[PATCH 0/3] libvdimm: Support namespace enumeration while locked

[PATCH 0/3] libvdimm: Support namespace enumeration while locked

[Dan Williams]

The NVDIMM_FAMILY_INTEL DSM path for manipulating labels explicitly
allows reading the namespace label area while the DIMM is locked. In
contrast the ACPI _LSR method for reading labels only indicates if the
label area is locked and returns no data.

It naturally follows that if the namespace label area is locked, the
DIMM is locked since the driver is prevented from discovering namespace
boundaries.

On platforms where both mechanisms are available, prefer the
NVDIMM_FAMILY_INTEL implementation and support enumerating namespace
boundaries while the DIMM is locked.

In addition to ACPI capability detection and the libnvdimm core change
to fail probe of locked namespaces, provide a capability to inject this
semantic difference via nfit_test.

---

Dan Williams (3):
      libnvdimm: Introduce locked DIMM capacity support
      acpi, nfit: Prefer _DSM over _LSR for namespace label reads
      tools/testing/nvdimm: Make DSM failure code injection an override


 drivers/acpi/nfit/core.c         |   12 +++++++++
 drivers/nvdimm/dimm.c            |   24 +++++++++++++++++--
 drivers/nvdimm/namespace_devs.c  |   23 ++++++++++++++++++
 drivers/nvdimm/nd.h              |    1 +
 tools/testing/nvdimm/test/nfit.c |   49 +++++++++++++++++++++++++-------------
 5 files changed, 89 insertions(+), 20 deletions(-)
_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm

[PATCH 0/3] libvdimm: Support namespace enumeration while locked

[Dan Williams]

The NVDIMM_FAMILY_INTEL DSM path for manipulating labels explicitly
allows reading the namespace label area while the DIMM is locked. In
contrast the ACPI _LSR method for reading labels only indicates if the
label area is locked and returns no data.

It naturally follows that if the namespace label area is locked, the
DIMM is locked since the driver is prevented from discovering namespace
boundaries.

On platforms where both mechanisms are available, prefer the
NVDIMM_FAMILY_INTEL implementation and support enumerating namespace
boundaries while the DIMM is locked.

In addition to ACPI capability detection and the libnvdimm core change
to fail probe of locked namespaces, provide a capability to inject this
semantic difference via nfit_test.

---

Dan Williams (3):
      libnvdimm: Introduce locked DIMM capacity support
      acpi, nfit: Prefer _DSM over _LSR for namespace label reads
      tools/testing/nvdimm: Make DSM failure code injection an override


 drivers/acpi/nfit/core.c         |   12 +++++++++
 drivers/nvdimm/dimm.c            |   24 +++++++++++++++++--
 drivers/nvdimm/namespace_devs.c  |   23 ++++++++++++++++++
 drivers/nvdimm/nd.h              |    1 +
 tools/testing/nvdimm/test/nfit.c |   49 +++++++++++++++++++++++++-------------
 5 files changed, 89 insertions(+), 20 deletions(-)

[PATCH 0/3] libvdimm: Support namespace enumeration while locked

[Dan Williams]

The NVDIMM_FAMILY_INTEL DSM path for manipulating labels explicitly
allows reading the namespace label area while the DIMM is locked. In
contrast the ACPI _LSR method for reading labels only indicates if the
label area is locked and returns no data.

It naturally follows that if the namespace label area is locked, the
DIMM is locked since the driver is prevented from discovering namespace
boundaries.

On platforms where both mechanisms are available, prefer the
NVDIMM_FAMILY_INTEL implementation and support enumerating namespace
boundaries while the DIMM is locked.

In addition to ACPI capability detection and the libnvdimm core change
to fail probe of locked namespaces, provide a capability to inject this
semantic difference via nfit_test.

---

Dan Williams (3):
      libnvdimm: Introduce locked DIMM capacity support
      acpi, nfit: Prefer _DSM over _LSR for namespace label reads
      tools/testing/nvdimm: Make DSM failure code injection an override


 drivers/acpi/nfit/core.c         |   12 +++++++++
 drivers/nvdimm/dimm.c            |   24 +++++++++++++++++--
 drivers/nvdimm/namespace_devs.c  |   23 ++++++++++++++++++
 drivers/nvdimm/nd.h              |    1 +
 tools/testing/nvdimm/test/nfit.c |   49 +++++++++++++++++++++++++-------------
 5 files changed, 89 insertions(+), 20 deletions(-)

[PATCH 1/3] libnvdimm: Introduce locked DIMM capacity support

[Dan Williams]

When a DIMM is locked its namespace label area may not be. Introduce the
distinction of locked namespaces to allow namespace enumeration while
the capacity is locked.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 drivers/nvdimm/dimm.c           |   24 ++++++++++++++++++++++--
 drivers/nvdimm/namespace_devs.c |   23 +++++++++++++++++++++++
 drivers/nvdimm/nd.h             |    1 +
 3 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/drivers/nvdimm/dimm.c b/drivers/nvdimm/dimm.c
index 233907889f96..6c8fb7590838 100644
--- a/drivers/nvdimm/dimm.c
+++ b/drivers/nvdimm/dimm.c
@@ -34,6 +34,9 @@ static int nvdimm_probe(struct device *dev)
 		return rc;
 	}
 
