Add basic USDT support for s390x (iovisor#2266)

Approach and code shamelessly borrowed from "422db709: Add basic USDT
support for powerpc64".

src/cc/usdt.h

@@ -81,6 +81,7 @@ class Argument {
  friend class ArgumentParser;
  friend class ArgumentParser_aarch64;
  friend class ArgumentParser_powerpc64;
+ friend class ArgumentParser_s390x;
  friend class ArgumentParser_x64;
 };
 
@@ -130,6 +131,12 @@ class ArgumentParser_powerpc64 : public ArgumentParser {
  ArgumentParser_powerpc64(const char *arg) : ArgumentParser(arg) {}
 };
 
+class ArgumentParser_s390x : public ArgumentParser {
+public:
+ bool parse(Argument *dest);
+ ArgumentParser_s390x(const char *arg) : ArgumentParser(arg) {}
+};
+
 class ArgumentParser_x64 : public ArgumentParser {
 private:
  enum Register {

src/cc/usdt/usdt.cc

@@ -40,6 +40,8 @@ Location::Location(uint64_t addr, const std::string &bin_path, const char *arg_f
  ArgumentParser_aarch64 parser(arg_fmt);
 #elif __powerpc64__
  ArgumentParser_powerpc64 parser(arg_fmt);
+#elif __s390x__
+ ArgumentParser_s390x parser(arg_fmt);
 #else
  ArgumentParser_x64 parser(arg_fmt);
 #endif

src/cc/usdt/usdt_args.cc

@@ -285,6 +285,59 @@ bool ArgumentParser_powerpc64::parse(Argument *dest) {
  return true;
 }
 
+bool ArgumentParser_s390x::parse(Argument *dest) {
+ if (done())
+ return false;
+
+ bool matched;
+ std::cmatch matches;
+#define S390X_IMM "(-?[0-9]+)"
+ std::regex arg_n_regex("^" S390X_IMM "@");
+ // <imm>
+ std::regex arg_op_regex_imm("^" S390X_IMM "(?: +|$)");
+ // %r<N>
+#define S390X_REG "%r([0-9]|1[0-5])"
+ std::regex arg_op_regex_reg("^" S390X_REG "(?: +|$)");
+ // <disp>(%r<N>,%r<N>)
+ std::regex arg_op_regex_mem("^" S390X_IMM "?\\(" S390X_REG
+ "(?:," S390X_REG ")?\\)(?: +|$)");
+#undef S390X_IMM
+#undef S390X_REG
+
+ matched = std::regex_search(arg_ + cur_pos_, matches, arg_n_regex);
+ if (matched) {
+ dest->arg_size_ = stoi(matches.str(1));
+ cur_pos_ += matches.length(0);
+
+ if (std::regex_search(arg_ + cur_pos_, matches, arg_op_regex_imm)) {
+ dest->constant_ = stoi(matches.str(1));
+ } else if (std::regex_search(arg_ + cur_pos_, matches, arg_op_regex_reg)) {
+ dest->base_register_name_ = "gprs[" + matches.str(1) + "]";
+ } else if (std::regex_search(arg_ + cur_pos_, matches, arg_op_regex_mem)) {
+ if (matches.length(1) > 0) {
+ dest->deref_offset_ = stoi(matches.str(1));
+ }
+ dest->base_register_name_ = "gprs[" + matches.str(2) + "]";
+ if (matches.length(3) > 0) {
+ dest->index_register_name_ = "gprs[" + matches.str(3) + "]";
+ }
+ } else {
+ matched = false;
+ }
+ }
+
+ if (!matched) {
+ print_error(cur_pos_);
+ skip_until_whitespace_from(cur_pos_);
+ skip_whitespace_from(cur_pos_);
+ return false;
+ }
+
+ cur_pos_ += matches.length(0);
+ skip_whitespace_from(cur_pos_);
+ return true;
+}
+
 ssize_t ArgumentParser_x64::parse_identifier(ssize_t pos,
  optional<std::string> *result) {
  if (isalpha(arg_[pos]) || arg_[pos] == '_') {

tests/cc/test_usdt_args.cc

@@ -58,6 +58,8 @@ TEST_CASE("test usdt argument parsing", "[usdt]") {
  USDT::ArgumentParser_aarch64 parser("4@[x32,200]");
 #elif __powerpc64__
  USDT::ArgumentParser_powerpc64 parser("4@-12(42)");
+#elif __s390x__
+ USDT::ArgumentParser_s390x parser("4@-12(%r42)");
 #elif defined(__x86_64__)
  USDT::ArgumentParser_x64 parser("4@i%ra+1r");
 #endif
@@ -121,6 +123,50 @@ TEST_CASE("test usdt argument parsing", "[usdt]") {
  verify_register(parser, 2, 1097);
  verify_register(parser, 4, "gpr[30]", 108);
  verify_register(parser, -2, "gpr[31]", -4);
+#elif __s390x__
+ USDT::ArgumentParser_s390x parser(
+ "-4@%r0 8@%r0 8@0 4@0(%r0) -2@0(%r0) "
+ "1@%r0 -2@%r3 -8@9 -1@0(%r4) -4@16(%r6) "
+ "2@%r7 4@%r11 4@-67 8@-16(%r15) 1@-52(%r11) "
+ "-8@%r4 -8@%r14 2@-11 -2@14(%r13) -8@-32(%r12) "
+ "4@%r5 2@%r11 -8@-693 -1@-23(%r10) 4@28(%r9) "
+ "-2@%r3 -4@%r8 2@1097 4@108(%r7) -2@-4(%r6)");
+
+ verify_register(parser, -4, "gprs[0]");
+ verify_register(parser, 8, "gprs[0]");
+ verify_register(parser, 8, 0);
+ verify_register(parser, 4, "gprs[0]", 0);
+ verify_register(parser, -2, "gprs[0]", 0);
+
+ verify_register(parser, 1, "gprs[0]");
+ verify_register(parser, -2, "gprs[3]");
+ verify_register(parser, -8, 9);
+ verify_register(parser, -1, "gprs[4]", 0);
+ verify_register(parser, -4, "gprs[6]", 16);
+
+ verify_register(parser, 2, "gprs[7]");
+ verify_register(parser, 4, "gprs[11]");
+ verify_register(parser, 4, -67);
+ verify_register(parser, 8, "gprs[15]", -16);
+ verify_register(parser, 1, "gprs[11]", -52);
+
+ verify_register(parser, -8, "gprs[4]");
+ verify_register(parser, -8, "gprs[14]");
+ verify_register(parser, 2, -11);
+ verify_register(parser, -2, "gprs[13]", 14);
+ verify_register(parser, -8, "gprs[12]", -32);
+
+ verify_register(parser, 4, "gprs[5]");
+ verify_register(parser, 2, "gprs[11]");
+ verify_register(parser, -8, -693);
+ verify_register(parser, -1, "gprs[10]", -23);
+ verify_register(parser, 4, "gprs[9]", 28);
+
+ verify_register(parser, -2, "gprs[3]");
+ verify_register(parser, -4, "gprs[8]");
+ verify_register(parser, 2, 1097);
+ verify_register(parser, 4, "gprs[7]", 108);
+ verify_register(parser, -2, "gprs[6]", -4);
 #elif defined(__x86_64__)
  USDT::ArgumentParser_x64 parser(
  "-4@$0 8@$1234 %rdi %rax %rsi "