Re: [PATCH v4 2/3] spapr_numa: create a vcpu associativity helper

[Daniel Henrique Barboza]

On 9/4/20 7:02 AM, Greg Kurz wrote:
> On Thu,  3 Sep 2020 22:04:38 -0300
> Daniel Henrique Barboza <danielhb413@gmail.com> wrote:
>
> > The work to be done in h_home_node_associativity() intersects
> > with what is already done in spapr_numa_fixup_cpu_dt(). This
> > patch creates a new helper, spapr_numa_get_vcpu_assoc(), to
> > be used for both spapr_numa_fixup_cpu_dt() and
> > h_home_node_associativity().
> >
> > While we're at it, use memcpy() instead of loop assignment
> > to created the returned array.
> >
> > Signed-off-by: Daniel Henrique Barboza <danielhb413@gmail.com>
> > ---
>
> Hi Daniel,
>
> A few comments below.
>
> >   hw/ppc/spapr_numa.c | 30 ++++++++++++++++++++----------
> >   1 file changed, 20 insertions(+), 10 deletions(-)
> >
> > diff --git a/hw/ppc/spapr_numa.c b/hw/ppc/spapr_numa.c
> > index 368c1a494d..980a6488bf 100644
> > --- a/hw/ppc/spapr_numa.c
> > +++ b/hw/ppc/spapr_numa.c
> > @@ -71,13 +71,15 @@ void spapr_numa_write_associativity_dt(SpaprMachineState 
> > *spapr, void *fdt,
> >                         sizeof(spapr->numa_assoc_array[nodeid]))));
> >   }
> >   
> > -int spapr_numa_fixup_cpu_dt(SpaprMachineState *spapr, void *fdt,
> > -                            int offset, PowerPCCPU *cpu)
> > +static uint32_t *spapr_numa_get_vcpu_assoc(SpaprMachineState *spapr,
> > +                                          PowerPCCPU *cpu,
> > +                                          uint *vcpu_assoc_size)
> >   {
> > -    uint vcpu_assoc_size = NUMA_ASSOC_SIZE + 1;
> > -    uint32_t vcpu_assoc[vcpu_assoc_size];
> > +    uint32_t *vcpu_assoc = NULL;
>
> You don't need to initialize this pointer since it is assigned a value
> unconditionally just below.
>
> >       int index = spapr_get_vcpu_id(cpu);
> > -    int i;
> > +
> > +    *vcpu_assoc_size = (NUMA_ASSOC_SIZE + 1) * sizeof(uint32_t);
>
> It's a bit weird to return something that is definitely a compile
> time constant by reference... What about introducing a macro ?
>
> #define VCPU_NUMA_ASSOC_SIZE (NUMA_ASSOC_SIZE + 1)
>
> > +    vcpu_assoc = g_malloc(*vcpu_assoc_size);
> >   
>
>      vcpu_assoc = g_new(uint32_t, VCPU_NUMA_ASSOC_SIZE);
>
> >       /*
> >        * VCPUs have an extra 'cpu_id' value in ibm,associativity
> > @@ -86,16 +88,24 @@ int spapr_numa_fixup_cpu_dt(SpaprMachineState *spapr, void 
> > *fdt,
> >        * cpu_id last.
> >        */
> >       vcpu_assoc[0] = cpu_to_be32(MAX_DISTANCE_REF_POINTS + 1);
> > +    memcpy(vcpu_assoc + 1, spapr->numa_assoc_array[cpu->node_id],
> > +           MAX_DISTANCE_REF_POINTS);
> > +    vcpu_assoc[MAX_DISTANCE_REF_POINTS + 1] = cpu_to_be32(index);
> >   
>
>      memcpy(vcpu_assoc + 1, spapr->numa_assoc_array[cpu->node_id],
>             (VPCU_ASSOC_SIZE - 2) * sizeof(uint32_t));
>      vcpu_assoc[VCPU_ASSOC_SIZE - 1] = cpu_to_be32(index);
>
> I personally find more clear than using MAX_DISTANCE_REF_POINTS in an array
> that was just allocated with NUMA_ASSOC_SIZE... one has to check spapr.h
> to see that NUMA_ASSOC_SIZE == MAX_DISTANCE_REF_POINTS + 1


That all makes sense to me. I'll introduce a VCPU_ASSOC_SIZE in spapr_numa.h
and use it when operating the associativity for vcpus, both in this patch
and also in patch 3.


Thanks,


DHB


> > -    for (i = 1; i <= MAX_DISTANCE_REF_POINTS; i++) {
> > -        vcpu_assoc[i] = spapr->numa_assoc_array[cpu->node_id][i];
> > -    }
> > +    return vcpu_assoc;
> > +}
> > +
> > +int spapr_numa_fixup_cpu_dt(SpaprMachineState *spapr, void *fdt,
> > +                            int offset, PowerPCCPU *cpu)
> > +{
> > +    g_autofree uint32_t *vcpu_assoc = NULL;
> > +    uint vcpu_assoc_size;
> >   
> > -    vcpu_assoc[vcpu_assoc_size - 1] = cpu_to_be32(index);
> > +    vcpu_assoc = spapr_numa_get_vcpu_assoc(spapr, cpu, &vcpu_assoc_size);
> >   
> >       /* Advertise NUMA via ibm,associativity */
> >       return fdt_setprop(fdt, offset, "ibm,associativity",
> > -                       vcpu_assoc, sizeof(vcpu_assoc));
> > +                       vcpu_assoc, vcpu_assoc_size);
>
>      return fdt_setprop(fdt, offset, "ibm,associativity",
>                         vcpu_assoc, VCPU_NUMA_ASSOC_SIZE * sizeof(uint32_t));
>
> >   }
> >   
> >