| Message ID | 20180114192349.22212-3-joserz@linux.vnet.ibm.com |
|---|---|
| State | New |
| Headers | show |
| Series | Small fixes for SMT guests in Power9 | expand |
On Sun, Jan 14, 2018 at 05:23:49PM -0200, Jose Ricardo Ziviani wrote: > QEMU currently checks whether SMT passed is valid or not. However, it > doesn't check if KVM supports such mode when kvm is enabled. That's not really true - the attempt to actually set the vsmt mode in KVM later on in spapr_set_vsmt_mode() will fail if KVM can't support the number of threads. The error added here might be a bit easier to understand, since it doesn't refer to vsmt modes, which might just confuse the issue. The change isn't urgent, though. > This patch relies on KVM_CAP_PPC_SMT_POSSIBLE to make it sure that QEMU > will either set a valid SMT mode or warn an error message and quit. > > Signed-off-by: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com> > --- > hw/ppc/spapr.c | 10 ++++++++++ > target/ppc/kvm.c | 5 +++++ > target/ppc/kvm_ppc.h | 6 ++++++ > 3 files changed, 21 insertions(+) > > diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c > index d1acfe8858..aed4d25fc4 100644 > --- a/hw/ppc/spapr.c > +++ b/hw/ppc/spapr.c > @@ -2261,12 +2261,22 @@ static void spapr_set_vsmt_mode(sPAPRMachineState *spapr, Error **errp) > "on a pseries machine"); > goto out; > } > + > if (!is_power_of_2(smp_threads)) { > error_setg(&local_err, "Cannot support %d threads/core on a pseries " > "machine because it must be a power of 2", smp_threads); > goto out; > } > > + if (kvm_enabled() && kvmppc_cap_smt_possible() > 0) { > + if ((kvmppc_cap_smt_possible() & smp_threads) != smp_threads) { > + error_setg(&local_err, "KVM does not support %d threads/core.", > + smp_threads); > + kvmppc_hint_smt_possible(&local_err); > + goto out; > + } > + } I'd like to see a fallback for kernels that don't support the smt_possible cap and vsmt mode setting (for those, we must have smp_threads <= kvm_smt). > + > /* Detemine the VSMT mode to use: */ > if (vsmt_user) { > if (spapr->vsmt < smp_threads) { > diff --git a/target/ppc/kvm.c b/target/ppc/kvm.c > index 914be687e7..4a8ff4d63c 100644 > --- a/target/ppc/kvm.c > +++ b/target/ppc/kvm.c > @@ -2456,6 +2456,11 @@ bool kvmppc_has_cap_mmu_hash_v3(void) > return cap_mmu_hash_v3; > } > > +int kvmppc_cap_smt_possible(void) > +{ > + return cap_ppc_smt_possible; > +} > + > PowerPCCPUClass *kvm_ppc_get_host_cpu_class(void) > { > uint32_t host_pvr = mfpvr(); > diff --git a/target/ppc/kvm_ppc.h b/target/ppc/kvm_ppc.h > index ecb55493cc..2221850723 100644 > --- a/target/ppc/kvm_ppc.h > +++ b/target/ppc/kvm_ppc.h > @@ -59,6 +59,7 @@ bool kvmppc_has_cap_fixup_hcalls(void); > bool kvmppc_has_cap_htm(void); > bool kvmppc_has_cap_mmu_radix(void); > bool kvmppc_has_cap_mmu_hash_v3(void); > +int kvmppc_cap_smt_possible(void); > int kvmppc_enable_hwrng(void); > int kvmppc_put_books_sregs(PowerPCCPU *cpu); > PowerPCCPUClass *kvm_ppc_get_host_cpu_class(void); > @@ -290,6 +291,11 @@ static inline bool kvmppc_has_cap_mmu_hash_v3(void) > return false; > } > > +static inline int kvmppc_cap_smt_possible(void) > +{ > + return 0; > +} > + > static inline int kvmppc_enable_hwrng(void) > { > return -1;
diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c index d1acfe8858..aed4d25fc4 100644 --- a/hw/ppc/spapr.c +++ b/hw/ppc/spapr.c @@ -2261,12 +2261,22 @@ static void spapr_set_vsmt_mode(sPAPRMachineState *spapr, Error **errp) "on a pseries machine"); goto out; } + if (!is_power_of_2(smp_threads)) { error_setg(&local_err, "Cannot support %d threads/core on a pseries " "machine because it must be a power of 2", smp_threads); goto out; } + if (kvm_enabled() && kvmppc_cap_smt_possible() > 0) { + if ((kvmppc_cap_smt_possible() & smp_threads) != smp_threads) { + error_setg(&local_err, "KVM does not support %d threads/core.", + smp_threads); + kvmppc_hint_smt_possible(&local_err); + goto out; + } + } + /* Detemine the VSMT mode to use: */ if (vsmt_user) { if (spapr->vsmt < smp_threads) { diff --git a/target/ppc/kvm.c b/target/ppc/kvm.c index 914be687e7..4a8ff4d63c 100644 --- a/target/ppc/kvm.c +++ b/target/ppc/kvm.c @@ -2456,6 +2456,11 @@ bool kvmppc_has_cap_mmu_hash_v3(void) return cap_mmu_hash_v3; } +int kvmppc_cap_smt_possible(void) +{ + return cap_ppc_smt_possible; +} + PowerPCCPUClass *kvm_ppc_get_host_cpu_class(void) { uint32_t host_pvr = mfpvr(); diff --git a/target/ppc/kvm_ppc.h b/target/ppc/kvm_ppc.h index ecb55493cc..2221850723 100644 --- a/target/ppc/kvm_ppc.h +++ b/target/ppc/kvm_ppc.h @@ -59,6 +59,7 @@ bool kvmppc_has_cap_fixup_hcalls(void); bool kvmppc_has_cap_htm(void); bool kvmppc_has_cap_mmu_radix(void); bool kvmppc_has_cap_mmu_hash_v3(void); +int kvmppc_cap_smt_possible(void); int kvmppc_enable_hwrng(void); int kvmppc_put_books_sregs(PowerPCCPU *cpu); PowerPCCPUClass *kvm_ppc_get_host_cpu_class(void); @@ -290,6 +291,11 @@ static inline bool kvmppc_has_cap_mmu_hash_v3(void) return false; } +static inline int kvmppc_cap_smt_possible(void) +{ + return 0; +} + static inline int kvmppc_enable_hwrng(void) { return -1;
QEMU currently checks whether SMT passed is valid or not. However, it doesn't check if KVM supports such mode when kvm is enabled. This patch relies on KVM_CAP_PPC_SMT_POSSIBLE to make it sure that QEMU will either set a valid SMT mode or warn an error message and quit. Signed-off-by: Jose Ricardo Ziviani <joserz@linux.vnet.ibm.com> --- hw/ppc/spapr.c | 10 ++++++++++ target/ppc/kvm.c | 5 +++++ target/ppc/kvm_ppc.h | 6 ++++++ 3 files changed, 21 insertions(+)