[openmp] Allow all possible argument separators in TSAN_OPTIONS

[upsj]

Summary

Currently, the parser used to tokenize the TSAN_OPTIONS in libomp uses
only spaces as separators, even though TSAN in compiler-rt supports
other separators like ':' or ','. This causes TSAN used with OpenMP to
spuriously display the following warning:

Warning: please export TSAN_OPTIONS='ignore_noninstrumented_modules=1'
to avoid false positive reports from the OpenMP runtime!

when TSAN_OPTIONS contains multiple arguments and
ignore_noninstrumented_modules follows a non-space separator.

CTest uses ':' to separate sanitizer options by default, which is
where I encountered this issue first in practice.
The documentation for other sanitizers mentions ':' as separator,
but TSAN only lists spaces, which is probably where this mismatch originated.

---

I wasn't sure where to best add tests for this bugfix, so I wanted to leave it at this stage until I can get some pointers in the right direction :)

[protze.joachim]

Thanks for submitting this patch!

The archer tool has it's own test sub-directory. I think, it's best to add the test there.

[protze.joachim]

Can you add a test?

[upsj]

Sure, I just had to figure out how the LLVM test system works :) (and fix an off-by-one error in the tokenization)
I added a new test with disabled suppression and modified the default suppression to see that the tokenization works.

[upsj]

Shorten test checks for clang-format

[protze.joachim]

LGTM

Can you push the patch or should I?

[upsj]

I don't have commit access, so you will have to do it.

openmp/tools/archer/ompt-tsan.cpp

 #define __STDC_FORMAT_MACROS
 #endif
 
+#include <algorithm>
 #include <atomic>
 #include <cassert>
 #include <cstdlib>

   TsanFlags(const char *env) : ignore_noninstrumented_modules(0) {
     if (env) {
       std::vector<std::string> tokens;
-      std::string token;
       std::string str(env);
-      std::istringstream iss(str);
-      while (std::getline(iss, token, ' '))
-        tokens.push_back(token);
+      auto end = str.end();
+      auto it = str.begin();
+      auto is_sep = [](char c) {
+        return c == ' ' || c == ',' || c == ':' || c == '\n' || c == '\t' ||
+               c == '\r';
+      };
+      while (it != end) {
+        auto next_it = std::find_if(it, end, is_sep);
+        tokens.emplace_back(it, next_it);
+        it = next_it + 1;
+      }
 
-      for (std::vector<std::string>::iterator it = tokens.begin();
-           it != tokens.end(); ++it) {
+      for (const auto &token : tokens) {
         // we are interested in ignore_noninstrumented_modules to print a
         // warning
-        if (sscanf(it->c_str(), "ignore_noninstrumented_modules=%d",
+        if (sscanf(token.c_str(), "ignore_noninstrumented_modules=%d",
                    &ignore_noninstrumented_modules))
           continue;
       }