[Qemu-devel] [PATCH 1/3] target-s390x: fix CC computation for LOAD POSITIVE instructions

[Qemu-devel] [PATCH 1/3] target-s390x: fix CC computation for LOAD POSITIVE instructions

[Aurelien Jarno]

LOAD POSITIVE instructions (LPR, LPGR and LPGFR) set the following
condition code:
  0: Result zero; no overflow
  1: --
  2: Result greater than zero; no overflow
  3: Overflow

The current code wrongly returns 1 instead of 2 in case of a result
greater than 0. This patches fixes that. This fixes the marshalling of
the value '0L' in Python.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
---
 target-s390x/cc_helper.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/target-s390x/cc_helper.c b/target-s390x/cc_helper.c
index 00bc883..bfce3f1 100644
--- a/target-s390x/cc_helper.c
+++ b/target-s390x/cc_helper.c
@@ -195,7 +195,7 @@ static uint32_t cc_calc_abs_64(int64_t dst)
     if ((uint64_t)dst == 0x8000000000000000ULL) {
         return 3;
     } else if (dst) {
-        return 1;
+        return 2;
     } else {
         return 0;
     }
@@ -296,7 +296,7 @@ static uint32_t cc_calc_abs_32(int32_t dst)
     if ((uint32_t)dst == 0x80000000UL) {
         return 3;
     } else if (dst) {
-        return 1;
+        return 2;
     } else {
         return 0;
     }
-- 
2.1.4

[Qemu-devel] [PATCH 2/3] target-s390x: optimize (negative-) abs computation

[Aurelien Jarno]

Now that movcond exists, it's easy to write (negative-) absolute value
using TCG code instead of an helper.

Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
---
 target-s390x/helper.h     |  2 --
 target-s390x/int_helper.c | 22 ----------------------
 target-s390x/translate.c  | 16 ++++++++++++++--
 3 files changed, 14 insertions(+), 26 deletions(-)

diff --git a/target-s390x/helper.h b/target-s390x/helper.h
index 8d2c859..c4ec3f1 100644
--- a/target-s390x/helper.h
+++ b/target-s390x/helper.h
@@ -17,8 +17,6 @@ DEF_HELPER_4(mvst, i64, env, i64, i64, i64)
 DEF_HELPER_5(ex, i32, env, i32, i64, i64, i64)
 DEF_HELPER_FLAGS_1(abs_i32, TCG_CALL_NO_RWG_SE, i32, s32)
 DEF_HELPER_FLAGS_1(nabs_i32, TCG_CALL_NO_RWG_SE, s32, s32)
-DEF_HELPER_FLAGS_1(abs_i64, TCG_CALL_NO_RWG_SE, i64, s64)
-DEF_HELPER_FLAGS_1(nabs_i64, TCG_CALL_NO_RWG_SE, s64, s64)
 DEF_HELPER_FLAGS_4(stam, TCG_CALL_NO_WG, void, env, i32, i64, i32)
 DEF_HELPER_FLAGS_4(lam, TCG_CALL_NO_WG, void, env, i32, i64, i32)
 DEF_HELPER_4(mvcle, i32, env, i32, i64, i32)
diff --git a/target-s390x/int_helper.c b/target-s390x/int_helper.c
index cb8dd98..f53723d 100644
--- a/target-s390x/int_helper.c
+++ b/target-s390x/int_helper.c
@@ -135,28 +135,6 @@ int32_t HELPER(nabs_i32)(int32_t val)
     }
 }
 
-/* absolute value 64-bit */
-uint64_t HELPER(abs_i64)(int64_t val)
-{
-    HELPER_LOG("%s: val 0x%" PRIx64 "\n", __func__, val);
-
-    if (val < 0) {
-        return -val;
-    } else {
-        return val;
-    }
-}
-
-/* negative absolute value 64-bit */
-int64_t HELPER(nabs_i64)(int64_t val)
-{
-    if (val < 0) {
-        return val;
-    } else {
-        return -val;
-    }
-}
-
 /* count leading zeros, for find leftmost one */
 uint64_t HELPER(clz)(uint64_t v)
 {
diff --git a/target-s390x/translate.c b/target-s390x/translate.c
index c297938..5898adb 100644
--- a/target-s390x/translate.c
+++ b/target-s390x/translate.c
@@ -1304,7 +1304,13 @@ static ExitStatus help_branch(DisasContext *s, DisasCompare *c,
 
 static ExitStatus op_abs(DisasContext *s, DisasOps *o)
 {
-    gen_helper_abs_i64(o->out, o->in2);
+    TCGv_i64 z, n;
+    z = tcg_const_i64(0);
+    n = tcg_temp_new_i64();
+    tcg_gen_neg_i64(n, o->in2);
+    tcg_gen_movcond_i64(TCG_COND_LT, o->out, o->in2, z, n, o->in2);
+    tcg_temp_free_i64(n);
+    tcg_temp_free_i64(z);
     return NO_EXIT;
 }
 
@@ -2675,7 +2681,13 @@ static ExitStatus op_msdb(DisasContext *s, DisasOps *o)
 
 static ExitStatus op_nabs(DisasContext *s, DisasOps *o)
 {
-    gen_helper_nabs_i64(o->out, o->in2);
+    TCGv_i64 z, n;
+    z = tcg_const_i64(0);
+    n = tcg_temp_new_i64();
+    tcg_gen_neg_i64(n, o->in2);
+    tcg_gen_movcond_i64(TCG_COND_GE, o->out, o->in2, z, n, o->in2);
+    tcg_temp_free_i64(n);
+    tcg_temp_free_i64(z);
     return NO_EXIT;
 }
 
