diff mbox series

[21/21] softfloat: Remove fallback rule from pickNaN()

Message ID 20241025141254.2141506-22-peter.maydell@linaro.org
State New
Headers show
Series softfloat: Set 2-NaN propagation rule in float_status, not at compile time | expand

Commit Message

Peter Maydell Oct. 25, 2024, 2:12 p.m. UTC 
Now that all targets have been converted to explicitly set a NaN
propagation rule, we can remove the set of target ifdefs (which now
list every target) and clean up the references to fallback behaviour
for float_2nan_prop_none.

The "default" case in the switch will catch any remaining places
where status->float_2nan_prop_rule was not set by the target.

Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
 include/fpu/softfloat-types.h  | 10 +++-------
 fpu/softfloat-specialize.c.inc | 23 +++--------------------
 2 files changed, 6 insertions(+), 27 deletions(-)

Comments

Philippe Mathieu-Daudé Oct. 26, 2024, 4:31 a.m. UTC | #1 
On 25/10/24 11:12, Peter Maydell wrote:
> Now that all targets have been converted to explicitly set a NaN
> propagation rule, we can remove the set of target ifdefs (which now
> list every target) and clean up the references to fallback behaviour
> for float_2nan_prop_none.
> 
> The "default" case in the switch will catch any remaining places
> where status->float_2nan_prop_rule was not set by the target.
> 
> Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
> ---
>   include/fpu/softfloat-types.h  | 10 +++-------
>   fpu/softfloat-specialize.c.inc | 23 +++--------------------
>   2 files changed, 6 insertions(+), 27 deletions(-)

Reviewed-by: Philippe Mathieu-Daudé <philmd@linaro.org>
Richard Henderson Oct. 28, 2024, 12:28 p.m. UTC | #2 
On 10/25/24 15:12, Peter Maydell wrote:
> Now that all targets have been converted to explicitly set a NaN
> propagation rule, we can remove the set of target ifdefs (which now
> list every target) and clean up the references to fallback behaviour
> for float_2nan_prop_none.
> 
> The "default" case in the switch will catch any remaining places
> where status->float_2nan_prop_rule was not set by the target.
> 
> Signed-off-by: Peter Maydell<peter.maydell@linaro.org>
> ---
>   include/fpu/softfloat-types.h  | 10 +++-------
>   fpu/softfloat-specialize.c.inc | 23 +++--------------------
>   2 files changed, 6 insertions(+), 27 deletions(-)

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

r~
diff mbox series

Patch

diff --git a/include/fpu/softfloat-types.h b/include/fpu/softfloat-types.h
index 5cd5a0d0ae1..8f39691dfd0 100644
--- a/include/fpu/softfloat-types.h
+++ b/include/fpu/softfloat-types.h
@@ -178,13 +178,9 @@  typedef enum __attribute__((__packed__)) {
  * If default_nan_mode is enabled then it is valid not to set a
  * NaN propagation rule, because the softfloat code guarantees
  * not to try to pick a NaN to propagate in default NaN mode.
- *
- * For transition, currently the 'none' rule will cause us to
- * fall back to picking the propagation rule based on the existing
- * ifdef ladder. When all targets are converted it will be an error
- * not to set the rule in float_status unless in default_nan_mode,
- * and we will assert if we need to handle an input NaN and no
- * rule was selected.
+ * When not in default-NaN mode, it is an error for the target
+ * not to set the rule in float_status, and we will assert if
+ * we need to handle an input NaN and no rule was selected.
  */
 typedef enum __attribute__((__packed__)) {
     /* No propagation rule specified */
diff --git a/fpu/softfloat-specialize.c.inc b/fpu/softfloat-specialize.c.inc
index 254bbd67168..b5a32080505 100644
--- a/fpu/softfloat-specialize.c.inc
+++ b/fpu/softfloat-specialize.c.inc
@@ -390,32 +390,15 @@  bool float32_is_signaling_nan(float32 a_, float_status *status)
 static int pickNaN(FloatClass a_cls, FloatClass b_cls,
                    bool aIsLargerSignificand, float_status *status)
 {
-    Float2NaNPropRule rule = status->float_2nan_prop_rule;
-
     /*
      * We guarantee not to require the target to tell us how to
      * pick a NaN if we're always returning the default NaN.
+     * But if we're not in default-NaN mode then the target must
+     * specify via set_float_2nan_prop_rule().
      */
     assert(!status->default_nan_mode);
 
-    if (rule == float_2nan_prop_none) {
-        /* target didn't set the rule: fall back to old ifdef choices */
-#if defined(TARGET_AVR) || defined(TARGET_HEXAGON) \
-    || defined(TARGET_RISCV) || defined(TARGET_SH4) \
-    || defined(TARGET_TRICORE) || defined(TARGET_ARM) || defined(TARGET_MIPS) \
-    || defined(TARGET_LOONGARCH64) || defined(TARGET_HPPA) \
-    || defined(TARGET_S390X) || defined(TARGET_PPC) || defined(TARGET_M68K) \
-    || defined(TARGET_SPARC) || defined(TARGET_XTENSA) \
-    || defined(TARGET_I386) || defined(TARGET_ALPHA) \
-    || defined(TARGET_MICROBLAZE) || defined(TARGET_OPENRISC) \
-    || defined(TARGET_RX)
-        g_assert_not_reached();
-#else
-        rule = float_2nan_prop_x87;
-#endif
-    }
-
-    switch (rule) {
+    switch (status->float_2nan_prop_rule) {
     case float_2nan_prop_s_ab:
         if (is_snan(a_cls)) {
             return 0;