gdb: Fix issue with Clang CLI macros

Clang up to version 15 (current) adds macros that were defined in the
command line or by "other means", according to the Dwarf specification,
after the last DW_MACRO_end_file, instead of before the first
DW_MACRO_start_file, as the specification dictates.  When GDB reads the
macros after the last file is closed, the macros never end up "in scope"
and so we can't print them.  This has been submitted as a bug to Clang
developers (https://github.com/llvm/llvm-project/issues/54506), and PR
macros/29034 was opened for GDB to keep track of this.

Seeing as there is no expected date for it to be fixed, add a workaround
for all current versions of Clang.  The workaround detects when
the main file would be closed and if the producer is Clang, and turns
that operation into a noop, so we keep a reference to the current_file
as those macros are read.

A test case was added to confirm the functionality, and the KFAIL for
running gdb.base/macro-source-path when using clang.

Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=29034
Approved-By: Simon Marchi <simon.marchi@efficios.com>

5 files changed, 33 insertions, 6 deletions

diff --git a/gdb/dwarf2/cu.h b/gdb/dwarf2/cu.h
index 23cb3d2..51638e7 100644
--- a/gdb/dwarf2/cu.h
+++ b/gdb/dwarf2/cu.h
@@ -264,6 +264,7 @@ public:
   bool producer_is_icc : 1;
   bool producer_is_icc_lt_14 : 1;
   bool producer_is_codewarrior : 1;
+  bool producer_is_clang : 1;
 
   /* When true, the file that we're processing is known to have
      debugging info for C++ namespaces.  GCC 3.3.x did not produce
diff --git a/gdb/dwarf2/macro.c b/gdb/dwarf2/macro.c
index 38c0fdf..7224c45 100644
--- a/gdb/dwarf2/macro.c
+++ b/gdb/dwarf2/macro.c
@@ -445,7 +445,7 @@ dwarf_decode_macro_bytes (dwarf2_per_objfile *per_objfile,
 			  struct dwarf2_section_info *str_section,
 			  struct dwarf2_section_info *str_offsets_section,
 			  gdb::optional<ULONGEST> str_offsets_base,
-			  htab_t include_hash)
+			  htab_t include_hash, struct dwarf2_cu *cu)
 {
   struct objfile *objfile = per_objfile->objfile;
   enum dwarf_macro_record_type macinfo_type;
@@ -672,6 +672,17 @@ dwarf_decode_macro_bytes (dwarf2_per_objfile *per_objfile,
 	  if (! current_file)
 	    complaint (_("macro debug info has an unmatched "
 			 "`close_file' directive"));
+	  else if (current_file->included_by == nullptr
+		   && producer_is_clang (cu))
+	    {
+	      /* Clang, until the current version, misplaces some macro
+		 definitions - such as ones defined in the command line,
+		 putting them after the last DW_MACRO_end_file instead of
+		 before the first DW_MACRO_start_file.  Since at the time
+		 of writing there is no clang version with this bug fixed,
+		 we check for any clang producer.  This should be changed
+		 to producer_is_clang_lt_XX when possible. */
+	    }
 	  else
 	    {
 	      current_file = current_file->included_by;
@@ -751,7 +762,7 @@ dwarf_decode_macro_bytes (dwarf2_per_objfile *per_objfile,
 					  current_file, lh, section,
 					  section_is_gnu, is_dwz, offset_size,
 					  str_section, str_offsets_section,
-					  str_offsets_base, include_hash);
+					  str_offsets_base, include_hash, cu);
 
 		htab_remove_elt (include_hash, (void *) new_mac_ptr);
 	      }
@@ -795,7 +806,7 @@ dwarf_decode_macros (dwarf2_per_objfile *per_objfile,
 		     unsigned int offset, struct dwarf2_section_info *str_section,
 		     struct dwarf2_section_info *str_offsets_section,
 		     gdb::optional<ULONGEST> str_offsets_base,
-		     int section_is_gnu)
+		     int section_is_gnu, struct dwarf2_cu *cu)
 {
   bfd *abfd;
   const gdb_byte *mac_ptr, *mac_end;
@@ -956,5 +967,5 @@ dwarf_decode_macros (dwarf2_per_objfile *per_objfile,
   dwarf_decode_macro_bytes (per_objfile, builder, abfd, mac_ptr, mac_end,
 			    current_file, lh, section, section_is_gnu, 0,
 			    offset_size, str_section, str_offsets_section,
-			    str_offsets_base, include_hash.get ());
+			    str_offsets_base, include_hash.get (), cu);
 }
diff --git a/gdb/dwarf2/macro.h b/gdb/dwarf2/macro.h
index e8c33a34..02753ef 100644
--- a/gdb/dwarf2/macro.h
+++ b/gdb/dwarf2/macro.h
@@ -31,6 +31,6 @@ extern void dwarf_decode_macros (dwarf2_per_objfile *per_objfile,
 				 dwarf2_section_info *str_section,
 				 dwarf2_section_info *str_offsets_section,
 				 gdb::optional<ULONGEST> str_offsets_base,
-				 int section_is_gnu);
+				 int section_is_gnu, struct dwarf2_cu *cu);
 
 #endif /* GDB_DWARF2_MACRO_H */
diff --git a/gdb/dwarf2/read.c b/gdb/dwarf2/read.c
index 7a5745b..60e120a 100644
--- a/gdb/dwarf2/read.c
+++ b/gdb/dwarf2/read.c
@@ -9484,6 +9484,16 @@ producer_is_gcc_lt_4_3 (struct dwarf2_cu *cu)
   return cu->producer_is_gcc_lt_4_3;
 }
 
+/* See dwarf2/read.h.  */
+bool
+producer_is_clang (struct dwarf2_cu *cu)
+{
+  if (!cu->checked_producer)
+    check_producer (cu);
+
+  return cu->producer_is_clang;
+}
+
 static file_and_directory &
 find_file_and_directory (struct die_info *die, struct dwarf2_cu *cu)
 {
@@ -13342,6 +13352,8 @@ check_producer (struct dwarf2_cu *cu)
     }
   else if (startswith (cu->producer, "CodeWarrior S12/L-ISA"))
     cu->producer_is_codewarrior = true;
+  else if (producer_is_clang (cu->producer, &major, &minor))
+    cu->producer_is_clang = true;
   else
     {
       /* For other non-GCC compilers, expect their behavior is DWARF version
@@ -23366,7 +23378,7 @@ dwarf_decode_macros (struct dwarf2_cu *cu, unsigned int offset,
 
   dwarf_decode_macros (per_objfile, builder, section, lh,
 		       offset_size, offset, str_section, str_offsets_section,
-		       str_offsets_base, section_is_gnu);
+		       str_offsets_base, section_is_gnu, cu);
 }
 
 /* Return the .debug_loc section to use for CU.
diff --git a/gdb/dwarf2/read.h b/gdb/dwarf2/read.h
index 5f01fbc..dd71826 100644
--- a/gdb/dwarf2/read.h
+++ b/gdb/dwarf2/read.h
@@ -761,4 +761,7 @@ extern void dwarf2_get_section_info (struct objfile *,
 				     asection **, const gdb_byte **,
 				     bfd_size_type *);
 
+/* Return true if the producer of the inferior is clang.  */
+extern bool producer_is_clang (struct dwarf2_cu *cu);
+
 #endif /* DWARF2READ_H */