+	/* reset locked, to be validated below... */
+	nvdimm_clear_locked(dev);
+
 	ndd = kzalloc(sizeof(*ndd), GFP_KERNEL);
 	if (!ndd)
 		return -ENOMEM;
@@ -48,12 +51,30 @@ static int nvdimm_probe(struct device *dev)
 	get_device(dev);
 	kref_init(&ndd->kref);
 
+	/*
+	 * EACCES failures reading the namespace label-area-properties
+	 * are interpreted as the DIMM capacity being locked but the
+	 * namespace labels themselves being accessible.
+	 */
 	rc = nvdimm_init_nsarea(ndd);
-	if (rc == -EACCES)
+	if (rc == -EACCES) {
+		/*
+		 * See nvdimm_namespace_common_probe() where we fail to
+		 * allow namespaces to probe while the DIMM is locked,
+		 * but we do allow for namespace enumeration.
+		 */
 		nvdimm_set_locked(dev);
+		rc = 0;
+	}
 	if (rc)
 		goto err;
 
+	/*
+	 * EACCES failures reading the namespace label-data are
+	 * interpreted as the label area being locked in addition to the
+	 * DIMM capacity. We fail the dimm probe to prevent regions from
+	 * attempting to parse the label area.
+	 */
 	rc = nvdimm_init_config_data(ndd);
 	if (rc == -EACCES)
 		nvdimm_set_locked(dev);
@@ -72,7 +93,6 @@ static int nvdimm_probe(struct device *dev)
 		if (rc == 0)
 			nvdimm_set_aliasing(dev);
 	}
-	nvdimm_clear_locked(dev);
 	nvdimm_bus_unlock(dev);
 
 	if (rc)
diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c
index 28afdd668905..cb322f2bc605 100644
--- a/drivers/nvdimm/namespace_devs.c
+++ b/drivers/nvdimm/namespace_devs.c
@@ -1144,6 +1144,26 @@ resource_size_t nvdimm_namespace_capacity(struct nd_namespace_common *ndns)
 }
 EXPORT_SYMBOL(nvdimm_namespace_capacity);
 
+bool nvdimm_namespace_locked(struct nd_namespace_common *ndns)
+{
+	int i;
+	bool locked = false;
+	struct device *dev = &ndns->dev;
+	struct nd_region *nd_region = to_nd_region(dev->parent);
+
+	for (i = 0; i < nd_region->ndr_mappings; i++) {
+		struct nd_mapping *nd_mapping = &nd_region->mapping[i];
+		struct nvdimm *nvdimm = nd_mapping->nvdimm;
+
+		if (test_bit(NDD_LOCKED, &nvdimm->flags)) {
+			dev_dbg(dev, "%s locked\n", nvdimm_name(nvdimm));
+			locked = true;
+		}
+	}
+	return locked;
+}
+EXPORT_SYMBOL(nvdimm_namespace_locked);
+
 static ssize_t size_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
@@ -1695,6 +1715,9 @@ struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev)
 		}
 	}
 
+	if (nvdimm_namespace_locked(ndns))
+		return ERR_PTR(-EACCES);
+
 	size = nvdimm_namespace_capacity(ndns);
 	if (size < ND_MIN_NAMESPACE_SIZE) {
 		dev_dbg(&ndns->dev, "%pa, too small must be at least %#x\n",
diff --git a/drivers/nvdimm/nd.h b/drivers/nvdimm/nd.h
index 32e0364b48b9..9d17abd9f8d0 100644
--- a/drivers/nvdimm/nd.h
+++ b/drivers/nvdimm/nd.h
@@ -357,6 +357,7 @@ struct resource *nvdimm_allocate_dpa(struct nvdimm_drvdata *ndd,
 		struct nd_label_id *label_id, resource_size_t start,
 		resource_size_t n);
 resource_size_t nvdimm_namespace_capacity(struct nd_namespace_common *ndns);
+bool nvdimm_namespace_locked(struct nd_namespace_common *ndns);
 struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev);
 int nvdimm_namespace_attach_btt(struct nd_namespace_common *ndns);
 int nvdimm_namespace_detach_btt(struct nd_btt *nd_btt);

_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm

[PATCH 1/3] libnvdimm: Introduce locked DIMM capacity support

[Dan Williams]

When a DIMM is locked its namespace label area may not be. Introduce the
distinction of locked namespaces to allow namespace enumeration while
the capacity is locked.

