target/sh4: Implement prctl_unalign_sigbus

Leave TARGET_ALIGNED_ONLY set, but use the new CPUState
flag to set MO_UNALN for the instructions that the kernel
handles in the unaligned trap.

The Linux kernel does not handle all memory operations: no
floating-point and no MAC.

Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
Reviewed-by: Laurent Vivier <laurent@vivier.eu>
Message-Id: <20211227150127.2659293-7-richard.henderson@linaro.org>
Signed-off-by: Laurent Vivier <laurent@vivier.eu>

2 files changed, 38 insertions, 16 deletions

diff --git a/target/sh4/cpu.h b/target/sh4/cpu.h
index 4cfb109f56..fb9dd9db2f 100644
--- a/target/sh4/cpu.h
+++ b/target/sh4/cpu.h
@@ -83,6 +83,7 @@
 #define DELAY_SLOT_RTE         (1 << 2)
 
 #define TB_FLAG_PENDING_MOVCA  (1 << 3)
+#define TB_FLAG_UNALIGN        (1 << 4)
 
 #define GUSA_SHIFT             4
 #ifdef CONFIG_USER_ONLY
@@ -373,6 +374,9 @@ static inline void cpu_get_tb_cpu_state(CPUSH4State *env, target_ulong *pc,
             | (env->sr & ((1u << SR_MD) | (1u << SR_RB)))      /* Bits 29-30 */
             | (env->sr & (1u << SR_FD))                        /* Bit 15 */
             | (env->movcal_backup ? TB_FLAG_PENDING_MOVCA : 0); /* Bit 3 */
+#ifdef CONFIG_USER_ONLY
+    *flags |= TB_FLAG_UNALIGN * !env_cpu(env)->prctl_unalign_sigbus;
+#endif
 }
 
 #endif /* SH4_CPU_H */
diff --git a/target/sh4/translate.c b/target/sh4/translate.c
index ce5d674a52..50493c61ea 100644
--- a/target/sh4/translate.c
+++ b/target/sh4/translate.c
@@ -50,8 +50,10 @@ typedef struct DisasContext {
 
 #if defined(CONFIG_USER_ONLY)
 #define IS_USER(ctx) 1
+#define UNALIGN(C)   (ctx->tbflags & TB_FLAG_UNALIGN ? MO_UNALN : MO_ALIGN)
 #else
 #define IS_USER(ctx) (!(ctx->tbflags & (1u << SR_MD)))
+#define UNALIGN(C)   0
 #endif
 
