On Thu, Jan 02, 2020 at 01:21:09PM +0530, Ganesh Goudar wrote: > From: Aravinda Prasad > > This patch adds support in QEMU to handle "ibm,nmi-register" > and "ibm,nmi-interlock" RTAS calls. > > The machine check notification address is saved when the > OS issues "ibm,nmi-register" RTAS call. > > This patch also handles the case when multiple processors > experience machine check at or about the same time by > handling "ibm,nmi-interlock" call. In such cases, as per > PAPR, subsequent processors serialize waiting for the first > processor to issue the "ibm,nmi-interlock" call. The second > processor that also received a machine check error waits > till the first processor is done reading the error log. > The first processor issues "ibm,nmi-interlock" call > when the error log is consumed. > > [Move fwnmi registration to .apply hook] > Signed-off-by: Ganesh Goudar > Signed-off-by: Aravinda Prasad > --- > hw/ppc/spapr_caps.c | 6 +++++ > hw/ppc/spapr_rtas.c | 58 ++++++++++++++++++++++++++++++++++++++++++ > include/hw/ppc/spapr.h | 7 ++++- > 3 files changed, 70 insertions(+), 1 deletion(-) > > diff --git a/hw/ppc/spapr_caps.c b/hw/ppc/spapr_caps.c > index 3001098601..e922419cfb 100644 > --- a/hw/ppc/spapr_caps.c > +++ b/hw/ppc/spapr_caps.c > @@ -502,6 +502,12 @@ static void cap_fwnmi_mce_apply(SpaprMachineState *spapr, uint8_t val, > if (!val) { > return; /* Disabled by default */ > } > + > + if (!spapr->fwnmi_calls_registered && !kvmppc_set_fwnmi()) { So, we definitely need the kvmppc_set_fwnmi() call here. But in the case where we *do* have KVM, but the call fails, we should fail the .apply hook, rather than ignoring it silently. As we've discussed although TCG behaviour with fwnmi isn't 100% correct, it's close enough to pass for most purposes - so it's reasonable to continue if the cap is selected. But if the cap is selected and we're running with KVM we *must* enable the capability in KVM or we're not providing the environment the user requested. > + /* Register ibm,nmi-register and ibm,nmi-interlock RTAS calls */ > + spapr_fwnmi_register(); We discussed registering the hypercalls here, but I thought after that I suggested just always registering them, and having them bail out when called if the cap is not set. I see that you've implemented that bailout for register, though not for interlock. I think that's simpler than registering them here. > + spapr->fwnmi_calls_registered = true; > + } > } > > SpaprCapabilityInfo capability_table[SPAPR_CAP_NUM] = { > diff --git a/hw/ppc/spapr_rtas.c b/hw/ppc/spapr_rtas.c > index 2c066a372d..54b142f35b 100644 > --- a/hw/ppc/spapr_rtas.c > +++ b/hw/ppc/spapr_rtas.c > @@ -400,6 +400,56 @@ static void rtas_get_power_level(PowerPCCPU *cpu, SpaprMachineState *spapr, > rtas_st(rets, 1, 100); > } > > +static void rtas_ibm_nmi_register(PowerPCCPU *cpu, > + SpaprMachineState *spapr, > + uint32_t token, uint32_t nargs, > + target_ulong args, > + uint32_t nret, target_ulong rets) > +{ > + hwaddr rtas_addr; > + > + if (spapr_get_cap(spapr, SPAPR_CAP_FWNMI_MCE) == SPAPR_CAP_OFF) { > + rtas_st(rets, 0, RTAS_OUT_NOT_SUPPORTED); > + return; > + } > + > + rtas_addr = spapr_get_rtas_addr(); > + if (!rtas_addr) { > + rtas_st(rets, 0, RTAS_OUT_NOT_SUPPORTED); > + return; > + } > + > + spapr->guest_machine_check_addr = rtas_ld(args, 1); > + rtas_st(rets, 0, RTAS_OUT_SUCCESS); > +} > + > +static void rtas_ibm_nmi_interlock(PowerPCCPU *cpu, > + SpaprMachineState *spapr, > + uint32_t token, uint32_t nargs, > + target_ulong args, > + uint32_t nret, target_ulong rets) > +{ > + if (spapr->guest_machine_check_addr == -1) { > + /* NMI register not called */ > + rtas_st(rets, 0, RTAS_OUT_PARAM_ERROR); > + return; > + } > + > + if (spapr->mc_status != cpu->vcpu_id) { > + /* The vCPU that hit the NMI should invoke "ibm,nmi-interlock" */ > + rtas_st(rets, 0, RTAS_OUT_PARAM_ERROR); > + return; > + } > + > + /* > + * vCPU issuing "ibm,nmi-interlock" is done with NMI handling, > + * hence unset mc_status. > + */ > + spapr->mc_status = -1; > + qemu_cond_signal(&spapr->mc_delivery_cond); > + rtas_st(rets, 0, RTAS_OUT_SUCCESS); > +} > + > static struct rtas_call { > const char *name; > spapr_rtas_fn fn; > @@ -503,6 +553,14 @@ hwaddr spapr_get_rtas_addr(void) > return (hwaddr)fdt32_to_cpu(*rtas_data); > } > > +void spapr_fwnmi_register(void) > +{ > + spapr_rtas_register(RTAS_IBM_NMI_REGISTER, "ibm,nmi-register", > + rtas_ibm_nmi_register); > + spapr_rtas_register(RTAS_IBM_NMI_INTERLOCK, "ibm,nmi-interlock", > + rtas_ibm_nmi_interlock); > +} > + > static void core_rtas_register_types(void) > { > spapr_rtas_register(RTAS_DISPLAY_CHARACTER, "display-character", > diff --git a/include/hw/ppc/spapr.h b/include/hw/ppc/spapr.h > index 652a5514e8..a90e677cc3 100644 > --- a/include/hw/ppc/spapr.h > +++ b/include/hw/ppc/spapr.h > @@ -218,6 +218,8 @@ struct SpaprMachineState { > > unsigned gpu_numa_id; > SpaprTpmProxy *tpm_proxy; > + > + bool fwnmi_calls_registered; > }; > > #define H_SUCCESS 0 > @@ -656,8 +658,10 @@ target_ulong spapr_hypercall(PowerPCCPU *cpu, target_ulong opcode, > #define RTAS_IBM_REMOVE_PE_DMA_WINDOW (RTAS_TOKEN_BASE + 0x28) > #define RTAS_IBM_RESET_PE_DMA_WINDOW (RTAS_TOKEN_BASE + 0x29) > #define RTAS_IBM_SUSPEND_ME (RTAS_TOKEN_BASE + 0x2A) > +#define RTAS_IBM_NMI_REGISTER (RTAS_TOKEN_BASE + 0x2B) > +#define RTAS_IBM_NMI_INTERLOCK (RTAS_TOKEN_BASE + 0x2C) > > -#define RTAS_TOKEN_MAX (RTAS_TOKEN_BASE + 0x2B) > +#define RTAS_TOKEN_MAX (RTAS_TOKEN_BASE + 0x2D) > > /* RTAS ibm,get-system-parameter token values */ > #define RTAS_SYSPARM_SPLPAR_CHARACTERISTICS 20 > @@ -908,4 +912,5 @@ void spapr_check_pagesize(SpaprMachineState *spapr, hwaddr pagesize, > > void spapr_set_all_lpcrs(target_ulong value, target_ulong mask); > hwaddr spapr_get_rtas_addr(void); > +void spapr_fwnmi_register(void); > #endif /* HW_SPAPR_H */ -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson