cxl/acpi: Minimize granularity for x1 interleaves
authorDan Williams <dan.j.williams@intel.com>
Sat, 23 Jul 2022 00:56:09 +0000 (17:56 -0700)
committerDan Williams <dan.j.williams@intel.com>
Mon, 1 Aug 2022 22:36:33 +0000 (15:36 -0700)
The kernel enforces that region granularity is >= to the top-level
interleave-granularity for the given CXL window. However, when the CXL
window interleave is x1, i.e. non-interleaved at the host bridge level,
then the specified granularity does not matter. Override the window
specified granularity to the CXL minimum so that any valid region
granularity is >= to the root granularity.

Reported-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
Reviewed-by: Vishal Verma <vishal.l.verma@intel.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>
Link: https://lore.kernel.org/r/165853776917.2430596.16823264262010844458.stgit@dwillia2-xfh.jf.intel.com
[djbw: add CXL_DECODER_MIN_GRANULARITY per vishal]
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
drivers/cxl/acpi.c
drivers/cxl/cxl.h

index eb43626..fb64968 100644 (file)
@@ -140,6 +140,12 @@ static int cxl_parse_cfmws(union acpi_subtable_headers *header, void *arg,
                .end = res->end,
        };
        cxld->interleave_ways = ways;
+       /*
+        * Minimize the x1 granularity to advertise support for any
+        * valid region granularity
+        */
+       if (ways == 1)
+               ig = CXL_DECODER_MIN_GRANULARITY;
        cxld->interleave_granularity = ig;
 
        rc = cxl_decoder_add(cxld, target_map);
index 969953c..bc604b7 100644 (file)
@@ -248,6 +248,8 @@ enum cxl_decoder_type {
  */
 #define CXL_DECODER_MAX_INTERLEAVE 16
 
+#define CXL_DECODER_MIN_GRANULARITY 256
+
 /**
  * struct cxl_decoder - Common CXL HDM Decoder Attributes
  * @dev: this decoder's device