SystemTap: new parser

Signed-off-by: Masatake YAMATO <yamato@redhat.com>

Units/parser-systemtap.r/functions.d/args.ctags

@@ -0,0 +1,2 @@
+--sort=no
+--fields=+ne

Units/parser-systemtap.r/functions.d/expected.tags

@@ -0,0 +1,4 @@
+matched_str	input.stp	/^function matched_str:string() { return matched(0) }$/;"	f	line:24	end:24
+matched	input.stp	/^function matched:string(n:long)$/;"	f	line:39	end:82
+ngroups	input.stp	/^function ngroups:long()$/;"	f	line:96	end:105
+__asmlinkage_int_arg	input.stp	/^function __asmlinkage_int_arg:long(n:long)$/;"	f	line:116	end:120

Units/parser-systemtap.r/functions.d/input.stp

@@ -0,0 +1,120 @@
+#
+# Taken from /usr/share/systemtap/tapset/regex.stp and registers.stp
+#
+// Regular expression subexpression tapset
+// Copyright (C) 2017 Serhei Makarov
+// Copyright (C) 2013 Red Hat, Inc.
+//
+// This file is part of systemtap, and is free software.  You can
+// redistribute it and/or modify it under the terms of the GNU General
+// Public License (GPL); either version 2, or (at your option) any
+// later version.
+
+%{
+#define STAP_NEED_CONTEXT_SUBEXPRESSION 1
+%}
+
+/**
+ * sfunction matched_str - Return the last matched string.
+ *
+ * Description: returns the string matched by the last successful
+ * use of the =~ regexp matching operator. Returns an error if the
+ * last use of =~ led to a failed match.
+ */
+function matched_str:string() { return matched(0) }
+
+/**
+ * sfunction matched - Return a given matched subexpression.
+ *
+ * @n: index to the subexpression to return. 0 corresponds to the
+ * entire regular expression.
+ *
+ * Description: returns the content of the n'th subexpression of the
+ * last successful use of the =~ regex matching operator. Returns an
+ * empty string if the n'th subexpression was not matched (e.g. due to
+ * alternation). Throws an error if the last use of =~ was a failed
+ * match, or if fewer than n subexpressions are present in the
+ * original regexp.
+ */
+function matched:string(n:long)
+%{ /* pure */ /* unprivileged */ /* pragma:tagged_dfa */
+  int start_ix, end_ix; // indices into tag buffer
+  int start, end, length; // actual coordinate values
+
+  if (!CONTEXT->last_match.result) {
+    snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
+             "Attempted to get subexpression %lld from failed match", (long long) STAP_ARG_n);
+    CONTEXT->last_error = CONTEXT->error_buffer;
+  }
+
+  start_ix = 2 * STAP_ARG_n, end_ix = start_ix + 1;
+
+  if (end_ix >= CONTEXT->last_match.num_final_tags) {
+    snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
+             "Attempted to get nonexistent subexpression %lld", (long long) STAP_ARG_n);
+    CONTEXT->last_error = CONTEXT->error_buffer;
+  }
+
+  start = CONTEXT->last_match.tag_vals[start_ix];
+  end = CONTEXT->last_match.tag_vals[end_ix];
+  // _stp_printf ("**DEBUG** Extracted subexpression #%lld:(%d,%d) from %d to %d\n", STAP_ARG_n, start_ix, end_ix, start, end);
+
+  if (start < 0 || end < 0) {
+    // If indices are negative, the group was not matched. Return empty string:
+    start = end = 0;
+    // snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
+    //          "Unknown coordinates for subexpression %lld", STAP_ARG_n);
+    // CONTEXT->last_error = CONTEXT->error_buffer;
+  }
+
+  if (start > end) {
+    // This should not happen.
+    snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
+             "BUG: inverted coordinates for subexpression %lld", (long long) STAP_ARG_n);
+    CONTEXT->last_error = CONTEXT->error_buffer;
+  }
+
+  length = end - start;
+
+  // TODOXXX assert (start <= strlen(matched_str)) ??
+  // XXX: Must add 1 to length to account for NUL byte in strlcpy().
+  strlcpy(STAP_RETVALUE, CONTEXT->last_match.matched_str + start, length + 1);
+%}
+
+/**
+ * sfunction ngroups - Number of subexpressions in the last match.
+ *
+ * Description: returns the number of subexpressions from the
+ * last successful use of the =~ regex matching operator.
+ *
+ * Note that this number includes subexpressions which are present in
+ * the regex but did not match any string; for example, given the
+ * regex "a|(b)", the subexpressions will count the group for (b)
+ * regardless of whether it matched a string or not. Throws an error
+ * if the last use of =~ was a failed match.
+ */
+function ngroups:long()
+%{ /* pure */ /* unprivileged */ /* pragma:tagged_dfa */
+  if (!CONTEXT->last_match.result) {
+    snprintf(CONTEXT->error_buffer, sizeof(CONTEXT->error_buffer),
+             "Attempted to get subexpression count from failed match");
+    CONTEXT->last_error = CONTEXT->error_buffer;
+  }
+
+  STAP_RETVALUE = CONTEXT->last_match.num_final_tags / 2;
+%}
+
+// XXX: perhaps implement matched_start, matched_end to get indices?
+// XXX: some kind of find-replace functionality?
+// XXX: some kind of splitting / multiple-match functionality?
+
+/*
+ * In nd_syscall.return probes, we sometimes need @entry() values. To
+ * ensure we get the argument in the correct mode, we need a function
+ * that calls asmlinkage() first.
+ */
+function __asmlinkage_int_arg:long(n:long)
+{
+	asmlinkage()
+	return int_arg(n)
+}

