Merge commit '673e3051b0ed' from llvm.org/main into next

Author: git apple-llvm automerger

flang-rt/include/flang-rt/runtime/io-error.h

@@ -67,6 +67,17 @@ class IoErrorHandler : public Terminator {
   RT_API_ATTRS int GetIoStat() const { return ioStat_; }
   RT_API_ATTRS bool GetIoMsg(char *, std::size_t);
 
+  // Sets the HasEnd flag so that EOF isn't fatal; used to peek ahead
+  RT_API_ATTRS bool SetHasEnd(bool yes = true) {
+    bool oldValue{(flags_ & hasEnd) != 0};
+    if (yes) {
+      flags_ |= hasEnd;
+    } else {
+      flags_ &= ~hasEnd;
+    }
+    return oldValue;
+  }
+
 private:
   enum Flag : std::uint8_t {
     hasIoStat = 1, // IOSTAT=

flang-rt/include/flang-rt/runtime/io-stmt.h

@@ -703,6 +703,13 @@ class ChildListIoStatementState : public ChildIoStatementState<DIR>,
   using ListDirectedStatementState<DIR>::GetNextDataEdit;
   RT_API_ATTRS bool AdvanceRecord(int = 1);
   RT_API_ATTRS int EndIoStatement();
+  RT_API_ATTRS bool CanAdvance() {
+    return DIR == Direction::Input &&
+        (canAdvance_ || this->mutableModes().inNamelist);
+  }
+
+private:
+  bool canAdvance_{false};
 };
 
 template <Direction DIR>

flang-rt/lib/runtime/descriptor-io.cpp

@@ -47,9 +47,11 @@ static RT_API_ATTRS common::optional<bool> DefinedFormattedIo(
     const typeInfo::DerivedType &derived,
     const typeInfo::SpecialBinding &special,
     const SubscriptValue subscripts[]) {
-  // Look at the next data edit descriptor.  If this is list-directed I/O, the
-  // "maxRepeat=0" argument will prevent the input from advancing over an
+  // Look at the next data edit descriptor.  If this is list-directed input,
+  // the "maxRepeat=0" argument will prevent the input from advancing over an
   // initial '(' that shouldn't be consumed now as the start of a real part.
+  // It also allows reaching EOF without crashing, since the EOF only matters
+  // if a child READ is actually performed.
   common::optional<DataEdit> peek{io.GetNextDataEdit(/*maxRepeat=*/0)};
   if (peek &&
       (peek->descriptor == DataEdit::DefinedDerivedType ||

flang-rt/lib/runtime/edit-input.cpp

@@ -53,11 +53,13 @@ static RT_API_ATTRS bool EditBOZInput(
     IoStatementState &io, const DataEdit &edit, void *n, std::size_t bytes) {
   // Skip leading white space & zeroes
   common::optional<int> remaining{io.CueUpInput(edit)};
-  auto start{io.GetConnectionState().positionInRecord};
+  const ConnectionState &connection{io.GetConnectionState()};
+  auto leftTabLimit{connection.leftTabLimit.value_or(0)};
+  auto start{connection.positionInRecord - leftTabLimit};
   common::optional<char32_t> next{io.NextInField(remaining, edit)};
   if (next.value_or('?') == '0') {
     do {
-      start = io.GetConnectionState().positionInRecord;
+      start = connection.positionInRecord - leftTabLimit;
       next = io.NextInField(remaining, edit);
     } while (next && *next == '0');
   }
@@ -447,7 +449,9 @@ static RT_API_ATTRS ScannedRealInput ScanRealInput(
     }
     // In list-directed input, a bad exponent is not consumed.
     auto nextBeforeExponent{next};
-    auto startExponent{io.GetConnectionState().positionInRecord};
+    const ConnectionState &connection{io.GetConnectionState()};
+    auto leftTabLimit{connection.leftTabLimit.value_or(0)};
+    auto startExponent{connection.positionInRecord - leftTabLimit};
     bool hasGoodExponent{false};
     if (next) {
       if (isHexadecimal) {

