Author: markj
Date: Tue Jun 3 01:06:31 2014
New Revision: 266987
URL: http://svnweb.freebsd.org/changeset/base/266987
Log:
4477 DTrace should speak JSON
illumos/illumos-gate@f497f9fe231e0e400f339c84a7d80c4aae2ac4d5
Modified:
vendor-sys/illumos/dist/uts/common/dtrace/dtrace.c
vendor-sys/illumos/dist/uts/common/sys/dtrace.h
Changes in other areas also in this revision:
Added:
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.general.d
(contents, props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.general.d.out
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.strsize.d
(contents, props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.strsize.d.out
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.usdt.c (contents,
props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.usdt.d (contents,
props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/tst.usdt.d.out
vendor/illumos/dist/cmd/dtrace/test/tst/common/json/usdt.d (contents, props
changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/strtoll/
vendor/illumos/dist/cmd/dtrace/test/tst/common/strtoll/err.BaseTooLarge.d
(contents, props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/strtoll/err.BaseTooSmall.d
(contents, props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/strtoll/tst.strtoll.d
(contents, props changed)
vendor/illumos/dist/cmd/dtrace/test/tst/common/strtoll/tst.strtoll.d.out
Modified:
vendor/illumos/dist/cmd/dtrace/test/tst/common/aggs/tst.subr.d
vendor/illumos/dist/cmd/dtrace/test/tst/common/privs/tst.func_access.ksh
vendor/illumos/dist/lib/libdtrace/common/dt_open.c
Modified: vendor-sys/illumos/dist/uts/common/dtrace/dtrace.c
==============================================================================
--- vendor-sys/illumos/dist/uts/common/dtrace/dtrace.c Tue Jun 3 00:55:15
2014 (r266986)
+++ vendor-sys/illumos/dist/uts/common/dtrace/dtrace.c Tue Jun 3 01:06:31
2014 (r266987)
@@ -92,6 +92,7 @@
#include <sys/zone.h>
#include <sys/socket.h>
#include <netinet/in.h>
+#include "strtolctype.h"
/*
* DTrace Tunable Variables
@@ -859,6 +860,58 @@ dtrace_vcanload(void *src, dtrace_diftyp
}
/*
+ * Convert a string to a signed integer using safe loads.
+ *
+ * NOTE: This function uses various macros from strtolctype.h to manipulate
+ * digit values, etc -- these have all been checked to ensure they make
+ * no additional function calls.
+ */
+static int64_t
+dtrace_strtoll(char *input, int base, size_t limit)
+{
+ uintptr_t pos = (uintptr_t)input;
+ int64_t val = 0;
+ int x;
+ boolean_t neg = B_FALSE;
+ char c, cc, ccc;
+ uintptr_t end = pos + limit;
+
+ /*
+ * Consume any whitespace preceding digits.
+ */
+ while ((c = dtrace_load8(pos)) == ' ' || c == '\t')
+ pos++;
+
+ /*
+ * Handle an explicit sign if one is present.
+ */
+ if (c == '-' || c == '+') {
+ if (c == '-')
+ neg = B_TRUE;
+ c = dtrace_load8(++pos);
+ }
+
+ /*
+ * Check for an explicit hexadecimal prefix ("0x" or "0X") and skip it
+ * if present.
+ */
+ if (base == 16 && c == '0' && ((cc = dtrace_load8(pos + 1)) == 'x' ||
+ cc == 'X') && isxdigit(ccc = dtrace_load8(pos + 2))) {
+ pos += 2;
+ c = ccc;
+ }
+
+ /*
+ * Read in contiguous digits until the first non-digit character.
+ */
+ for (; pos < end && c != '\0' && lisalnum(c) && (x = DIGIT(c)) < base;
+ c = dtrace_load8(++pos))
+ val = val * base + x;
+
+ return (neg ? -val : val);
+}
+
+/*
* Compare two strings using safe loads.
