[PATCH v4 31/45] target/arm: Reset streaming sve state on exception boundaries

[Richard Henderson <richard.henderson@linaro.org>]

We can handle both exception entry and exception return by
hooking into aarch64_sve_change_el.

Reviewed-by: Peter Maydell <peter.maydell@linaro.org>
Signed-off-by: Richard Henderson <richard.henderson@linaro.org>
---
 target/arm/helper.c | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/target/arm/helper.c b/target/arm/helper.c
index 976e414eda..2dc1f95c6d 100644
--- a/target/arm/helper.c
+++ b/target/arm/helper.c
@@ -11754,6 +11754,19 @@ void aarch64_sve_change_el(CPUARMState *env, int old_el,
         return;
     }
 
+    old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64;
+    new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64;
+
+    /*
+     * Both AArch64.TakeException and AArch64.ExceptionReturn
+     * invoke ResetSVEState when taking an exception from, or
+     * returning to, AArch32 state when PSTATE.SM is enabled.
+     */
+    if (old_a64 != new_a64 && FIELD_EX64(env->svcr, SVCR, SM)) {
+        arm_reset_sve_state(env);
+        return;
+    }
+
     /*
      * DDI0584A.d sec 3.2: "If SVE instructions are disabled or trapped
      * at ELx, or not available because the EL is in AArch32 state, then
@@ -11766,10 +11779,8 @@ void aarch64_sve_change_el(CPUARMState *env, int old_el,
      * we already have the correct register contents when encountering the
      * vq0->vq0 transition between EL0->EL1.
      */
-    old_a64 = old_el ? arm_el_is_aa64(env, old_el) : el0_a64;
     old_len = (old_a64 && !sve_exception_el(env, old_el)
                ? sve_vqm1_for_el(env, old_el) : 0);
-    new_a64 = new_el ? arm_el_is_aa64(env, new_el) : el0_a64;
     new_len = (new_a64 && !sve_exception_el(env, new_el)
                ? sve_vqm1_for_el(env, new_el) : 0);
 
-- 
2.34.1