Merge user/peter/kinfo branch as of r185547 into head.

This changes struct kinfo_filedesc and kinfo_vmentry such that they are
same on both 32 and 64 bit platforms like i386/amd64 and won't require
sysctl wrapping.

Two new OIDs are assigned.  The old ones are available under
COMPAT_FREEBSD7 - but it isn't that simple.  The superceded interface
was never actually released on 7.x.

The other main change is to pack the data passed to userland via the
sysctl.  kf_structsize and kve_structsize are reduced for the copyout.
If you have a process with 100,000+ sockets open, the unpacked records
require a 132MB+ copyout.  With packing, it is "only" ~35MB.  (Still
seriously unpleasant, but not quite as devastating).  A similar problem
exists for the vmentry structure - have lots and lots of shared libraries
and small mmaps and its copyout gets expensive too.

My immediate problem is valgrind.  It traditionally achieves this
functionality by parsing procfs output, in a packed format.  Secondly, when
tracing 32 bit binaries on amd64 under valgrind, it uses a cross compiled
32 bit binary which ran directly into the differing data structures in 32
vs 64 bit mode.  (valgrind uses this to track file descriptor operations
and this therefore affected every single 32 bit binary)

I've added two utility functions to libutil to unpack the structures into
a fixed record length and to make it a little more convenient to use.

1 files changed, 77 insertions, 8 deletions

diff --git a/sys/sys/user.h b/sys/sys/user.h
index a12669f..93a78d3 100644
--- a/sys/sys/user.h
+++ b/sys/sys/user.h
@@ -277,20 +277,55 @@ struct user {
 #define	KF_FLAG_DIRECT		0x00000040
 #define	KF_FLAG_HASLOCK		0x00000080
 
-struct kinfo_file {
+/*
+ * Old format.  Has variable hidden padding due to alignment.
+ * This is a compatability hack for pre-build 7.1 packages.
+ */
+#if defined(__amd64__)
+#define	KINFO_OFILE_SIZE	1328
+#endif
+#if defined(__i386__)
+#define	KINFO_OFILE_SIZE	1324
+#endif
+
+struct kinfo_ofile {
 	int	kf_structsize;			/* Size of kinfo_file. */
 	int	kf_type;			/* Descriptor type. */
 	int	kf_fd;				/* Array index. */
 	int	kf_ref_count;			/* Reference count. */
 	int	kf_flags;			/* Flags. */
+	/* XXX Hidden alignment padding here on amd64 */
 	off_t	kf_offset;			/* Seek location. */
 	int	kf_vnode_type;			/* Vnode type. */
 	int	kf_sock_domain;			/* Socket domain. */
 	int	kf_sock_type;			/* Socket type. */
 	int	kf_sock_protocol;		/* Socket protocol. */
-	char	kf_path[PATH_MAX];		/* Path to file, if any. */
+	char	kf_path[PATH_MAX];	/* Path to file, if any. */
+	struct sockaddr_storage kf_sa_local;	/* Socket address. */
+	struct sockaddr_storage	kf_sa_peer;	/* Peer address. */
+};
+
+#if defined(__amd64__) || defined(__i386__)
+#define	KINFO_FILE_SIZE	1392
+#endif
+
+struct kinfo_file {
+	int	kf_structsize;			/* Variable size of record. */
+	int	kf_type;			/* Descriptor type. */
+	int	kf_fd;				/* Array index. */
+	int	kf_ref_count;			/* Reference count. */
+	int	kf_flags;			/* Flags. */
+	int	_kf_pad0;			/* Round to 64 bit alignment */
+	uint64_t kf_offset;			/* Seek location. */
+	int	kf_vnode_type;			/* Vnode type. */
+	int	kf_sock_domain;			/* Socket domain. */
+	int	kf_sock_type;			/* Socket type. */
+	int	kf_sock_protocol;		/* Socket protocol. */
 	struct sockaddr_storage kf_sa_local;	/* Socket address. */
 	struct sockaddr_storage	kf_sa_peer;	/* Peer address. */
+	int	_kf_ispare[16];			/* Space for more stuff. */
+	/* Truncated before copyout in sysctl */
+	char	kf_path[PATH_MAX];		/* Path to file, if any. */
 };
 
