[PATCH 0/4] cxl/test: Add unit testing for extended linear cache

[PATCH 0/4] cxl/test: Add unit testing for extended linear cache

[Dave Jiang]

The series adds support for mocking of extended linear cache support
in cxl_test. A module parameter is introduced to activate the mocking.
Tests will be added to CXL CLI user tools for regression testing.

This series depends on the patch [1] that provides the extended linear
cache size via sysfs.

[1]: https://lore.kernel.org/linux-cxl/20251022203052.4078527-1-dave.jiang@intel.com/

Dave Jiang (4):
  cxl/test: Standardize CXL auto region size
  cxl/test: Add cxl_test CFMWS support for extended linear cache
  cxl/test: Add support for acpi extended linear cache
  cxl: Adjust offset calculation for poison injection

 drivers/cxl/core/region.c     |  2 ++
 tools/testing/cxl/Kbuild      |  1 +
 tools/testing/cxl/test/cxl.c  | 55 ++++++++++++++++++++++++++++++++---
 tools/testing/cxl/test/mock.c | 20 +++++++++++++
 tools/testing/cxl/test/mock.h |  3 ++
 5 files changed, 77 insertions(+), 4 deletions(-)


base-commit: 659163e60a142e8c3bd66f9865e1f8ffcb8cbd9b
-- 
2.51.0

[PATCH 1/4] cxl/test: Standardize CXL auto region size

[Dave Jiang]

Create a global define for the size of the mock CXL auto region used
in cxl_test. Remove the declared size in mock_init_hdm_decoder()
function.

Signed-off-by: Dave Jiang <dave.jiang@intel.com>
---
 tools/testing/cxl/test/cxl.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/tools/testing/cxl/test/cxl.c b/tools/testing/cxl/test/cxl.c
index 2d135ca533d0..d53b72bc034f 100644
--- a/tools/testing/cxl/test/cxl.c
+++ b/tools/testing/cxl/test/cxl.c
@@ -26,6 +26,8 @@ static int interleave_arithmetic;
 #define NR_CXL_PORT_DECODERS 8
 #define NR_BRIDGES (NR_CXL_HOST_BRIDGES + NR_CXL_SINGLE_HOST + NR_CXL_RCH)
 
+#define MOCK_AUTO_REGION_SIZE SZ_512M
+
 static struct platform_device *cxl_acpi;
 static struct platform_device *cxl_host_bridge[NR_CXL_HOST_BRIDGES];
 #define NR_MULTI_ROOT (NR_CXL_HOST_BRIDGES * NR_CXL_ROOT_PORTS)
@@ -738,7 +740,6 @@ static void mock_init_hdm_decoder(struct cxl_decoder *cxld)
 	struct cxl_endpoint_decoder *cxled;
 	struct cxl_switch_decoder *cxlsd;
 	struct cxl_port *port, *iter;
-	const int size = SZ_512M;
 	struct cxl_memdev *cxlmd;
 	struct cxl_dport *dport;
 	struct device *dev;
@@ -783,7 +784,7 @@ static void mock_init_hdm_decoder(struct cxl_decoder *cxld)
 	base = window->base_hpa;
 	cxld->hpa_range = (struct range) {
 		.start = base,
-		.end = base + size - 1,
+		.end = base + MOCK_AUTO_REGION_SIZE - 1,
 	};
 
 	cxld->interleave_ways = 2;
@@ -792,7 +793,8 @@ static void mock_init_hdm_decoder(struct cxl_decoder *cxld)
 	cxld->flags = CXL_DECODER_F_ENABLE;
 	cxled->state = CXL_DECODER_STATE_AUTO;
 	port->commit_end = cxld->id;
-	devm_cxl_dpa_reserve(cxled, 0, size / cxld->interleave_ways, 0);
+	devm_cxl_dpa_reserve(cxled, 0,
+			     MOCK_AUTO_REGION_SIZE / cxld->interleave_ways, 0);
 	cxld->commit = mock_decoder_commit;
 	cxld->reset = mock_decoder_reset;
 
@@ -841,7 +843,7 @@ static void mock_init_hdm_decoder(struct cxl_decoder *cxld)
 		cxld->interleave_granularity = 4096;
 		cxld->hpa_range = (struct range) {
 			.start = base,
-			.end = base + size - 1,
+			.end = base + MOCK_AUTO_REGION_SIZE - 1,
 		};
 		put_device(dev);
 	}
-- 
2.51.0