*/
static int
@@ -3335,6 +3388,463 @@ dtrace_dif_variable(dtrace_mstate_t *mst
}
}
+
+typedef enum dtrace_json_state {
+ DTRACE_JSON_REST = 1,
+ DTRACE_JSON_OBJECT,
+ DTRACE_JSON_STRING,
+ DTRACE_JSON_STRING_ESCAPE,
+ DTRACE_JSON_STRING_ESCAPE_UNICODE,
+ DTRACE_JSON_COLON,
+ DTRACE_JSON_COMMA,
+ DTRACE_JSON_VALUE,
+ DTRACE_JSON_IDENTIFIER,
+ DTRACE_JSON_NUMBER,
+ DTRACE_JSON_NUMBER_FRAC,
+ DTRACE_JSON_NUMBER_EXP,
+ DTRACE_JSON_COLLECT_OBJECT
+} dtrace_json_state_t;
+
+/*
+ * This function possesses just enough knowledge about JSON to extract a single
+ * value from a JSON string and store it in the scratch buffer. It is able
+ * to extract nested object values, and members of arrays by index.
+ *
+ * elemlist is a list of JSON keys, stored as packed NUL-terminated strings, to
+ * be looked up as we descend into the object tree. e.g.
+ *
+ * foo[0].bar.baz[32] --> "foo" NUL "0" NUL "bar" NUL "baz" NUL "32" NUL
+ * with nelems = 5.
+ *
+ * The run time of this function must be bounded above by strsize to limit the
+ * amount of work done in probe context. As such, it is implemented as a
+ * simple state machine, reading one character at a time using safe loads
+ * until we find the requested element, hit a parsing error or run off the
+ * end of the object or string.
+ *
+ * As there is no way for a subroutine to return an error without interrupting
+ * clause execution, we simply return NULL in the event of a missing key or any
+ * other error condition. Each NULL return in this function is commented with
+ * the error condition it represents -- parsing or otherwise.
+ *
+ * The set of states for the state machine closely matches the JSON
+ * specification (http://json.org/). Briefly:
+ *
+ * DTRACE_JSON_REST:
+ * Skip whitespace until we find either a top-level Object, moving
+ * to DTRACE_JSON_OBJECT; or an Array, moving to DTRACE_JSON_VALUE.
+ *
+ * DTRACE_JSON_OBJECT:
+ * Locate the next key String in an Object. Sets a flag to denote
+ * the next String as a key string and moves to DTRACE_JSON_STRING.
+ *
+ * DTRACE_JSON_COLON:
+ * Skip whitespace until we find the colon that separates key Strings
+ * from their values. Once found, move to DTRACE_JSON_VALUE.
+ *
+ * DTRACE_JSON_VALUE:
+ * Detects the type of the next value (String, Number, Identifier, Object
+ * or Array) and routes to the states that process that type. Here we also
+ * deal with the element selector list if we are requested to traverse down
+ * into the object tree.
+ *
+ * DTRACE_JSON_COMMA:
+ * Skip whitespace until we find the comma that separates key-value pairs
+ * in Objects (returning to DTRACE_JSON_OBJECT) or values in Arrays
+ * (similarly DTRACE_JSON_VALUE). All following literal value processing
+ * states return to this state at the end of their value, unless otherwise
+ * noted.
+ *
+ * DTRACE_JSON_NUMBER, DTRACE_JSON_NUMBER_FRAC, DTRACE_JSON_NUMBER_EXP:
+ * Processes a Number literal from the JSON, including any exponent
+ * component that may be present. Numbers are returned as strings, which
+ * may be passed to strtoll() if an integer is required.
+ *
+ * DTRACE_JSON_IDENTIFIER:
+ * Processes a "true", "false" or "null" literal in the JSON.
