KVM: s390: index kvm->arch.idle_mask by vcpu_idx
Subsystems (4)
PATCH 1/1 KVM: s390: index kvm->arch.idle_mask by vcpu_idx
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.
--- 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(-)
arch/s390/include/asm/kvm_host.h +1 -0
@@ -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;
arch/s390/kvm/interrupt.c +6 -6
@@ -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 ;
arch/s390/kvm/kvm-s390.c +1 -1
@@ -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);
arch/s390/kvm/kvm-s390.h +1 -1
@@ -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
Signature · 1 line
2.25.1
Quoted 15 lines · Halil Pasic
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_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_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_idx. The latter is preferable for obvious reasons.
Quoted 7 lines
Let us make switch from indexing
kvm->arch.idle_maskbyvcpu_idto indexing it byvcpu_idx. To keep gisa_int.kicked_mask indexed by the same index asidle_masklets make the same change for it as well.Signed-off-by Halil Pasic <[email protected]>Fixes 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")Cc <[email protected]> # 3.15+
Will apply and wait for the nightly run.
Quoted 91 lines
--- 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(-)arch/s390/include/asm/kvm_host.h +1
-0@@ -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;arch/s390/kvm/interrupt.c +6
-6@@ -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 ;arch/s390/kvm/kvm-s390.c +1
-1@@ -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);arch/s390/kvm/kvm-s390.h +1
-1@@ -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
Quoted 2 lines · On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic <[email protected]> wrote:
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis 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
Quoted 5 lines
Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_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.
Quoted 8 lines
do_stuff(vcpu); }is not legit. The trouble is, we do actually use
kvm->arch.idle_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_idx. The latter is preferable for obvious reasons.
Quoted 6 lines
Let us make switch from indexing
kvm->arch.idle_maskbyvcpu_idto indexing it byvcpu_idx. To keep gisa_int.kicked_mask indexed by the same index asidle_masklets make the same change for it as well.Signed-off-by Halil Pasic <[email protected]>Fixes 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")
otherwise looks good to me.
Quoted 91 lines
Cc <[email protected]> # 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(-)arch/s390/include/asm/kvm_host.h +1
-0@@ -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;arch/s390/kvm/interrupt.c +6
-6@@ -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 ;arch/s390/kvm/kvm-s390.c +1
-1@@ -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);arch/s390/kvm/kvm-s390.h +1
-1@@ -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
Quoted 14 lines · Claudio Imbrenda
Quoted 2 lines · On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic <[email protected]> wrote:
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.why?
maybe add a simple explanation of why
vcpu_idxandvcpu_idcan be different, namely:
KVM decides thevcpu_idx, userspace decides thevcpu_id, thus the two might not matchQuoted 1 line
Quoted 9 lines
Quoted 4 lines
Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_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_vcpuexpects anvcpu_idx, not anvcpu_id.Quoted 1 line
do_stuff(vcpu);
I will modify the patch description accordingly before sending to Paolo. Thanks for noticing.
Quoted 110 lines
Quoted 7 lines
}
is not legit. The trouble is, we do actually usekvm->arch.idle_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_idx. The latter is preferable for obvious reasons.Quoted 6 lines
Let us make switch from indexing
kvm->arch.idle_maskbyvcpu_idto indexing it byvcpu_idx. To keep gisa_int.kicked_mask indexed by the same index asidle_masklets make the same change for it as well.Signed-off-by Halil Pasic <[email protected]>Fixes 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")otherwise looks good to me.