Signed-off-by: Dan Williams <dan.j.williams-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/nvdimm/dimm.c           |   24 ++++++++++++++++++++++--
 drivers/nvdimm/namespace_devs.c |   23 +++++++++++++++++++++++
 drivers/nvdimm/nd.h             |    1 +
 3 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/drivers/nvdimm/dimm.c b/drivers/nvdimm/dimm.c
index 233907889f96..6c8fb7590838 100644
--- a/drivers/nvdimm/dimm.c
+++ b/drivers/nvdimm/dimm.c
@@ -34,6 +34,9 @@ static int nvdimm_probe(struct device *dev)
 		return rc;
 	}
 
+	/* reset locked, to be validated below... */
+	nvdimm_clear_locked(dev);
+
 	ndd = kzalloc(sizeof(*ndd), GFP_KERNEL);
 	if (!ndd)
 		return -ENOMEM;
@@ -48,12 +51,30 @@ static int nvdimm_probe(struct device *dev)
 	get_device(dev);
 	kref_init(&ndd->kref);
 
+	/*
+	 * EACCES failures reading the namespace label-area-properties
+	 * are interpreted as the DIMM capacity being locked but the
+	 * namespace labels themselves being accessible.
+	 */
 	rc = nvdimm_init_nsarea(ndd);
-	if (rc == -EACCES)
+	if (rc == -EACCES) {
+		/*
+		 * See nvdimm_namespace_common_probe() where we fail to
+		 * allow namespaces to probe while the DIMM is locked,
+		 * but we do allow for namespace enumeration.
+		 */
 		nvdimm_set_locked(dev);
+		rc = 0;
+	}
 	if (rc)
 		goto err;
 
+	/*
+	 * EACCES failures reading the namespace label-data are
+	 * interpreted as the label area being locked in addition to the
+	 * DIMM capacity. We fail the dimm probe to prevent regions from
+	 * attempting to parse the label area.
+	 */
 	rc = nvdimm_init_config_data(ndd);
 	if (rc == -EACCES)
 		nvdimm_set_locked(dev);
@@ -72,7 +93,6 @@ static int nvdimm_probe(struct device *dev)
 		if (rc == 0)
 			nvdimm_set_aliasing(dev);
 	}
-	nvdimm_clear_locked(dev);
 	nvdimm_bus_unlock(dev);
 
 	if (rc)
diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c
index 28afdd668905..cb322f2bc605 100644
--- a/drivers/nvdimm/namespace_devs.c
+++ b/drivers/nvdimm/namespace_devs.c
@@ -1144,6 +1144,26 @@ resource_size_t nvdimm_namespace_capacity(struct nd_namespace_common *ndns)
 }
 EXPORT_SYMBOL(nvdimm_namespace_capacity);
 
+bool nvdimm_namespace_locked(struct nd_namespace_common *ndns)
+{
+	int i;
+	bool locked = false;
+	struct device *dev = &ndns->dev;
+	struct nd_region *nd_region = to_nd_region(dev->parent);
+
+	for (i = 0; i < nd_region->ndr_mappings; i++) {
+		struct nd_mapping *nd_mapping = &nd_region->mapping[i];
+		struct nvdimm *nvdimm = nd_mapping->nvdimm;
+
+		if (test_bit(NDD_LOCKED, &nvdimm->flags)) {
+			dev_dbg(dev, "%s locked\n", nvdimm_name(nvdimm));
+			locked = true;
+		}
+	}
+	return locked;
+}
+EXPORT_SYMBOL(nvdimm_namespace_locked);
+
 static ssize_t size_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
@@ -1695,6 +1715,9 @@ struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev)
 		}
 	}
 
+	if (nvdimm_namespace_locked(ndns))
+		return ERR_PTR(-EACCES);
+
 	size = nvdimm_namespace_capacity(ndns);
 	if (size < ND_MIN_NAMESPACE_SIZE) {
 		dev_dbg(&ndns->dev, "%pa, too small must be at least %#x\n",
diff --git a/drivers/nvdimm/nd.h b/drivers/nvdimm/nd.h
index 32e0364b48b9..9d17abd9f8d0 100644
--- a/drivers/nvdimm/nd.h
+++ b/drivers/nvdimm/nd.h
@@ -357,6 +357,7 @@ struct resource *nvdimm_allocate_dpa(struct nvdimm_drvdata *ndd,
 		struct nd_label_id *label_id, resource_size_t start,
 		resource_size_t n);
 resource_size_t nvdimm_namespace_capacity(struct nd_namespace_common *ndns);
+bool nvdimm_namespace_locked(struct nd_namespace_common *ndns);
 struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev);
 int nvdimm_namespace_attach_btt(struct nd_namespace_common *ndns);
 int nvdimm_namespace_detach_btt(struct nd_btt *nd_btt);

[PATCH 1/3] libnvdimm: Introduce locked DIMM capacity support

[Dan Williams]

When a DIMM is locked its namespace label area may not be. Introduce the
distinction of locked namespaces to allow namespace enumeration while
the capacity is locked.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 drivers/nvdimm/dimm.c           |   24 ++++++++++++++++++++++--
 drivers/nvdimm/namespace_devs.c |   23 +++++++++++++++++++++++
 drivers/nvdimm/nd.h             |    1 +
 3 files changed, 46 insertions(+), 2 deletions(-)

diff --git a/drivers/nvdimm/dimm.c b/drivers/nvdimm/dimm.c
index 233907889f96..6c8fb7590838 100644
--- a/drivers/nvdimm/dimm.c
+++ b/drivers/nvdimm/dimm.c
@@ -34,6 +34,9 @@ static int nvdimm_probe(struct device *dev)
 		return rc;
 	}
 
+	/* reset locked, to be validated below... */
+	nvdimm_clear_locked(dev);
+
 	ndd = kzalloc(sizeof(*ndd), GFP_KERNEL);
 	if (!ndd)
 		return -ENOMEM;
@@ -48,12 +51,30 @@ static int nvdimm_probe(struct device *dev)
 	get_device(dev);
 	kref_init(&ndd->kref);
 
+	/*
+	 * EACCES failures reading the namespace label-area-properties
+	 * are interpreted as the DIMM capacity being locked but the
+	 * namespace labels themselves being accessible.
+	 */
 	rc = nvdimm_init_nsarea(ndd);
-	if (rc == -EACCES)
+	if (rc == -EACCES) {
+		/*
+		 * See nvdimm_namespace_common_probe() where we fail to
+		 * allow namespaces to probe while the DIMM is locked,
+		 * but we do allow for namespace enumeration.
+		 */
 		nvdimm_set_locked(dev);
+		rc = 0;
+	}
 	if (rc)
 		goto err;
 
+	/*
+	 * EACCES failures reading the namespace label-data are
+	 * interpreted as the label area being locked in addition to the
+	 * DIMM capacity. We fail the dimm probe to prevent regions from
+	 * attempting to parse the label area.
+	 */
 	rc = nvdimm_init_config_data(ndd);
 	if (rc == -EACCES)
 		nvdimm_set_locked(dev);
@@ -72,7 +93,6 @@ static int nvdimm_probe(struct device *dev)
 		if (rc == 0)
 			nvdimm_set_aliasing(dev);
 	}
-	nvdimm_clear_locked(dev);
 	nvdimm_bus_unlock(dev);
 
 	if (rc)
diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c
index 28afdd668905..cb322f2bc605 100644
--- a/drivers/nvdimm/namespace_devs.c
+++ b/drivers/nvdimm/namespace_devs.c
@@ -1144,6 +1144,26 @@ resource_size_t nvdimm_namespace_capacity(struct nd_namespace_common *ndns)
 }
 EXPORT_SYMBOL(nvdimm_namespace_capacity);
 
+bool nvdimm_namespace_locked(struct nd_namespace_common *ndns)
+{
+	int i;
+	bool locked = false;
+	struct device *dev = &ndns->dev;
+	struct nd_region *nd_region = to_nd_region(dev->parent);
+
+	for (i = 0; i < nd_region->ndr_mappings; i++) {
+		struct nd_mapping *nd_mapping = &nd_region->mapping[i];
+		struct nvdimm *nvdimm = nd_mapping->nvdimm;
+
+		if (test_bit(NDD_LOCKED, &nvdimm->flags)) {
+			dev_dbg(dev, "%s locked\n", nvdimm_name(nvdimm));
+			locked = true;
+		}
+	}
+	return locked;
+}
+EXPORT_SYMBOL(nvdimm_namespace_locked);
+
 static ssize_t size_show(struct device *dev,
 		struct device_attribute *attr, char *buf)
 {
@@ -1695,6 +1715,9 @@ struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev)
 		}
 	}
 
+	if (nvdimm_namespace_locked(ndns))
+		return ERR_PTR(-EACCES);
+
 	size = nvdimm_namespace_capacity(ndns);
 	if (size < ND_MIN_NAMESPACE_SIZE) {
 		dev_dbg(&ndns->dev, "%pa, too small must be at least %#x\n",
diff --git a/drivers/nvdimm/nd.h b/drivers/nvdimm/nd.h
index 32e0364b48b9..9d17abd9f8d0 100644
--- a/drivers/nvdimm/nd.h
+++ b/drivers/nvdimm/nd.h
@@ -357,6 +357,7 @@ struct resource *nvdimm_allocate_dpa(struct nvdimm_drvdata *ndd,
 		struct nd_label_id *label_id, resource_size_t start,
 		resource_size_t n);
 resource_size_t nvdimm_namespace_capacity(struct nd_namespace_common *ndns);
+bool nvdimm_namespace_locked(struct nd_namespace_common *ndns);
 struct nd_namespace_common *nvdimm_namespace_common_probe(struct device *dev);
 int nvdimm_namespace_attach_btt(struct nd_namespace_common *ndns);
 int nvdimm_namespace_detach_btt(struct nd_btt *nd_btt);

[PATCH 2/3] acpi, nfit: Prefer _DSM over _LSR for namespace label reads

[Dan Williams]

The _LSR method indicates locked status via error-code-3 returned in the
_LSR payload. When any error is returned the payload of _LSR is
truncated to a zero-length buffer.

The _DSM path in comparison allows system software to retrieve the
locked status *and* namespace label area contents.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 drivers/acpi/nfit/core.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
index e2235ed3e4be..5ce3b0221073 100644
--- a/drivers/acpi/nfit/core.c
+++ b/drivers/acpi/nfit/core.c
@@ -1694,7 +1694,7 @@ static int acpi_nfit_add_dimm(struct acpi_nfit_desc *acpi_desc,
 {
 	struct acpi_device *adev, *adev_dimm;
 	struct device *dev = acpi_desc->dev;
-	unsigned long dsm_mask;
+	unsigned long dsm_mask, label_mask;
 	const guid_t *guid;
 	int i;
 	int family = -1;
@@ -1766,6 +1766,16 @@ static int acpi_nfit_add_dimm(struct acpi_nfit_desc *acpi_desc,
 					1ULL << i))
 			set_bit(i, &nfit_mem->dsm_mask);
 
+	/*
+	 * Prefer the NVDIMM_FAMILY_INTEL label read commands if present
+	 * due to their better semantics handling locked capacity.
+	 */
+	label_mask = 1 << ND_CMD_GET_CONFIG_SIZE | 1 << ND_CMD_GET_CONFIG_DATA
+		| 1 << ND_CMD_SET_CONFIG_DATA;
+	if (family == NVDIMM_FAMILY_INTEL
+			&& (dsm_mask & label_mask) == label_mask)
+		return 0;
+
 	if (acpi_nvdimm_has_method(adev_dimm, "_LSI")
 			&& acpi_nvdimm_has_method(adev_dimm, "_LSR")) {
 		dev_dbg(dev, "%s: has _LSR\n", dev_name(&adev_dimm->dev));

_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm

[PATCH 2/3] acpi, nfit: Prefer _DSM over _LSR for namespace label reads

[Dan Williams]

The _LSR method indicates locked status via error-code-3 returned in the
_LSR payload. When any error is returned the payload of _LSR is
truncated to a zero-length buffer.

The _DSM path in comparison allows system software to retrieve the
locked status *and* namespace label area contents.

Signed-off-by: Dan Williams <dan.j.williams-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 drivers/acpi/nfit/core.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
index e2235ed3e4be..5ce3b0221073 100644
--- a/drivers/acpi/nfit/core.c
+++ b/drivers/acpi/nfit/core.c
@@ -1694,7 +1694,7 @@ static int acpi_nfit_add_dimm(struct acpi_nfit_desc *acpi_desc,
 {
 	struct acpi_device *adev, *adev_dimm;
 	struct device *dev = acpi_desc->dev;
-	unsigned long dsm_mask;
+	unsigned long dsm_mask, label_mask;
 	const guid_t *guid;
 	int i;
 	int family = -1;
@@ -1766,6 +1766,16 @@ static int acpi_nfit_add_dimm(struct acpi_nfit_desc *acpi_desc,
 					1ULL << i))
 			set_bit(i, &nfit_mem->dsm_mask);
 
+	/*
+	 * Prefer the NVDIMM_FAMILY_INTEL label read commands if present
+	 * due to their better semantics handling locked capacity.
+	 */
+	label_mask = 1 << ND_CMD_GET_CONFIG_SIZE | 1 << ND_CMD_GET_CONFIG_DATA
+		| 1 << ND_CMD_SET_CONFIG_DATA;
+	if (family == NVDIMM_FAMILY_INTEL
+			&& (dsm_mask & label_mask) == label_mask)
+		return 0;
+
 	if (acpi_nvdimm_has_method(adev_dimm, "_LSI")
 			&& acpi_nvdimm_has_method(adev_dimm, "_LSR")) {
 		dev_dbg(dev, "%s: has _LSR\n", dev_name(&adev_dimm->dev));

[PATCH 2/3] acpi, nfit: Prefer _DSM over _LSR for namespace label reads

[Dan Williams]

The _LSR method indicates locked status via error-code-3 returned in the
_LSR payload. When any error is returned the payload of _LSR is
truncated to a zero-length buffer.

The _DSM path in comparison allows system software to retrieve the
locked status *and* namespace label area contents.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 drivers/acpi/nfit/core.c |   12 +++++++++++-
 1 file changed, 11 insertions(+), 1 deletion(-)

diff --git a/drivers/acpi/nfit/core.c b/drivers/acpi/nfit/core.c
index e2235ed3e4be..5ce3b0221073 100644
--- a/drivers/acpi/nfit/core.c
+++ b/drivers/acpi/nfit/core.c
@@ -1694,7 +1694,7 @@ static int acpi_nfit_add_dimm(struct acpi_nfit_desc *acpi_desc,
 {
 	struct acpi_device *adev, *adev_dimm;
 	struct device *dev = acpi_desc->dev;
-	unsigned long dsm_mask;
+	unsigned long dsm_mask, label_mask;
 	const guid_t *guid;
 	int i;
 	int family = -1;
@@ -1766,6 +1766,16 @@ static int acpi_nfit_add_dimm(struct acpi_nfit_desc *acpi_desc,
 					1ULL << i))
 			set_bit(i, &nfit_mem->dsm_mask);
 