+ *
+ * DTRACE_JSON_STRING, DTRACE_JSON_STRING_ESCAPE,
+ * DTRACE_JSON_STRING_ESCAPE_UNICODE:
+ * Processes a String literal from the JSON, whether the String denotes
+ * a key, a value or part of a larger Object. Handles all escape sequences
+ * present in the specification, including four-digit unicode characters,
+ * but merely includes the escape sequence without converting it to the
+ * actual escaped character. If the String is flagged as a key, we
+ * move to DTRACE_JSON_COLON rather than DTRACE_JSON_COMMA.
+ *
+ * DTRACE_JSON_COLLECT_OBJECT:
+ * This state collects an entire Object (or Array), correctly handling
+ * embedded strings. If the full element selector list matches this nested
+ * object, we return the Object in full as a string. If not, we use this
+ * state to skip to the next value at this level and continue processing.
+ *
+ * NOTE: This function uses various macros from strtolctype.h to manipulate
+ * digit values, etc -- these have all been checked to ensure they make
+ * no additional function calls.
+ */
+static char *
+dtrace_json(uint64_t size, uintptr_t json, char *elemlist, int nelems,
+ char *dest)
+{
+ dtrace_json_state_t state = DTRACE_JSON_REST;
+ int64_t array_elem = INT64_MIN;
+ int64_t array_pos = 0;
+ uint8_t escape_unicount = 0;
+ boolean_t string_is_key = B_FALSE;
+ boolean_t collect_object = B_FALSE;
+ boolean_t found_key = B_FALSE;
+ boolean_t in_array = B_FALSE;
+ uint32_t braces = 0, brackets = 0;
+ char *elem = elemlist;
+ char *dd = dest;
+ uintptr_t cur;
+
+ for (cur = json; cur < json + size; cur++) {
+ char cc = dtrace_load8(cur);
+ if (cc == '\0')
+ return (NULL);
+
+ switch (state) {
+ case DTRACE_JSON_REST:
+ if (isspace(cc))
+ break;
+
+ if (cc == '{') {
+ state = DTRACE_JSON_OBJECT;
+ break;
+ }
+
+ if (cc == '[') {
+ in_array = B_TRUE;
+ array_pos = 0;
+ array_elem = dtrace_strtoll(elem, 10, size);
+ found_key = array_elem == 0 ? B_TRUE : B_FALSE;
+ state = DTRACE_JSON_VALUE;
+ break;
+ }
+
+ /*
+ * ERROR: expected to find a top-level object or array.
+ */
+ return (NULL);
+ case DTRACE_JSON_OBJECT:
+ if (isspace(cc))
+ break;
+
+ if (cc == '"') {
+ state = DTRACE_JSON_STRING;
+ string_is_key = B_TRUE;
+ break;
+ }
+
+ /*
+ * ERROR: either the object did not start with a key
+ * string, or we've run off the end of the object
+ * without finding the requested key.
+ */
+ return (NULL);
+ case DTRACE_JSON_STRING:
+ if (cc == '\\') {
+ *dd++ = '\\';
+ state = DTRACE_JSON_STRING_ESCAPE;
+ break;
+ }
+
+ if (cc == '"') {
+ if (collect_object) {
+ /*
+ * We don't reset the dest here, as
+ * the string is part of a larger
+ * object being collected.
+ */
+ *dd++ = cc;
+ collect_object = B_FALSE;
+ state = DTRACE_JSON_COLLECT_OBJECT;
+ break;
+ }
+ *dd = '\0';
+ dd = dest; /* reset string buffer */
+ if (string_is_key) {
+ if (dtrace_strncmp(dest, elem,
+ size) == 0)
+ found_key = B_TRUE;
+ } else if (found_key) {
+ if (nelems > 1) {
+ /*
+ * We expected an object, not
+ * this string.
