gold: Support 386 TLS code sequences without PLT

There are extensions to 386 psABI:

https://groups.google.com/forum/#!topic/ia32-abi/awsRSvJOJfs

to call tls_get_addr via GOT:

call *___tls_get_addr@GOT(%reg)

where EBX register isn't required as GOT base.

Since direct call is 4-byte long and indirect call, is 5-byte long, the
extra one byte must be handled properly.

For general dynamic model, 7-byte lea instruction before call
instruction is replaced by 6-byte one to make room for indirect call.
For local dynamic model, we simply use 5-byte indirect call.

TLS linker optimization is updated to recognize new instruction
patterns.  For local dynamic model to local exec model transition,
we generate a 6-byte lea instruction as nop, instead of a 1-byte nop
plus a 4-byte lea instruction.

	PR gold/20308
	* i386.cc (Target_i386::Relocate::relocate): Allow
	R_386_GOT32X relocation against ___tls_get_addr.
	(Target_i386::Relocate::tls_gd_to_ie): Support indirect
	call to __tls_get_addr.
	(Target_i386::Relocate::tls_gd_to_le): Likewise.
	(Target_i386::Relocate::tls_ld_to_le): Likewise.
	* testsuite/Makefile.am (check_PROGRAMS): Add pr20308a_test,
	pr20308b_test, pr20308c_test, pr20308d_test, pr20308e_test.
	(pr20308a_test_SOURCES): New.
	(pr20308a_test_DEPENDENCIES): Likewise.
	(pr20308a_test_CFLAGS): Likewise.
	(pr20308a_test_LDFLAGS): Likewise.
	(pr20308a_test_LDADD): Likewise.
	(pr20308b_test_SOURCES): Likewise.
	(pr20308b_test_DEPENDENCIES): Likewise.
	(pr20308b_test_CFLAGS): Likewise.
	(pr20308b_test_LDFLAGS): Likewise.
	(pr20308b_test_LDADD): Likewise.
	(pr20308c_test_SOURCES): Likewise.
	(pr20308c_test_DEPENDENCIES): Likewise.
	(pr20308c_test_CFLAGS): Likewise.
	(pr20308c_test_LDFLAGS): Likewise.
	(pr20308c_test_LDADD): Likewise.
	(pr20308d_test_SOURCES): Likewise.
	(pr20308d_test_DEPENDENCIES): Likewise.
	(pr20308d_test_CFLAGS): Likewise.
	(pr20308d_test_LDFLAGS): Likewise.
	(pr20308d_test_LDADD): Likewise.
	(pr20308e_test_SOURCES): Likewise.
	(pr20308e_test_DEPENDENCIES): Likewise.
	(pr20308e_test_CFLAGS): Likewise.
	(pr20308e_test_LDFLAGS): Likewise.
	(pr20308e_test_LDADD): Likewise.
	(pr20308a.so): Likewise.
	(pr20308b.so): Likewise.
	(pr20308_gd.o): Likewise.
	(pr20308_ld.o): Likewise.
	(MOSTLYCLEANFILES): Add pr20308a.so pr20308b.so.
	* testsuite/Makefile.in: Regenerated.
	* testsuite/pr20308_def.c: New file.
	* testsuite/pr20308_gd.S: Likewise.
	* testsuite/pr20308_ld.S: Likewise.
	* testsuite/pr20308_main.c: Likewise.

1 files changed, 54 insertions, 18 deletions

diff --git a/gold/i386.cc b/gold/i386.cc
index 87e6a6d..28864cd 100644
--- a/gold/i386.cc
+++ b/gold/i386.cc
@@ -2790,6 +2790,7 @@ Target_i386::Relocate::relocate(const Relocate_info<32, false>* relinfo,
   if (this->skip_call_tls_get_addr_)
     {
       if ((r_type != elfcpp::R_386_PLT32
+	   && r_type != elfcpp::R_386_GOT32X
 	   && r_type != elfcpp::R_386_PC32)
 	  || gsym == NULL
 	  || strcmp(gsym->name(), "___tls_get_addr") != 0)
@@ -3318,9 +3319,11 @@ Target_i386::Relocate::tls_gd_to_le(const Relocate_info<32, false>* relinfo,
 				    unsigned char* view,
 				    section_size_type view_size)
 {
-  // leal foo(,%reg,1),%eax; call ___tls_get_addr
+  // leal foo(,%ebx,1),%eax; call ___tls_get_addr@PLT
   //  ==> movl %gs:0,%eax; subl $foo@tpoff,%eax
-  // leal foo(%reg),%eax; call ___tls_get_addr
+  // leal foo(%ebx),%eax; call ___tls_get_addr@PLT
+  //  ==> movl %gs:0,%eax; subl $foo@tpoff,%eax
+  // leal foo(%reg),%eax; call *___tls_get_addr@GOT(%reg)
   //  ==> movl %gs:0,%eax; subl $foo@tpoff,%eax
 
   tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size, -2);
@@ -3328,10 +3331,12 @@ Target_i386::Relocate::tls_gd_to_le(const Relocate_info<32, false>* relinfo,
 
   unsigned char op1 = view[-1];
   unsigned char op2 = view[-2];
+  unsigned char op3 = view[4];
 
   tls::check_tls(relinfo, relnum, rel.get_r_offset(),
 		 op2 == 0x8d || op2 == 0x04);
-  tls::check_tls(relinfo, relnum, rel.get_r_offset(), view[4] == 0xe8);
+  tls::check_tls(relinfo, relnum, rel.get_r_offset(),
+		 op3 == 0xe8 || op3 == 0xff);
 
   int roff = 5;
 
