[PATCH 1/2] KVM: PPC: Book3S HV: Allow creating max number of VCPUs on POWER9

Commit 1e175d2 ("KVM: PPC: Book3S HV: Pack VCORE IDs to access full
VCPU ID space", 2018-07-25) allowed use of VCPU IDs up to
KVM_MAX_VCPU_ID on POWER9 in all guest SMT modes and guest emulated
hardware SMT modes.  However, with the current definition of
KVM_MAX_VCPU_ID, a guest SMT mode of 1 and an emulated SMT mode of 8,
it is only possible to create KVM_MAX_VCPUS / 2 VCPUS, because
threads_per_subcore is 4 on POWER9 CPUs.  (Using an emulated SMT mode
of 8 is useful when migrating VMs to or from POWER8 hosts.)

This increases KVM_MAX_VCPU_ID to 8 * KVM_MAX_VCPUS when HV KVM is
configured in, so that a full complement of KVM_MAX_VCPUS VCPUs can
be created on POWER9 in all guest SMT modes and emulated hardware
SMT modes.

Signed-off-by: Paul Mackerras <redacted>
---
This and the next patch apply on my kvm-ppc-next branch, which
includes Sam Bobroff's patch "KVM: PPC: Book3S HV: Pack VCORE IDs to
access full VCPU ID space".

 arch/powerpc/include/asm/kvm_host.h | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/arch/powerpc/include/asm/kvm_host.h b/arch/powerpc/include/asm/kvm_host.h
index 5b9e660..906bcbdf 100644
--- a/arch/powerpc/include/asm/kvm_host.h
+++ b/arch/powerpc/include/asm/kvm_host.h

@@ -42,7 +42,14 @@
 #define KVM_USER_MEM_SLOTS	512
 
 #include <asm/cputhreads.h>
-#define KVM_MAX_VCPU_ID                (threads_per_subcore * KVM_MAX_VCORES)
+
+#ifdef CONFIG_KVM_BOOK3S_HV_POSSIBLE
+#include <asm/kvm_book3s_asm.h>		/* for MAX_SMT_THREADS */
+#define KVM_MAX_VCPU_ID		(MAX_SMT_THREADS * KVM_MAX_VCORES)
+
+#else
+#define KVM_MAX_VCPU_ID		KVM_MAX_VCPUS
+#endif /* CONFIG_KVM_BOOK3S_HV_POSSIBLE */
 
 #define __KVM_HAVE_ARCH_INTC_INITIALIZED
 

-- 
2.7.4

On Thu, Jul 26, 2018 at 03:43:54PM +1000, Paul Mackerras wrote:

Commit 1e175d2 ("KVM: PPC: Book3S HV: Pack VCORE IDs to access full
VCPU ID space", 2018-07-25) allowed use of VCPU IDs up to
KVM_MAX_VCPU_ID on POWER9 in all guest SMT modes and guest emulated
hardware SMT modes.  However, with the current definition of
KVM_MAX_VCPU_ID, a guest SMT mode of 1 and an emulated SMT mode of 8,
it is only possible to create KVM_MAX_VCPUS / 2 VCPUS, because
threads_per_subcore is 4 on POWER9 CPUs.  (Using an emulated SMT mode
of 8 is useful when migrating VMs to or from POWER8 hosts.)

This increases KVM_MAX_VCPU_ID to 8 * KVM_MAX_VCPUS when HV KVM is
configured in, so that a full complement of KVM_MAX_VCPUS VCPUs can
be created on POWER9 in all guest SMT modes and emulated hardware
SMT modes.

Signed-off-by: Paul Mackerras <redacted>
---
This and the next patch apply on my kvm-ppc-next branch, which
includes Sam Bobroff's patch "KVM: PPC: Book3S HV: Pack VCORE IDs to
access full VCPU ID space".

Thanks!

I've tested kvm-ppc-next with all three above patches, and I'm able to
access the entire VCPU ID space (tested using CONFIG_NR_CPUS 1024 in the
host, rather than the default 2048, as that's the limit imposed by
QEMU). So:

Tested-by: Sam Bobroff <redacted>