diff mbox series

[05/11] x86: respect ".arch nonop" when selecting which NOPs to emit

Message ID a0856079-0bf9-20d7-bd62-8f5f20584251@suse.com
State New
Headers
Series x86: NOP emission adjustments |

Checks

Context Check Description
linaro-tcwg-bot/tcwg_binutils_check--master-aarch64 success Testing passed
linaro-tcwg-bot/tcwg_binutils_check--master-arm success Testing passed
linaro-tcwg-bot/tcwg_binutils_build--master-arm success Testing passed
linaro-tcwg-bot/tcwg_binutils_build--master-aarch64 success Testing passed

Commit Message

Jan Beulich Sept. 27, 2023, 3:50 p.m. UTC 
  Making GENERIC64 a special case was never correct; prior to the
generalization of ".arch .no*" to cover all ISA extensions other
processor families supporting long NOPs should have been covered as
well. When introducing ".arch .nonops" (among others) it wasn't
apparent that a hidden implication of .cpunop not being possible to
separately turn off existed here. Seeing that the two large case label
blocks in the 2nd switch() already had identical behavior, simply
collapse all of the (useful) case labels into a single "default" one.
diff mbox series

Patch

--- a/gas/config/tc-i386.c
+++ b/gas/config/tc-i386.c
@@ -802,7 +802,7 @@  static const char *cpu_arch_name = NULL;
 static char *cpu_sub_arch_name = NULL;
 
 /* CPU feature flags.  */
-static i386_cpu_flags cpu_arch_flags = CPU_UNKNOWN_FLAGS;
+i386_cpu_flags cpu_arch_flags = CPU_UNKNOWN_FLAGS;
 
 /* If we have selected a cpu we are generating instructions for.  */
 static int cpu_arch_tune_set = 0;
@@ -1463,7 +1463,8 @@  i386_generate_nops (fragS *fragP, char *
       patt = fragP->tc_frag_data.code == CODE_64BIT ? f64_patt : f32_patt;
       if (fragP->tc_frag_data.isa == PROCESSOR_UNKNOWN)
 	{
-	  /* PROCESSOR_UNKNOWN means that all ISAs may be used.  */
+	  /* PROCESSOR_UNKNOWN means that all ISAs may be used, unless
+	     explicitly disabled.  */
 	  switch (fragP->tc_frag_data.tune)
 	    {
 	    case PROCESSOR_UNKNOWN:
@@ -1485,8 +1486,10 @@  i386_generate_nops (fragS *fragP, char *
 	    case PROCESSOR_BD:
 	    case PROCESSOR_ZNVER:
 	    case PROCESSOR_BT:
-	      patt = alt_patt;
+	      if (fragP->tc_frag_data.cpunop)
+		patt = alt_patt;
 	      break;
+
 	    case PROCESSOR_I386:
 	    case PROCESSOR_I486:
 	    case PROCESSOR_PENTIUM:
@@ -1508,35 +1511,13 @@  i386_generate_nops (fragS *fragP, char *
 	      abort ();
 	      break;
 
-	    case PROCESSOR_I386:
-	    case PROCESSOR_I486:
-	    case PROCESSOR_PENTIUM:
-	    case PROCESSOR_IAMCU:
-	    case PROCESSOR_K6:
-	    case PROCESSOR_ATHLON:
-	    case PROCESSOR_K8:
-	    case PROCESSOR_AMDFAM10:
-	    case PROCESSOR_BD:
-	    case PROCESSOR_ZNVER:
-	    case PROCESSOR_BT:
-	    case PROCESSOR_GENERIC32:
+	    default:
 	      /* We use cpu_arch_isa_flags to check if we CAN optimize
 		 with nops.  */
 	      if (fragP->tc_frag_data.isa_flags.bitfield.cpunop)
 		patt = alt_patt;
 	      break;
-	    case PROCESSOR_PENTIUMPRO:
-	    case PROCESSOR_PENTIUM4:
-	    case PROCESSOR_NOCONA:
-	    case PROCESSOR_CORE:
-	    case PROCESSOR_CORE2:
-	    case PROCESSOR_COREI7:
-	      if (fragP->tc_frag_data.isa_flags.bitfield.cpunop)
-		patt = alt_patt;
-	      break;
-	    case PROCESSOR_GENERIC64:
-	      patt = alt_patt;
-	      break;
+
 	    case PROCESSOR_NONE:
 	      abort ();
 	    }
--- a/gas/config/tc-i386.h
+++ b/gas/config/tc-i386.h
@@ -260,6 +260,7 @@  enum processor_type
   PROCESSOR_NONE
 };
 
+extern i386_cpu_flags cpu_arch_flags;
 extern enum processor_type cpu_arch_tune;
 extern enum processor_type cpu_arch_isa;
 extern i386_cpu_flags cpu_arch_isa_flags;
@@ -295,6 +296,7 @@  struct i386_tc_frag_data
   unsigned int mf_type : 3;
   unsigned int classified : 1;
   unsigned int branch_type : 3;
+  unsigned int cpunop : 1;
 };
 
 /* We need to emit the right NOP pattern in .align frags.  This is
@@ -310,6 +312,7 @@  struct i386_tc_frag_data
      (FRAGP)->tc_frag_data.isa = cpu_arch_isa;			\
      (FRAGP)->tc_frag_data.isa_flags = cpu_arch_isa_flags;	\
      (FRAGP)->tc_frag_data.tune = cpu_arch_tune;		\
+     (FRAGP)->tc_frag_data.cpunop = cpu_arch_flags.bitfield.cpunop; \
      (FRAGP)->tc_frag_data.code = i386_flag_code;		\
      (FRAGP)->tc_frag_data.max_bytes = (MAX_BYTES);		\
      (FRAGP)->tc_frag_data.length = 0;				\
--- a/gas/testsuite/gas/i386/x86-64.exp
+++ b/gas/testsuite/gas/i386/x86-64.exp
@@ -109,6 +109,8 @@  run_dump_test "x86-64-nops-1-g64"
 run_dump_test "x86-64-nops-1-k8"
 run_dump_test "x86-64-nops-1-core2"
 run_dump_test "x86-64-nops-1-pentium"
+run_dump_test "x86-64-nops-1a-g64"
+run_dump_test "x86-64-nops-1a-core2"
 run_dump_test "x86-64-nops-2"
 run_dump_test "x86-64-nops-3"
 run_dump_test "x86-64-nops-4"
--- /dev/null
+++ b/gas/testsuite/gas/i386/x86-64-nops-1a-core2.d
@@ -0,0 +1,5 @@ 
+#as: -march=core2+nonop
+#source: nops-1.s
+#objdump: -drw
+#name: x86-64 -march=core2+nonop nops 1
+#dump: x86-64-nops-1-pentium.d
--- /dev/null
+++ b/gas/testsuite/gas/i386/x86-64-nops-1a-g64.d
@@ -0,0 +1,5 @@ 
+#as: -march=generic64+nonop
+#source: nops-1.s
+#objdump: -drw
+#name: x86-64 -march=generic64+nonop nops 1
+#dump: x86-64-nops-1-pentium.d