Re: [PATCH 1/4] cxl/test: Standardize CXL auto region size

[Jonathan Cameron]

On Fri, 31 Oct 2025 10:32:21 -0700
Dave Jiang <dave.jiang@intel.com> wrote:

> Create a global define for the size of the mock CXL auto region used
> in cxl_test. Remove the declared size in mock_init_hdm_decoder()
> function.
> 
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Seems reasonable.
Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com>

Re: [PATCH 1/4] cxl/test: Standardize CXL auto region size

[Alison Schofield]

On Fri, Oct 31, 2025 at 10:32:21AM -0700, Dave Jiang wrote:
> Create a global define for the size of the mock CXL auto region used
> in cxl_test. Remove the declared size in mock_init_hdm_decoder()
> function.

Tested-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>

[PATCH 2/4] cxl/test: Add cxl_test CFMWS support for extended linear cache

[Dave Jiang]

Add a module parameter to allow activation of extended linear cache
on the auto region for cxl_test. The current platform implementation
for extended linear cache is 1:1 of DRAM and CXL memory. A CFMWS is
created with the size of both memory together where DRAM takes the
first part of the memory range and CXL covers the second part. The
current CXL auto region on cxl_test consists of 2 256M devices that
creates a 512M region. The new extended linear cache setup will have
512M DRAM and 512M CXL memory for a total of 1G CFMWS. The hardware
decoders must have their starting offset moved to after the DRAM region
to handle the CXL regions.

Signed-off-by: Dave Jiang <dave.jiang@intel.com>
---
 tools/testing/cxl/test/cxl.c | 24 ++++++++++++++++++++++++
 1 file changed, 24 insertions(+)

diff --git a/tools/testing/cxl/test/cxl.c b/tools/testing/cxl/test/cxl.c
index d53b72bc034f..4e9007f2d510 100644
--- a/tools/testing/cxl/test/cxl.c
+++ b/tools/testing/cxl/test/cxl.c
@@ -15,6 +15,7 @@
 #include "mock.h"
 
 static int interleave_arithmetic;
+static bool extended_linear_cache;
 
 #define FAKE_QTG_ID	42
 
@@ -428,6 +429,24 @@ static struct cxl_mock_res *alloc_mock_res(resource_size_t size, int align)
 	return res;
 }
 
+/*
+ * Only update CFMWS0 as this is used by the auto region.
+ */
+static void cfmws_elc_update(struct acpi_cedt_cfmws *window, int index)
+{
+	if (!extended_linear_cache)
+		return;
+
+	if (index != 0)
+		return;
+
+	/*
+	 * The window size should be 2x of the CXL region size where half is
+	 * DRAM and half is CXL
+	 */
+	window->window_size = MOCK_AUTO_REGION_SIZE * 2;
+}
+
 static int populate_cedt(void)
 {
 	struct cxl_mock_res *res;
@@ -452,6 +471,7 @@ static int populate_cedt(void)
 	for (i = cfmws_start; i <= cfmws_end; i++) {
 		struct acpi_cedt_cfmws *window = mock_cfmws[i];
 
+		cfmws_elc_update(window, i);
 		res = alloc_mock_res(window->window_size, SZ_256M);
 		if (!res)
 			return -ENOMEM;
@@ -782,6 +802,8 @@ static void mock_init_hdm_decoder(struct cxl_decoder *cxld)
 	}
 
 	base = window->base_hpa;
+	if (extended_linear_cache)
+		base += MOCK_AUTO_REGION_SIZE;
 	cxld->hpa_range = (struct range) {
 		.start = base,
 		.end = base + MOCK_AUTO_REGION_SIZE - 1,
@@ -1608,6 +1630,8 @@ static __exit void cxl_test_exit(void)
 
 module_param(interleave_arithmetic, int, 0444);
 MODULE_PARM_DESC(interleave_arithmetic, "Modulo:0, XOR:1");
+module_param(extended_linear_cache, bool, 0444);
+MODULE_PARM_DESC(extended_linear_cache, "Enable extended linear cache support");
 module_init(cxl_test_init);
 module_exit(cxl_test_exit);
 MODULE_LICENSE("GPL v2");
-- 
2.51.0