Reviewed-by Claudio Imbrenda <[email protected]>Quoted 91 lines
Cc <[email protected]> # 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(-)arch/s390/include/asm/kvm_host.h +1
-0@@ -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;arch/s390/kvm/interrupt.c +6
-6@@ -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 ;arch/s390/kvm/kvm-s390.c +1
-1@@ -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);arch/s390/kvm/kvm-s390.h +1
-1@@ -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
Quoted 5 lines · On Fri, 27 Aug 2021 18:36:48 +0200 Christian Borntraeger…
Quoted 4 lines · Claudio Imbrenda
Quoted 1 line · On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic <[email protected]> wrote:
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true,
s/vcp_id/vcpu_id/
Quoted 15 lines
Quoted 15 lines
Quoted 1 line
it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why
vcpu_idxandvcpu_idcan be different, namely:
KVM decides thevcpu_idx, userspace decides thevcpu_id, thus the two might not matchQuoted 5 lines
Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id);
Quoted 7 lines
Quoted 4 lines
you can also add a sentence to clarify that
kvm_get_vcpuexpects anvcpu_idx, not anvcpu_id.Quoted 1 line
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
Quoted 29 lines · Halil Pasic
Quoted 5 lines · On Fri, 27 Aug 2021 18:36:48 +0200 Christian Borntraeger…
Quoted 4 lines · Claudio Imbrenda
Quoted 1 line · On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic <[email protected]> wrote:
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true,s/vcp_id/vcpu_id/
Quoted 18 lines
Quoted 18 lines
Quoted 1 line
it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why
vcpu_idxandvcpu_idcan be different, namely:
KVM decides thevcpu_idx, userspace decides thevcpu_id, thus the two might not matchQuoted 5 lines
Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_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_vcpuexpects anvcpu_idx, not anvcpu_id.
Quoted 7 lines
Quoted 4 lines
Quoted 1 line
Quoted 1 line
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).
I already queued, but I think this is not a big deal.
Quoted 4 lines · On Fri, 27 Aug 2021 16:06:16 +0200 Claudio Imbrenda <[email protected]>…
Quoted 1 line · On Fri, 27 Aug 2021 14:54:29 +0200 Halil Pasic <[email protected]> wrote:
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true,
s/vcp_id/vcpu_id/
Quoted 8 lines
Quoted 1 line
it may not always be, and we must not rely on this.
why?
maybe add a simple explanation of why
vcpu_idxandvcpu_idcan be different, namely:
KVM decides thevcpu_idx, userspace decides thevcpu_id, thus the two might not match
Not sure that is a good explanation. A quote fromDocumentation/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.
Quoted 9 lines
Quoted 6 lines
Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_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_vcpuexpects anvcpu_idx, not anvcpu_id.
maybe ...
Quoted 4 lines
Quoted 4 lines
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.
[..]
Quoted 4 lines
otherwise looks good to me.
Reviewed-by Claudio Imbrenda <[email protected]>
Thanks for your reveiew!
Halil
[..]
Quoted 111 lines · Halil Pasic
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_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_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_idx. The latter is preferable for obvious reasons.Let us make switch from indexing
kvm->arch.idle_maskbyvcpu_idto indexing it byvcpu_idx. To keep gisa_int.kicked_mask indexed by the same index asidle_masklets make the same change for it as well.Signed-off-by Halil Pasic <[email protected]>Fixes 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")Cc <[email protected]> # 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(-)arch/s390/include/asm/kvm_host.h +1
-0@@ -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;arch/s390/kvm/interrupt.c +6
-6@@ -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 ;arch/s390/kvm/kvm-s390.c +1
-1@@ -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);arch/s390/kvm/kvm-s390.h +1
-1@@ -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)
Quoted 1 line
base-commit:
77dd11439b86e3f7990e4c0c9e0b67dca82750ba
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
Quoted 1 line
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
Quoted 9 lines
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); }
Quoted 5 lines
is not legit. The trouble is, we do actually use
kvm->arch.idle_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_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
Quoted 97 lines
Let us make switch from indexing