 /*
@@ -313,11 +348,18 @@ struct kinfo_file {
 #define	KVME_FLAG_COW		0x00000001
 #define	KVME_FLAG_NEEDS_COPY	0x00000002
 
-struct kinfo_vmentry {
+#if defined(__amd64__)
+#define	KINFO_OVMENTRY_SIZE	1168
+#endif
+#if defined(__i386__)
+#define	KINFO_OVMENTRY_SIZE	1128
+#endif
+
+struct kinfo_ovmentry {
 	int	 kve_structsize;		/* Size of kinfo_vmmapentry. */
 	int	 kve_type;			/* Type of map entry. */
-	void	*kve_start;			/* Starting pointer. */
-	void	*kve_end;			/* Finishing pointer. */
+	void	*kve_start;			/* Starting address. */
+	void	*kve_end;			/* Finishing address. */
 	int	 kve_flags;			/* Flags on map entry. */
 	int	 kve_resident;			/* Number of resident pages. */
 	int	 kve_private_resident;		/* Number of private pages. */
@@ -327,11 +369,35 @@ struct kinfo_vmentry {
 	char	 kve_path[PATH_MAX];		/* Path to VM obj, if any. */
 	void	*_kve_pspare[8];		/* Space for more stuff. */
 	off_t	 kve_offset;			/* Mapping offset in object */
-	uint64_t kve_fileid;			/* inode number of vnode */
+	uint64_t kve_fileid;			/* inode number if vnode */
 	dev_t	 kve_fsid;			/* dev_t of vnode location */
 	int	 _kve_ispare[3];		/* Space for more stuff. */
 };
 
+#if defined(__amd64__) || defined(__i386__)
+#define	KINFO_VMENTRY_SIZE	1160
+#endif
+
+struct kinfo_vmentry {
+	int	 kve_structsize;		/* Variable size of record. */
+	int	 kve_type;			/* Type of map entry. */
+	uint64_t kve_start;			/* Starting address. */
+	uint64_t kve_end;			/* Finishing address. */
+	uint64_t kve_offset;			/* Mapping offset in object */
+	uint64_t kve_fileid;			/* inode number if vnode */
+	uint32_t kve_fsid;			/* dev_t of vnode location */
+	int	 kve_flags;			/* Flags on map entry. */
+	int	 kve_resident;			/* Number of resident pages. */
+	int	 kve_private_resident;		/* Number of private pages. */
+	int	 kve_protection;		/* Protection bitmask. */
+	int	 kve_ref_count;			/* VM obj ref count. */
+	int	 kve_shadow_count;		/* VM obj shadow count. */
+	int	 _kve_pad0;			/* 64bit align next field */
+	int	 _kve_ispare[16];		/* Space for more stuff. */
+	/* Truncated before copyout in sysctl */
+	char	 kve_path[PATH_MAX];		/* Path to VM obj, if any. */
+};
+
 /*
  * The KERN_PROC_KSTACK sysctl allows a process to dump the kernel stacks of
  * another process as a series of entries.  Each stack is represented by a
@@ -343,12 +409,15 @@ struct kinfo_vmentry {
 #define	KKST_STATE_SWAPPED	1		/* Stack swapped out. */
 #define	KKST_STATE_RUNNING	2		/* Stack ephemeral. */
 
+#if defined(__amd64__) || defined(__i386__)
+#define	KINFO_KSTACK_SIZE	1096
+#endif
+
 struct kinfo_kstack {
 	lwpid_t	 kkst_tid;			/* ID of thread. */
 	int	 kkst_state;			/* Validity of stack. */
 	char	 kkst_trace[KKST_MAXLEN];	/* String representing stack. */
-	void	*_kkst_pspare[8];		/* Space for more stuff. */
-	int	 _kkst_ispare[8];		/* Space for more stuff. */
+	int	 _kkst_ispare[16];		/* Space for more stuff. */
 };
 
 #endif