+ */
+ return (NULL);
+ }
+ return (dest);
+ }
+ state = string_is_key ? DTRACE_JSON_COLON :
+ DTRACE_JSON_COMMA;
+ string_is_key = B_FALSE;
+ break;
+ }
+
+ *dd++ = cc;
+ break;
+ case DTRACE_JSON_STRING_ESCAPE:
+ *dd++ = cc;
+ if (cc == 'u') {
+ escape_unicount = 0;
+ state = DTRACE_JSON_STRING_ESCAPE_UNICODE;
+ } else {
+ state = DTRACE_JSON_STRING;
+ }
+ break;
+ case DTRACE_JSON_STRING_ESCAPE_UNICODE:
+ if (!isxdigit(cc)) {
+ /*
+ * ERROR: invalid unicode escape, expected
+ * four valid hexidecimal digits.
+ */
+ return (NULL);
+ }
+
+ *dd++ = cc;
+ if (++escape_unicount == 4)
+ state = DTRACE_JSON_STRING;
+ break;
+ case DTRACE_JSON_COLON:
+ if (isspace(cc))
+ break;
+
+ if (cc == ':') {
+ state = DTRACE_JSON_VALUE;
+ break;
+ }
+
+ /*
+ * ERROR: expected a colon.
+ */
+ return (NULL);
+ case DTRACE_JSON_COMMA:
+ if (isspace(cc))
+ break;
+
+ if (cc == ',') {
+ if (in_array) {
+ state = DTRACE_JSON_VALUE;
+ if (++array_pos == array_elem)
+ found_key = B_TRUE;
+ } else {
+ state = DTRACE_JSON_OBJECT;
+ }
+ break;
+ }
+
+ /*
+ * ERROR: either we hit an unexpected character, or
+ * we reached the end of the object or array without
+ * finding the requested key.
+ */
+ return (NULL);
+ case DTRACE_JSON_IDENTIFIER:
+ if (islower(cc)) {
+ *dd++ = cc;
+ break;
+ }
+
+ *dd = '\0';
+ dd = dest; /* reset string buffer */
+
+ if (dtrace_strncmp(dest, "true", 5) == 0 ||
+ dtrace_strncmp(dest, "false", 6) == 0 ||
+ dtrace_strncmp(dest, "null", 5) == 0) {
+ if (found_key) {
+ if (nelems > 1) {
+ /*
+ * ERROR: We expected an object,
+ * not this identifier.
+ */
+ return (NULL);
+ }
+ return (dest);
+ } else {
+ cur--;
+ state = DTRACE_JSON_COMMA;
+ break;
+ }
+ }
+
+ /*
+ * ERROR: we did not recognise the identifier as one
+ * of those in the JSON specification.
+ */
+ return (NULL);
+ case DTRACE_JSON_NUMBER:
+ if (cc == '.') {
+ *dd++ = cc;
+ state = DTRACE_JSON_NUMBER_FRAC;
+ break;
+ }
+
+ if (cc == 'x' || cc == 'X') {
+ /*
+ * ERROR: specification explicitly excludes
+ * hexidecimal or octal numbers.
+ */
+ return (NULL);
+ }
+
+ /* FALLTHRU */
+ case DTRACE_JSON_NUMBER_FRAC:
+ if (cc == 'e' || cc == 'E') {
+ *dd++ = cc;
+ state = DTRACE_JSON_NUMBER_EXP;
+ break;
+ }
+
+ if (cc == '+' || cc == '-') {
+ /*
+ * ERROR: expect sign as part of exponent only.
+ */
+ return (NULL);
+ }
+ /* FALLTHRU */
+ case DTRACE_JSON_NUMBER_EXP:
+ if (isdigit(cc) || cc == '+' || cc == '-') {
+ *dd++ = cc;
+ break;
+ }
+
+ *dd = '\0';
+ dd = dest; /* reset string buffer */
+ if (found_key) {
+ if (nelems > 1) {
+ /*
+ * ERROR: We expected an object, not
+ * this number.
+ */
+ return (NULL);
+ }
+ return (dest);
+ }
+
+ cur--;
+ state = DTRACE_JSON_COMMA;
+ break;
+ case DTRACE_JSON_VALUE:
+ if (isspace(cc))
+ break;
+
+ if (cc == '{' || cc == '[') {
+ if (nelems > 1 && found_key) {
+ in_array = cc == '[' ? B_TRUE : B_FALSE;
+ /*
+ * If our element selector directs us
+ * to descend into this nested object,
+ * then move to the next selector
+ * element in the list and restart the
+ * state machine.
