While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array") Cc: stable@vger.kernel.org # 3.15+ --- arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS); + /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv; diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT); - set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask); + set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT); - clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask); + clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len)
static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) { - int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus); + int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { - vcpu = kvm_get_vcpu(kvm, vcpu_id); + for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) { + vcpu = kvm_get_vcpu(kvm, vcpu_idx); if (psw_ioint_disabled(vcpu)) continue; deliverable_mask &= (u8)(vcpu->arch.sie_block->gcr[6] >> 24); if (deliverable_mask) { /* lately kicked but not yet running */ - if (test_and_set_bit(vcpu_id, gi->kicked_mask)) + if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return; diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask); + clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu)
static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) { - return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask); + return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
On 27.08.21 14:54, Halil Pasic wrote:
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array") Cc: stable@vger.kernel.org # 3.15+
Reviewed-by: Christian Bornträger borntraeger@de.ibm.com Will apply and wait for the nightly run.
arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS);
- /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv;
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT);
- set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT);
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len) static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) {
- int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus);
- int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_id);
- for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_idx);
if (test_and_set_bit(vcpu_id, gi->kicked_mask))
if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return;diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu) static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) {
- return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic pasic@linux.ibm.com wrote:
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why vcpu_idx and vcpu_id can be different, namely: KVM decides the vcpu_idx, userspace decides the vcpu_id, thus the two might not match
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id);
you can also add a sentence to clarify that kvm_get_vcpu expects an vcpu_idx, not an vcpu_id.
do_stuff(vcpu);} is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")
otherwise looks good to me.
Reviewed-by: Claudio Imbrenda imbrenda@linux.ibm.com
Cc: stable@vger.kernel.org # 3.15+
arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS);
- /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv;
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT);
- set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask);
} static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT);
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask);
} static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len) static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) {
- int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus);
- int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_id);
- for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_idx);
if (test_and_set_bit(vcpu_id, gi->kicked_mask))
if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return;diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu) static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) {
- return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask);
} static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
On 27.08.21 16:06, Claudio Imbrenda wrote:
On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic pasic@linux.ibm.com wrote:
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why vcpu_idx and vcpu_id can be different, namely: KVM decides the vcpu_idx, userspace decides the vcpu_id, thus the two might not match
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id);
you can also add a sentence to clarify that kvm_get_vcpu expects an vcpu_idx, not an vcpu_id.
do_stuff(vcpu);
I will modify the patch description accordingly before sending to Paolo. Thanks for noticing.
} is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")
otherwise looks good to me.
Reviewed-by: Claudio Imbrenda imbrenda@linux.ibm.com
Cc: stable@vger.kernel.org # 3.15+
arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS);
- /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv;
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT);
- set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT);
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len) static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) {
- int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus);
- int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_id);
- for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_idx);
if (test_and_set_bit(vcpu_id, gi->kicked_mask))
if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return;diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu) static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) {
- return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
On Fri, 27 Aug 2021 18:36:48 +0200 Christian Borntraeger borntraeger@de.ibm.com wrote:
On 27.08.21 16:06, Claudio Imbrenda wrote:
On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic pasic@linux.ibm.com wrote:
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true,
s/vcp_id/vcpu_id/
it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why vcpu_idx and vcpu_id can be different, namely: KVM decides the vcpu_idx, userspace decides the vcpu_id, thus the two might not match
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id);
you can also add a sentence to clarify that kvm_get_vcpu expects an vcpu_idx, not an vcpu_id.
do_stuff(vcpu);I will modify the patch description accordingly before sending to Paolo. Thanks for noticing.
Can you also please fix the typo I pointed out above (in the first line of the long description).
Thanks! Halil
On Fri, 27 Aug 2021 16:06:16 +0200 Claudio Imbrenda imbrenda@linux.ibm.com wrote:
On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic pasic@linux.ibm.com wrote:
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true,
s/vcp_id/vcpu_id/
it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why vcpu_idx and vcpu_id can be different, namely: KVM decides the vcpu_idx, userspace decides the vcpu_id, thus the two might not match
Not sure that is a good explanation. A quote from Documentation/virt/kvm/api.rst:
""" 4.7 KVM_CREATE_VCPU -------------------
:Capability: basic :Architectures: all :Type: vm ioctl :Parameters: vcpu id (apic id on x86) :Returns: vcpu fd on success, -1 on error
This API adds a vcpu to a virtual machine. No more than max_vcpus may be added. The vcpu id is an integer in the range [0, max_vcpu_id).