 /* Target-specific values for ctx->base.is_jmp.  */
@@ -495,7 +497,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_addi_i32(addr, REG(B11_8), B3_0 * 4);
-            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx, MO_TEUL);
+            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx,
+                                MO_TEUL | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -503,7 +506,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_addi_i32(addr, REG(B7_4), B3_0 * 4);
-            tcg_gen_qemu_ld_i32(REG(B11_8), addr, ctx->memidx, MO_TESL);
+            tcg_gen_qemu_ld_i32(REG(B11_8), addr, ctx->memidx,
+                                MO_TESL | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -558,19 +562,23 @@ static void _decode_opc(DisasContext * ctx)
         tcg_gen_qemu_st_i32(REG(B7_4), REG(B11_8), ctx->memidx, MO_UB);
 	return;
     case 0x2001:		/* mov.w Rm,@Rn */
-        tcg_gen_qemu_st_i32(REG(B7_4), REG(B11_8), ctx->memidx, MO_TEUW);
+        tcg_gen_qemu_st_i32(REG(B7_4), REG(B11_8), ctx->memidx,
+                            MO_TEUW | UNALIGN(ctx));
 	return;
     case 0x2002:		/* mov.l Rm,@Rn */
-        tcg_gen_qemu_st_i32(REG(B7_4), REG(B11_8), ctx->memidx, MO_TEUL);
+        tcg_gen_qemu_st_i32(REG(B7_4), REG(B11_8), ctx->memidx,
+                            MO_TEUL | UNALIGN(ctx));
 	return;
     case 0x6000:		/* mov.b @Rm,Rn */
         tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx, MO_SB);
 	return;
     case 0x6001:		/* mov.w @Rm,Rn */
-        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx, MO_TESW);
+        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx,
+                            MO_TESW | UNALIGN(ctx));
 	return;
     case 0x6002:		/* mov.l @Rm,Rn */
-        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx, MO_TESL);
+        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx,
+                            MO_TESL | UNALIGN(ctx));
 	return;
     case 0x2004:		/* mov.b Rm,@-Rn */
 	{
@@ -586,7 +594,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_subi_i32(addr, REG(B11_8), 2);
-            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx, MO_TEUW);
+            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx,
+                                MO_TEUW | UNALIGN(ctx));
 	    tcg_gen_mov_i32(REG(B11_8), addr);
 	    tcg_temp_free(addr);
 	}
@@ -595,7 +604,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_subi_i32(addr, REG(B11_8), 4);
-            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx, MO_TEUL);
+            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx,
+                                MO_TEUL | UNALIGN(ctx));
 	    tcg_gen_mov_i32(REG(B11_8), addr);
         tcg_temp_free(addr);
 	}
@@ -606,12 +616,14 @@ static void _decode_opc(DisasContext * ctx)
 		tcg_gen_addi_i32(REG(B7_4), REG(B7_4), 1);
 	return;
     case 0x6005:		/* mov.w @Rm+,Rn */
-        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx, MO_TESW);
+        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx,
+                            MO_TESW | UNALIGN(ctx));
 	if ( B11_8 != B7_4 )
 		tcg_gen_addi_i32(REG(B7_4), REG(B7_4), 2);
 	return;
     case 0x6006:		/* mov.l @Rm+,Rn */
-        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx, MO_TESL);
+        tcg_gen_qemu_ld_i32(REG(B11_8), REG(B7_4), ctx->memidx,
+                            MO_TESL | UNALIGN(ctx));
 	if ( B11_8 != B7_4 )
 		tcg_gen_addi_i32(REG(B7_4), REG(B7_4), 4);
 	return;
@@ -627,7 +639,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_add_i32(addr, REG(B11_8), REG(0));
-            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx, MO_TEUW);
+            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx,
+                                MO_TEUW | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -635,7 +648,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_add_i32(addr, REG(B11_8), REG(0));
-            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx, MO_TEUL);
+            tcg_gen_qemu_st_i32(REG(B7_4), addr, ctx->memidx,
+                                MO_TEUL | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -651,7 +665,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_add_i32(addr, REG(B7_4), REG(0));
-            tcg_gen_qemu_ld_i32(REG(B11_8), addr, ctx->memidx, MO_TESW);
+            tcg_gen_qemu_ld_i32(REG(B11_8), addr, ctx->memidx,
+                                MO_TESW | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -659,7 +674,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_add_i32(addr, REG(B7_4), REG(0));
-            tcg_gen_qemu_ld_i32(REG(B11_8), addr, ctx->memidx, MO_TESL);
+            tcg_gen_qemu_ld_i32(REG(B11_8), addr, ctx->memidx,
+                                MO_TESL | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -1253,7 +1269,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_addi_i32(addr, REG(B7_4), B3_0 * 2);
-            tcg_gen_qemu_st_i32(REG(0), addr, ctx->memidx, MO_TEUW);
+            tcg_gen_qemu_st_i32(REG(0), addr, ctx->memidx,
+                                MO_TEUW | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;
@@ -1269,7 +1286,8 @@ static void _decode_opc(DisasContext * ctx)
 	{
 	    TCGv addr = tcg_temp_new();
 	    tcg_gen_addi_i32(addr, REG(B7_4), B3_0 * 2);
-            tcg_gen_qemu_ld_i32(REG(0), addr, ctx->memidx, MO_TESW);
+            tcg_gen_qemu_ld_i32(REG(0), addr, ctx->memidx,
+                                MO_TESW | UNALIGN(ctx));
 	    tcg_temp_free(addr);
 	}
 	return;