Re: [PATCH 2/4] cxl/test: Add cxl_test CFMWS support for extended linear cache

[Jonathan Cameron]

On Fri, 31 Oct 2025 10:32:22 -0700
Dave Jiang <dave.jiang@intel.com> wrote:

> Add a module parameter to allow activation of extended linear cache
> on the auto region for cxl_test. The current platform implementation
> for extended linear cache is 1:1 of DRAM and CXL memory. A CFMWS is
> created with the size of both memory together where DRAM takes the
> first part of the memory range and CXL covers the second part. The
> current CXL auto region on cxl_test consists of 2 256M devices that
> creates a 512M region. The new extended linear cache setup will have
> 512M DRAM and 512M CXL memory for a total of 1G CFMWS. The hardware
> decoders must have their starting offset moved to after the DRAM region
> to handle the CXL regions.
> 
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com>

Trivial comment though - feel free to ignore.
> ---
>  tools/testing/cxl/test/cxl.c | 24 ++++++++++++++++++++++++
>  1 file changed, 24 insertions(+)
> 
> diff --git a/tools/testing/cxl/test/cxl.c b/tools/testing/cxl/test/cxl.c
> index d53b72bc034f..4e9007f2d510 100644
> --- a/tools/testing/cxl/test/cxl.c
> +++ b/tools/testing/cxl/test/cxl.c
> @@ -15,6 +15,7 @@
>  #include "mock.h"
>  
>  static int interleave_arithmetic;
> +static bool extended_linear_cache;
>  
>  #define FAKE_QTG_ID	42
>  
> @@ -428,6 +429,24 @@ static struct cxl_mock_res *alloc_mock_res(resource_size_t size, int align)
>  	return res;
>  }
>  
> +/*
> + * Only update CFMWS0 as this is used by the auto region.
Single line comment syntax seems fine here.  Slightly less scrolling always
preferred!

> + */
> +static void cfmws_elc_update(struct acpi_cedt_cfmws *window, int index)
> +{
> +	if (!extended_linear_cache)
> +		return;
> +
> +	if (index != 0)
> +		return;
> +
> +	/*
> +	 * The window size should be 2x of the CXL region size where half is
> +	 * DRAM and half is CXL
> +	 */
> +	window->window_size = MOCK_AUTO_REGION_SIZE * 2;
> +}
> +
>  static int populate_cedt(void)
>  {
>  	struct cxl_mock_res *res;
> @@ -452,6 +471,7 @@ static int populate_cedt(void)
>  	for (i = cfmws_start; i <= cfmws_end; i++) {
>  		struct acpi_cedt_cfmws *window = mock_cfmws[i];
>  
> +		cfmws_elc_update(window, i);
>  		res = alloc_mock_res(window->window_size, SZ_256M);
>  		if (!res)
>  			return -ENOMEM;
> @@ -782,6 +802,8 @@ static void mock_init_hdm_decoder(struct cxl_decoder *cxld)
>  	}
>  
>  	base = window->base_hpa;
> +	if (extended_linear_cache)
> +		base += MOCK_AUTO_REGION_SIZE;
>  	cxld->hpa_range = (struct range) {
>  		.start = base,
>  		.end = base + MOCK_AUTO_REGION_SIZE - 1,
> @@ -1608,6 +1630,8 @@ static __exit void cxl_test_exit(void)
>  
>  module_param(interleave_arithmetic, int, 0444);
>  MODULE_PARM_DESC(interleave_arithmetic, "Modulo:0, XOR:1");
> +module_param(extended_linear_cache, bool, 0444);
> +MODULE_PARM_DESC(extended_linear_cache, "Enable extended linear cache support");
>  module_init(cxl_test_init);
>  module_exit(cxl_test_exit);
>  MODULE_LICENSE("GPL v2");

Re: [PATCH 2/4] cxl/test: Add cxl_test CFMWS support for extended linear cache

[Alison Schofield]

On Fri, Oct 31, 2025 at 10:32:22AM -0700, Dave Jiang wrote:
> Add a module parameter to allow activation of extended linear cache
> on the auto region for cxl_test. The current platform implementation
> for extended linear cache is 1:1 of DRAM and CXL memory. A CFMWS is
> created with the size of both memory together where DRAM takes the
> first part of the memory range and CXL covers the second part. The
> current CXL auto region on cxl_test consists of 2 256M devices that
> creates a 512M region. The new extended linear cache setup will have
> 512M DRAM and 512M CXL memory for a total of 1G CFMWS. The hardware
> decoders must have their starting offset moved to after the DRAM region
> to handle the CXL regions.

Tested-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>

[PATCH 3/4] cxl/test: Add support for acpi extended linear cache

[Dave Jiang]

Add the mock wrappers for hmat_get_extended_linear_cache_size() in order
to emulate the ACPI helper function for the regions that are mock'd by
cxl_test.

Signed-off-by: Dave Jiang <dave.jiang@intel.com>
---
 tools/testing/cxl/Kbuild      |  1 +
 tools/testing/cxl/test/cxl.c  | 21 +++++++++++++++++++++
 tools/testing/cxl/test/mock.c | 20 ++++++++++++++++++++
 tools/testing/cxl/test/mock.h |  3 +++
 4 files changed, 45 insertions(+)

diff --git a/tools/testing/cxl/Kbuild b/tools/testing/cxl/Kbuild
index 0d5ce4b74b9f..65e9ac227aee 100644
--- a/tools/testing/cxl/Kbuild
+++ b/tools/testing/cxl/Kbuild
@@ -11,6 +11,7 @@ ldflags-y += --wrap=cxl_rcd_component_reg_phys
 ldflags-y += --wrap=cxl_endpoint_parse_cdat
 ldflags-y += --wrap=cxl_dport_init_ras_reporting
 ldflags-y += --wrap=devm_cxl_endpoint_decoders_setup
+ldflags-y += --wrap=hmat_get_extended_linear_cache_size
 
 DRIVERS := ../../../drivers
 CXL_SRC := $(DRIVERS)/cxl
diff --git a/tools/testing/cxl/test/cxl.c b/tools/testing/cxl/test/cxl.c
index 4e9007f2d510..b537f65e4c11 100644
--- a/tools/testing/cxl/test/cxl.c
+++ b/tools/testing/cxl/test/cxl.c
@@ -613,6 +613,25 @@ mock_acpi_evaluate_integer(acpi_handle handle, acpi_string pathname,
 	return AE_OK;
 }
 
+static int
+mock_hmat_get_extended_linear_cache_size(struct resource *backing_res,
+					 int nid, resource_size_t *cache_size)
+{
+	struct acpi_cedt_cfmws *window = mock_cfmws[0];
+	struct resource cfmws0_res =
+		DEFINE_RES_MEM(window->base_hpa, window->window_size);
+
+	if (!extended_linear_cache ||
+	    !resource_contains(&cfmws0_res, backing_res)) {
+		return hmat_get_extended_linear_cache_size(backing_res,
+							   nid, cache_size);
+	}
+
+	*cache_size = MOCK_AUTO_REGION_SIZE;
+
+	return 0;
+}
+
 static struct pci_bus mock_pci_bus[NR_BRIDGES];
 static struct acpi_pci_root mock_pci_root[ARRAY_SIZE(mock_pci_bus)] = {
 	[0] = {
@@ -1141,6 +1160,8 @@ static struct cxl_mock_ops cxl_mock_ops = {
 	.devm_cxl_port_enumerate_dports = mock_cxl_port_enumerate_dports,
 	.cxl_endpoint_parse_cdat = mock_cxl_endpoint_parse_cdat,
 	.devm_cxl_add_dport_by_dev = mock_cxl_add_dport_by_dev,
+	.hmat_get_extended_linear_cache_size =
+		mock_hmat_get_extended_linear_cache_size,
 	.list = LIST_HEAD_INIT(cxl_mock_ops.list),
 };
 
diff --git a/tools/testing/cxl/test/mock.c b/tools/testing/cxl/test/mock.c
index 995269a75cbd..bff5853717f0 100644
--- a/tools/testing/cxl/test/mock.c
+++ b/tools/testing/cxl/test/mock.c
@@ -111,6 +111,26 @@ acpi_status __wrap_acpi_evaluate_integer(acpi_handle handle,
 }
 EXPORT_SYMBOL(__wrap_acpi_evaluate_integer);
 