The recommended max_vcpus value can be retrieved using the KVM_CAP_NR_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. The maximum possible value for max_vcpus can be retrieved using the KVM_CAP_MAX_VCPUS of the KVM_CHECK_EXTENSION ioctl() at run-time. """
Based on that and a quick look at the code, it looks to me like the set of valid vcpu_id values are a subset of the range of vcpu_idx-es, i.e. that kvm could decide to choose vcpu_id for the value of vcpu_idx. I don't think it should, but it could. Were the set of valid vcpu_id values not a subset of the set of supported vcpu_idx values, then one could argue that this is why.
I didn't want to get into explaining the why, I just wanted to state the fact.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id);
you can also add a sentence to clarify that kvm_get_vcpu expects an vcpu_idx, not an vcpu_id.
maybe ...
do_stuff(vcpu);} is not legit. The trouble is, we do actually use kvm->arch.idle_mask
... s/legit./legit, because kvm_get_vcpu() expects a vcpu_idx and not a vcpu_id.
But I agree kvm_get_vcpu(kvm, vcpu_id); does not scream BUG at me while kvm_get_vcpu_by_idx(kvm, vcpu_id) would look much more suspicious.
[..]
otherwise looks good to me.
Reviewed-by: Claudio Imbrenda imbrenda@linux.ibm.com
Thanks for your reveiew!
Halil
[..]
On 27.08.21 14:54, Halil Pasic wrote:
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array") Cc: stable@vger.kernel.org # 3.15+
arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS);
- /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv;
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT);
- set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT);
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len) static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) {
- int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus);
- int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_id);
- for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_idx);
if (test_and_set_bit(vcpu_id, gi->kicked_mask))
if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return;diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu) static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) {
- return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
Reviewed-by: Michael Mueller mimu@linux.ibm.com
I did also run some tests and have prepared a qemu that will choose vcpu_id's that are not equal to the index in the kernel.
Michael
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
I'm just backporting this fix to SLES 12 SP5, and I've noticed that there is still this code in __floating_irq_kick():
/* find idle VCPUs first, then round robin */ sigcpu = find_first_bit(fi->idle_mask, online_vcpus); /* ... round robin loop removed ... dst_vcpu = kvm_get_vcpu(kvm, sigcpu);
It seems to me that this does exactly the thing that is not legit, but I'm no expert. Did I miss something?
Petr T
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array") Cc: stable@vger.kernel.org # 3.15+
arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS);
- /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv;
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT);
- set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT);
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len) static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) {
- int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus);
- int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_id);
- for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_idx);
if (test_and_set_bit(vcpu_id, gi->kicked_mask))
if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return;diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu) static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) {
- return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
On Mon, 31 Jan 2022 11:13:18 +0100 Petr Tesařík ptesarik@suse.cz wrote:
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
I'm just backporting this fix to SLES 12 SP5, and I've noticed that there is still this code in __floating_irq_kick():
/* find idle VCPUs first, then round robin */ sigcpu = find_first_bit(fi->idle_mask, online_vcpus); /* ... round robin loop removed ... dst_vcpu = kvm_get_vcpu(kvm, sigcpu);
It seems to me that this does exactly the thing that is not legit, but I'm no expert. Did I miss something?
We made that legit by making the N-th bit in idle_mask correspond to the vcpu whose vcpu_idx == N. The second argument of kvm_get_vcpu() is the vcpu_idx. IMHO that ain't super-intuitive because it ain't spelled out.
So before this was a mismatch (with a vcpu_id based bitmap we would have to use kvm_get_vcpu_by_id()), and with this patch applied this code becomes legit because both idle_mask and kvm_get_vcpu() operate with vcpu_idx.
Does that make sense?
I'm sorry the commit message did not convey this clearly enough...
