[lldb] Fix slide calculation in LLDBMemoryReader

The slide calculation was incorrectly negating an unsigned integer and
passing it to a call to SaturatingAdd. The section load address should
always be larger than the section file address, so change the underflow
check to a comparison instead.

(cherry picked from commit 61f28df)

Author: augusto2112

lldb/source/Plugins/LanguageRuntime/Swift/LLDBMemoryReader.cpp

@@ -643,17 +643,17 @@ LLDBMemoryReader::resolveRemoteAddress(uint64_t address) const {
   auto sec_file_address = sec->GetFileAddress();
   auto sec_load_address = sec->GetLoadBaseAddress(&m_process.GetTarget());
 
-  bool overflow = false;
-  auto slide =
-      llvm::SaturatingAdd(sec_load_address, -sec_file_address, &overflow);
-  if (overflow) {
+  if (sec_load_address < sec_file_address) {
     LLDB_LOG(log,
-             "[MemoryReader] section load address {0:x} - file address {1:x} "
-             "overflows",
+             "[MemoryReader] section load address {0:x} is smaller than "
+             "section file address {1:x}",
              sec_load_address, sec_file_address);
     return {};
   }
 
+  auto slide = sec_load_address - sec_file_address;
+
+  bool overflow = false;
   auto virtual_address = llvm::SaturatingAdd(file_address, slide, &overflow);
   if (overflow) {
     LLDB_LOG(log, "[MemoryReader] file address {0:x} + slide {1:x} overflows",