-- 
2.1.4

Re: [Qemu-devel] [PATCH 2/3] target-s390x: optimize (negative-) abs computation

[Richard Henderson]

On 05/18/2015 06:39 AM, Aurelien Jarno wrote:
> Now that movcond exists, it's easy to write (negative-) absolute value
> using TCG code instead of an helper.
> 
> Cc: Richard Henderson <rth@twiddle.net>
> Cc: Alexander Graf <agraf@suse.de>
> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
> ---
>  target-s390x/helper.h     |  2 --
>  target-s390x/int_helper.c | 22 ----------------------
>  target-s390x/translate.c  | 16 ++++++++++++++--
>  3 files changed, 14 insertions(+), 26 deletions(-)


Reviewed-by: Richard Henderson <rth@twiddle.net>


r~

[Qemu-devel] [PATCH 3/3] target-s390x: remove unused helpers

[Aurelien Jarno]

Cc: Richard Henderson <rth@twiddle.net>
Cc: Alexander Graf <agraf@suse.de>
Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
---
 target-s390x/helper.h     |  2 --
 target-s390x/int_helper.c | 20 --------------------
 2 files changed, 22 deletions(-)

diff --git a/target-s390x/helper.h b/target-s390x/helper.h
index c4ec3f1..e6f2afb 100644
--- a/target-s390x/helper.h
+++ b/target-s390x/helper.h
@@ -15,8 +15,6 @@ DEF_HELPER_4(clst, i64, env, i64, i64, i64)
 DEF_HELPER_4(mvpg, void, env, i64, i64, i64)
 DEF_HELPER_4(mvst, i64, env, i64, i64, i64)
 DEF_HELPER_5(ex, i32, env, i32, i64, i64, i64)
-DEF_HELPER_FLAGS_1(abs_i32, TCG_CALL_NO_RWG_SE, i32, s32)
-DEF_HELPER_FLAGS_1(nabs_i32, TCG_CALL_NO_RWG_SE, s32, s32)
 DEF_HELPER_FLAGS_4(stam, TCG_CALL_NO_WG, void, env, i32, i64, i32)
 DEF_HELPER_FLAGS_4(lam, TCG_CALL_NO_WG, void, env, i32, i64, i32)
 DEF_HELPER_4(mvcle, i32, env, i32, i64, i32)
diff --git a/target-s390x/int_helper.c b/target-s390x/int_helper.c
index f53723d..2c2b3f6 100644
--- a/target-s390x/int_helper.c
+++ b/target-s390x/int_helper.c
@@ -115,26 +115,6 @@ uint64_t HELPER(divu64)(CPUS390XState *env, uint64_t ah, uint64_t al,
     return ret;
 }
 
-/* absolute value 32-bit */
-uint32_t HELPER(abs_i32)(int32_t val)
-{
-    if (val < 0) {
-        return -val;
-    } else {
-        return val;
-    }
-}
-
-/* negative absolute value 32-bit */
-int32_t HELPER(nabs_i32)(int32_t val)
-{
-    if (val < 0) {
-        return val;
-    } else {
-        return -val;
-    }
-}
-
 /* count leading zeros, for find leftmost one */
 uint64_t HELPER(clz)(uint64_t v)
 {
-- 
2.1.4

Re: [Qemu-devel] [PATCH 3/3] target-s390x: remove unused helpers

[Richard Henderson]

On 05/18/2015 06:40 AM, Aurelien Jarno wrote:
> Cc: Richard Henderson <rth@twiddle.net>
> Cc: Alexander Graf <agraf@suse.de>
> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
> ---
>  target-s390x/helper.h     |  2 --
>  target-s390x/int_helper.c | 20 --------------------
>  2 files changed, 22 deletions(-)


Reviewed-by: Richard Henderson <rth@twiddle.net>


r~

Re: [Qemu-devel] [PATCH 1/3] target-s390x: fix CC computation for LOAD POSITIVE instructions

[Richard Henderson]

On 05/18/2015 06:39 AM, Aurelien Jarno wrote:
> LOAD POSITIVE instructions (LPR, LPGR and LPGFR) set the following
> condition code:
>   0: Result zero; no overflow
>   1: --
>   2: Result greater than zero; no overflow
>   3: Overflow
> 
> The current code wrongly returns 1 instead of 2 in case of a result
> greater than 0. This patches fixes that. This fixes the marshalling of
> the value '0L' in Python.
> 
> Cc: Richard Henderson <rth@twiddle.net>
> Cc: Alexander Graf <agraf@suse.de>
> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>
> ---
>  target-s390x/cc_helper.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)

Reviewed-by: Richard Henderson <rth@twiddle.net>


r~

Re: [Qemu-devel] [PATCH 1/3] target-s390x: fix CC computation for LOAD POSITIVE instructions

[Alexander Graf]

On 05/18/2015 03:39 PM, Aurelien Jarno wrote:
> LOAD POSITIVE instructions (LPR, LPGR and LPGFR) set the following
> condition code:
>    0: Result zero; no overflow
>    1: --
>    2: Result greater than zero; no overflow
>    3: Overflow
>
> The current code wrongly returns 1 instead of 2 in case of a result
> greater than 0. This patches fixes that. This fixes the marshalling of
> the value '0L' in Python.
>
> Cc: Richard Henderson <rth@twiddle.net>
> Cc: Alexander Graf <agraf@suse.de>
> Signed-off-by: Aurelien Jarno <aurelien@aurel32.net>

Thanks, applied all to s390-next.


Alex