern "C" {
 #ifdef	__STDC__
 #if (_POSIX_C_SOURCE > 2) || defined(_XPG4_2)
 extern void *mmap(void *, size_t, int, int, int, off_t);
+extern int mmapobj(int, uint_t, void *, uint_t *, void *);
 extern int munmap(void *, size_t);
 extern int mprotect(void *, size_t, int);
 extern int msync(void *, size_t, int);
@@ -172,6 +236,7 @@ extern void *mmap64(void *, size_t, int, int, int, off64_t);
 #else	/* (_POSIX_C_SOURCE > 2) || defined(_XPG4_2) */
 extern caddr_t mmap(caddr_t, size_t, int, int, int, off_t);
 extern int munmap(caddr_t, size_t);
+extern int mmapobj(int, uint_t, mmapobj_result_t *, uint_t *, void *);
 extern int mprotect(caddr_t, size_t, int);
 extern int msync(caddr_t, size_t, int);
 extern int mlock(caddr_t, size_t);
@@ -211,6 +276,7 @@ extern int posix_madvise(void *, size_t, int);
 #else	/* __STDC__ */
 extern caddr_t mmap();
 extern int munmap();
+extern int mmapobj();
 extern int mprotect();
 extern int mincore();
 extern int memcntl();
diff --git a/usr/src/uts/common/sys/mmapobj.h b/usr/src/uts/common/sys/mmapobj.h
new file mode 100644
index 0000000000..fd3f8f16f4
--- /dev/null
+++ b/usr/src/uts/common/sys/mmapobj.h
@@ -0,0 +1,52 @@
+/*
+ * CDDL HEADER START
+ *
+ * The contents of this file are subject to the terms of the
+ * Common Development and Distribution License (the "License").
+ * You may not use this file except in compliance with the License.
+ *
+ * You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
+ * or http://www.opensolaris.org/os/licensing.
+ * See the License for the specific language governing permissions
+ * and limitations under the License.
+ *
+ * When distributing Covered Code, include this CDDL HEADER in each
+ * file and include the License file at usr/src/OPENSOLARIS.LICENSE.
+ * If applicable, add the following below this CDDL HEADER, with the
+ * fields enclosed by brackets "[]" replaced with your own identifying
+ * information: Portions Copyright [yyyy] [name of copyright owner]
+ *
+ * CDDL HEADER END
+ */
+/*
+ * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
+ * Use is subject to license terms.
+ */
+
+#ifndef _SYS_MMAPOBJ_H
+#define	_SYS_MMAPOBJ_H
+
+#ifdef	__cplusplus
+extern "C" {
+#endif
+
+/*
+ * Number of mmapobj_result_t structures created on stack for quick allocation.
+ * More will be manually kmem_alloc'ed if needed.
+ * Should be enough for most ELF objects.
+ */
+#define	LIBVA_CACHED_SEGS 3
+
+#ifdef _KERNEL
+extern void lib_va_init(void);
+extern void mmapobj_unmap(mmapobj_result_t *, int, int, ushort_t);
+#endif
+
+extern int mmapobj(vnode_t *, uint_t, mmapobj_result_t *, uint_t *, size_t,
+    cred_t *);
+
+#ifdef	__cplusplus
+}
+#endif
+
+#endif	/* _SYS_MMAPOBJ_H */
diff --git a/usr/src/uts/common/sys/syscall.h b/usr/src/uts/common/sys/syscall.h
index 779672e7f5..5b186aaf06 100644
--- a/usr/src/uts/common/sys/syscall.h
+++ b/usr/src/uts/common/sys/syscall.h
@@ -30,8 +30,6 @@
 #ifndef	_SYS_SYSCALL_H
 #define	_SYS_SYSCALL_H
 
-#pragma ident	"%Z%%M%	%I%	%E% SMI"
-
 #ifdef	__cplusplus
 extern "C" {
 #endif
@@ -319,6 +317,7 @@ extern "C" {
 #define	SYS_lxstat	124
 #define	SYS_fxstat	125
 #define	SYS_xmknod	126
+#define	SYS_mmapobj	127
 #define	SYS_setrlimit	128
 #define	SYS_getrlimit	129
 #define	SYS_lchown	130
diff --git a/usr/src/uts/common/sys/vmem.h b/usr/src/uts/common/sys/vmem.h
index abcc8b26bc..a76d0fbed7 100644
--- a/usr/src/uts/common/sys/vmem.h
+++ b/usr/src/uts/common/sys/vmem.h
@@ -19,15 +19,13 @@
  * CDDL HEADER END
  */
 /*
- * Copyright 2007 Sun Microsystems, Inc.  All rights reserved.
+ * Copyright 2008 Sun Microsystems, Inc.  All rights reserved.
  * Use is subject to license terms.
  */
 
 #ifndef _SYS_VMEM_H
 #define	_SYS_VMEM_H
 
-#pragma ident	"%Z%%M%	%I%	%E% SMI"
-
 #include <sys/types.h>
 
 #ifdef	__cplusplus
@@ -62,6 +60,12 @@ extern "C" {
  */
 #define	VM_ABORT	0x00002000
 
+/*
+ * VM_ENDALLOC requests that large addresses be preferred in allocations.
+ * Has no effect if VM_NEXTFIT is active.
+ */
+#define	VM_ENDALLOC	0x00004000
+
 #define	VM_FLAGS	0x0000FFFF
 
 /*