Re: [Qemu-devel] [PATCH v12 5/6] ppc: spapr: Handle "ibm, nmi-register" and "ibm, nmi-interlock" RTAS calls

[Greg Kurz]

On Fri, 30 Aug 2019 14:44:07 +0530
Aravinda Prasad <address@hidden> wrote:

> This patch adds support in QEMU to handle "ibm,nmi-register"
> and "ibm,nmi-interlock" RTAS calls and sets the default
> value of SPAPR_CAP_FWNMI_MCE to SPAPR_CAP_ON for machine
> type 4.2.
> 
> 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.
> 
> Signed-off-by: Aravinda Prasad <address@hidden>
> ---
>  hw/ppc/spapr.c         |   12 +++++++++++-
>  hw/ppc/spapr_rtas.c    |   50 
> ++++++++++++++++++++++++++++++++++++++++++++++++
>  include/hw/ppc/spapr.h |    5 ++++-
>  3 files changed, 65 insertions(+), 2 deletions(-)
> 
> diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
> index 9f2e5d2..1c0908e 100644
> --- a/hw/ppc/spapr.c
> +++ b/hw/ppc/spapr.c
> @@ -2941,6 +2941,15 @@ static void spapr_machine_init(MachineState *machine)
>  
>          /* Resize rtas blob to accommodate error log */
>          spapr->rtas_size = RTAS_ERROR_LOG_MAX;
> +
> +        /* Set fwnmi capability in KVM */
> +        if (kvmppc_set_fwnmi() < 0) {
> +            error_report("Could not enable FWNMI capability");
> +            exit(1);
> +        }

Hmm... shouldn't this be performed only when the guest
calls "ibm,nmi-register" ?

> +
> +        /* Register ibm,nmi-register and ibm,nmi-interlock RTAS calls */
> +        spapr_fwnmi_register();
>      }
>  
>      spapr->rtas_blob = g_malloc(spapr->rtas_size);
> @@ -4508,7 +4517,7 @@ static void spapr_machine_class_init(ObjectClass *oc, 
> void *data)
>      smc->default_caps.caps[SPAPR_CAP_NESTED_KVM_HV] = SPAPR_CAP_OFF;
>      smc->default_caps.caps[SPAPR_CAP_LARGE_DECREMENTER] = SPAPR_CAP_ON;
>      smc->default_caps.caps[SPAPR_CAP_CCF_ASSIST] = SPAPR_CAP_OFF;
> -    smc->default_caps.caps[SPAPR_CAP_FWNMI_MCE] = SPAPR_CAP_OFF;
> +    smc->default_caps.caps[SPAPR_CAP_FWNMI_MCE] = SPAPR_CAP_ON;
>      spapr_caps_add_properties(smc, &error_abort);
>      smc->irq = &spapr_irq_dual;
>      smc->dr_phb_enabled = true;
> @@ -4582,6 +4591,7 @@ static void 
> spapr_machine_4_1_class_options(MachineClass *mc)
>      smc->linux_pci_probe = false;
>      compat_props_add(mc->compat_props, hw_compat_4_1, hw_compat_4_1_len);
>      compat_props_add(mc->compat_props, compat, G_N_ELEMENTS(compat));
> +    smc->default_caps.caps[SPAPR_CAP_FWNMI_MCE] = SPAPR_CAP_OFF;
>  }
>  
>  DEFINE_SPAPR_MACHINE(4_1, "4.1", false);
> diff --git a/hw/ppc/spapr_rtas.c b/hw/ppc/spapr_rtas.c
> index d8fb8a8..d892583 100644
> --- a/hw/ppc/spapr_rtas.c
> +++ b/hw/ppc/spapr_rtas.c
> @@ -400,6 +400,48 @@ 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 = spapr_get_rtas_addr();
> +
> +    if (!rtas_addr) {
> +        rtas_st(rets, 0, RTAS_OUT_NOT_SUPPORTED);
> +        return;
> +    }
> +
> +    if (spapr_get_cap(spapr, SPAPR_CAP_FWNMI_MCE) == SPAPR_CAP_OFF) {
> +        rtas_st(rets, 0, RTAS_OUT_NOT_SUPPORTED);
> +        return;
> +    }
> +
> +    spapr->guest_machine_check_addr = rtas_ld(args, 1);

Doesn't this need some sanity checks ? At least error out on -1
which has a special meaning in the code and cannot really be used
as a valid instruction address.

Also PAPR+ says:

R1–7.3.14–6. For the FWNMI option: The Real/Logical address of the
registered OS Machine Check and System Reset routines must be in the
 first 32 MB of the OS’s memory address space.


And only at this point you may enable the cap in KVM since the
guest has decided to use FWNMI.

My concern is: what happens when the guest reboots ? We set
guest_machine_check_addr back to -1 during machine reset but
KVM still assumes the guest has enabled FWNMI... I see that
enabling FWNMI boils down to setting a kvm->arch.fwnmi_enabled
to true... what about providing a way to disable FWNMI ?

> +    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);
> +    } else {
> +        /*
> +         * 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;
> @@ -544,6 +586,14 @@ 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 ffefde7..dada821 100644
> --- a/include/hw/ppc/spapr.h
> +++ b/include/hw/ppc/spapr.h
> @@ -655,8 +655,10 @@ 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 +910,5 @@ 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 */
>