From 682ecc1745123d8c8cb324cce151accff2a311ed Mon Sep 17 00:00:00 2001
From: Michael Matz <matz@suse.de>
Date: Fri, 14 Oct 2016 17:32:10 +0200
Subject: [PATCH] arm64: Fix -run

With -run the call instruction and a defined function can be
far away, if the function is defined in the executable itself,
not in the to be compiled code.  So we always need PLT slots
for -run, not just for undefined symbols.
---
 arm64-gen.c | 16 ++--------------
 tccelf.c    | 18 ++++++++++--------
 2 files changed, 12 insertions(+), 22 deletions(-)

diff --git a/arm64-gen.c b/arm64-gen.c
index 5f52e1f2..cd6faac5 100644
--- a/arm64-gen.c
+++ b/arm64-gen.c
@@ -592,22 +592,10 @@ ST_FUNC void store(int r, SValue *sv)
 
 static void arm64_gen_bl_or_b(int b)
 {
-    // Currently TCC's linker does not generate a PLT when TCC is invoked
-    // with "-run". This means functions can be out of range if we try to
-    // call them in the usual way. Until the linker is fixed, work around
-    // this by using R_AARCH64_MOVW_UABS_G* relocations; see arm64_sym.
-    int avoid_call26 = 1;
-
     if ((vtop->r & (VT_VALMASK | VT_LVAL)) == VT_CONST) {
         assert(!b && (vtop->r & VT_SYM));
-        if (avoid_call26) {
-            arm64_sym(30, vtop->sym, 0);
-            o(0xd63f03c0); // blr x30
-        }
-        else {
-            greloc(cur_text_section, vtop->sym, ind, R_AARCH64_CALL26);
-            o(0x94000000); // bl .
-        }
+	greloc(cur_text_section, vtop->sym, ind, R_AARCH64_CALL26);
+	o(0x94000000); // bl .
     }
     else
         o(0xd61f0000 | (uint32_t)!b << 21 | intr(gv(RC_R30)) << 5); // br/blr
diff --git a/tccelf.c b/tccelf.c
index f0a32417..212ba08c 100644
--- a/tccelf.c
+++ b/tccelf.c
@@ -826,7 +826,8 @@ ST_FUNC void relocate_section(TCCState *s1, Section *s)
         case R_AARCH64_CALL26:
 	    /* This check must match the one in build_got_entries, testing
 	       if we really need a PLT slot.  */
-	    if (sym->st_shndx == SHN_UNDEF)
+	    if (sym->st_shndx == SHN_UNDEF ||
+		s1->output_type == TCC_OUTPUT_MEMORY)
 	        /* We've put the PLT slot offset into r_addend when generating
 		   it, and that's what we must use as relocation value (adjusted
 		   by section offset of course).  */
@@ -1423,19 +1424,20 @@ ST_FUNC void build_got_entries(TCCState *s1)
 
 	    case R_AARCH64_JUMP26:
 	    case R_AARCH64_CALL26:
-                if (!s1->got)
-                    build_got(s1);
-                sym_index = ELFW(R_SYM)(rel->r_info);
-                sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
-                if (sym->st_shndx == SHN_UNDEF) {
+		if (!s1->got)
+		    build_got(s1);
+		sym_index = ELFW(R_SYM)(rel->r_info);
+		sym = &((ElfW(Sym) *)symtab_section->data)[sym_index];
+		if (sym->st_shndx == SHN_UNDEF ||
+		    s1->output_type == TCC_OUTPUT_MEMORY) {
 		    unsigned long ofs;
 		    reloc_type = R_AARCH64_JUMP_SLOT;
-                    ofs = put_got_entry(s1, reloc_type, sym->st_size,
+		    ofs = put_got_entry(s1, reloc_type, sym->st_size,
 					sym->st_info, sym_index);
 		    /* We store the place of the generated PLT slot
 		       in our addend.  */
 		    rel->r_addend += ofs;
-                }
+		}
 		break;
 #elif defined(TCC_TARGET_C67)
             case R_C60_GOT32: