[PATCH v4 0/4] cxl: Dirty shutdown followups

[PATCH v4 0/4] cxl: Dirty shutdown followups

[Davidlohr Bueso]

Changes from v3 (https://lore.kernel.org/linux-cxl/20250219062832.237881-1-dave@stgolabs.net/):
  - pickup reviews
  - simplify patch 1 (DaveJ)
  - change -1 to ULLONG_MAX in patch 3 (Ira)
  - patch 4 replace int for u8 for shutdown state (DaveJ)

Hi,

Some followup patches to the GPF work. First two patches are from feedback
provided by DaveJ. The third patch adds a $platform/dirty_shutdown sysfs
attribute to expose the count to userspace. Fourth patch adds support
emulating the set shutdown state command for the mock device.

Applies against the -next branch of cxl.git.

Thanks!

Davidlohr Bueso (4):
  cxl/pci: Introduce cxl_gpf_get_dvsec()
  cxl/pmem: Rename cxl_dirty_shutdown_state()
  cxl/pmem: Export dirty shutdown count via sysfs
  tools/testing/cxl: Set Shutdown State support

 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 25 +++++-
 drivers/cxl/core/pci.c                        | 30 ++++---
 drivers/cxl/cxl.h                             |  3 +
 drivers/cxl/cxlmem.h                          | 15 +++-
 drivers/cxl/pmem.c                            | 79 ++++++++++++++++---
 tools/testing/cxl/test/mem.c                  | 23 ++++++
 8 files changed, 166 insertions(+), 23 deletions(-)

-- 
2.39.5

[PATCH 1/4] cxl/pci: Introduce cxl_gpf_get_dvsec()

[Davidlohr Bueso]

Add a helper to fetch the port/device GPF dvsecs. This is
currently only used for ports, but a later patch to export
dirty count to users will make use of the device one.

Reviewed-by: Li Ming <ming.li@zohomail.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 drivers/cxl/core/pci.c | 30 ++++++++++++++++++++----------
 drivers/cxl/cxl.h      |  2 ++
 2 files changed, 22 insertions(+), 10 deletions(-)

diff --git a/drivers/cxl/core/pci.c b/drivers/cxl/core/pci.c
index a5c65f79db18..96fecb799cbc 100644
--- a/drivers/cxl/core/pci.c
+++ b/drivers/cxl/core/pci.c
@@ -1072,6 +1072,22 @@ int cxl_pci_get_bandwidth(struct pci_dev *pdev, struct access_coordinate *c)
 #define GPF_TIMEOUT_BASE_MAX 2
 #define GPF_TIMEOUT_SCALE_MAX 7 /* 10 seconds */
 
+u16 cxl_gpf_get_dvsec(struct device *dev, bool is_port)
+{
+	u16 dvsec;
+
+	if (!dev_is_pci(dev))
+		return 0;
+
+	dvsec = pci_find_dvsec_capability(to_pci_dev(dev), PCI_VENDOR_ID_CXL,
+			is_port ? CXL_DVSEC_PORT_GPF : CXL_DVSEC_DEVICE_GPF);
+	if (!dvsec)
+		dev_warn(dev, "%s GPF DVSEC not present\n",
+			 is_port ? "Port" : "Device");
+	return dvsec;
+}
+EXPORT_SYMBOL_NS_GPL(cxl_gpf_get_dvsec, "CXL");
+
 static int update_gpf_port_dvsec(struct pci_dev *pdev, int dvsec, int phase)
 {
 	u64 base, scale;
@@ -1116,26 +1132,20 @@ int cxl_gpf_port_setup(struct device *dport_dev, struct cxl_port *port)
 {
 	struct pci_dev *pdev;
 
-	if (!dev_is_pci(dport_dev))
-		return 0;
-
-	pdev = to_pci_dev(dport_dev);
-	if (!pdev || !port)
+	if (!port)
 		return -EINVAL;
 
 	if (!port->gpf_dvsec) {
 		int dvsec;
 
-		dvsec = pci_find_dvsec_capability(pdev, PCI_VENDOR_ID_CXL,
-						  CXL_DVSEC_PORT_GPF);
-		if (!dvsec) {
-			pci_warn(pdev, "Port GPF DVSEC not present\n");
+		dvsec = cxl_gpf_get_dvsec(dport_dev, true);
+		if (!dvsec)
 			return -EINVAL;
-		}
 
 		port->gpf_dvsec = dvsec;
 	}
 
+	pdev = to_pci_dev(dport_dev);
 	update_gpf_port_dvsec(pdev, port->gpf_dvsec, 1);
 	update_gpf_port_dvsec(pdev, port->gpf_dvsec, 2);
 
diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
index 6baec4ba9141..29f2ab0d5bf6 100644
--- a/drivers/cxl/cxl.h
+++ b/drivers/cxl/cxl.h
@@ -901,4 +901,6 @@ bool cxl_endpoint_decoder_reset_detected(struct cxl_port *port);
 #define __mock static
 #endif
 
+u16 cxl_gpf_get_dvsec(struct device *dev, bool is_port);
+
 #endif /* __CXL_H__ */
-- 
2.39.5

Re: [PATCH 1/4] cxl/pci: Introduce cxl_gpf_get_dvsec()

[Dave Jiang]