@@ -3345,12 +3350,18 @@ Target_i386::Relocate::tls_gd_to_le(const Relocate_info<32, false>* relinfo,
     }
   else
     {
+      unsigned char reg = op1 & 7;
       tls::check_tls(relinfo, relnum, rel.get_r_offset(),
-		     (op1 & 0xf8) == 0x80 && (op1 & 7) != 4);
-      if (rel.get_r_offset() + 9 < view_size
-	  && view[9] == 0x90)
+		     ((op1 & 0xf8) == 0x80
+		      && reg != 4
+		      && reg != 0
+		      && (op3 == 0xe8 || (view[5] & 0x7) == reg)));
+      if (op3 == 0xff
+	  || (rel.get_r_offset() + 9 < view_size
+	      && view[9] == 0x90))
 	{
-	  // There is a trailing nop.  Use the size byte subl.
+	  // There is an indirect call or a trailing nop.  Use the size
+	  // byte subl.
 	  memcpy(view - 2, "\x65\xa1\0\0\0\0\x81\xe8\0\0\0", 12);
 	  roff = 6;
 	}
@@ -3381,20 +3392,24 @@ Target_i386::Relocate::tls_gd_to_ie(const Relocate_info<32, false>* relinfo,
 				    unsigned char* view,
 				    section_size_type view_size)
 {
-  // leal foo(,%ebx,1),%eax; call ___tls_get_addr
+  // leal foo(,%ebx,1),%eax; call ___tls_get_addr@PLT
   //  ==> movl %gs:0,%eax; addl foo@gotntpoff(%ebx),%eax
-  // leal foo(%ebx),%eax; call ___tls_get_addr; nop
+  // leal foo(%ebx),%eax; call ___tls_get_addr@PLT; nop
   //  ==> movl %gs:0,%eax; addl foo@gotntpoff(%ebx),%eax
+  // leal foo(%reg),%eax; call *___tls_get_addr@GOT(%reg)
+  //  ==> movl %gs:0,%eax; addl foo@gotntpoff(%reg),%eax
 
   tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size, -2);
   tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size, 9);
 
   unsigned char op1 = view[-1];
   unsigned char op2 = view[-2];
+  unsigned char op3 = view[4];
 
   tls::check_tls(relinfo, relnum, rel.get_r_offset(),
 		 op2 == 0x8d || op2 == 0x04);
-  tls::check_tls(relinfo, relnum, rel.get_r_offset(), view[4] == 0xe8);
+  tls::check_tls(relinfo, relnum, rel.get_r_offset(),
+		 op3 == 0xe8 || op3 == 0xff);
 
   int roff;
 
@@ -3408,10 +3423,14 @@ Target_i386::Relocate::tls_gd_to_ie(const Relocate_info<32, false>* relinfo,
     }
   else
     {
+      unsigned char reg = op1 & 7;
       tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size, 10);
       tls::check_tls(relinfo, relnum, rel.get_r_offset(),
-		     (op1 & 0xf8) == 0x80 && (op1 & 7) != 4);
-      tls::check_tls(relinfo, relnum, rel.get_r_offset(), view[9] == 0x90);
+		     ((op1 & 0xf8) == 0x80
+		      && reg != 4
+		      && reg != 0
+		      && ((op3 == 0xe8 && view[9] == 0x90)
+			   || (view[5] & 0x7) == reg)));
       roff = 6;
     }
 
@@ -3512,19 +3531,36 @@ Target_i386::Relocate::tls_ld_to_le(const Relocate_info<32, false>* relinfo,
 				    unsigned char* view,
 				    section_size_type view_size)
 {
-  // leal foo(%reg), %eax; call ___tls_get_addr
+  // leal foo(%ebx), %eax; call ___tls_get_addr@PLT
   // ==> movl %gs:0,%eax; nop; leal 0(%esi,1),%esi
+  // leal foo(%reg), %eax; call call *___tls_get_addr@GOT(%reg)
+  // ==> movl %gs:0,%eax; leal (%esi),%esi
 
   tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size, -2);
-  tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size, 9);
 
-  // FIXME: Does this test really always pass?
+  unsigned char op1 = view[-1];
+  unsigned char op2 = view[-2];
+  unsigned char op3 = view[4];
+
   tls::check_tls(relinfo, relnum, rel.get_r_offset(),
-		 view[-2] == 0x8d && view[-1] == 0x83);
+		 op3 == 0xe8 || op3 == 0xff);
+  tls::check_range(relinfo, relnum, rel.get_r_offset(), view_size,
+		   op3 == 0xe8 ? 9 : 10);
+
+  // FIXME: Does this test really always pass?
+  tls::check_tls(relinfo, relnum, rel.get_r_offset(), op2 == 0x8d);
 
-  tls::check_tls(relinfo, relnum, rel.get_r_offset(), view[4] == 0xe8);
+  unsigned char reg = op1 & 7;
+  tls::check_tls(relinfo, relnum, rel.get_r_offset(),
+		 ((op1 & 0xf8) == 0x80
+		  && reg != 4
+		  && reg != 0
+		  && (op3 == 0xe8 || (view[5] & 0x7) == reg)));
 
-  memcpy(view - 2, "\x65\xa1\0\0\0\0\x90\x8d\x74\x26\0", 11);
+  if (op3 == 0xe8)
+    memcpy(view - 2, "\x65\xa1\0\0\0\0\x90\x8d\x74\x26\0", 11);
+  else
+    memcpy(view - 2, "\x65\xa1\0\0\0\0\x8d\xb6\0\0\0\0", 12);
 
   // The next reloc should be a PLT32 reloc against __tls_get_addr.
   // We can skip it.