+	/*
+	 * Prefer the NVDIMM_FAMILY_INTEL label read commands if present
+	 * due to their better semantics handling locked capacity.
+	 */
+	label_mask = 1 << ND_CMD_GET_CONFIG_SIZE | 1 << ND_CMD_GET_CONFIG_DATA
+		| 1 << ND_CMD_SET_CONFIG_DATA;
+	if (family == NVDIMM_FAMILY_INTEL
+			&& (dsm_mask & label_mask) == label_mask)
+		return 0;
+
 	if (acpi_nvdimm_has_method(adev_dimm, "_LSI")
 			&& acpi_nvdimm_has_method(adev_dimm, "_LSR")) {
 		dev_dbg(dev, "%s: has _LSR\n", dev_name(&adev_dimm->dev));

[PATCH 3/3] tools/testing/nvdimm: Make DSM failure code injection an override

[Dan Williams]

In order to emulate the behavior of the NVDIMM_FAMILY_INTEL DSMs
nfit_test needs the ability to execute the DSM and then override the
return code.

Split the current return code injection from get_dimm() and apply at
after the function has executed to override the return status.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 tools/testing/nvdimm/test/nfit.c |   49 +++++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 17 deletions(-)

diff --git a/tools/testing/nvdimm/test/nfit.c b/tools/testing/nvdimm/test/nfit.c
index 4ea385be528f..6f19d6e28f5f 100644
--- a/tools/testing/nvdimm/test/nfit.c
+++ b/tools/testing/nvdimm/test/nfit.c
@@ -882,6 +882,16 @@ static int nd_intel_test_cmd_set_lss_status(struct nfit_test *t,
 	return 0;
 }
 
+static int override_return_code(int dimm, unsigned int func, int rc)
+{
+	if ((1 << func) & dimm_fail_cmd_flags[dimm]) {
+		if (dimm_fail_cmd_code[dimm])
+			return dimm_fail_cmd_code[dimm];
+		return -EIO;
+	}
+	return rc;
+}
+
 static int get_dimm(struct nfit_mem *nfit_mem, unsigned int func)
 {
 	int i;
@@ -892,13 +902,6 @@ static int get_dimm(struct nfit_mem *nfit_mem, unsigned int func)
 			break;
 	if (i >= ARRAY_SIZE(handle))
 		return -ENXIO;
-
-	if ((1 << func) & dimm_fail_cmd_flags[i]) {
-		if (dimm_fail_cmd_code[i])
-			return dimm_fail_cmd_code[i];
-		return -EIO;
-	}
-
 	return i;
 }
 
@@ -937,48 +940,59 @@ static int nfit_test_ctl(struct nvdimm_bus_descriptor *nd_desc,
 
 			switch (func) {
 			case ND_INTEL_ENABLE_LSS_STATUS:
-				return nd_intel_test_cmd_set_lss_status(t,
+				rc = nd_intel_test_cmd_set_lss_status(t,
 						buf, buf_len);
+				break;
 			case ND_INTEL_FW_GET_INFO:
-				return nd_intel_test_get_fw_info(t, buf,
+				rc = nd_intel_test_get_fw_info(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_START_UPDATE:
-				return nd_intel_test_start_update(t, buf,
+				rc = nd_intel_test_start_update(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_SEND_DATA:
-				return nd_intel_test_send_data(t, buf,
+				rc = nd_intel_test_send_data(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_FINISH_UPDATE:
-				return nd_intel_test_finish_fw(t, buf,
+				rc = nd_intel_test_finish_fw(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_FINISH_QUERY:
-				return nd_intel_test_finish_query(t, buf,
+				rc = nd_intel_test_finish_query(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_SMART:
-				return nfit_test_cmd_smart(buf, buf_len,
+				rc = nfit_test_cmd_smart(buf, buf_len,
 						&t->smart[i - t->dcr_idx]);
+				break;
 			case ND_INTEL_SMART_THRESHOLD:
-				return nfit_test_cmd_smart_threshold(buf,
+				rc = nfit_test_cmd_smart_threshold(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx]);
+				break;
 			case ND_INTEL_SMART_SET_THRESHOLD:
-				return nfit_test_cmd_smart_set_threshold(buf,
+				rc = nfit_test_cmd_smart_set_threshold(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx],
 						&t->smart[i - t->dcr_idx],
 						&t->pdev.dev, t->dimm_dev[i]);
+				break;
 			case ND_INTEL_SMART_INJECT:
-				return nfit_test_cmd_smart_inject(buf,
+				rc = nfit_test_cmd_smart_inject(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx],
 						&t->smart[i - t->dcr_idx],
 						&t->pdev.dev, t->dimm_dev[i]);
+				break;
 			default:
 				return -ENOTTY;
 			}
+			return override_return_code(i, func, rc);
 		}
 
 		if (!test_bit(cmd, &cmd_mask)
@@ -1004,6 +1018,7 @@ static int nfit_test_ctl(struct nvdimm_bus_descriptor *nd_desc,
 		default:
 			return -ENOTTY;
 		}
+		return override_return_code(i, func, rc);
 	} else {
 		struct ars_state *ars_state = &t->ars_state;
 		struct nd_cmd_pkg *call_pkg = buf;

