[binutils-gdb] RISC-V: Re-define mapping symbol $x to the file elf architecture attribute

[Nelson Chu]

https://sourceware.org/git/gitweb.cgi?p=binutils-gdb.git;h=ade87b8e6293f146f754629101cf4e367aec5da9

commit ade87b8e6293f146f754629101cf4e367aec5da9
Author: Andrew Oates <andrew@andrewoates.com>
Date:   Mon Feb 24 15:36:54 2025 +0800

    RISC-V: Re-define mapping symbol $x to the file elf architecture attribute
    
    The mapping symbol "$x" without an ISA string "means using ISA
    configuration from ELF attribute."[1].  Currently the code does not
    reset the subset_list.  This means that a previous mapping symbol that
    overrides the ISA string will continue to be used, rather than the
    default string set in the ELF file's .riscv.attributes section.  This
    can cause incorrect or failed instruction decodings.
    
    In practice, this causes problems when disassembling code generated by
    LLVM, which (unlike gas) does not emit explicit mapping symbols at the
    start of each section.
    
    This change stores the default architecture string seen at the beginning
    of disassembly in the global parse data struct, and restores that to
    subset_list whenever a bare "$x" symbol is seen.
    
    [1] https://github.com/riscv-non-isa/riscv-elf-psabi-doc/blob/master/riscv-elf.adoc#mapping-symbol
    
    Before this patch, the mapping-x.s was dumped as,
    
    00000000 <.text>:
       0:   00000013                nop
       4:   0001                    .insn   2, 0x0001
       6:   0001                    .insn   2, 0x0001
    
    Which is caused by the definiation of $x was conflict with the psABI.

Diff:
---
 gas/testsuite/gas/riscv/mapping-x.d | 13 +++++++++++++
 gas/testsuite/gas/riscv/mapping-x.s |  6 ++++++
 opcodes/riscv-dis.c                 | 19 +++++++++++++------
 3 files changed, 32 insertions(+), 6 deletions(-)

diff --git a/gas/testsuite/gas/riscv/mapping-x.d b/gas/testsuite/gas/riscv/mapping-x.d
new file mode 100644
index 00000000000..ccf9ef762e7
--- /dev/null
+++ b/gas/testsuite/gas/riscv/mapping-x.d
@@ -0,0 +1,13 @@
+#as:
+#source: mapping-x.s
+#objdump: -d
+
+.*:[ 	]+file format .*
+
+
+Disassembly of section .text:
+
+0+000 <.text>:
+[ 	]+[0-9a-f]+:[ 	]+00000013[ 	]+nop
+[ 	]+[0-9a-f]+:[ 	]+0001[ 	]+nop
+[ 	]+[0-9a-f]+:[ 	]+0001[ 	]+nop
diff --git a/gas/testsuite/gas/riscv/mapping-x.s b/gas/testsuite/gas/riscv/mapping-x.s
new file mode 100644
index 00000000000..f17cf2295ca
--- /dev/null
+++ b/gas/testsuite/gas/riscv/mapping-x.s
@@ -0,0 +1,6 @@
+.attribute arch, "rv32ic"
+.option arch, -c
+.insn 4, 0x00000013
+$x:
+.insn 2, 0x0001
+.insn 2, 0x0001
diff --git a/opcodes/riscv-dis.c b/opcodes/riscv-dis.c
index 9a6349931e7..c5e93620b7d 100644
--- a/opcodes/riscv-dis.c
+++ b/opcodes/riscv-dis.c
@@ -52,6 +52,9 @@ struct riscv_private_data
   enum riscv_spec_class default_priv_spec;
   /* Used for architecture parser.  */
   riscv_parse_subset_t riscv_rps_dis;
+  /* Default architecture string for the object file.  It will be changed once
+     elf architecture attribute exits.  This is used for mapping symbol $x.  */
+  const char* default_arch;
   /* Used for mapping symbols.  */
   int last_map_symbol;
   bfd_vma last_stop_offset;
@@ -1065,10 +1068,14 @@ riscv_update_map_state (int n,
     return;
 
   name = bfd_asymbol_name(info->symtab[n]);
-  if (strcmp (name, "$x") == 0)
-    *state = MAP_INSN;
-  else if (strcmp (name, "$d") == 0)
+  if (strcmp (name, "$d") == 0)
     *state = MAP_DATA;
+  else if (strcmp (name, "$x") == 0)
+    {
+      *state = MAP_INSN;
+      riscv_release_subset_list (pd->riscv_rps_dis.subset_list);
+      riscv_parse_subset (&pd->riscv_rps_dis, pd->default_arch);
+    }
   else if (strncmp (name, "$xrv", 4) == 0)
     {
       *state = MAP_INSN;
@@ -1380,7 +1387,7 @@ riscv_init_disasm_info (struct disassemble_info *info)
   pd->riscv_rps_dis.xlen = &pd->xlen;
   pd->riscv_rps_dis.isa_spec = &pd->default_isa_spec;
   pd->riscv_rps_dis.check_unknown_prefixed_ext = false;
-  const char *default_arch = "rv64gc";
+  pd->default_arch = "rv64gc";
   if (info->section != NULL)
     {
       bfd *abfd = info->section->owner;
@@ -1398,12 +1405,12 @@ riscv_init_disasm_info (struct disassemble_info *info)
 						      attr[Tag_b].i,
 						      attr[Tag_c].i,
 						      &pd->default_priv_spec);
-	      default_arch = attr[Tag_RISCV_arch].s;
+	      pd->default_arch = attr[Tag_RISCV_arch].s;
 	    }
 	}
     }
   riscv_release_subset_list (pd->riscv_rps_dis.subset_list);
-  riscv_parse_subset (&pd->riscv_rps_dis, default_arch);
+  riscv_parse_subset (&pd->riscv_rps_dis, pd->default_arch);
 
   pd->last_map_symbol = -1;
   pd->last_stop_offset = 0;