Units/parser-systemtap.r/macros.d/args.ctags

@@ -0,0 +1,2 @@
+--sort=no
+--fields=+ne

Units/parser-systemtap.r/macros.d/expected.tags

@@ -0,0 +1,3 @@
+m0	input.stpm	/^@define m0$/;"	m	line:4	end:7
+m1	input.stpm	/^@define m1(a0) %($/;"	m	line:12	end:14
+m2	input.stpm	/^@define m2(a1,$/;"	m	line:19	end:22

Units/parser-systemtap.r/macros.d/input.stpm

@@ -0,0 +1,22 @@
+#
+# sharp Comment
+#
+@define m0
+%(
+  0
+%)
+
+/* 
+ * C comment 
+ */
+@define m1(a0) %(
+    @a0
+%)
+
+//
+// C++  comment
+//
+@define m2(a1,
+	   a2) %(
+ %( @a1 + @a2 %)
+%)

Units/parser-systemtap.r/probes.d/args.ctags

@@ -0,0 +1,2 @@
+--sort=no
+--fields=+ne

Units/parser-systemtap.r/probes.d/expected.tags

@@ -0,0 +1,38 @@
+tcp_get_info_rto	input.stp	/^function tcp_get_info_rto:long(sock:long)$/;"	f	line:33	end:44
+tcp_get_info_snd_cwnd	input.stp	/^function tcp_get_info_snd_cwnd:long(sock:long)$/;"	f	line:50	end:60
+tcp_ts_get_info_state	input.stp	/^function tcp_ts_get_info_state:long(sock:long)$/;"	f	line:78	end:83
+__tcp_sock_dport	input.stp	/^function __tcp_sock_dport:long (sock:long)$/;"	f	line:86	end:93
+__get_skb_tcphdr_new	input.stp	/^@__private30 function __get_skb_tcphdr_new:long(skb:long)$/;"	f	line:96	end:107
+__get_skb_tcphdr	input.stp	/^function __get_skb_tcphdr:long(skb:long)$/;"	f	line:110	end:118
+__tcp_skb_urg	input.stp	/^function __tcp_skb_urg:long (tcphdr:long)$/;"	f	line:121	end:124
+__tcp_skb_ack	input.stp	/^function __tcp_skb_ack:long (tcphdr:long)$/;"	f	line:127	end:130
+__tcp_skb_psh	input.stp	/^function __tcp_skb_psh:long (tcphdr:long)$/;"	f	line:133	end:136
+__tcp_skb_rst	input.stp	/^function __tcp_skb_rst:long (tcphdr:long)$/;"	f	line:139	end:142
+__tcp_skb_syn	input.stp	/^function __tcp_skb_syn:long (tcphdr:long)$/;"	f	line:145	end:148
+__tcp_skb_fin	input.stp	/^function __tcp_skb_fin:long (tcphdr:long)$/;"	f	line:151	end:154
+__tcp_skb_sport	input.stp	/^function __tcp_skb_sport:long (tcphdr:long)$/;"	f	line:157	end:160
+__tcp_skb_dport	input.stp	/^function __tcp_skb_dport:long (tcphdr:long){$/;"	f	line:163	end:165
+__tcp_sock_sport	input.stp	/^function __tcp_sock_sport:long (sock:long)$/;"	f	line:168	end:174
+sockstate	input.stp	/^@__private30 global sockstate[12]$/;"	v	line:176
+tcp_sockstate_str	input.stp	/^function tcp_sockstate_str:string (state:long) {$/;"	f	line:192	end:194