+int __wrap_hmat_get_extended_linear_cache_size(struct resource *backing_res,
+					       int nid,
+					       resource_size_t *cache_size)
+{
+	int index, rc;
+	struct cxl_mock_ops *ops = get_cxl_mock_ops(&index);
+
+	if (ops)
+		rc = ops->hmat_get_extended_linear_cache_size(backing_res, nid,
+							      cache_size);
+	else
+		rc = hmat_get_extended_linear_cache_size(backing_res, nid,
+							 cache_size);
+
+	put_cxl_mock_ops(index);
+
+	return rc;
+}
+EXPORT_SYMBOL_GPL(__wrap_hmat_get_extended_linear_cache_size);
+
 struct acpi_pci_root *__wrap_acpi_pci_find_root(acpi_handle handle)
 {
 	int index;
diff --git a/tools/testing/cxl/test/mock.h b/tools/testing/cxl/test/mock.h
index 4ed932e76aae..bfd36cbdbae2 100644
--- a/tools/testing/cxl/test/mock.h
+++ b/tools/testing/cxl/test/mock.h
@@ -25,6 +25,9 @@ struct cxl_mock_ops {
 	void (*cxl_endpoint_parse_cdat)(struct cxl_port *port);
 	struct cxl_dport *(*devm_cxl_add_dport_by_dev)(struct cxl_port *port,
 						       struct device *dport_dev);
+	int (*hmat_get_extended_linear_cache_size)(struct resource *backing_res,
+						   int nid,
+						   resource_size_t *cache_size);
 };
 
 void register_cxl_mock_ops(struct cxl_mock_ops *ops);
-- 
2.51.0

Re: [PATCH 3/4] cxl/test: Add support for acpi extended linear cache

[Jonathan Cameron]

On Fri, 31 Oct 2025 10:32:23 -0700
Dave Jiang <dave.jiang@intel.com> wrote:

> Add the mock wrappers for hmat_get_extended_linear_cache_size() in order
> to emulate the ACPI helper function for the regions that are mock'd by
> cxl_test.
> 
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>
LGTM
Reviewed-by: Jonathan Cameron <jonathan.cameron@huawei.com>

Re: [PATCH 3/4] cxl/test: Add support for acpi extended linear cache

[Alison Schofield]

On Fri, Oct 31, 2025 at 10:32:23AM -0700, Dave Jiang wrote:
> Add the mock wrappers for hmat_get_extended_linear_cache_size() in order
> to emulate the ACPI helper function for the regions that are mock'd by
> cxl_test.

Tested-by: Alison Schofield <alison.schofield@intel.com>
Reviewed-by: Alison Schofield <alison.schofield@intel.com>

[PATCH 4/4] cxl: Adjust offset calculation for poison injection

[Dave Jiang]

The HPA to DPA translation for poison injection assumes that the
base address starts from where the CXL region begins. When the
extended linear cache is active, the offset can be within the DRAM
region. Adjust the offset so that it correctly reflects the offset
within the CXL region.

Signed-off-by: Dave Jiang <dave.jiang@intel.com>
---
 drivers/cxl/core/region.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
index 095f5dcd17a1..eaf085aa44c4 100644
--- a/drivers/cxl/core/region.c
+++ b/drivers/cxl/core/region.c
@@ -3731,6 +3731,7 @@ static int cxl_region_debugfs_poison_inject(void *data, u64 offset)
 	if (validate_region_offset(cxlr, offset))
 		return -EINVAL;
 
+	offset -= cxlr->params.cache_size;
 	rc = region_offset_to_dpa_result(cxlr, offset, &result);
 	if (rc || !result.cxlmd || result.dpa == ULLONG_MAX) {
 		dev_dbg(&cxlr->dev,
@@ -3763,6 +3764,7 @@ static int cxl_region_debugfs_poison_clear(void *data, u64 offset)
 	if (validate_region_offset(cxlr, offset))
 		return -EINVAL;
 
+	offset -= cxlr->params.cache_size;
 	rc = region_offset_to_dpa_result(cxlr, offset, &result);
 	if (rc || !result.cxlmd || result.dpa == ULLONG_MAX) {
 		dev_dbg(&cxlr->dev,
-- 
2.51.0

Re: [PATCH 4/4] cxl: Adjust offset calculation for poison injection

[Alison Schofield]

On Fri, Oct 31, 2025 at 10:32:24AM -0700, Dave Jiang wrote:
> The HPA to DPA translation for poison injection assumes that the
> base address starts from where the CXL region begins. When the
> extended linear cache is active, the offset can be within the DRAM
> region. Adjust the offset so that it correctly reflects the offset
> within the CXL region.

