Vishal Verma wrote:
> Add a cxl_region object to libcxl that represents a CXL region. CXL
> regions are made up of one or more cxl_memdev 'targets'. The
> relationship between a target and a region is conveyed with a
> cxl_memdev_mapping object.
>
> CXL regions are childeren of root decoders, and are organized as such.
> Mapping objects are childeren of a CXL region. Introduce the two
> classes of objects themselves, and common accessors related to them.
>
> Cc: Dan Williams <dan.j.willi...@intel.com>
> Signed-off-by: Vishal Verma <vishal.l.ve...@intel.com>
> ---
> cxl/lib/private.h | 34 ++++
> cxl/lib/libcxl.c | 421 +++++++++++++++++++++++++++++++++++++++++++--
> cxl/libcxl.h | 41 +++++
> .clang-format | 2 +
> cxl/lib/libcxl.sym | 20 +++
> 5 files changed, 508 insertions(+), 10 deletions(-)
>
> diff --git a/cxl/lib/private.h b/cxl/lib/private.h
> index 832a815..d58a73b 100644
> --- a/cxl/lib/private.h
> +++ b/cxl/lib/private.h
> @@ -116,7 +116,41 @@ struct cxl_decoder {
> bool accelmem_capable;
> bool locked;
> enum cxl_decoder_target_type target_type;
> + int regions_init;
> struct list_head targets;
> + struct list_head regions;
> +};
> +
> +enum cxl_region_commit {
> + CXL_REGION_COMMIT_UNKNOWN = -1,
> + CXL_REGION_DECOMMIT = 0,
I had been calling the opposite of the commit operation "reset()". How
about CXL_DECODE_COMMIT, and CXL_DECODE_RESET since it's the "decode"
not the "region" that's being mutated.
However, this isn't an exported definition, only CXL_REGION_COMMIT
appears in code, and nothing else in this patch looks like it needs
updating, so just leave it for now.
Reviewed-by: Dan Williams <dan.j.willi...@intel.com>
