[lldb] Teach FuncUnwinders about discontinuous functions (#133072)

The main change here is that we're now able to correctly look up plans
for these functions. Previously, due to caching, we could end up with
one entry covering most of the address space (because part of the
function was at the beginning and one at the end). Now, we can correctly
recognise that the part in between does not belong to that function, and
we can create a different FuncUnwinders instance for it. It doesn't help
the discontinuous function much (its plan will still be garbled), but
we can at least properly unwind out of the simple functions in between.

Fixing the unwind plans for discontinuous functions requires handling
each unwind source specially, and this setup allows us to make the
transition incrementally.

Author: labath

lldb/include/lldb/Symbol/FuncUnwinders.h

@@ -13,9 +13,9 @@ class UnwindTable;
 class FuncUnwinders {
 public:
   // FuncUnwinders objects are used to track UnwindPlans for a function (named
-  // or not - really just an address range)
+  // or not - really just a set of address ranges)
 
-  // We'll record four different UnwindPlans for each address range:
+  // We'll record four different UnwindPlans for each function:
   //
   //   1. Unwinding from a call site (a valid exception throw location)
   //      This is often sourced from the eh_frame exception handling info
@@ -31,7 +31,8 @@ class FuncUnwinders {
   // instructions are finished for migrating breakpoints past the stack frame
   // setup instructions when we don't have line table information.
 
-  FuncUnwinders(lldb_private::UnwindTable &unwind_table, AddressRange range);
+  FuncUnwinders(lldb_private::UnwindTable &unwind_table, Address addr,
+                AddressRanges ranges);
 
   ~FuncUnwinders();
 
@@ -54,7 +55,9 @@ class FuncUnwinders {
   const Address &GetFunctionStartAddress() const;
 
   bool ContainsAddress(const Address &addr) const {
-    return m_range.ContainsFileAddress(addr);
+    return llvm::any_of(m_ranges, [&](const AddressRange range) {
+      return range.ContainsFileAddress(addr);
+    });
   }
 
   // A function may have a Language Specific Data Area specified -- a block of
@@ -113,6 +116,15 @@ class FuncUnwinders {
       Thread &thread, const lldb::UnwindPlanSP &a, const lldb::UnwindPlanSP &b);
 
   UnwindTable &m_unwind_table;
+
+  /// Start address of the function described by this object.
+  Address m_addr;
+
+  /// The address ranges of the function.
+  AddressRanges m_ranges;
+
+  /// The smallest address range covering the entire function.
+  /// DEPRECATED: Use m_ranges instead.
   AddressRange m_range;
 
   std::recursive_mutex m_mutex;

lldb/include/lldb/Symbol/UnwindTable.h

@@ -66,8 +66,7 @@ class UnwindTable {
   void Dump(Stream &s);
 
   void Initialize();
-  std::optional<AddressRange> GetAddressRange(const Address &addr,
-                                              const SymbolContext &sc);
+  AddressRanges GetAddressRanges(const Address &addr, const SymbolContext &sc);
 
   typedef std::map<lldb::addr_t, lldb::FuncUnwindersSP> collection;
   typedef collection::iterator iterator;

lldb/source/Symbol/FuncUnwinders.cpp

@@ -31,15 +31,29 @@
 using namespace lldb;
 using namespace lldb_private;
 
-/// constructor
-
-FuncUnwinders::FuncUnwinders(UnwindTable &unwind_table, AddressRange range)
-    : m_unwind_table(unwind_table), m_range(range), m_mutex(),
-      m_unwind_plan_assembly_sp(), m_unwind_plan_eh_frame_sp(),
-      m_unwind_plan_eh_frame_augmented_sp(), m_unwind_plan_compact_unwind(),
-      m_unwind_plan_arm_unwind_sp(), m_unwind_plan_fast_sp(),
-      m_unwind_plan_arch_default_sp(),
-      m_unwind_plan_arch_default_at_func_entry_sp(),
+static AddressRange CollapseRanges(llvm::ArrayRef<AddressRange> ranges) {
+  if (ranges.empty())
+    return AddressRange();
+  if (ranges.size() == 1)
+    return ranges[0];
+
+  Address lowest_addr = ranges[0].GetBaseAddress();
+  addr_t highest_addr = lowest_addr.GetFileAddress() + ranges[0].GetByteSize();
+  for (const AddressRange &range : ranges.drop_front()) {
+    Address range_begin = range.GetBaseAddress();
+    addr_t range_end = range_begin.GetFileAddress() + range.GetByteSize();
+    if (range_begin.GetFileAddress() < lowest_addr.GetFileAddress())
+      lowest_addr = range_begin;
+    if (range_end > highest_addr)
+      highest_addr = range_end;
+  }
+  return AddressRange(lowest_addr, highest_addr - lowest_addr.GetFileAddress());
+}
+
+FuncUnwinders::FuncUnwinders(UnwindTable &unwind_table, Address addr,
+                             AddressRanges ranges)
+    : m_unwind_table(unwind_table), m_addr(std::move(addr)),
+      m_ranges(std::move(ranges)), m_range(CollapseRanges(m_ranges)),
       m_tried_unwind_plan_assembly(false), m_tried_unwind_plan_eh_frame(false),
       m_tried_unwind_plan_object_file(false),
       m_tried_unwind_plan_debug_frame(false),
@@ -511,9 +525,7 @@ Address &FuncUnwinders::GetFirstNonPrologueInsn(Target &target) {
   return m_first_non_prologue_insn;
 }
 