+ */
+ while (*elem != '\0')
+ elem++;
+ elem++; /* skip the inter-element NUL */
+ nelems--;
+ dd = dest;
+ if (in_array) {
+ state = DTRACE_JSON_VALUE;
+ array_pos = 0;
+ array_elem = dtrace_strtoll(
+ elem, 10, size);
+ found_key = array_elem == 0 ?
+ B_TRUE : B_FALSE;
+ } else {
+ found_key = B_FALSE;
+ state = DTRACE_JSON_OBJECT;
+ }
+ break;
+ }
+
+ /*
+ * Otherwise, we wish to either skip this
+ * nested object or return it in full.
+ */
+ if (cc == '[')
+ brackets = 1;
+ else
+ braces = 1;
+ *dd++ = cc;
+ state = DTRACE_JSON_COLLECT_OBJECT;
+ break;
+ }
+
+ if (cc == '"') {
+ state = DTRACE_JSON_STRING;
+ break;
+ }
+
+ if (islower(cc)) {
+ /*
+ * Here we deal with true, false and null.
+ */
+ *dd++ = cc;
+ state = DTRACE_JSON_IDENTIFIER;
+ break;
+ }
+
+ if (cc == '-' || isdigit(cc)) {
+ *dd++ = cc;
+ state = DTRACE_JSON_NUMBER;
+ break;
+ }
+
+ /*
+ * ERROR: unexpected character at start of value.
+ */
+ return (NULL);
+ case DTRACE_JSON_COLLECT_OBJECT:
+ if (cc == '\0')
+ /*
+ * ERROR: unexpected end of input.
+ */
+ return (NULL);
+
+ *dd++ = cc;
+ if (cc == '"') {
+ collect_object = B_TRUE;
+ state = DTRACE_JSON_STRING;
+ break;
+ }
+
+ if (cc == ']') {
+ if (brackets-- == 0) {
+ /*
+ * ERROR: unbalanced brackets.
+ */
+ return (NULL);
+ }
+ } else if (cc == '}') {
+ if (braces-- == 0) {
+ /*
+ * ERROR: unbalanced braces.
+ */
+ return (NULL);
+ }
+ } else if (cc == '{') {
+ braces++;
+ } else if (cc == '[') {
+ brackets++;
+ }
+
+ if (brackets == 0 && braces == 0) {
+ if (found_key) {
+ *dd = '\0';
+ return (dest);
+ }
+ dd = dest; /* reset string buffer */
+ state = DTRACE_JSON_COMMA;
+ }
+ break;
+ }
+ }
+ return (NULL);
+}
+
/*
* Emulate the execution of DTrace ID subroutines invoked by the call opcode.
* Notice that we don't bother validating the proper number of arguments or
@@ -4035,6 +4545,65 @@ dtrace_dif_subr(uint_t subr, uint_t rd,
break;
}
+ case DIF_SUBR_JSON: {
+ uint64_t size = state->dts_options[DTRACEOPT_STRSIZE];
+ uintptr_t json = tupregs[0].dttk_value;
+ size_t jsonlen = dtrace_strlen((char *)json, size);
+ uintptr_t elem = tupregs[1].dttk_value;
+ size_t elemlen = dtrace_strlen((char *)elem, size);
+
+ char *dest = (char *)mstate->dtms_scratch_ptr;
+ char *elemlist = (char *)mstate->dtms_scratch_ptr + jsonlen + 1;
+ char *ee = elemlist;
+ int nelems = 1;
+ uintptr_t cur;
+
+ if (!dtrace_canload(json, jsonlen + 1, mstate, vstate) ||
+ !dtrace_canload(elem, elemlen + 1, mstate, vstate)) {
+ regs[rd] = NULL;
+ break;
+ }
+
+ if (!DTRACE_INSCRATCH(mstate, jsonlen + 1 + elemlen + 1)) {
+ DTRACE_CPUFLAG_SET(CPU_DTRACE_NOSCRATCH);
+ regs[rd] = NULL;
+ break;
+ }
+
+ /*
+ * Read the element selector and split it up into a packed list
+ * of strings.