+tcp_ts_get_info_snd_ssthresh	input.stp	/^function tcp_ts_get_info_snd_ssthresh:long(sock:long)$/;"	f	line:199	end:209
+tcp_ts_get_info_rcv_mss	input.stp	/^function tcp_ts_get_info_rcv_mss:long(sock:long)$/;"	f	line:213	end:224
+__sockopt	input.stp	/^@__private30 global __sockopt[18]$/;"	v	line:248
+tcp_sockopt_str	input.stp	/^function tcp_sockopt_str:string (optname:long)$/;"	f	line:281	end:285
+__ipv6_sockopt	input.stp	/^@__private30 global __ipv6_sockopt[55]$/;"	v	line:349
+tcp_ipv6_sockopt_str	input.stp	/^function tcp_ipv6_sockopt_str:string (optname:long)$/;"	f	line:467	end:471
+tcp.sendmsg	input.stp	/^probe tcp.sendmsg = kernel.function("tcp_sendmsg") {$/;"	p	line:483	end:488
+tcp.sendmsg.return	input.stp	/^probe tcp.sendmsg.return = kernel.function("tcp_sendmsg").return {$/;"	p	line:498	end:501
+tcp.recvmsg	input.stp	/^probe tcp.recvmsg = kernel.function("tcp_recvmsg") {$/;"	p	line:516	end:525
+tcp.recvmsg.return	input.stp	/^probe tcp.recvmsg.return = kernel.function("tcp_recvmsg").return {$/;"	p	line:540	end:548
+tcp.disconnect	input.stp	/^probe tcp.disconnect = kernel.function("tcp_disconnect") {$/;"	p	line:564	end:573
+tcp.disconnect.return	input.stp	/^probe tcp.disconnect.return = kernel.function("tcp_disconnect").return {$/;"	p	line:583	end:586
+tcp.setsockopt	input.stp	/^probe tcp.setsockopt = tcp.ipv4.setsockopt, tcp.ipv6.setsockopt$/;"	p	line:601	end:603
+tcp.ipv4.setsockopt	input.stp	/^probe tcp.ipv4.setsockopt = kernel.function("tcp_setsockopt")$/;"	p	line:604	end:613
+tcp.ipv6.setsockopt	input.stp	/^probe tcp.ipv6.setsockopt = kernel.function("ipv6_setsockopt")!,$/;"	p	line:614	end:624
+tcp.setsockopt.return	input.stp	/^probe tcp.setsockopt.return = tcp.ipv4.setsockopt.return,$/;"	p	line:634	end:637
+tcp.ipv4.setsockopt.return	input.stp	/^probe tcp.ipv4.setsockopt.return = kernel.function("tcp_setsockopt").return$/;"	p	line:638	end:642
+tcp.ipv6.setsockopt.return	input.stp	/^probe tcp.ipv6.setsockopt.return = kernel.function("ipv6_setsockopt").return!,$/;"	p	line:643	end:648
+tcp.receive	input.stp	/^probe tcp.receive = tcp.ipv4.receive, tcp.ipv6.receive$/;"	p	line:667	end:669
+tcp.ipv4.receive	input.stp	/^probe tcp.ipv4.receive = kernel.function("tcp_v4_rcv")$/;"	p	line:671	end:694
+tcp.ipv6.receive	input.stp	/^probe tcp.ipv6.receive = kernel.function("tcp_v6_rcv")!,$/;"	p	line:696	end:730