-const Address &FuncUnwinders::GetFunctionStartAddress() const {
-  return m_range.GetBaseAddress();
-}
+const Address &FuncUnwinders::GetFunctionStartAddress() const { return m_addr; }
 
 lldb::UnwindAssemblySP
 FuncUnwinders::GetUnwindAssemblyProfiler(Target &target) {

lldb/source/Symbol/UnwindTable.cpp

@@ -91,30 +91,35 @@ void UnwindTable::ModuleWasUpdated() {
 
 UnwindTable::~UnwindTable() = default;
 
-std::optional<AddressRange>
-UnwindTable::GetAddressRange(const Address &addr, const SymbolContext &sc) {
+AddressRanges UnwindTable::GetAddressRanges(const Address &addr,
+                                            const SymbolContext &sc) {
   AddressRange range;
 
   // First check the unwind info from the object file plugin
   if (m_object_file_unwind_up &&
       m_object_file_unwind_up->GetAddressRange(addr, range))
-    return range;
+    return {range};
 
   // Check the symbol context
-  if (sc.GetAddressRange(eSymbolContextFunction | eSymbolContextSymbol, 0,
-                         false, range) &&
-      range.GetBaseAddress().IsValid())
-    return range;
+  AddressRanges result;
+  for (size_t idx = 0;
+       sc.GetAddressRange(eSymbolContextFunction | eSymbolContextSymbol, idx,
+                          false, range) &&
+       range.GetBaseAddress().IsValid();
+       ++idx)
+    result.push_back(range);
+  if (!result.empty())
+    return result;
 
   // Does the eh_frame unwind info has a function bounds for this addr?
   if (m_eh_frame_up && m_eh_frame_up->GetAddressRange(addr, range))
-    return range;
+    return {range};
 
   // Try debug_frame as well
   if (m_debug_frame_up && m_debug_frame_up->GetAddressRange(addr, range))
-    return range;
+    return {range};
 
-  return std::nullopt;
+  return {};
 }
 
 FuncUnwindersSP
@@ -140,14 +145,14 @@ UnwindTable::GetFuncUnwindersContainingAddress(const Address &addr,
       return pos->second;
   }
 
-  auto range_or = GetAddressRange(addr, sc);
-  if (!range_or)
+  AddressRanges ranges = GetAddressRanges(addr, sc);
+  if (ranges.empty())
     return nullptr;
 
-  FuncUnwindersSP func_unwinder_sp(new FuncUnwinders(*this, *range_or));
-  m_unwinds.insert(insert_pos,
-                   std::make_pair(range_or->GetBaseAddress().GetFileAddress(),
-                                  func_unwinder_sp));
+  auto func_unwinder_sp = std::make_shared<FuncUnwinders>(*this, addr, ranges);
+  for (const AddressRange &range : ranges)
+    m_unwinds.emplace_hint(insert_pos, range.GetBaseAddress().GetFileAddress(),
+                           func_unwinder_sp);
   return func_unwinder_sp;
 }
 
@@ -159,11 +164,11 @@ FuncUnwindersSP UnwindTable::GetUncachedFuncUnwindersContainingAddress(
     const Address &addr, const SymbolContext &sc) {
   Initialize();
 
-  auto range_or = GetAddressRange(addr, sc);
-  if (!range_or)
+  AddressRanges ranges = GetAddressRanges(addr, sc);
+  if (ranges.empty())
     return nullptr;
 
-  return std::make_shared<FuncUnwinders>(*this, *range_or);
+  return std::make_shared<FuncUnwinders>(*this, addr, std::move(ranges));
 }
 
 void UnwindTable::Dump(Stream &s) {