[PATCH v2 10/10] whpx: x86: optimise MMIO vmexit by only restoring needed registers

[Mohamed Mediouni]

Signed-off-by: Mohamed Mediouni <moha...@unpredictable.fr>
---
 include/system/whpx-accel-ops.h |  8 +++++---
 target/i386/whpx/whpx-all.c     | 12 ++++++++++--
 2 files changed, 15 insertions(+), 5 deletions(-)

diff --git a/include/system/whpx-accel-ops.h b/include/system/whpx-accel-ops.h
index ed9d4c49f4..469a3a4350 100644
--- a/include/system/whpx-accel-ops.h
+++ b/include/system/whpx-accel-ops.h
@@ -22,11 +22,13 @@ void whpx_cpu_synchronize_post_reset(CPUState *cpu);
 void whpx_cpu_synchronize_post_init(CPUState *cpu);
 void whpx_cpu_synchronize_pre_loadvm(CPUState *cpu);
 
+/* subset of runtime state for faster returns from vmexit */
+#define WHPX_SET_FAST_RUNTIME_STATE   1
 /* state subset only touched by the VCPU itself during runtime */
-#define WHPX_SET_RUNTIME_STATE   1
+#define WHPX_SET_RUNTIME_STATE   2
 /* state subset modified during VCPU reset */
-#define WHPX_SET_RESET_STATE     2
+#define WHPX_SET_RESET_STATE     3
 /* full state set, modified during initialization or on vmload */
-#define WHPX_SET_FULL_STATE      3
+#define WHPX_SET_FULL_STATE      4
 
 #endif /* TARGET_I386_WHPX_ACCEL_OPS_H */
diff --git a/target/i386/whpx/whpx-all.c b/target/i386/whpx/whpx-all.c
index 0e8fb0e72e..61251a2835 100644
--- a/target/i386/whpx/whpx-all.c
+++ b/target/i386/whpx/whpx-all.c
@@ -420,6 +420,13 @@ void whpx_set_registers(CPUState *cpu, int level)
         vcxt.values[idx].Segment = whpx_seg_q2h(&env->segs[i], v86, r86);
     }
 
+    /*
+     * This is a hot path called on every MMIO access.
+     */
+    if (level <= WHPX_SET_FAST_RUNTIME_STATE) {
+        goto skip_to_set_registers;
+    }
+
     assert(idx == WHvX64RegisterLdtr);
     vcxt.values[idx++].Segment = whpx_seg_q2h(&env->ldt, 0, 0);
 
@@ -529,10 +536,11 @@ void whpx_set_registers(CPUState *cpu, int level)
 
     assert(idx == RTL_NUMBER_OF(whpx_register_names));
 
+    skip_to_set_registers:
     hr = whp_dispatch.WHvSetVirtualProcessorRegisters(
         whpx->partition, cpu->cpu_index,
         whpx_register_names,
-        RTL_NUMBER_OF(whpx_register_names),
+        idx,
         &vcxt.values[0]);
 
     if (FAILED(hr)) {
@@ -771,7 +779,7 @@ static int emulate_instruction(CPUState *cpu, const uint8_t 
*insn_bytes, size_t
     whpx_get_registers(cpu);
     decode_instruction_stream(env, &decode, &stream);
     exec_instruction(env, &decode);
-    whpx_set_registers(cpu, WHPX_SET_RUNTIME_STATE);
+    whpx_set_registers(cpu, WHPX_SET_FAST_RUNTIME_STATE);
 
     return 0;
 }
-- 
2.50.1 (Apple Git-155)