Re: [Qemu-devel] [PATCH v3 1/2] memhp: extend address auto assignment to support gaps

[Bharata B Rao <bharata@linux.vnet.ibm.com>]

On Tue, Sep 29, 2015 at 04:53:28PM +0200, Igor Mammedov wrote:
> setting gap to TRUE will make sparse DIMM
> address auto allocation, leaving gaps between
> a new DIMM address and preceeding existing DIMM.
> 
> Signed-off-by: Igor Mammedov <imammedo@redhat.com>
> ---
> v2:
>   * make 'gap' boolean and just add 1 byte gap
>     when calculating new_addr for DIMM.
> ---
>  hw/i386/pc.c             |  3 ++-
>  hw/mem/pc-dimm.c         | 15 +++++++++------
>  hw/ppc/spapr.c           |  2 +-
>  include/hw/mem/pc-dimm.h |  7 ++++---
>  4 files changed, 16 insertions(+), 11 deletions(-)
> 
> diff --git a/hw/i386/pc.c b/hw/i386/pc.c
> index 461c128..ef02736 100644
> --- a/hw/i386/pc.c
> +++ b/hw/i386/pc.c
> @@ -1644,7 +1644,8 @@ static void pc_dimm_plug(HotplugHandler *hotplug_dev,
>          goto out;
>      }
> 
> -    pc_dimm_memory_plug(dev, &pcms->hotplug_memory, mr, align, &local_err);
> +    pc_dimm_memory_plug(dev, &pcms->hotplug_memory, mr, align, false,
> +                        &local_err);
>      if (local_err) {
>          goto out;
>      }
> diff --git a/hw/mem/pc-dimm.c b/hw/mem/pc-dimm.c
> index bb04862..6cc6ac3 100644
> --- a/hw/mem/pc-dimm.c
> +++ b/hw/mem/pc-dimm.c
> @@ -32,7 +32,8 @@ typedef struct pc_dimms_capacity {
>  } pc_dimms_capacity;
> 
>  void pc_dimm_memory_plug(DeviceState *dev, MemoryHotplugState *hpms,
> -                         MemoryRegion *mr, uint64_t align, Error **errp)
> +                         MemoryRegion *mr, uint64_t align, bool gap,
> +                         Error **errp)
>  {
>      int slot;
>      MachineState *machine = MACHINE(qdev_get_machine());
> @@ -48,7 +49,7 @@ void pc_dimm_memory_plug(DeviceState *dev, MemoryHotplugState *hpms,
> 
>      addr = pc_dimm_get_free_addr(hpms->base,
>                                   memory_region_size(&hpms->mr),
> -                                 !addr ? NULL : &addr, align,
> +                                 !addr ? NULL : &addr, align, gap,
>                                   memory_region_size(mr), &local_err);
>      if (local_err) {
>          goto out;
> @@ -287,8 +288,8 @@ static int pc_dimm_built_list(Object *obj, void *opaque)
> 
>  uint64_t pc_dimm_get_free_addr(uint64_t address_space_start,
>                                 uint64_t address_space_size,
> -                               uint64_t *hint, uint64_t align, uint64_t size,
> -                               Error **errp)
> +                               uint64_t *hint, uint64_t align, bool gap,
> +                               uint64_t size, Error **errp)
>  {
>      GSList *list = NULL, *item;
>      uint64_t new_addr, ret = 0;
> @@ -333,13 +334,15 @@ uint64_t pc_dimm_get_free_addr(uint64_t address_space_start,
>              goto out;
>          }
> 
> -        if (ranges_overlap(dimm->addr, dimm_size, new_addr, size)) {
> +        if (ranges_overlap(dimm->addr, dimm_size, new_addr,
> +                           size + (gap ? 1 : 0))) {
>              if (hint) {
>                  DeviceState *d = DEVICE(dimm);
>                  error_setg(errp, "address range conflicts with '%s'", d->id);
>                  goto out;
>              }
> -            new_addr = QEMU_ALIGN_UP(dimm->addr + dimm_size, align);
> +            new_addr = QEMU_ALIGN_UP(dimm->addr + dimm_size + (gap ? 1 : 0),
> +                                     align);

This change of adding 1 byte gap will break PowerPC memory hotplug in
its current form.

Currently we divide hotpluggable memory region into chuncks of 256M
and there is a DR connector object for each such chunk. The DR connector
object maintains/controls the state transitions of that memory chunk
as per PAPR specifications.

Now after this 1 byte gap, we end up having a 64K alignment (default
page size) for the DIMM address and end up having an address for which
there is no DR connector object. I will have to revisit the parts of
the code in PowerPC that creates DR connector objects and looks them up
by address.

Regards,
Bharata.