kvm->arch.idle_maskbyvcpu_idto indexing it byvcpu_idx. To keep gisa_int.kicked_mask indexed by the same index asidle_masklets make the same change for it as well.Signed-off-by Halil Pasic <[email protected]>Fixes 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")Cc <[email protected]> # 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(-)arch/s390/include/asm/kvm_host.h +1
-0@@ -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;arch/s390/kvm/interrupt.c +6
-6@@ -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 ;arch/s390/kvm/kvm-s390.c +1
-1@@ -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);arch/s390/kvm/kvm-s390.h +1
-1@@ -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
Quoted 21 lines · On Mon, 31 Jan 2022 11:13:18 +0100 Petr Tesařík <[email protected]> wrote:
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
Quoted 9 lines
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); }Quoted 5 lines
is not legit. The trouble is, we do actually use
kvm->arch.idle_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_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():
Quoted 8 lines
/* 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
Quoted 99 lines
Petr T
Quoted 97 lines
Let us make switch from indexing
kvm->arch.idle_maskbyvcpu_idto indexing it byvcpu_idx. To keep gisa_int.kicked_mask indexed by the same index asidle_masklets make the same change for it as well.Signed-off-by Halil Pasic <[email protected]>Fixes 1ee0bc559dc3 ("KVM: s390: get rid of local_int array")Cc <[email protected]> # 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(-)arch/s390/include/asm/kvm_host.h +1
-0@@ -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;arch/s390/kvm/interrupt.c +6
-6@@ -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 ;arch/s390/kvm/kvm-s390.c +1
-1@@ -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);arch/s390/kvm/kvm-s390.h +1
-1@@ -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):
Quoted 37 lines
Quoted 21 lines · On Mon, 31 Jan 2022 11:13:18 +0100 Petr Tesařík <[email protected]> wrote:
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
Quoted 9 lines
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); }Quoted 5 lines
is not legit. The trouble is, we do actually use
kvm->arch.idle_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_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():Quoted 8 lines
/* 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_maskcorrespond to the vcpu whosevcpu_idx== N. The second argument ofkvm_get_vcpu()is thevcpu_idx. IMHO that ain't super-intuitive because it ain't spelled out.
Quoted 6 lines
So before this was a mismatch (with a
vcpu_idbased bitmap we would have to usekvm_get_vcpu_by_id()), and with this patch applied this code becomes legit because bothidle_maskandkvm_get_vcpu()operate withvcpu_idx.Does that make sense?
Yes!
Quoted 1 line
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
Quoted 45 lines · On Mon, 31 Jan 2022 13:09:26 +0100 Petr Tesařík <[email protected]> wrote:
Hi Halil,
Dne 31. 01. 22 v 12:53 Halil Pasic napsal(a):
Quoted 42 lines
Quoted 21 lines · On Mon, 31 Jan 2022 11:13:18 +0100 Petr Tesařík <[email protected]> wrote:
Hi Halil,
Dne 27. 08. 21 v 14:54 Halil Pasic napsal(a):
Quoted 9 lines
While in practice
vcpu->vcpu_idx==vcpu->vcp_idis often true, it may not always be, and we must not rely on this.Currently
kvm->arch.idle_maskis indexed byvcpu_id, which implies that code likefor_each_set_bit(vcpu_id, kvm->arch.idle_mask, online_vcpus) { vcpu = kvm_get_vcpu(kvm, vcpu_id); do_stuff(vcpu); }Quoted 5 lines
is not legit. The trouble is, we do actually use
kvm->arch.idle_masklike this. To fix this problem we have two options. Either usekvm_get_vcpu_by_id(vcpu_id), which would loop to find the rightvcpu_id, or switch to indexing viavcpu_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():Quoted 8 lines
/* 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_maskcorrespond to the vcpu whosevcpu_idx== N. The second argument ofkvm_get_vcpu()is thevcpu_idx. IMHO that ain't super-intuitive because it ain't spelled out.So before this was a mismatch (with a
vcpu_idbased bitmap we would have to usekvm_get_vcpu_by_id()), and with this patch applied this code becomes legit because bothidle_maskandkvm_get_vcpu()operate withvcpu_idx.
Quoted 8 lines
Quoted 1 line
Does that make sense?
Yes!
Quoted 1 line
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_idandvcpu_idxbeing so similar I got confused.
No problem at all!
Quoted 3 lines
In short, there's no bug now, indeed. Thanks for your patience.
Thank you for being mindful when backporting!
Regards,
Halil