Re: [PATCH v2 1/8] kvm: Set cpu_single_env only once
On 2012-02-11 11:02, Blue Swirl wrote:
On Fri, Feb 10, 2012 at 18:31, Jan Kiszka jan.kis...@siemens.com wrote:
As we have thread-local cpu_single_env now and KVM uses exactly one
thread per VCPU, we can drop the cpu_single_env updates from the loop
and initialize this variable only once during setup.
I don't think this is correct. Maybe you missed the part that sets
cpu_single_env to NULL, which I think is to annoy broken code that
assumes that some CPU state is always globally available. This is not
true for monitor context.
I did check this before changing, and I see no such need. Particularly
as this old debugging help prevents valid use case.
Jan
Signed-off-by: Jan Kiszka jan.kis...@siemens.com
---
cpus.c|1 +
kvm-all.c |5 -
2 files changed, 1 insertions(+), 5 deletions(-)
diff --git a/cpus.c b/cpus.c
index f45a438..d0c8340 100644
--- a/cpus.c
+++ b/cpus.c
@@ -714,6 +714,7 @@ static void *qemu_kvm_cpu_thread_fn(void *arg)
qemu_mutex_lock(qemu_global_mutex);
qemu_thread_get_self(env-thread);
env-thread_id = qemu_get_thread_id();
+cpu_single_env = env;
r = kvm_init_vcpu(env);
if (r 0) {
diff --git a/kvm-all.c b/kvm-all.c
index c4babda..e2cbc03 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -1118,8 +1118,6 @@ int kvm_cpu_exec(CPUState *env)
return EXCP_HLT;
}
-cpu_single_env = env;
-
do {
if (env-kvm_vcpu_dirty) {
kvm_arch_put_registers(env, KVM_PUT_RUNTIME_STATE);
@@ -1136,13 +1134,11 @@ int kvm_cpu_exec(CPUState *env)
*/
qemu_cpu_kick_self();
}
-cpu_single_env = NULL;
qemu_mutex_unlock_iothread();
run_ret = kvm_vcpu_ioctl(env, KVM_RUN, 0);
qemu_mutex_lock_iothread();
-cpu_single_env = env;
kvm_arch_post_run(env, run);
kvm_flush_coalesced_mmio_buffer();
@@ -1206,7 +1202,6 @@ int kvm_cpu_exec(CPUState *env)
}
env-exit_request = 0;
-cpu_single_env = NULL;
return ret;
}
--
1.7.3.4
signature.asc
Description: OpenPGP digital signature
Re: [PATCH v2 1/8] kvm: Set cpu_single_env only once
On Sat, Feb 11, 2012 at 10:06, Jan Kiszka jan.kis...@web.de wrote:
On 2012-02-11 11:02, Blue Swirl wrote:
On Fri, Feb 10, 2012 at 18:31, Jan Kiszka jan.kis...@siemens.com wrote:
As we have thread-local cpu_single_env now and KVM uses exactly one
thread per VCPU, we can drop the cpu_single_env updates from the loop
and initialize this variable only once during setup.
I don't think this is correct. Maybe you missed the part that sets
cpu_single_env to NULL, which I think is to annoy broken code that
assumes that some CPU state is always globally available. This is not
true for monitor context.
I did check this before changing, and I see no such need. Particularly
as this old debugging help prevents valid use case.
It looks like monitor code is safe now. But in several places there
are checks like this in pc.c:
DeviceState *cpu_get_current_apic(void)
{
if (cpu_single_env) {
return cpu_single_env-apic_state;
} else {
return NULL;
}
}
In cpu-exec.c, there are these lines:
/* fail safe : never use cpu_single_env outside cpu_exec() */
cpu_single_env = NULL;
I think using cpu_single_env is an indication of a problem, like poor
code, layering violation or poor API (vmport). What is your use case?
Jan
Signed-off-by: Jan Kiszka jan.kis...@siemens.com
---
cpus.c | 1 +
kvm-all.c | 5 -
2 files changed, 1 insertions(+), 5 deletions(-)
diff --git a/cpus.c b/cpus.c
index f45a438..d0c8340 100644
--- a/cpus.c
+++ b/cpus.c
@@ -714,6 +714,7 @@ static void *qemu_kvm_cpu_thread_fn(void *arg)
qemu_mutex_lock(qemu_global_mutex);
qemu_thread_get_self(env-thread);
env-thread_id = qemu_get_thread_id();
+ cpu_single_env = env;
r = kvm_init_vcpu(env);
if (r 0) {
diff --git a/kvm-all.c b/kvm-all.c
index c4babda..e2cbc03 100644
--- a/kvm-all.c
+++ b/kvm-all.c
@@ -1118,8 +1118,6 @@ int kvm_cpu_exec(CPUState *env)
return EXCP_HLT;
}
- cpu_single_env = env;
-
do {
if (env-kvm_vcpu_dirty) {
kvm_arch_put_registers(env, KVM_PUT_RUNTIME_STATE);
@@ -1136,13 +1134,11 @@ int kvm_cpu_exec(CPUState *env)
*/
qemu_cpu_kick_self();
}
- cpu_single_env = NULL;
qemu_mutex_unlock_iothread();
run_ret = kvm_vcpu_ioctl(env, KVM_RUN, 0);
qemu_mutex_lock_iothread();
- cpu_single_env = env;
kvm_arch_post_run(env, run);
kvm_flush_coalesced_mmio_buffer();
@@ -1206,7 +1202,6 @@ int kvm_cpu_exec(CPUState *env)
}
env-exit_request = 0;
- cpu_single_env = NULL;
return ret;
}
--
1.7.3.4
--
To unsubscribe from this list: send the line unsubscribe kvm in
the body of a message to majord...@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH v2 1/8] kvm: Set cpu_single_env only once
On 2012-02-11 12:25, Blue Swirl wrote:
On Sat, Feb 11, 2012 at 10:06, Jan Kiszka jan.kis...@web.de wrote:
On 2012-02-11 11:02, Blue Swirl wrote:
On Fri, Feb 10, 2012 at 18:31, Jan Kiszka jan.kis...@siemens.com wrote:
As we have thread-local cpu_single_env now and KVM uses exactly one
thread per VCPU, we can drop the cpu_single_env updates from the loop
and initialize this variable only once during setup.
I don't think this is correct. Maybe you missed the part that sets
cpu_single_env to NULL, which I think is to annoy broken code that
assumes that some CPU state is always globally available. This is not
true for monitor context.
I did check this before changing, and I see no such need. Particularly
as this old debugging help prevents valid use case.
It looks like monitor code is safe now. But in several places there
are checks like this in pc.c:
DeviceState *cpu_get_current_apic(void)
{
if (cpu_single_env) {
return cpu_single_env-apic_state;
} else {
return NULL;
}
}
In cpu-exec.c, there are these lines:
/* fail safe : never use cpu_single_env outside cpu_exec() */
cpu_single_env = NULL;
That's legacy stuff from the pre-io-thread times. Nowadays,
cpu_single_env is logically either always valid (KVM) or switching
seamlessly between the VCPUs (TCG).
I think using cpu_single_env is an indication of a problem, like poor
code, layering violation or poor API (vmport). What is your use case?
We have a few poor ABIs like the VMware stuff or the KVM VAPIC, so we
have to live with it already for that use cases. Moreover, cpus.c use it
internally, like for vm_stop, to find out the caller context.
Jan
signature.asc
Description: OpenPGP digital signature