flang-rt/lib/runtime/io-stmt.cpp

@@ -880,6 +880,9 @@ ListDirectedStatementState<Direction::Input>::GetNextDataEdit(
     edit.descriptor = DataEdit::ListDirectedImaginaryPart;
   }
   auto fastField{io.GetUpcomingFastAsciiField()};
+  // Reaching EOF is okay when peeking at list-directed defined input;
+  // pretend that there's an END= in that case.
+  bool oldHasEnd{maxRepeat == 0 && !io.GetIoErrorHandler().SetHasEnd()};
   auto ch{io.GetNextNonBlank(byteCount, &fastField)};
   if (ch && *ch == comma && eatComma_) {
     // Consume comma & whitespace after previous item.
@@ -890,19 +893,23 @@ ListDirectedStatementState<Direction::Input>::GetNextDataEdit(
     ch = io.GetNextNonBlank(byteCount, &fastField);
   }
   eatComma_ = true;
-  if (!ch) {
-    return common::nullopt;
+  if (maxRepeat == 0 && !oldHasEnd) {
+    io.GetIoErrorHandler().SetHasEnd(false);
   }
-  if (*ch == '/') {
+  if (!ch) { // EOF
+    if (maxRepeat == 0) {
+      return edit; // DataEdit::ListDirected for look-ahead
+    } else {
+      return common::nullopt;
+    }
+  } else if (*ch == '/') {
     hitSlash_ = true;
     edit.descriptor = DataEdit::ListDirectedNullValue;
     return edit;
-  }
-  if (*ch == comma) { // separator: null value
+  } else if (*ch == comma) { // separator: null value
     edit.descriptor = DataEdit::ListDirectedNullValue;
     return edit;
-  }
-  if (imaginaryPart_) { // can't repeat components
+  } else if (imaginaryPart_) { // can't repeat components
     return edit;
   }
   if (*ch >= '0' && *ch <= '9' && fastField.MightBeRepetitionCount()) {
@@ -1103,10 +1110,19 @@ ChildListIoStatementState<DIR>::ChildListIoStatementState(
     : ChildIoStatementState<DIR>{child, sourceFile, sourceLine} {
 #if !defined(RT_DEVICE_AVOID_RECURSION)
   if constexpr (DIR == Direction::Input) {
-    if (auto *listInput{child.parent()
+    if (const auto *listInput{child.parent()
                 .get_if<ListDirectedStatementState<Direction::Input>>()}) {
       this->set_eatComma(listInput->eatComma());
       this->namelistGroup_ = listInput->namelistGroup();
+      if (auto *childListInput{child.parent()
+                  .get_if<ChildListIoStatementState<Direction::Input>>()}) {
+        // Child list input whose parent is child list input: can advance
+        // if the parent can.
+        this->canAdvance_ = childListInput->CanAdvance();
+      } else {
+        // Child list input of top-level list input: can advance.
+        this->canAdvance_ = true;
+      }
     }
   }
 #else
@@ -1117,12 +1133,7 @@ ChildListIoStatementState<DIR>::ChildListIoStatementState(
 template <Direction DIR>
 bool ChildListIoStatementState<DIR>::AdvanceRecord(int n) {
 #if !defined(RT_DEVICE_AVOID_RECURSION)
-  // Allow child NAMELIST input to advance
-  if (DIR == Direction::Input && this->mutableModes().inNamelist) {
-    return this->child().parent().AdvanceRecord(n);
-  } else {
-    return false;
-  }
+  return this->CanAdvance() && this->child().parent().AdvanceRecord(n);
 #else
   this->ReportUnsupportedChildIo();
 #endif

flang/docs/Extensions.md

@@ -928,6 +928,17 @@ print *, [(j,j=1,10)]
   and the portable interpretation across the most common Fortran
   compilers.
 
+* `NAMELIST` child input statements are allowed to advance to further
+  input records.
+  Further, advancement is allowed when the parent input statement is
+  a non-child (top level) list-directed input statement, or, recursively,
+  an intermediate child list-directed input statement that can advance.
+  This means that non-`NAMELIST` list-directed child input statements are
+  not allowed to advance when they have an ancestor formatted input statement
+  that is not list-directed and there is no intervening `NAMELIST`.
+  This design allows format-driven input with `DT` editing to retain
+  control over advancement in child input, while otherwise allowing it.
+
 ## De Facto Standard Features
 
 * `EXTENDS_TYPE_OF()` returns `.TRUE.` if both of its arguments have the