+ */
+ for (cur = elem; cur < elem + elemlen; cur++) {
+ char cc = dtrace_load8(cur);
+
+ if (cur == elem && cc == '[') {
+ /*
+ * If the first element selector key is
+ * actually an array index then ignore the
+ * bracket.
+ */
+ continue;
+ }
+
+ if (cc == ']')
+ continue;
+
+ if (cc == '.' || cc == '[') {
+ nelems++;
+ cc = '\0';
+ }
+
+ *ee++ = cc;
+ }
+ *ee++ = '\0';
+
+ if ((regs[rd] = (uintptr_t)dtrace_json(size, json, elemlist,
+ nelems, dest)) != NULL)
+ mstate->dtms_scratch_ptr += jsonlen + 1;
+ break;
+ }
+
case DIF_SUBR_TOUPPER:
case DIF_SUBR_TOLOWER: {
uintptr_t s = tupregs[0].dttk_value;
@@ -4342,6 +4911,28 @@ case DIF_SUBR_GETMAJOR:
break;
}
+ case DIF_SUBR_STRTOLL: {
+ uintptr_t s = tupregs[0].dttk_value;
+ uint64_t size = state->dts_options[DTRACEOPT_STRSIZE];
+ int base = 10;
+
+ if (nargs > 1) {
+ if ((base = tupregs[1].dttk_value) <= 1 ||
+ base > ('z' - 'a' + 1) + ('9' - '0' + 1)) {
+ *flags |= CPU_DTRACE_ILLOP;
+ break;
+ }
+ }
+
+ if (!dtrace_strcanload(s, size, mstate, vstate)) {
+ regs[rd] = INT64_MIN;
+ break;
+ }
+
+ regs[rd] = dtrace_strtoll((char *)s, base, size);
+ break;
+ }
+
case DIF_SUBR_LLTOSTR: {
int64_t i = (int64_t)tupregs[0].dttk_value;
uint64_t val, digit;
@@ -8861,7 +9452,9 @@ dtrace_difo_validate_helper(dtrace_difo_
subr == DIF_SUBR_INET_NTOA ||
subr == DIF_SUBR_INET_NTOA6 ||
subr == DIF_SUBR_INET_NTOP ||
+ subr == DIF_SUBR_JSON ||
subr == DIF_SUBR_LLTOSTR ||
+ subr == DIF_SUBR_STRTOLL ||
subr == DIF_SUBR_RINDEX ||
subr == DIF_SUBR_STRCHR ||
subr == DIF_SUBR_STRJOIN ||
Modified: vendor-sys/illumos/dist/uts/common/sys/dtrace.h
==============================================================================
--- vendor-sys/illumos/dist/uts/common/sys/dtrace.h Tue Jun 3 00:55:15
2014 (r266986)
+++ vendor-sys/illumos/dist/uts/common/sys/dtrace.h Tue Jun 3 01:06:31
2014 (r266987)
@@ -289,8 +289,10 @@ typedef enum dtrace_probespec {
#define DIF_SUBR_TOUPPER 44
#define DIF_SUBR_TOLOWER 45
#define DIF_SUBR_GETF 46
+#define DIF_SUBR_JSON 47
+#define DIF_SUBR_STRTOLL 48
-#define DIF_SUBR_MAX 46 /* max subroutine value
*/
+#define DIF_SUBR_MAX 48 /* max subroutine value
*/
typedef uint32_t dif_instr_t;
_______________________________________________
[email protected] mailing list
http://lists.freebsd.org/mailman/listinfo/svn-src-all
To unsubscribe, send any mail to "[email protected]"
