Skip to contentSkip to conversation

sparc32 : Signed integer division overflow

PATCH In review Olivier Danet 2 participants 3 messages Mar 19, 2014 – Mar 20, 2014

lore
Share & tools
lore

Olivier Danet

author
#0 lore raw

PATCH v2 1/1 sparc32 : Signed integer division overflow

The signed integer division -0x8000_0000_0000_0000 / -1 must be handled separately to avoid an overflow on the QEMU host.

Signed-off-by Olivier Danet <[email protected]>
---
 target-sparc/helper.c | 13 ++++++++-----
 1 file changed, 8 insertions(+), 5 deletions(-)
target-sparc/helper.c +8 -5
index f3c7fbf..3822c48 100644
@@ -122,12 +122,15 @@ static target_ulong helper_sdiv_common(CPUSPARCState *env, target_ulong a,
     if (x1 == 0) {
         cpu_restore_state(CPU(cpu), GETPC());
         helper_raise_exception(env, TT_DIV_ZERO);
-    }
-
-    x0 = x0 / x1;
-    if ((int32_t) x0 != x0) {
-        x0 = x0 < 0 ? 0x80000000 : 0x7fffffff;
+    } else if (x1 == -1 && x0 == INT64_MIN) {
+        x0 = 0x7fffffff;
         overflow = 1;
+    } else {
+        x0 = x0 / x1;
+        if ((int32_t) x0 != x0) {
+            x0 = x0 < 0 ? 0x80000000 : 0x7fffffff;
+            overflow = 1;
+        }
     }
 
     if (cc) {
Signature · 1 line
1.8.1.5
#2 lore raw
In reply to Olivier Danet

target-sparc/helper.c static target_ulong helper_sdiv_common(CPUSPARCState *env, target_ulong a, · line 133

Quoted 4 lines · Olivier Danet
target-sparc/helper.c +4 -0
+        if ((int32_t) x0 != x0) {
+            x0 = x0 < 0 ? 0x80000000 : 0x7fffffff;
+            overflow = 1;
+        }

Actually, there's an existing bug here that we might as well fix.

## ... the 32-bit result is sign-extended to 64 bits and ## written into register R[rd].

That 0x80000000 is not going to promote properly to 64-bits. We might as well go ahead and use INT32_MIN and INT32_MAX here at the same time.

r~