Re: [PATCH 3/4 v4] cxl/core: Enable Region creation on x86 with LMH

["Cheatham, Benjamin" <benjamin.cheatham@amd.com>]

On 7/24/2025 9:20 AM, Fabio M. De Francesco wrote:
> The CXL Fixed Memory Window Structure (CFMWS) describes zero or more Host
> Physical Address (HPA) windows that are associated with each CXL Host
> Bridge. Each window represents a contiguous HPA that may be interleaved
> with one or more targets (CXL v3.1 - 9.18.1.3).

Update to 3.2 spec? Sorry I forgot to mention it earlier, but you'll probably
want to do this for the whole series.

> 
> The Low Memory Hole (LMH) of x86 is a range of addresses of physical low
> memory to which systems cannot send transactions. In some cases the size
> of that hole is not compatible with the CXL hardware decoder constraint
> that the size is always aligned to 256M * Interleave Ways.

Spec ref here.

> 
> On those systems, BIOS publishes CFMWS which communicate the active System
> Physical Address (SPA) ranges that map to a subset of the Host Physical
> Address (HPA) ranges. The SPA range trims out the hole, and capacity in
> the endpoint is lost with no SPA to map to CXL HPA in that hole.
> 
> In the early stages of CXL Regions construction and attach on platforms
> with Low Memory Holes, cxl_add_to_region() fails and returns an error
> because it can't find any CXL Window that matches a given CXL Endpoint
> Decoder.
> 
> Detect a Low Memory Hole by comparing Root Decoders and Endpoint Decoders
> ranges with the use of arch_match_{spa,region}() helpers.
> 
> Match Root Decoders and CXL Regions with corresponding CXL Endpoint
> Decoders. Currently a Low Memory Holes would prevent the matching functions
> to return true.
> 
> Construct CXL Regions with HPA range's end adjusted to the matching SPA.
> 
> Allow the attach target process to complete by allowing Regions to not
> fit with alignment constraints (i.e., alignment to NIW * 256M rule).
> 
> Cc: Alison Schofield <alison.schofield@intel.com>
> Cc: Dan Williams <dan.j.williams@intel.com>
> Cc: Dave Jiang <dave.jiang@intel.com>
> Cc: Ira Weiny <ira.weiny@intel.com>
> Signed-off-by: Fabio M. De Francesco <fabio.m.de.francesco@linux.intel.com>
> ---
>  drivers/cxl/core/region.c | 53 +++++++++++++++++++++++++++++++++------
>  tools/testing/cxl/Kbuild  |  1 +
>  2 files changed, 46 insertions(+), 8 deletions(-)
> 
> diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
> index f607e7f97184..b7fdf9c4393d 100644
> --- a/drivers/cxl/core/region.c
> +++ b/drivers/cxl/core/region.c
> @@ -12,6 +12,7 @@
>  #include <linux/memory-tiers.h>
>  #include <cxlmem.h>
>  #include <cxl.h>
> +#include "platform.h"
>  #include "core.h"
>  
>  /**
> @@ -834,6 +835,8 @@ static int match_free_decoder(struct device *dev, const void *data)
>  static bool region_res_match_cxl_range(const struct cxl_region_params *p,
>  				       struct range *range)
>  {
> +	struct cxl_decoder *cxld;
> +
>  	if (!p->res)
>  		return false;
>  
> @@ -842,8 +845,15 @@ static bool region_res_match_cxl_range(const struct cxl_region_params *p,
>  	 * to be fronted by the DRAM range in current known implementation.
>  	 * This assumption will be made until a variant implementation exists.
>  	 */
> -	return p->res->start + p->cache_size == range->start &&
> -		p->res->end == range->end;
> +	if (p->res->start + p->cache_size == range->start &&
> +	    p->res->end == range->end)
> +		return true;
> +
> +	cxld = container_of(range, struct cxl_decoder, hpa_range);
> +	if (platform_region_contains(p, cxld))
> +		return true;
> +
> +	return false;

Can just return result of platform_region_contains() here.