Regards, Halil
Petr T
Let us make switch from indexing kvm->arch.idle_mask by vcpu_id to indexing it by vcpu_idx. To keep gisa_int.kicked_mask indexed by the same index as idle_mask lets make the same change for it as well.
Signed-off-by: Halil Pasic pasic@linux.ibm.com Fixes: 1ee0bc559dc3 ("KVM: s390: get rid of local_int array") Cc: stable@vger.kernel.org # 3.15+
arch/s390/include/asm/kvm_host.h | 1 + arch/s390/kvm/interrupt.c | 12 ++++++------ arch/s390/kvm/kvm-s390.c | 2 +- arch/s390/kvm/kvm-s390.h | 2 +- 4 files changed, 9 insertions(+), 8 deletions(-)
diff --git a/arch/s390/include/asm/kvm_host.h b/arch/s390/include/asm/kvm_host.h index 161a9e12bfb8..630eab0fa176 100644 --- a/arch/s390/include/asm/kvm_host.h +++ b/arch/s390/include/asm/kvm_host.h @@ -957,6 +957,7 @@ struct kvm_arch{ atomic64_t cmma_dirty_pages; /* subset of available cpu features enabled by user space */ DECLARE_BITMAP(cpu_feat, KVM_S390_VM_CPU_FEAT_NR_BITS);
- /* indexed by vcpu_idx */ DECLARE_BITMAP(idle_mask, KVM_MAX_VCPUS); struct kvm_s390_gisa_interrupt gisa_int; struct kvm_s390_pv pv;
diff --git a/arch/s390/kvm/interrupt.c b/arch/s390/kvm/interrupt.c index d548d60caed2..16256e17a544 100644 --- a/arch/s390/kvm/interrupt.c +++ b/arch/s390/kvm/interrupt.c @@ -419,13 +419,13 @@ static unsigned long deliverable_irqs(struct kvm_vcpu *vcpu) static void __set_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_set_cpuflags(vcpu, CPUSTAT_WAIT);
- set_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- set_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __unset_cpu_idle(struct kvm_vcpu *vcpu) { kvm_s390_clear_cpuflags(vcpu, CPUSTAT_WAIT);
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static void __reset_intercept_indicators(struct kvm_vcpu *vcpu) @@ -3050,18 +3050,18 @@ int kvm_s390_get_irq_state(struct kvm_vcpu *vcpu, __u8 __user *buf, int len) static void __airqs_kick_single_vcpu(struct kvm *kvm, u8 deliverable_mask) {
- int vcpu_id, online_vcpus = atomic_read(&kvm->online_vcpus);
- int vcpu_idx, online_vcpus = atomic_read(&kvm->online_vcpus); struct kvm_s390_gisa_interrupt *gi = &kvm->arch.gisa_int; struct kvm_vcpu *vcpu;
- for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_id);
- for_each_set_bit(vcpu_idx, kvm->arch.idle_mask, online_vcpus) {
vcpu = kvm_get_vcpu(kvm, vcpu_idx);
if (test_and_set_bit(vcpu_id, gi->kicked_mask))
if (test_and_set_bit(vcpu_idx, gi->kicked_mask)) return; kvm_s390_vcpu_wakeup(vcpu); return;diff --git a/arch/s390/kvm/kvm-s390.c b/arch/s390/kvm/kvm-s390.c index 4527ac7b5961..8580543c5bc3 100644 --- a/arch/s390/kvm/kvm-s390.c +++ b/arch/s390/kvm/kvm-s390.c @@ -4044,7 +4044,7 @@ static int vcpu_pre_run(struct kvm_vcpu *vcpu) kvm_s390_patch_guest_per_regs(vcpu); }
- clear_bit(vcpu->vcpu_id, vcpu->kvm->arch.gisa_int.kicked_mask);
- clear_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.gisa_int.kicked_mask);
vcpu->arch.sie_block->icptcode = 0; cpuflags = atomic_read(&vcpu->arch.sie_block->cpuflags); diff --git a/arch/s390/kvm/kvm-s390.h b/arch/s390/kvm/kvm-s390.h index 9fad25109b0d..ecd741ee3276 100644 --- a/arch/s390/kvm/kvm-s390.h +++ b/arch/s390/kvm/kvm-s390.h @@ -79,7 +79,7 @@ static inline int is_vcpu_stopped(struct kvm_vcpu *vcpu) static inline int is_vcpu_idle(struct kvm_vcpu *vcpu) {
- return test_bit(vcpu->vcpu_id, vcpu->kvm->arch.idle_mask);
- return test_bit(kvm_vcpu_get_idx(vcpu), vcpu->kvm->arch.idle_mask); }
static inline int kvm_is_ucontrol(struct kvm *kvm)
base-commit: 77dd11439b86e3f7990e4c0c9e0b67dca82750ba
Hi Halil,
Dne 31. 01. 22 v 12:53 Halil Pasic napsal(a):
On Mon, 31 Jan 2022 11:13:18 +0100 Petr Tesařík ptesarik@suse.cz wrote:
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
I'm just backporting this fix to SLES 12 SP5, and I've noticed that there is still this code in __floating_irq_kick():
/* find idle VCPUs first, then round robin */ sigcpu = find_first_bit(fi->idle_mask, online_vcpus); /* ... round robin loop removed ... dst_vcpu = kvm_get_vcpu(kvm, sigcpu);
It seems to me that this does exactly the thing that is not legit, but I'm no expert. Did I miss something?