Dave, good catch!

This breakage was added in 6.18-rc1. Please consider it for a pull
request before the end of 6.18.

Here's justification and Fixes Tag:

Without this patch, when a user intends to inject poison at an offset
in region, it can inject poison into a location in a device that is
not mapped by that region.

The test case in cxl-poison-elc.sh that exposed this issue, injected
poison 0x20000000 which should have been the first byte of the CXL
half of an Extended Linear Cache (ELC) region. Instead, it injected
poison at the next byte of a device beyond the regions mapping.

Fixes: c3dd67681c70 ("cxl/region: Add inject and clear poison by region offset")

Reviewed-by: Alison Schofield <alison.schofield@intel.com>

Re: [PATCH 4/4] cxl: Adjust offset calculation for poison injection

[Dave Jiang]

On 10/31/25 1:58 PM, Alison Schofield wrote:
> On Fri, Oct 31, 2025 at 10:32:24AM -0700, Dave Jiang wrote:
>> The HPA to DPA translation for poison injection assumes that the
>> base address starts from where the CXL region begins. When the
>> extended linear cache is active, the offset can be within the DRAM
>> region. Adjust the offset so that it correctly reflects the offset
>> within the CXL region.
> 
> Dave, good catch!
> 
> This breakage was added in 6.18-rc1. Please consider it for a pull
> request before the end of 6.18.
> 
> Here's justification and Fixes Tag:
> 
> Without this patch, when a user intends to inject poison at an offset
> in region, it can inject poison into a location in a device that is
> not mapped by that region.
> 
> The test case in cxl-poison-elc.sh that exposed this issue, injected
> poison 0x20000000 which should have been the first byte of the CXL
> half of an Extended Linear Cache (ELC) region. Instead, it injected
> poison at the next byte of a device beyond the regions mapping.
> 
> Fixes: c3dd67681c70 ("cxl/region: Add inject and clear poison by region offset")
> 
> Reviewed-by: Alison Schofield <alison.schofield@intel.com>
> 
Yeah I can move it to fixes instead

Re: [PATCH 4/4] cxl: Adjust offset calculation for poison injection

[Dave Jiang]

On 10/31/25 10:32 AM, Dave Jiang wrote:
> The HPA to DPA translation for poison injection assumes that the
> base address starts from where the CXL region begins. When the
> extended linear cache is active, the offset can be within the DRAM
> region. Adjust the offset so that it correctly reflects the offset
> within the CXL region.
> 
> Signed-off-by: Dave Jiang <dave.jiang@intel.com>

This specific patch is applied to cxl/fixes
b6cfddd26ec55e865b4715f73e9bbb17a15091ed

> ---
>  drivers/cxl/core/region.c | 2 ++
>  1 file changed, 2 insertions(+)
> 
> diff --git a/drivers/cxl/core/region.c b/drivers/cxl/core/region.c
> index 095f5dcd17a1..eaf085aa44c4 100644
> --- a/drivers/cxl/core/region.c
> +++ b/drivers/cxl/core/region.c
> @@ -3731,6 +3731,7 @@ static int cxl_region_debugfs_poison_inject(void *data, u64 offset)
>  	if (validate_region_offset(cxlr, offset))
>  		return -EINVAL;
>  
> +	offset -= cxlr->params.cache_size;
>  	rc = region_offset_to_dpa_result(cxlr, offset, &result);
>  	if (rc || !result.cxlmd || result.dpa == ULLONG_MAX) {
>  		dev_dbg(&cxlr->dev,
> @@ -3763,6 +3764,7 @@ static int cxl_region_debugfs_poison_clear(void *data, u64 offset)
>  	if (validate_region_offset(cxlr, offset))
>  		return -EINVAL;
>  
> +	offset -= cxlr->params.cache_size;
>  	rc = region_offset_to_dpa_result(cxlr, offset, &result);
>  	if (rc || !result.cxlmd || result.dpa == ULLONG_MAX) {
>  		dev_dbg(&cxlr->dev,