[PATCH 1/2] target/s390x: Use mutable temporary value for op_ts

[PATCH 1/2] target/s390x: Use mutable temporary value for op_ts

[Ilya Leoshkevich]

From: Ido Plat <ido.plat@ibm.com>

Otherwise TCG would assume the register that holds t1 would be constant
and reuse whenever it needs the value within it.

Reviewed-by: Ilya Leoshkevich <iii@linux.ibm.com>
[iii: Adjust a newline and capitalization]
Signed-off-by: Ido Plat <ido.plat@ibm.com>
---
 target/s390x/tcg/translate.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/target/s390x/tcg/translate.c b/target/s390x/tcg/translate.c
index 0d0c672c959..3fdddac7684 100644
--- a/target/s390x/tcg/translate.c
+++ b/target/s390x/tcg/translate.c
@@ -4781,8 +4781,9 @@ static DisasJumpType op_trXX(DisasContext *s, DisasOps *o)
 
 static DisasJumpType op_ts(DisasContext *s, DisasOps *o)
 {
-    TCGv_i32 t1 = tcg_constant_i32(0xff);
+    TCGv_i32 t1 = tcg_temp_new_i32();
 
+    tcg_gen_movi_i32(t1, 0xff);
     tcg_gen_atomic_xchg_i32(t1, o->in2, t1, get_mem_index(s), MO_UB);
     tcg_gen_extract_i32(cc_op, t1, 7, 1);
     set_cc_static(s);
-- 
2.44.0

[PATCH 2/2] tests/tcg/s390x: Test TEST AND SET

[Ilya Leoshkevich]

Add a small test to prevent regressions.

Signed-off-by: Ilya Leoshkevich <iii@linux.ibm.com>
---
 tests/tcg/s390x/Makefile.target |  1 +
 tests/tcg/s390x/ts.c            | 35 +++++++++++++++++++++++++++++++++
 2 files changed, 36 insertions(+)
 create mode 100644 tests/tcg/s390x/ts.c

diff --git a/tests/tcg/s390x/Makefile.target b/tests/tcg/s390x/Makefile.target
index e2aba2ec274..a8f86c94498 100644
--- a/tests/tcg/s390x/Makefile.target
+++ b/tests/tcg/s390x/Makefile.target
@@ -47,6 +47,7 @@ TESTS+=add-logical-with-carry
 TESTS+=lae
 TESTS+=cvd
 TESTS+=cvb
+TESTS+=ts
 
 cdsg: CFLAGS+=-pthread
 cdsg: LDFLAGS+=-pthread
diff --git a/tests/tcg/s390x/ts.c b/tests/tcg/s390x/ts.c
new file mode 100644
index 00000000000..441faf30d98
--- /dev/null
+++ b/tests/tcg/s390x/ts.c
@@ -0,0 +1,35 @@
+/*
+ * Test the TEST AND SET instruction.
+ *
+ * SPDX-License-Identifier: GPL-2.0-or-later
+ */
+#include <assert.h>
+#include <stdlib.h>
+
+static int ts(char *p)
+{
+    int cc;
+
+    asm("ts %[p]\n"
+        "ipm %[cc]"
+        : [cc] "=r" (cc)
+        , [p] "+Q" (*p)
+        : : "cc");
+
+    return (cc >> 28) & 3;
+}
+
+int main(void)
+{
+    char c;
+
+    c = 0x80;
+    assert(ts(&c) == 1);
+    assert(c == 0xff);
+
+    c = 0x7f;
+    assert(ts(&c) == 0);
+    assert(c == 0xff);
+
+    return EXIT_SUCCESS;
+}
-- 
2.44.0

Re: [PATCH 1/2] target/s390x: Use mutable temporary value for op_ts

[Thomas Huth]

On 18/03/2024 17.26, Ilya Leoshkevich wrote:
> From: Ido Plat <ido.plat@ibm.com>
> 
> Otherwise TCG would assume the register that holds t1 would be constant
> and reuse whenever it needs the value within it.
> 
> Reviewed-by: Ilya Leoshkevich <iii@linux.ibm.com>
> [iii: Adjust a newline and capitalization]
> Signed-off-by: Ido Plat <ido.plat@ibm.com>

  Hi Ilya,

please add your Signed-off-by as well when the patch is sent out by you.

  Thanks,
   Thomas

Re: [PATCH 1/2] target/s390x: Use mutable temporary value for op_ts

[Richard Henderson]

On 3/18/24 06:26, Ilya Leoshkevich wrote:
> From: Ido Plat <ido.plat@ibm.com>
> 
> Otherwise TCG would assume the register that holds t1 would be constant
> and reuse whenever it needs the value within it.
> 
> Reviewed-by: Ilya Leoshkevich <iii@linux.ibm.com>
> [iii: Adjust a newline and capitalization]
> Signed-off-by: Ido Plat <ido.plat@ibm.com>
> ---
>   target/s390x/tcg/translate.c | 3 ++-
>   1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/target/s390x/tcg/translate.c b/target/s390x/tcg/translate.c
> index 0d0c672c959..3fdddac7684 100644
> --- a/target/s390x/tcg/translate.c
> +++ b/target/s390x/tcg/translate.c
> @@ -4781,8 +4781,9 @@ static DisasJumpType op_trXX(DisasContext *s, DisasOps *o)
>   
>   static DisasJumpType op_ts(DisasContext *s, DisasOps *o)
>   {
> -    TCGv_i32 t1 = tcg_constant_i32(0xff);
> +    TCGv_i32 t1 = tcg_temp_new_i32();
>   
> +    tcg_gen_movi_i32(t1, 0xff);
>       tcg_gen_atomic_xchg_i32(t1, o->in2, t1, get_mem_index(s), MO_UB);

Better as

     TCGv_i32 ff = tcg_constant_i32(0xff);
     TCGv_i32 t1 = tcg_temp_new_i32();

     tcg_gen_atomic_xchg_i32(t1, o->in2, ff, get_mem_index(s), MO_UB);

With that,
Reviewed-by: Richard Henderson <richard.henderson@linaro.org>


r~