Skip to contentSkip to conversation

target/i386: Correct misplaced break statement in gen_shiftd_rm_T1()

RFC In review Philippe Mathieu-Daudé 3 participants 4 messages Jul 19, 2019

lore
Share & tools
lore
#0 lore raw

RFC 1/1 target/i386: Correct misplaced break statement in gen_shiftd_rm_T1()

Reported by GCC9 when building with CFLAG -Wimplicit-fallthrough=2:

CC target/i386/translate.o
target/i386/translate.c: In function ‘gen_shiftd_rm_T1’:
target/i386/translate.c:1785:12: error: this statement may fall through [-Werror=implicit-fallthrough=]
1785 | if (is_right) {
| ^
target/i386/translate.c:1810:5: note: here

   1810 |     default:
        |     ^~~~~~~

cc1: all warnings being treated as errors

Fixes f437d0a3c24
---
 target/i386/translate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
target/i386/translate.c +1 -1
index 03150a86e2..4b2b5937ca 100644
@@ -1805,8 +1805,8 @@ static void gen_shiftd_rm_T1(DisasContext *s, TCGMemOp ot, int op1,
             tcg_gen_shri_i64(s->tmp0, s->tmp0, 32);
             tcg_gen_shri_i64(s->T0, s->T0, 32);
         }
-        break;
 #endif
+        break;
     default:
         tcg_gen_subi_tl(s->tmp0, count, 1);
         if (is_right) {
Signature · 1 line
2.20.1
#1 lore raw
Quoted 24 lines · Philippe Mathieu-Daudé

Reported by GCC9 when building with CFLAG -Wimplicit-fallthrough=2:

CC target/i386/translate.o
target/i386/translate.c: In function ‘gen_shiftd_rm_T1’:
target/i386/translate.c:1785:12: error: this statement may fall through [-Werror=implicit-fallthrough=]
1785 | if (is_right) {
| ^
target/i386/translate.c:1810:5: note: here

   1810 |     default:
        |     ^~~~~~~

cc1: all warnings being treated as errors

Fixes f437d0a3c24
---
 target/i386/translate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
target/i386/translate.c +0 -0
index 03150a86e2..4b2b5937ca 100644

target/i386/translate.c static void gen_shiftd_rm_T1(DisasContext *s, TCGMemOp ot, int op1, · line 1812

Quoted 9 lines
target/i386/translate.c +1 -1
@@ -1805,8 +1805,8 @@ static void gen_shiftd_rm_T1(DisasContext *s, TCGMemOp ot, int op1,
             tcg_gen_shri_i64(s->tmp0, s->tmp0, 32);
             tcg_gen_shri_i64(s->T0, s->T0, 32);
         }
-        break;
 #endif
+        break;
     default:
         tcg_gen_subi_tl(s->tmp0, count, 1);
         if (is_right) {

I haven't looked closely at the code but I would guess that the fallthrough is intended, because the default label has an "ot == MO_16" condition.

It certainly needs more comments... :(

Paolo

#2 lore raw
Quoted 31 lines · Paolo Bonzini
Quoted 29 lines · Philippe Mathieu-Daudé

Reported by GCC9 when building with CFLAG -Wimplicit-fallthrough=2:

CC target/i386/translate.o
target/i386/translate.c: In function ‘gen_shiftd_rm_T1’:
target/i386/translate.c:1785:12: error: this statement may fall through [-Werror=implicit-fallthrough=]
1785 | if (is_right) {
| ^
target/i386/translate.c:1810:5: note: here

   1810 |     default:
        |     ^~~~~~~

cc1: all warnings being treated as errors

Fixes f437d0a3c24
---
 target/i386/translate.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)
target/i386/translate.c +1 -1
index 03150a86e2..4b2b5937ca 100644
@@ -1805,8 +1805,8 @@ static void gen_shiftd_rm_T1(DisasContext *s, TCGMemOp ot, int op1,
             tcg_gen_shri_i64(s->tmp0, s->tmp0, 32);
             tcg_gen_shri_i64(s->T0, s->T0, 32);
         }
-        break;
 #endif
+        break;
Quoted 8 lines

target/i386/translate.c static void gen_shiftd_rm_T1(DisasContext *s, TCGMemOp ot, int op1, · line 1812

Quoted 4 lines
target/i386/translate.c +0 -0
     default:
         tcg_gen_subi_tl(s->tmp0, count, 1);
         if (is_right) {

I haven't looked closely at the code but I would guess that the fallthrough is intended, because the default label has an "ot == MO_16" condition.

Yeah, this code is really weird -- if TARGET_X86_64 then MO_16 falls through into MO_32, but if only i386 then MO_16 falls through into the default case ?!?

thanks
-- PMM

#3 lore raw
Quoted 6 lines · Peter Maydell
Quoted 3 lines

I haven't looked closely at the code but I would guess that the fallthrough is intended, because the default label has an "ot == MO_16" condition.

Yeah, this code is really weird -- if TARGET_X86_64 then MO_16 falls through into MO_32, but if only i386 then MO_16 falls through into the default case ?!?

Yes, and in either case MO_16 falls through into the 32-bit code. However, the i386 32-bit version and the x86-64 64-bit version are unified into a single piece of code for TARGET_LONG_BITS-bit operands.

Almost, because you still need that ugly special case for MO_16 in the default label.

Paolo