The CTR_EL0 register has some bits which allow the implementation to
tell the guest that it does not need to do cache maintenance for
data-to-instruction coherence and instruction-to-data coherence.
QEMU doesn't emulate caches and so our cache maintenance insns are
all NOPs.
We already have some models of specific CPUs where we set these bits
(e.g. the Neoverse V1), but the 'max' CPU still uses the settings it
inherits from Cortex-A57. Set the bits for 'max' as well, so the
guest doesn't need to do unnecessary work.
Signed-off-by: Peter Maydell <peter.maydell@linaro.org>
---
This is worthwhile anyway; it also works around what Marc Z
and I think is a KVM bug where booting the L2 guest hangs
if L0 thinks it needs to do cache maintenance ops, when
running all this under QEMU's FEAT_NV/FEAT_NV2 emulation.
---
target/arm/tcg/cpu64.c | 10 ++++++++++
1 file changed, 10 insertions(+)
diff --git a/target/arm/tcg/cpu64.c b/target/arm/tcg/cpu64.c
index fcda99e1583..40e7a45166f 100644
--- a/target/arm/tcg/cpu64.c
+++ b/target/arm/tcg/cpu64.c
@@ -1105,6 +1105,16 @@ void aarch64_max_tcg_initfn(Object *obj)
u = FIELD_DP32(u, CLIDR_EL1, LOUU, 0);
cpu->clidr = u;
+ /*
+ * Set CTR_EL0.DIC and IDC to tell the guest it doesnt' need to
+ * do any cache maintenance for data-to-instruction or
+ * instruction-to-guest coherence. (Our cache ops are nops.)
+ */
+ t = cpu->ctr;
+ t = FIELD_DP64(t, CTR_EL0, IDC, 1);
+ t = FIELD_DP64(t, CTR_EL0, DIC, 1);
+ cpu->ctr = t;
+
t = cpu->isar.id_aa64isar0;
t = FIELD_DP64(t, ID_AA64ISAR0, AES, 2); /* FEAT_PMULL */
t = FIELD_DP64(t, ID_AA64ISAR0, SHA1, 1); /* FEAT_SHA1 */
--
2.34.1