>  }
>  
>  static int match_auto_decoder(struct device *dev, const void *data)
> @@ -1763,6 +1773,7 @@ static int match_switch_and_ep_decoders(struct device *dev, const void *data)
>  {
>  	const struct cxl_endpoint_decoder *cxled = data;
>  	struct cxl_switch_decoder *cxlsd;
> +	struct cxl_root_decoder *cxlrd;
>  	const struct range *r1, *r2;
>  
>  	if (!is_switch_decoder(dev))
> @@ -1772,8 +1783,13 @@ static int match_switch_and_ep_decoders(struct device *dev, const void *data)
>  	r1 = &cxlsd->cxld.hpa_range;
>  	r2 = &cxled->cxld.hpa_range;
>  
> -	if (is_root_decoder(dev))
> -		return range_contains(r1, r2);
> +	if (is_root_decoder(dev)) {
> +		if (range_contains(r1, r2))
> +			return 1;
> +		cxlrd = to_cxl_root_decoder(dev);
> +		if (platform_root_decoder_contains(cxlrd, cxled))
> +			return 1;
> +	}

I don't think it's possible, but the way this is written allows falling through
to the return statement below. This if statement should probably be: 

	if (is_root_decoder(dev)) {
		cxlrd = to_cxl_root_decoder(dev);
		return range_contains(r1, r2) || platform_root_decoder_contains(cxlrd, cxled));
	}

>  	return (r1->start == r2->start && r1->end == r2->end);
>  }
>  
> @@ -1990,7 +2006,7 @@ static int cxl_region_attach(struct cxl_region *cxlr,
>  	}
>  
>  	if (resource_size(cxled->dpa_res) * p->interleave_ways + p->cache_size !=
> -	    resource_size(p->res)) {
> +	    resource_size(p->res) && !platform_root_decoder_contains(cxlrd, cxled)) {
>  		dev_dbg(&cxlr->dev,
>  			"%s:%s-size-%#llx * ways-%d + cache-%#llx != region-size-%#llx\n",
>  			dev_name(&cxlmd->dev), dev_name(&cxled->cxld.dev),
> @@ -3230,7 +3246,12 @@ static int match_root_and_ep_decoders(struct device *dev, const void *data)
>  	r1 = &cxlrd->cxlsd.cxld.hpa_range;
>  	r2 = &cxled->cxld.hpa_range;
>  
> -	return range_contains(r1, r2);
> +	if (range_contains(r1, r2))
> +		return true;
> +	if (platform_root_decoder_contains(cxlrd, cxled))
> +		return true;
> +
> +	return false;

Can just return the || of the two above if statements.

>  }
>  
>  static struct cxl_decoder *
> @@ -3277,8 +3298,12 @@ static int match_region_and_ep_decoder(struct device *dev, const void *data)
>  	p = &cxlr->params;
>  
>  	guard(rwsem_read)(&cxl_region_rwsem);
> -	if (p->res && p->res->start == r->start && p->res->end == r->end)
> -		return 1;
> +	if (p->res) {
> +		if (p->res->start == r->start && p->res->end == r->end)
> +			return 1;
> +		if (platform_region_contains(p, &cxled->cxld))
> +			return 1;
> +	}

Same thing here.

>  
>  	return 0;
>  }
> @@ -3355,6 +3380,12 @@ static int __construct_region(struct cxl_region *cxlr,
>  	*res = DEFINE_RES_MEM_NAMED(hpa->start, range_len(hpa),
>  				    dev_name(&cxlr->dev));
>  
> +	/*
> +	 * Trim the HPA retrieved from hardware to fit the SPA mapped by the
> +	 * platform
> +	 */
> +	platform_res_adjust(res, cxled, cxlrd);
> +
>  	rc = cxl_extended_linear_cache_resize(cxlr, res);
>  	if (rc && rc != -EOPNOTSUPP) {
>  		/*
> @@ -3464,6 +3495,12 @@ int cxl_add_to_region(struct cxl_endpoint_decoder *cxled)
>  		cxl_find_region_by_range(cxlrd, cxled);
>  	if (!cxlr)
>  		cxlr = construct_region(cxlrd, cxled);
> +	else
> +		/*
> +		 * Adjust the Endpoint Decoder's dpa_res to fit the Region which
> +		 * it has to be attached to
> +		 */
> +		platform_res_adjust(NULL, cxled, cxlrd);

I'm 50/50 on whether these comments are unnecessary. The routine is pretty well documented
and also has an explanatory comment above the definition in platform.c. I think you
can probably remove them, but I'll defer to your/someone else's judgement here.

Thanks,
Ben