Build on Jeff Roberson's linker-set based dynamic per-CPU allocator

(DPCPU), as suggested by Peter Wemm, and implement a new per-virtual
network stack memory allocator.  Modify vnet to use the allocator
instead of monolithic global container structures (vinet, ...).  This
change solves many binary compatibility problems associated with
VIMAGE, and restores ELF symbols for virtualized global variables.

Each virtualized global variable exists as a "reference copy", and also
once per virtual network stack.  Virtualized global variables are
tagged at compile-time, placing the in a special linker set, which is
loaded into a contiguous region of kernel memory.  Virtualized global
variables in the base kernel are linked as normal, but those in modules
are copied and relocated to a reserved portion of the kernel's vnet
region with the help of a the kernel linker.

Virtualized global variables exist in per-vnet memory set up when the
network stack instance is created, and are initialized statically from
the reference copy.  Run-time access occurs via an accessor macro, which
converts from the current vnet and requested symbol to a per-vnet
address.  When "options VIMAGE" is not compiled into the kernel, normal
global ELF symbols will be used instead and indirection is avoided.

This change restores static initialization for network stack global
variables, restores support for non-global symbols and types, eliminates
the need for many subsystem constructors, eliminates large per-subsystem
structures that caused many binary compatibility issues both for
monitoring applications (netstat) and kernel modules, removes the
per-function INIT_VNET_*() macros throughout the stack, eliminates the
need for vnet_symmap ksym(2) munging, and eliminates duplicate
definitions of virtualized globals under VIMAGE_GLOBALS.

Bump __FreeBSD_version and update UPDATING.

Portions submitted by:  bz
Reviewed by:            bz, zec
Discussed with:         gnn, jamie, jeff, jhb, julian, sam
Suggested by:           peter
Approved by:            re (kensmith)

5 files changed, 185 insertions, 1 deletions

diff --git a/sys/ddb/db_expr.c b/sys/ddb/db_expr.c
index eaab874..424384c 100644
--- a/sys/ddb/db_expr.c
+++ b/sys/ddb/db_expr.c
@@ -51,7 +51,9 @@ db_term(db_expr_t *valuep)
 
 	t = db_read_token();
 	if (t == tIDENT) {
-	    if (!db_value_of_name(db_tok_string, valuep)) {
+	    if (!db_value_of_name(db_tok_string, valuep) &&
+		!db_value_of_name_pcpu(db_tok_string, valuep) &&
+		!db_value_of_name_vnet(db_tok_string, valuep)) {
 		db_error("Symbol not found\n");
 		/*NOTREACHED*/
 	    }
diff --git a/sys/ddb/db_sym.c b/sys/ddb/db_sym.c
index 9d043b3..1a95c11 100644
--- a/sys/ddb/db_sym.c
+++ b/sys/ddb/db_sym.c
@@ -32,10 +32,15 @@
 __FBSDID("$FreeBSD$");
 
 #include <sys/param.h>
+#include <sys/smp.h>
 #include <sys/systm.h>
+#include <sys/vimage.h>
+
+#include <net/vnet.h>
 
 #include <ddb/ddb.h>
 #include <ddb/db_sym.h>
+#include <ddb/db_variables.h>
 
 #include <opt_ddb.h>
 
@@ -56,6 +61,117 @@ static char		*db_qualify(c_db_sym_t sym, char *symtabname);
 static boolean_t	db_symbol_is_ambiguous(c_db_sym_t sym);
 static boolean_t	db_line_at_pc(c_db_sym_t, char **, int *, db_expr_t);
 
+static int db_cpu = -1;
+
+#ifdef VIMAGE
+extern uintptr_t	*__start_set_vnet;
+extern uintptr_t	*__stop_set_vnet;
+
+#define	VNET_START	(uintptr_t)&__start_set_vnet
+#define	VNET_STOP	(uintptr_t)&__stop_set_vnet
+
+static void *db_vnet = NULL;
+#endif
+
+/*
+ * Validate the CPU number used to interpret per-CPU variables so we can
+ * avoid later confusion if an invalid CPU is requested.
+ */
+int
+db_var_db_cpu(struct db_variable *vp, db_expr_t *valuep, int op)
+{
+
+	switch (op) {
+	case DB_VAR_GET:
+		*valuep = db_cpu;
+		return (1);
+
+	case DB_VAR_SET:
+		if (*(int *)valuep < -1 && *(int *)valuep > mp_maxid) {
+			db_printf("Invalid value: %d", *(int*)valuep);
+			return (0);
+		}
+		db_cpu = *(int *)valuep;
+		return (1);
+
+	default:
+		db_printf("db_var_db_cpu: unknown operation\n");
+		return (0);
+	}
+}
+
+/*
+ * Read-only variable reporting the current CPU, which is what we use when
+ * db_cpu is set to -1.
+ */
+int
+db_var_curcpu(struct db_variable *vp, db_expr_t *valuep, int op)
+{
+
+	switch (op) {
+	case DB_VAR_GET:
+		*valuep = curcpu;
+		return (1);
+
+	case DB_VAR_SET:
+		db_printf("Read-only variable.\n");
+		return (0);
+
+	default:
+		db_printf("db_var_curcpu: unknown operation\n");
+		return (0);
+	}
+}
+
+#ifdef VIMAGE
+/*
+ * Validate the virtual network pointer used to interpret per-vnet global
+ * variable expansion.  Right now we don't do much here, really we should
+ * walk the global vnet list to check it's an OK pointer.
+ */
+int
+db_var_db_vnet(struct db_variable *vp, db_expr_t *valuep, int op)
+{
+
+	switch (op) {
+	case DB_VAR_GET:
+		*valuep = (db_expr_t)db_vnet;
+		return (1);
+
+	case DB_VAR_SET:
+		db_vnet = *(void **)valuep;
+		return (1);
+
+	default:
+		db_printf("db_var_db_vnet: unknown operation\n");
+		return (0);
+	}
+}
+
+/*
+ * Read-only variable reporting the current vnet, which is what we use when
+ * db_vnet is set to NULL.
+ */
+int
+db_var_curvnet(struct db_variable *vp, db_expr_t *valuep, int op)
+{
+
+	switch (op) {
+	case DB_VAR_GET:
+		*valuep = (db_expr_t)curvnet;
+		return (1);
+
+	case DB_VAR_SET:
+		db_printf("Read-only variable.\n");
+		return (0);
+
+	default:
+		db_printf("db_var_curcpu: unknown operation\n");
+		return (0);
+	}
+}
+#endif
+
 /*
  * Add symbol table, with given name, to list of symbol tables.
  */
@@ -125,6 +241,59 @@ db_value_of_name(name, valuep)
 	return (TRUE);
 }
 
