Re: [Qemu-devel] [PATCH qemu v6 08/15] spapr_iommu: Introduce "enabled" state for TCE table

[Thomas Huth <thuth@redhat.com>]

Am Sat, 11 Apr 2015 01:24:37 +1000
schrieb Alexey Kardashevskiy <aik@ozlabs.ru>:

> Currently TCE tables are created once at start and their size never
> changes. We are going to change that by introducing a Dynamic DMA windows
> support where DMA configuration may change during the guest execution.
> 
> This changes spapr_tce_new_table() to create an empty stub object. Only
> LIOBN is assigned by the time of creation. It still will be called once
> at the owner object (VIO or PHB) creation.
> 
> This introduces an "enabled" state for TCE table objects with two
> helper functions - spapr_tce_table_enable()/spapr_tce_table_disable().
> spapr_tce_table_enable() receives TCE table parameters and allocates
> a guest view of the TCE table (in the user space or KVM).
> spapr_tce_table_disable() disposes the table.
> 
> Follow up patches will disable+enable tables on reset (system reset
> or DDW reset).
> 
> No visible change in behaviour is expected except the actual table
> will be reallocated every reset. We might optimize this later.
> 
> The other way to implement this would be dynamically create/remove
> the TCE table QOM objects but this would make migration impossible
> as migration expects all QOM objects to exist at the receiver
> so we have to have TCE table objects created when migration begins.
> 
> Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
> ---
> Changes:
> v6:
> * got rid of set_props()
> ---
>  hw/ppc/spapr_iommu.c    | 104 +++++++++++++++++++++++++++++++-----------------
>  hw/ppc/spapr_pci.c      |  16 +++++---
>  hw/ppc/spapr_pci_vfio.c |  10 ++---
>  hw/ppc/spapr_vio.c      |   9 ++---
>  include/hw/ppc/spapr.h  |  11 ++---
>  5 files changed, 93 insertions(+), 57 deletions(-)
> 
> diff --git a/hw/ppc/spapr_iommu.c b/hw/ppc/spapr_iommu.c
> index a14cdc4..64f20f2 100644
> --- a/hw/ppc/spapr_iommu.c
> +++ b/hw/ppc/spapr_iommu.c
> @@ -126,8 +126,47 @@ static MemoryRegionIOMMUOps spapr_iommu_ops = {
>  static int spapr_tce_table_realize(DeviceState *dev)
>  {
>      sPAPRTCETable *tcet = SPAPR_TCE_TABLE(dev);
> +
> +    QLIST_INSERT_HEAD(&spapr_tce_tables, tcet, list);
> +
> +    vmstate_register(DEVICE(tcet), tcet->liobn, &vmstate_spapr_tce_table,
> +                     tcet);
> +
> +    return 0;
> +}
> +
> +sPAPRTCETable *spapr_tce_new_table(DeviceState *owner, uint32_t liobn)
> +{
> +    sPAPRTCETable *tcet;
> +    char tmp[64];

Hm, any reason for that value "64"? 64 bytes seem much more than enough
here, if I count correctly, you only need 19 bytes max.

> +    if (spapr_tce_find_by_liobn(liobn)) {
> +        fprintf(stderr, "Attempted to create TCE table with duplicate"
> +                " LIOBN 0x%x\n", liobn);
> +        return NULL;
> +    }
> +
> +    tcet = SPAPR_TCE_TABLE(object_new(TYPE_SPAPR_TCE_TABLE));
> +    tcet->liobn = liobn;
> +
> +    snprintf(tmp, sizeof(tmp), "tce-table-%x", liobn);
> +    object_property_add_child(OBJECT(owner), tmp, OBJECT(tcet), NULL);
> +
> +    object_property_set_bool(OBJECT(tcet), true, "realized", NULL);
> +
> +    trace_spapr_iommu_new_table(tcet->liobn, tcet, tcet->table, tcet->fd);
> +
> +    return tcet;
> +}

 Thomas