Re: [for-5.0 3/4] spapr: Clean up RMA size calculation

[David Gibson <david@gibson.dropbear.id.au>]

[-- Attachment #1: Type: text/plain, Size: 5497 bytes --]

On Tue, Dec 03, 2019 at 02:44:06PM +1100, Alexey Kardashevskiy wrote:
> 
> 
> On 29/11/2019 12:35, David Gibson wrote:
> > Move the calculation of the Real Mode Area (RMA) size into a helper
> > function.  While we're there clean it up and correct it in a few ways:
> >   * Add comments making it clearer where the various constraints come from
> >   * Remove a pointless check that the RMA fits within Node 0 (we've just
> >     clamped it so that it does)
> >   * The 16GiB limit we apply is only correct for POWER8, but there is also
> >     a 1TiB limit that applies on POWER9.
> > 
> > Signed-off-by: David Gibson <david@gibson.dropbear.id.au>
> > ---
> >  hw/ppc/spapr.c | 57 +++++++++++++++++++++++++++++++-------------------
> >  1 file changed, 35 insertions(+), 22 deletions(-)
> > 
> > diff --git a/hw/ppc/spapr.c b/hw/ppc/spapr.c
> > index 52c39daa99..7efd4f2b85 100644
> > --- a/hw/ppc/spapr.c
> > +++ b/hw/ppc/spapr.c
> > @@ -2664,6 +2664,40 @@ static PCIHostState *spapr_create_default_phb(void)
> >      return PCI_HOST_BRIDGE(dev);
> >  }
> >  
> > +static hwaddr spapr_rma_size(SpaprMachineState *spapr, Error **errp)
> > +{
> > +    MachineState *machine = MACHINE(spapr);
> > +    hwaddr rma_size = machine->ram_size;
> > +    hwaddr node0_size = spapr_node0_size(machine);
> > +
> > +    /* RMA has to fit in the first NUMA node */
> > +    rma_size = MIN(rma_size, node0_size);
> > +
> > +    /*
> > +     * VRMA access is via a special 1TiB SLB mapping, so the RMA can
> > +     * never exceed that
> > +     */
> > +    rma_size = MIN(rma_size, TiB);
> > +
> > +    /*
> > +     * RMA size is controlled in hardware by LPCR[RMLS].  On POWER8
> 
> 
> RMA is controlled by LPCR on P8 but the RMLS bits on P9 are reserved
> (also reserved in PowerISA 3.0).
> 
> 
> > +     * the largest RMA that can be specified there is 16GiB
> 
> 
> The P8 user manual says:
> ===
> The following RMO sizes are available for the POWER8 processor.
> The RMLS[34:37] field in the LPCR defines the RMO sizes, as described below.
> 1000 - 32 MB
> 0011 - 64 MB
> 0111 - 128 MB
> 0100 - 256 MB
> 0010 - 1 GB
> 0001 - 16 GB
> 0000 - 256 GB
> ===
> 
> The maximum seems to be 256GiB.

Huh.  Ok, looks like I was wrong about where that 16GiB clamp came
from originally.  I wonder where it *did* come from, or if it's simply
wrong.

> > +     */
> > +    if (!ppc_type_check_compat(machine->cpu_type, CPU_POWERPC_LOGICAL_3_00,
> > +                               0, spapr->max_compat_pvr)) {
> > +        rma_size = MIN(rma_size, 16 * GiB);
> > +    }
> > +
> > +    if (rma_size < (MIN_RMA_SLOF * MiB)) {
> 
> 
> nit: it is time to redefine MIN_RMA_SLOF to use MiBs imho :)

Yeah, I'd thought about that too.  I've added a patch to clean that up.

> 
> 
> > +        error_setg(errp,
> > +"pSeries SLOF firmware requires >= %ldMiB guest RMA (Real Mode Area)",
> > +                   MIN_RMA_SLOF);
> 
> Something went wrong with formatting here.

Actually, that's intentional.  The idea is to keep the string on one
line for greppability, without going too far to the right.  I believe
that's a generally accepted exception to the usual formatting rules in
qemu.

> 
> Otherwise looks good. Thanks,
> 
> 
> 
> > +        return -1;
> > +    }
> > +
> > +    return rma_size;
> > +}
> > +
> >  /* pSeries LPAR / sPAPR hardware init */
> >  static void spapr_machine_init(MachineState *machine)
> >  {
> > @@ -2675,7 +2709,6 @@ static void spapr_machine_init(MachineState *machine)
> >      int i;
> >      MemoryRegion *sysmem = get_system_memory();
> >      MemoryRegion *ram = g_new(MemoryRegion, 1);
> > -    hwaddr node0_size = spapr_node0_size(machine);
> >      long load_limit, fw_size;
> >      char *filename;
> >      Error *resize_hpt_err = NULL;
> > @@ -2715,20 +2748,7 @@ static void spapr_machine_init(MachineState *machine)
> >          exit(1);
> >      }
> >  
> > -    spapr->rma_size = node0_size;
> > -
> > -    /* Actually we don't support unbounded RMA anymore since we added
> > -     * proper emulation of HV mode. The max we can get is 16G which
> > -     * also happens to be what we configure for PAPR mode so make sure
> > -     * we don't do anything bigger than that
> > -     */
> > -    spapr->rma_size = MIN(spapr->rma_size, 0x400000000ull);
> > -
> > -    if (spapr->rma_size > node0_size) {
> > -        error_report("Numa node 0 has to span the RMA (%#08"HWADDR_PRIx")",
> > -                     spapr->rma_size);
> > -        exit(1);
> > -    }
> > +    spapr->rma_size = spapr_rma_size(spapr, &error_fatal);
> >  
> >      /* Setup a load limit for the ramdisk leaving room for SLOF and FDT */
> >      load_limit = MIN(spapr->rma_size, RTAS_MAX_ADDR) - FW_OVERHEAD;
> > @@ -2956,13 +2976,6 @@ static void spapr_machine_init(MachineState *machine)
> >          }
> >      }
> >  
> > -    if (spapr->rma_size < (MIN_RMA_SLOF * MiB)) {
> > -        error_report(
> > -            "pSeries SLOF firmware requires >= %ldM guest RMA (Real Mode Area memory)",
> > -            MIN_RMA_SLOF);
> > -        exit(1);
> > -    }
> > -
> >      if (kernel_filename) {
> >          uint64_t lowaddr = 0;
> >  
> > 
> 

-- 
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

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]