+boolean_t
+db_value_of_name_pcpu(name, valuep)
+	const char	*name;
+	db_expr_t	*valuep;
+{
+	static char     tmp[256];
+	db_expr_t	value;
+	c_db_sym_t	sym;
+	int		cpu;
+
+	if (db_cpu != -1)
+		cpu = db_cpu;
+	else
+		cpu = curcpu;
+	snprintf(tmp, sizeof(tmp), "pcpu_entry_%s", name);
+	sym = db_lookup(tmp);
+	if (sym == C_DB_SYM_NULL)
+		return (FALSE);
+	db_symbol_values(sym, &name, &value);
+	if (value < DPCPU_START || value >= DPCPU_STOP)
+		return (FALSE);
+	*valuep = (db_expr_t)((uintptr_t)value + dpcpu_off[cpu]);
+	return (TRUE);
+}
+
+boolean_t
+db_value_of_name_vnet(name, valuep)
+	const char	*name;
+	db_expr_t	*valuep;
+{
+#ifdef VIMAGE
+	static char     tmp[256];
+	db_expr_t	value;
+	c_db_sym_t	sym;
+	struct vnet	*vnet;
+
+	if (db_vnet != NULL)
+		vnet = db_vnet;
+	else
+		vnet = curvnet;
+	snprintf(tmp, sizeof(tmp), "vnet_entry_%s", name);
+	sym = db_lookup(tmp);
+	if (sym == C_DB_SYM_NULL)
+		return (FALSE);
+	db_symbol_values(sym, &name, &value);
+	if (value < VNET_START || value >= VNET_STOP)
+		return (FALSE);
+	*valuep = (db_expr_t)((uintptr_t)value + vnet->vnet_data_base);
+	return (TRUE);
+#else
+	return (FALSE);
+#endif
+}
 
 /*
  * Lookup a symbol.
diff --git a/sys/ddb/db_variables.c b/sys/ddb/db_variables.c
index 706ebe6..69c11ae 100644
--- a/sys/ddb/db_variables.c
+++ b/sys/ddb/db_variables.c
@@ -46,6 +46,12 @@ static struct db_variable db_vars[] = {
 	{ "maxwidth",	&db_max_width, FCN_NULL },
 	{ "tabstops",	&db_tab_stop_width, FCN_NULL },
 	{ "lines",	&db_lines_per_page, FCN_NULL },
+	{ "curcpu",	NULL, db_var_curcpu },
+	{ "db_cpu",	NULL, db_var_db_cpu },
+#ifdef VIMAGE
+	{ "curvnet",	NULL, db_var_curvnet },
+	{ "db_vnet",	NULL, db_var_db_vnet },
+#endif
 };
 static struct db_variable *db_evars =
 	db_vars + sizeof(db_vars)/sizeof(db_vars[0]);
diff --git a/sys/ddb/db_variables.h b/sys/ddb/db_variables.h
index 1ff0e44..42a3bf9 100644
--- a/sys/ddb/db_variables.h
+++ b/sys/ddb/db_variables.h
@@ -52,6 +52,11 @@ struct db_variable {
 extern struct db_variable	db_regs[];	/* machine registers */
 extern struct db_variable	*db_eregs;
 
+extern db_varfcn_t	db_var_curcpu;		/* DPCPU default CPU */
+extern db_varfcn_t	db_var_curvnet;		/* Default vnet */
+extern db_varfcn_t	db_var_db_cpu;		/* DPCPU active CPU */
+extern db_varfcn_t	db_var_db_vnet;		/* Active vnet */
+
 int db_read_variable(struct db_variable *, db_expr_t *);
 int db_write_variable(struct db_variable *, db_expr_t);
 
diff --git a/sys/ddb/ddb.h b/sys/ddb/ddb.h
index 1afdfa3..e6d8a95 100644
--- a/sys/ddb/ddb.h
+++ b/sys/ddb/ddb.h
@@ -213,6 +213,8 @@ boolean_t	db_stop_at_pc(boolean_t *is_breakpoint);
 void		db_trace_self(void);
 int		db_trace_thread(struct thread *, int);
 int		db_value_of_name(const char *name, db_expr_t *valuep);
+int		db_value_of_name_pcpu(const char *name, db_expr_t *valuep);
+int		db_value_of_name_vnet(const char *name, db_expr_t *valuep);
 int		db_write_bytes(vm_offset_t addr, size_t size, char *data);
 void		db_command_register(struct command_table *, struct command *);
 void		db_command_unregister(struct command_table *, struct command *);