_______________________________________________
Linux-nvdimm mailing list
Linux-nvdimm@lists.01.org
https://lists.01.org/mailman/listinfo/linux-nvdimm

[PATCH 3/3] tools/testing/nvdimm: Make DSM failure code injection an override

[Dan Williams]

In order to emulate the behavior of the NVDIMM_FAMILY_INTEL DSMs
nfit_test needs the ability to execute the DSM and then override the
return code.

Split the current return code injection from get_dimm() and apply at
after the function has executed to override the return status.

Signed-off-by: Dan Williams <dan.j.williams-ral2JQCrhuEAvxtiuMwx3w@public.gmane.org>
---
 tools/testing/nvdimm/test/nfit.c |   49 +++++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 17 deletions(-)

diff --git a/tools/testing/nvdimm/test/nfit.c b/tools/testing/nvdimm/test/nfit.c
index 4ea385be528f..6f19d6e28f5f 100644
--- a/tools/testing/nvdimm/test/nfit.c
+++ b/tools/testing/nvdimm/test/nfit.c
@@ -882,6 +882,16 @@ static int nd_intel_test_cmd_set_lss_status(struct nfit_test *t,
 	return 0;
 }
 
+static int override_return_code(int dimm, unsigned int func, int rc)
+{
+	if ((1 << func) & dimm_fail_cmd_flags[dimm]) {
+		if (dimm_fail_cmd_code[dimm])
+			return dimm_fail_cmd_code[dimm];
+		return -EIO;
+	}
+	return rc;
+}
+
 static int get_dimm(struct nfit_mem *nfit_mem, unsigned int func)
 {
 	int i;
@@ -892,13 +902,6 @@ static int get_dimm(struct nfit_mem *nfit_mem, unsigned int func)
 			break;
 	if (i >= ARRAY_SIZE(handle))
 		return -ENXIO;
-
-	if ((1 << func) & dimm_fail_cmd_flags[i]) {
-		if (dimm_fail_cmd_code[i])
-			return dimm_fail_cmd_code[i];
-		return -EIO;
-	}
-
 	return i;
 }
 
@@ -937,48 +940,59 @@ static int nfit_test_ctl(struct nvdimm_bus_descriptor *nd_desc,
 
 			switch (func) {
 			case ND_INTEL_ENABLE_LSS_STATUS:
-				return nd_intel_test_cmd_set_lss_status(t,
+				rc = nd_intel_test_cmd_set_lss_status(t,
 						buf, buf_len);
+				break;
 			case ND_INTEL_FW_GET_INFO:
-				return nd_intel_test_get_fw_info(t, buf,
+				rc = nd_intel_test_get_fw_info(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_START_UPDATE:
-				return nd_intel_test_start_update(t, buf,
+				rc = nd_intel_test_start_update(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_SEND_DATA:
-				return nd_intel_test_send_data(t, buf,
+				rc = nd_intel_test_send_data(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_FINISH_UPDATE:
-				return nd_intel_test_finish_fw(t, buf,
+				rc = nd_intel_test_finish_fw(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_FINISH_QUERY:
-				return nd_intel_test_finish_query(t, buf,
+				rc = nd_intel_test_finish_query(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_SMART:
-				return nfit_test_cmd_smart(buf, buf_len,
+				rc = nfit_test_cmd_smart(buf, buf_len,
 						&t->smart[i - t->dcr_idx]);
+				break;
 			case ND_INTEL_SMART_THRESHOLD:
-				return nfit_test_cmd_smart_threshold(buf,
+				rc = nfit_test_cmd_smart_threshold(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx]);
+				break;
 			case ND_INTEL_SMART_SET_THRESHOLD:
-				return nfit_test_cmd_smart_set_threshold(buf,
+				rc = nfit_test_cmd_smart_set_threshold(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx],
 						&t->smart[i - t->dcr_idx],
 						&t->pdev.dev, t->dimm_dev[i]);
+				break;
 			case ND_INTEL_SMART_INJECT:
-				return nfit_test_cmd_smart_inject(buf,
+				rc = nfit_test_cmd_smart_inject(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx],
 						&t->smart[i - t->dcr_idx],
 						&t->pdev.dev, t->dimm_dev[i]);
+				break;
 			default:
 				return -ENOTTY;
 			}
+			return override_return_code(i, func, rc);
 		}
 
 		if (!test_bit(cmd, &cmd_mask)
@@ -1004,6 +1018,7 @@ static int nfit_test_ctl(struct nvdimm_bus_descriptor *nd_desc,
 		default:
 			return -ENOTTY;
 		}
+		return override_return_code(i, func, rc);
 	} else {
 		struct ars_state *ars_state = &t->ars_state;
 		struct nd_cmd_pkg *call_pkg = buf;

[PATCH 3/3] tools/testing/nvdimm: Make DSM failure code injection an override

[Dan Williams]

In order to emulate the behavior of the NVDIMM_FAMILY_INTEL DSMs
nfit_test needs the ability to execute the DSM and then override the
return code.

