[Qemu-devel] [PATCH 8/9] KVM: Rework of guest debug state writing

[Jan Kiszka]

So far we synchronized any dirty VCPU state back into the kernel before
updating the guest debug state. This was a tribute to a deficite in x86
kernels before 2.6.33. But as this is an arch-dependent issue, it is
better handle in the x86 part of KVM and remove the writeback point for
generic code. This also avoids overwriting the flushed state later on if
user space decides to change some more registers before resuming the
guest.

We furthermore need to reinject guest exceptions via the appropriate
mechanism. That is KVM_SET_GUEST_DEBUG for older kernels and
KVM_SET_VCPU_EVENTS for recent ones. Using both mechanisms at the same
time will cause state corruptions.


Signed-off-by: Jan Kiszka <address@hidden>
---
 kvm-all.c         |   12 ++++--------
 target-i386/kvm.c |   37 +++++++++++++++++++++++++++++++++----
 2 files changed, 37 insertions(+), 12 deletions(-)

diff --git a/kvm-all.c b/kvm-all.c
index 2237239..f0c5cf6 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -994,10 +994,6 @@ static void kvm_invoke_set_guest_debug(void *data)
     struct kvm_set_guest_debug_data *dbg_data = data;
     CPUState *env = dbg_data->env;
 
-    if (env->kvm_vcpu_dirty) {
-        kvm_arch_put_registers(env);
-        env->kvm_vcpu_dirty = 0;
-    }
     dbg_data->err = kvm_vcpu_ioctl(env, KVM_SET_GUEST_DEBUG, &dbg_data->dbg);
 }
 
@@ -1005,12 +1001,12 @@ int kvm_update_guest_debug(CPUState *env, unsigned long 
reinject_trap)
 {
     struct kvm_set_guest_debug_data data;
 
-    data.dbg.control = 0;
-    if (env->singlestep_enabled)
-        data.dbg.control = KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_SINGLESTEP;
+    data.dbg.control = reinject_trap;
 
+    if (env->singlestep_enabled) {
+        data.dbg.control |= KVM_GUESTDBG_ENABLE | KVM_GUESTDBG_SINGLESTEP;
+    }
     kvm_arch_update_guest_debug(env, &data.dbg);
-    data.dbg.control |= reinject_trap;
     data.env = env;
 
     on_vcpu(env, kvm_invoke_set_guest_debug, &data);
diff --git a/target-i386/kvm.c b/target-i386/kvm.c
index 8743f32..a4130d4 100644
--- a/target-i386/kvm.c
+++ b/target-i386/kvm.c
@@ -885,6 +885,32 @@ int kvm_arch_put_registers(CPUState *env)
     if (ret < 0)
         return ret;
 
+    /*
+     * Kernels before 2.6.33 (which correlates with !kvm_has_vcpu_events())
+     * overwrote flags.TF injected via SET_GUEST_DEBUG while updating GP regs.
+     * Work around this by updating the debug state once again if
+     * single-stepping is on.
+     * Another reason to call kvm_update_guest_debug here is a pending debug
+     * trap raise by the guest. On kernels without SET_VCPU_EVENTS we have to
+     * reinject them via SET_GUEST_DEBUG.
+     */
+    if (!kvm_has_vcpu_events() &&
+        (env->exception_injected != -1 || env->singlestep_enabled)) {
+        unsigned long reinject_trap = 0;
+
+        if (env->exception_injected == 1) {
+            reinject_trap = KVM_GUESTDBG_INJECT_DB;
+        } else if (env->exception_injected == 3) {
+            reinject_trap = KVM_GUESTDBG_INJECT_BP;
+        }
+        env->exception_injected = -1;
+
+        ret = kvm_update_guest_debug(env, reinject_trap);
+        if (ret < 0) {
+            return ret;
+        }
+    }
+
     return 0;
 }
 
@@ -1130,10 +1156,13 @@ int kvm_arch_debug(struct kvm_debug_exit_arch 
*arch_info)
     } else if (kvm_find_sw_breakpoint(cpu_single_env, arch_info->pc))
         handle = 1;
 
-    if (!handle)
-        kvm_update_guest_debug(cpu_single_env,
-                        (arch_info->exception == 1) ?
-                        KVM_GUESTDBG_INJECT_DB : KVM_GUESTDBG_INJECT_BP);
+    if (!handle) {
+        cpu_synchronize_state(cpu_single_env);
+        assert(cpu_single_env->exception_injected == -1);
+
+        cpu_single_env->exception_injected = arch_info->exception;
+        cpu_single_env->has_error_code = 0;
+    }
 
     return handle;
 }
-- 
1.6.0.2