We made that legit by making the N-th bit in idle_mask correspond to the vcpu whose vcpu_idx == N. The second argument of kvm_get_vcpu() is the vcpu_idx. IMHO that ain't super-intuitive because it ain't spelled out.
So before this was a mismatch (with a vcpu_id based bitmap we would have to use kvm_get_vcpu_by_id()), and with this patch applied this code becomes legit because both idle_mask and kvm_get_vcpu() operate with vcpu_idx.
Does that make sense?
Yes!
I'm sorry the commit message did not convey this clearly enough...
No, it's not your fault. I didn't pay enough attention to the change, and with vcpu_id and vcpu_idx being so similar I got confused.
In short, there's no bug now, indeed. Thanks for your patience.
Petr T
On Mon, 31 Jan 2022 13:09:26 +0100 Petr Tesařík ptesarik@suse.cz wrote:
Hi Halil,
Dne 31. 01. 22 v 12:53 Halil Pasic napsal(a):
On Mon, 31 Jan 2022 11:13:18 +0100 Petr Tesařík ptesarik@suse.cz wrote:
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
While in practice vcpu->vcpu_idx == vcpu->vcp_id is often true, it may not always be, and we must not rely on this.
Currently kvm->arch.idle_mask is indexed by vcpu_id, which implies that code like for_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); } is not legit. The trouble is, we do actually use kvm->arch.idle_mask like this. To fix this problem we have two options. Either use kvm_get_vcpu_by_id(vcpu_id), which would loop to find the right vcpu_id, or switch to indexing via vcpu_idx. The latter is preferable for obvious reasons.
I'm just backporting this fix to SLES 12 SP5, and I've noticed that there is still this code in __floating_irq_kick():
/* find idle VCPUs first, then round robin */ sigcpu = find_first_bit(fi->idle_mask, online_vcpus); /* ... round robin loop removed ... dst_vcpu = kvm_get_vcpu(kvm, sigcpu);
It seems to me that this does exactly the thing that is not legit, but I'm no expert. Did I miss something?
We made that legit by making the N-th bit in idle_mask correspond to the vcpu whose vcpu_idx == N. The second argument of kvm_get_vcpu() is the vcpu_idx. IMHO that ain't super-intuitive because it ain't spelled out.
So before this was a mismatch (with a vcpu_id based bitmap we would have to use kvm_get_vcpu_by_id()), and with this patch applied this code becomes legit because both idle_mask and kvm_get_vcpu() operate with vcpu_idx.
Does that make sense?
Yes!
I'm sorry the commit message did not convey this clearly enough...
No, it's not your fault. I didn't pay enough attention to the change, and with vcpu_id and vcpu_idx being so similar I got confused.
No problem at all!
In short, there's no bug now, indeed. Thanks for your patience.
Thank you for being mindful when backporting!
Regards, Halil
linux-stable-mirror@lists.linaro.org
-
Christian Borntraeger -
Claudio Imbrenda -
Halil Pasic -
Michael Mueller -
Petr Tesařík