Split the current return code injection from get_dimm() and apply at
after the function has executed to override the return status.

Signed-off-by: Dan Williams <dan.j.williams@intel.com>
---
 tools/testing/nvdimm/test/nfit.c |   49 +++++++++++++++++++++++++-------------
 1 file changed, 32 insertions(+), 17 deletions(-)

diff --git a/tools/testing/nvdimm/test/nfit.c b/tools/testing/nvdimm/test/nfit.c
index 4ea385be528f..6f19d6e28f5f 100644
--- a/tools/testing/nvdimm/test/nfit.c
+++ b/tools/testing/nvdimm/test/nfit.c
@@ -882,6 +882,16 @@ static int nd_intel_test_cmd_set_lss_status(struct nfit_test *t,
 	return 0;
 }
 
+static int override_return_code(int dimm, unsigned int func, int rc)
+{
+	if ((1 << func) & dimm_fail_cmd_flags[dimm]) {
+		if (dimm_fail_cmd_code[dimm])
+			return dimm_fail_cmd_code[dimm];
+		return -EIO;
+	}
+	return rc;
+}
+
 static int get_dimm(struct nfit_mem *nfit_mem, unsigned int func)
 {
 	int i;
@@ -892,13 +902,6 @@ static int get_dimm(struct nfit_mem *nfit_mem, unsigned int func)
 			break;
 	if (i >= ARRAY_SIZE(handle))
 		return -ENXIO;
-
-	if ((1 << func) & dimm_fail_cmd_flags[i]) {
-		if (dimm_fail_cmd_code[i])
-			return dimm_fail_cmd_code[i];
-		return -EIO;
-	}
-
 	return i;
 }
 
@@ -937,48 +940,59 @@ static int nfit_test_ctl(struct nvdimm_bus_descriptor *nd_desc,
 
 			switch (func) {
 			case ND_INTEL_ENABLE_LSS_STATUS:
-				return nd_intel_test_cmd_set_lss_status(t,
+				rc = nd_intel_test_cmd_set_lss_status(t,
 						buf, buf_len);
+				break;
 			case ND_INTEL_FW_GET_INFO:
-				return nd_intel_test_get_fw_info(t, buf,
+				rc = nd_intel_test_get_fw_info(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_START_UPDATE:
-				return nd_intel_test_start_update(t, buf,
+				rc = nd_intel_test_start_update(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_SEND_DATA:
-				return nd_intel_test_send_data(t, buf,
+				rc = nd_intel_test_send_data(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_FINISH_UPDATE:
-				return nd_intel_test_finish_fw(t, buf,
+				rc = nd_intel_test_finish_fw(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_FW_FINISH_QUERY:
-				return nd_intel_test_finish_query(t, buf,
+				rc = nd_intel_test_finish_query(t, buf,
 						buf_len, i - t->dcr_idx);
+				break;
 			case ND_INTEL_SMART:
-				return nfit_test_cmd_smart(buf, buf_len,
+				rc = nfit_test_cmd_smart(buf, buf_len,
 						&t->smart[i - t->dcr_idx]);
+				break;
 			case ND_INTEL_SMART_THRESHOLD:
-				return nfit_test_cmd_smart_threshold(buf,
+				rc = nfit_test_cmd_smart_threshold(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx]);
+				break;
 			case ND_INTEL_SMART_SET_THRESHOLD:
-				return nfit_test_cmd_smart_set_threshold(buf,
+				rc = nfit_test_cmd_smart_set_threshold(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx],
 						&t->smart[i - t->dcr_idx],
 						&t->pdev.dev, t->dimm_dev[i]);
+				break;
 			case ND_INTEL_SMART_INJECT:
-				return nfit_test_cmd_smart_inject(buf,
+				rc = nfit_test_cmd_smart_inject(buf,
 						buf_len,
 						&t->smart_threshold[i -
 							t->dcr_idx],
 						&t->smart[i - t->dcr_idx],
 						&t->pdev.dev, t->dimm_dev[i]);
+				break;
 			default:
 				return -ENOTTY;
 			}
+			return override_return_code(i, func, rc);
 		}
 
 		if (!test_bit(cmd, &cmd_mask)
@@ -1004,6 +1018,7 @@ static int nfit_test_ctl(struct nvdimm_bus_descriptor *nd_desc,
 		default:
 			return -ENOTTY;
 		}
+		return override_return_code(i, func, rc);
 	} else {
 		struct ars_state *ars_state = &t->ars_state;
 		struct nd_cmd_pkg *call_pkg = buf;