Bringing comments up to date and update SortedIndexes

Summary: Removing unnecessary links and updating some comments.

Test Plan: Existing tests

Reviewers: zasgar, #engineering

Reviewed By: zasgar, #engineering

Differential Revision: https://phab.corp.pixielabs.ai/D6881

GitOrigin-RevId: 39d710a6ae0f27a67d743cdb5eb60b2d5c2a27cf

src/stirling/bcc_bpf/protocol_inference.h

@@ -271,13 +271,16 @@ static __inline enum MessageType infer_http2_message(const char* buf, size_t cou
 }
 
 static __inline enum MessageType infer_dns_message(const char* buf, size_t count) {
-  // https://stackoverflow.com/questions/6794926/how-many-a-records-can-fit-in-a-single-dns-response
-
   const int kDNSHeaderSize = 12;
 
   // Use the maximum *guaranteed* UDP packet size as the max DNS message size.
   // UDP packets can be larger, but this is the typical maximum size for DNS.
   const int kMaxDNSMessageSize = 512;
+
+  // Maximum number of resource records.
+  // https://stackoverflow.com/questions/6794926/how-many-a-records-can-fit-in-a-single-dns-response
+  const int kMaxNumRR = 25;
+
   if (count < kDNSHeaderSize || count > kMaxDNSMessageSize) {
     return kUnknown;
   }
@@ -306,9 +309,8 @@ static __inline enum MessageType infer_dns_message(const char* buf, size_t count
     return kUnknown;
   }
 
-  // https://stackoverflow.com/questions/6794926/how-many-a-records-can-fit-in-a-single-dns-response
   uint32_t num_rr = num_questions + num_answers + num_auth + num_addl;
-  if (num_rr > 25) {
+  if (num_rr > kMaxNumRR) {
     return kUnknown;
   }
 

src/stirling/data_table.cc

@@ -46,14 +46,16 @@ namespace {
 // Note 2: There are different ways to define the reorder indexes.
 // Here we use the form where the result, idx, is used to sort x according to:
 //    { x[idx[0]], x[idx[1]], x[idx[2]], ... }
-// From https://stackoverflow.com/questions/1577475/c-sorting-and-keeping-track-of-indexes
-std::vector<size_t> SortedIndexes(const std::vector<uint64_t> v) {
-  // Initialize original index locations.
+std::vector<size_t> SortedIndexes(const std::vector<uint64_t>& v) {
+  // Create indices corresponding to v.
   std::vector<size_t> idx(v.size());
-  std::iota(idx.begin(), idx.end(), 0);
+  // Initialize idx = {0, 1, 2, 3, ... }
+  for (size_t i = 0; i < idx.size(); ++i) {
+    idx[i] = i;
+  }
 
-  // Sort indexes based on comparing values in v using std::stable_sort instead of std::sort
-  // to avoid unnecessary index re-orderings when v contains elements of equal values.
+  // Find the sorted indices by running a sort on idx, but using the values of v.
+  // Use std::stable_sort instead of std::sort to minimize churn in indices.
   std::stable_sort(idx.begin(), idx.end(), [&v](size_t i1, size_t i2) { return v[i1] < v[i2]; });
 
   return idx;

src/stirling/dynamic_tracing/code_gen.cc

@@ -283,7 +283,10 @@ StatusOr<std::vector<std::string>> GenPtrLenVariable(const PtrLenVariable& var)
 
   // Make sure we don't overrun the buffer by capping the length (also required for verifier).
   // Below, we want the size of blobXX->buf. We can do that with this trick:
-  // https://stackoverflow.com/questions/3553296/sizeof-single-struct-member-in-c
+  //   sizeof(((struct blobXX)0)->buf)
+  // This trick won't cause a null dereference because it's just for the compiler
+  // to find the size of the member; there is no real data access at run-time.
+  // See https://stackoverflow.com/questions/3553296/sizeof-single-struct-member-in-c
   code_lines.push_back(absl::Substitute(
       "uint8_t $0_truncate__ = $0 > sizeof(((struct blob$1*)0)->buf);", var.len_var_name(), size));
   code_lines.push_back(absl::Substitute(

src/stirling/k8s/run_on_k8s.sh

@@ -97,7 +97,6 @@ rm -rf ${LOGDIR:?}/*
 timestamp=$(date +%s)
 for pod in $pods; do
   # xargs removes the leading and trailing white spaces.
-  # See https://stackoverflow.com/a/12973694
   node_name="$(kubectl get pod "${pod}" -n "${NAMESPACE}" -o=custom-columns=:.spec.nodeName | xargs)"
   filename="${LOGDIR}/log$timestamp.$pod.${node_name}"
   kubectl logs -n "${NAMESPACE}" "$pod" > "$filename"

src/stirling/obj_tools/elf_tools.cc

@@ -325,7 +325,7 @@ std::vector<uint64_t> FindRetInsts(utils::u8string_view byte_code) {
   // code, we just provide a small buffer.
   // (Unfortunately, nullptr and 0 crashes.)
   constexpr int kBufSize = 32;
-  // Initialize array to zero. See more details at: https://stackoverflow.com/a/5591516.
+  // Initialize array to zero.
   char buf[kBufSize] = {};
 
   uint64_t pc = 0;

src/stirling/socket_trace_bpf_test.cc

@@ -66,7 +66,6 @@ constexpr std::string_view kNoProtocolMsg = R"(This is not an HTTP message)";
 
 // This test requires docker container with --pid=host so that the container's PID and the
 // host machine are identical.
-// See https://stackoverflow.com/questions/33328841/pid-mapping-between-docker-and-host
 
 // TODO(yzhao): Apply this pattern to other syscall pairs. An issue is that other syscalls do not
 // use scatter buffer. One approach would be to concatenate inner vector to a single string, and

src/stirling/testing/java/run_hello_world_on_gke.sh

@@ -2,9 +2,7 @@
 
 bazel run //src/stirling/testing/java:push_hello_world_image
 
-# Because simply apply the yaml file would not force re-pulling the image, as the tag has not been
-# changed. We have to delete and then apply.
-# We can use kubectl rollout restart:
-# https://stackoverflow.com/a/51835397
+# kubectl apply does not force re-pulling the image, as the tag has not changed.
+# We have to delete and then apply.
 kubectl delete -f src/stirling/testing/java/hello_world.yaml
 kubectl apply -f src/stirling/testing/java/hello_world.yaml