Re: [Qemu-devel] [PATCH 9/9] hostmem-ram: use whole path for memory region name with >= 3.1

[Igor Mammedov <imammedo@redhat.com>]

On Wed, 12 Sep 2018 16:55:31 +0400
Marc-André Lureau <marcandre.lureau@redhat.com> wrote:

> hostmem-file and hostmem-memfd use the whole object path

maybe add something along the lines:

consisting from user supplied 'id' and ...

> for the
> memory region name, but hostname-ram uses only the path component (the
> basename):

so path component would mean a concrete thing.

> 
> qemu -m 1024 -object memory-backend-ram,id=mem,size=1G -numa node,memdev=mem -monitor stdio
> (qemu) info ramblock
>               Block Name    PSize              Offset               Used              Total
>                      mem    4 KiB  0x0000000000000000 0x0000000040000000 0x0000000040000000
> 
> qemu -m 1024 -object memory-backend-file,id=mem,size=1G,mem-path=/tmp/foo -numa node,memdev=mem -monitor stdio
> (qemu) info ramblock
>               Block Name    PSize              Offset               Used              Total
>             /objects/mem    4 KiB  0x0000000000000000 0x0000000040000000 0x0000000040000000
> 
> qemu -m 1024 -object memory-backend-memfd,id=mem,size=1G -numa node,memdev=mem -monitor stdio
> (qemu) info ramblock
>               Block Name    PSize              Offset               Used              Total
>             /objects/mem    4 KiB  0x0000000000000000 0x0000000040000000 0x0000000040000000
> 
> Use the whole path name with >= 3.1. Having a consistent naming allow
> to migrate to different hostmem backends.
> 
> Signed-off-by: Marc-André Lureau <marcandre.lureau@redhat.com>
> ---
>  include/hw/compat.h    |  6 +++++-
>  backends/hostmem-ram.c | 47 ++++++++++++++++++++++++++++++++++++++----
>  2 files changed, 48 insertions(+), 5 deletions(-)
> 
> diff --git a/include/hw/compat.h b/include/hw/compat.h
> index 6f4d5fc647..8ce7a7057b 100644
> --- a/include/hw/compat.h
> +++ b/include/hw/compat.h
> @@ -2,7 +2,11 @@
>  #define HW_COMPAT_H
>  
>  #define HW_COMPAT_3_0 \
> -    /* empty */
> +    {\
> +        .driver   = "memory-backend-ram",\
> +        .property = "x-component-name",\
> +        .value    = "true",\
> +    },
>  
>  #define HW_COMPAT_2_12 \
>      {\
> diff --git a/backends/hostmem-ram.c b/backends/hostmem-ram.c
> index 7ddd08d370..a9eb99cf1b 100644
> --- a/backends/hostmem-ram.c
> +++ b/backends/hostmem-ram.c
> @@ -16,21 +16,56 @@
>  
>  #define TYPE_MEMORY_BACKEND_RAM "memory-backend-ram"
>  
> +typedef struct RamMemoryBackend {
> +    HostMemoryBackend parent;
> +
> +    bool component_name;
> +} RamMemoryBackend;
> +
> +#define RAM_BACKEND(obj) \
> +    OBJECT_CHECK(RamMemoryBackend, (obj), TYPE_MEMORY_BACKEND_RAM)
> +
> +static char *
> +ram_backend_get_name(RamMemoryBackend *self)
> +{
> +    /* < 3.1 use the component as memory region name */
I'd drop comment here, it belongs more to HW_COMPAT_...

> +    if (self->component_name) {
> +        return object_get_canonical_path_component(OBJECT(self));
> +    }
> +
> +    return object_get_canonical_path(OBJECT(self));
> +}
>  
>  static void
>  ram_backend_memory_alloc(HostMemoryBackend *backend, Error **errp)
>  {
> -    char *path;
> +    char *name;
>  
>      if (!backend->size) {
>          error_setg(errp, "can't create backend with size 0");
>          return;
>      }
>  
> -    path = object_get_canonical_path_component(OBJECT(backend));
> -    memory_region_init_ram_shared_nomigrate(&backend->mr, OBJECT(backend), path,
> +    name = ram_backend_get_name(RAM_BACKEND(backend));
> +    memory_region_init_ram_shared_nomigrate(&backend->mr, OBJECT(backend), name,
>                             backend->size, backend->share, errp);
> -    g_free(path);
> +    g_free(name);
> +}
> +
> +static bool
> +ram_backend_get_component_name(Object *obj, Error **errp)
> +{
> +    RamMemoryBackend *self = RAM_BACKEND(obj);
> +
> +    return self->component_name;
> +}
> +
> +static void
> +ram_backend_set_component_name(Object *obj, bool value, Error **errp)
> +{
> +    RamMemoryBackend *self = RAM_BACKEND(obj);
> +
> +    self->component_name = value;
>  }
>  
>  static void
> @@ -39,11 +74,15 @@ ram_backend_class_init(ObjectClass *oc, void *data)
>      HostMemoryBackendClass *bc = MEMORY_BACKEND_CLASS(oc);
>  
>      bc->alloc = ram_backend_memory_alloc;
> +    object_class_property_add_bool(oc, "x-component-name",
If we would convert initial RAM to backend/frontend model,
we would actually need to keep current ramblock naming scheme for 
ram_backend and even do this /i.e. non path variant/ for other
backends to keep migration stream compatible.

How about adding this property to generic memory-backend and
naming it something like 'x-use-canonical-path-for-ramblock-id'

I wonder if there is a point to use path for ramblock at all,
considering that it is single name space for all backends anyways.
Maybe by default we should just use 'id' everywhere?

> +        ram_backend_get_component_name,
> +        ram_backend_set_component_name, &error_abort);

 + object_class_property_set_description(...)

>  }
>  
>  static const TypeInfo ram_backend_info = {
>      .name = TYPE_MEMORY_BACKEND_RAM,
>      .parent = TYPE_MEMORY_BACKEND,
> +    .instance_size = sizeof(RamMemoryBackend),
>      .class_init = ram_backend_class_init,
>  };
>