[PULL 0/2] KVM: s390: last patches for next

[PULL 0/2] KVM: s390: last patches for next

[Christian Borntraeger]

Paolo,

here are two patches that are targetted against kvm/next for 3.15.
One is a bugfix for a performance regression of the ucontrol path,
but we can handle that via stable. No need to force that into 3.14.0
The other patch removes some BUG_ONs in inner loops that cant be
triggered anyway. Patches are against kvm/next

Christian


The following changes since commit 8fbb1daf3e8254afc17fc4490b69db00920197ae:

  Merge branch 'kvm-ppc-fix' into HEAD (2014-03-14 16:06:30 +0100)

are available in the git repository at:


  git://git.kernel.org/pub/scm/linux/kernel/git/kvms390/linux.git tags/kvm-s390-20140317

for you to fetch changes up to 2955c83f72801245afd0fe5c560cc75b82bea9aa:

  KVM: s390: Optimize ucontrol path (2014-03-17 11:06:51 +0100)

----------------------------------------------------------------
Two patches:
- one regression fix for reducing the amount of ucontrol userspace exits
- get rid of BUG_ONs in hot inner loops

----------------------------------------------------------------
Christian Borntraeger (1):
      KVM: s390: Optimize ucontrol path

Dominik Dingel (1):
      KVM: s390: Removing untriggerable BUG_ONs

 arch/s390/kvm/interrupt.c | 2 --
 arch/s390/kvm/kvm-s390.c  | 5 ++---
 2 files changed, 2 insertions(+), 5 deletions(-)

[PULL 1/2] KVM: s390: Removing untriggerable BUG_ONs

[Christian Borntraeger]

From: Dominik Dingel <dingel@linux.vnet.ibm.com>

The BUG_ON in kvm-s390.c is unreachable, as we get the vcpu per common code,
which itself does this from the private_data field of the file descriptor,
and there is no KVM_UNCREATE_VCPU.

The __{set,unset}_cpu_idle BUG_ONs are not triggerable because the vcpu
creation code already checks against KVM_MAX_VCPUS.

Signed-off-by: Dominik Dingel <dingel@linux.vnet.ibm.com>
Acked-by: Cornelia Huck <cornelia.huck@de.ibm.com>
Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
---
 arch/s390/kvm/interrupt.c | 2 --
 arch/s390/kvm/kvm-s390.c  | 2 --
 2 files changed, 4 deletions(-)

diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c
index 1d0f9d5..79d2e4f 100644
--- a/arch/s390/kvm/interrupt.c
+++ b/arch/s390/kvm/interrupt.c
@@ -114,14 +114,12 @@ static int __interrupt_is_deliverable(struct kvm_vcpu *vcpu,
 
 static void __set_cpu_idle(struct kvm_vcpu *vcpu)
 {
-	BUG_ON(vcpu->vcpu_id > KVM_MAX_VCPUS - 1);
 	atomic_set_mask(CPUSTAT_WAIT, &vcpu->arch.sie_block->cpuflags);
 	set_bit(vcpu->vcpu_id, vcpu->arch.local_int.float_int->idle_mask);
 }
 
 static void __unset_cpu_idle(struct kvm_vcpu *vcpu)
 {
-	BUG_ON(vcpu->vcpu_id > KVM_MAX_VCPUS - 1);
 	atomic_clear_mask(CPUSTAT_WAIT, &vcpu->arch.sie_block->cpuflags);
 	clear_bit(vcpu->vcpu_id, vcpu->arch.local_int.float_int->idle_mask);
 }
diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 9136f8d..6268357 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -949,8 +949,6 @@ int kvm_arch_vcpu_ioctl_run(struct kvm_vcpu *vcpu, struct kvm_run *kvm_run)
 
 	atomic_clear_mask(CPUSTAT_STOPPED, &vcpu->arch.sie_block->cpuflags);
 
-	BUG_ON(kvm_get_vcpu(vcpu->kvm, vcpu->vcpu_id) == NULL);
-
 	switch (kvm_run->exit_reason) {
 	case KVM_EXIT_S390_SIEIC:
 	case KVM_EXIT_UNKNOWN:
-- 
1.8.4.2

[PULL 2/2] KVM: s390: Optimize ucontrol path

[Christian Borntraeger]

Since commit 7c470539c95630c1f2a10f109e96f249730b75eb
(s390/kvm: avoid automatic sie reentry) we will run through the C code
of KVM on host interrupts instead of just reentering the guest. This
will result in additional ucontrol exits (at least HZ per second). Let
handle a 0 intercept in the kernel and dont return to userspace,
even if in ucontrol mode.

Signed-off-by: Christian Borntraeger <borntraeger@de.ibm.com>
Reviewed-by: Cornelia Huck <cornelia.huck@de.ibm.com>
CC: stable@vger.kernel.org
---
 arch/s390/kvm/kvm-s390.c | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c
index 6268357..7337c57 100644
--- a/arch/s390/kvm/kvm-s390.c
+++ b/arch/s390/kvm/kvm-s390.c
@@ -896,7 +896,8 @@ static int vcpu_post_run(struct kvm_vcpu *vcpu, int exit_reason)
 
 	if (rc == 0) {
 		if (kvm_is_ucontrol(vcpu->kvm))
-			rc = -EOPNOTSUPP;
+			/* Don't exit for host interrupts. */
+			rc = vcpu->arch.sie_block->icptcode ? -EOPNOTSUPP : 0;
 		else
 			rc = kvm_handle_sie_intercept(vcpu);
 	}
-- 
1.8.4.2

Re: [PULL 0/2] KVM: s390: last patches for next

[Paolo Bonzini]

Il 17/03/2014 12:13, Christian Borntraeger ha scritto:
> Paolo,
>
> here are two patches that are targetted against kvm/next for 3.15.
> One is a bugfix for a performance regression of the ucontrol path,
> but we can handle that via stable. No need to force that into 3.14.0
> The other patch removes some BUG_ONs in inner loops that cant be
> triggered anyway. Patches are against kvm/next
>
> Christian
>
>
> The following changes since commit 8fbb1daf3e8254afc17fc4490b69db00920197ae:
>
>   Merge branch 'kvm-ppc-fix' into HEAD (2014-03-14 16:06:30 +0100)
>
> are available in the git repository at:
>
>
>   git://git.kernel.org/pub/scm/linux/kernel/git/kvms390/linux.git tags/kvm-s390-20140317
>
> for you to fetch changes up to 2955c83f72801245afd0fe5c560cc75b82bea9aa:
>
>   KVM: s390: Optimize ucontrol path (2014-03-17 11:06:51 +0100)
>
> ----------------------------------------------------------------
> Two patches:
> - one regression fix for reducing the amount of ucontrol userspace exits
> - get rid of BUG_ONs in hot inner loops
>
> ----------------------------------------------------------------
> Christian Borntraeger (1):
>       KVM: s390: Optimize ucontrol path
>
> Dominik Dingel (1):
>       KVM: s390: Removing untriggerable BUG_ONs
>
>  arch/s390/kvm/interrupt.c | 2 --
>  arch/s390/kvm/kvm-s390.c  | 5 ++---
>  2 files changed, 2 insertions(+), 5 deletions(-)
>

Applied, thanks!

Paolo