On 2/19/25 6:36 PM, Davidlohr Bueso wrote:
> Add a helper to fetch the port/device GPF dvsecs. This is
> currently only used for ports, but a later patch to export
> dirty count to users will make use of the device one.
> 
> Reviewed-by: Li Ming <ming.li@zohomail.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> ---
>  drivers/cxl/core/pci.c | 30 ++++++++++++++++++++----------
>  drivers/cxl/cxl.h      |  2 ++
>  2 files changed, 22 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/cxl/core/pci.c b/drivers/cxl/core/pci.c
> index a5c65f79db18..96fecb799cbc 100644
> --- a/drivers/cxl/core/pci.c
> +++ b/drivers/cxl/core/pci.c
> @@ -1072,6 +1072,22 @@ int cxl_pci_get_bandwidth(struct pci_dev *pdev, struct access_coordinate *c)
>  #define GPF_TIMEOUT_BASE_MAX 2
>  #define GPF_TIMEOUT_SCALE_MAX 7 /* 10 seconds */
>  
> +u16 cxl_gpf_get_dvsec(struct device *dev, bool is_port)
> +{
> +	u16 dvsec;
> +
> +	if (!dev_is_pci(dev))
> +		return 0;
> +
> +	dvsec = pci_find_dvsec_capability(to_pci_dev(dev), PCI_VENDOR_ID_CXL,
> +			is_port ? CXL_DVSEC_PORT_GPF : CXL_DVSEC_DEVICE_GPF);
> +	if (!dvsec)
> +		dev_warn(dev, "%s GPF DVSEC not present\n",
> +			 is_port ? "Port" : "Device");
> +	return dvsec;
> +}
> +EXPORT_SYMBOL_NS_GPL(cxl_gpf_get_dvsec, "CXL");
> +
>  static int update_gpf_port_dvsec(struct pci_dev *pdev, int dvsec, int phase)
>  {
>  	u64 base, scale;
> @@ -1116,26 +1132,20 @@ int cxl_gpf_port_setup(struct device *dport_dev, struct cxl_port *port)
>  {
>  	struct pci_dev *pdev;
>  
> -	if (!dev_is_pci(dport_dev))
> -		return 0;
> -
> -	pdev = to_pci_dev(dport_dev);
> -	if (!pdev || !port)
> +	if (!port)
>  		return -EINVAL;
>  
>  	if (!port->gpf_dvsec) {
>  		int dvsec;
>  
> -		dvsec = pci_find_dvsec_capability(pdev, PCI_VENDOR_ID_CXL,
> -						  CXL_DVSEC_PORT_GPF);
> -		if (!dvsec) {
> -			pci_warn(pdev, "Port GPF DVSEC not present\n");
> +		dvsec = cxl_gpf_get_dvsec(dport_dev, true);
> +		if (!dvsec)
>  			return -EINVAL;
> -		}
>  
>  		port->gpf_dvsec = dvsec;
>  	}
>  
> +	pdev = to_pci_dev(dport_dev);
>  	update_gpf_port_dvsec(pdev, port->gpf_dvsec, 1);
>  	update_gpf_port_dvsec(pdev, port->gpf_dvsec, 2);
>  
> diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
> index 6baec4ba9141..29f2ab0d5bf6 100644
> --- a/drivers/cxl/cxl.h
> +++ b/drivers/cxl/cxl.h
> @@ -901,4 +901,6 @@ bool cxl_endpoint_decoder_reset_detected(struct cxl_port *port);
>  #define __mock static
>  #endif
>  
> +u16 cxl_gpf_get_dvsec(struct device *dev, bool is_port);
> +
>  #endif /* __CXL_H__ */

Re: [PATCH 1/4] cxl/pci: Introduce cxl_gpf_get_dvsec()

[Ira Weiny]

Davidlohr Bueso wrote:
> Add a helper to fetch the port/device GPF dvsecs. This is
> currently only used for ports, but a later patch to export
> dirty count to users will make use of the device one.
> 
> Reviewed-by: Li Ming <ming.li@zohomail.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Reviewed-by: Ira Weiny <ira.weiny@intel.com>

[snip]

Re: [PATCH 1/4] cxl/pci: Introduce cxl_gpf_get_dvsec()

[Jonathan Cameron]

On Wed, 19 Feb 2025 17:36:01 -0800
Davidlohr Bueso <dave@stgolabs.net> wrote:

> Add a helper to fetch the port/device GPF dvsecs. This is
> currently only used for ports, but a later patch to export
> dirty count to users will make use of the device one.
> 
> Reviewed-by: Li Ming <ming.li@zohomail.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

Maybe an enum for the port vs device thing would have been nice but
I'm not that bothered and a single use enum is bit nasty.

Re: [PATCH 1/4] cxl/pci: Introduce cxl_gpf_get_dvsec()

[Fan Ni]

On Wed, Feb 19, 2025 at 05:36:01PM -0800, Davidlohr Bueso wrote:
> Add a helper to fetch the port/device GPF dvsecs. This is
> currently only used for ports, but a later patch to export
> dirty count to users will make use of the device one.
> 
> Reviewed-by: Li Ming <ming.li@zohomail.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
> ---

Reviewed-by: Fan Ni <fan.ni@samsung.com>

>  drivers/cxl/core/pci.c | 30 ++++++++++++++++++++----------
>  drivers/cxl/cxl.h      |  2 ++
>  2 files changed, 22 insertions(+), 10 deletions(-)
> 
> diff --git a/drivers/cxl/core/pci.c b/drivers/cxl/core/pci.c
> index a5c65f79db18..96fecb799cbc 100644
> --- a/drivers/cxl/core/pci.c
> +++ b/drivers/cxl/core/pci.c
> @@ -1072,6 +1072,22 @@ int cxl_pci_get_bandwidth(struct pci_dev *pdev, struct access_coordinate *c)
>  #define GPF_TIMEOUT_BASE_MAX 2
>  #define GPF_TIMEOUT_SCALE_MAX 7 /* 10 seconds */
>  
> +u16 cxl_gpf_get_dvsec(struct device *dev, bool is_port)
> +{
> +	u16 dvsec;
> +
> +	if (!dev_is_pci(dev))
> +		return 0;
> +
> +	dvsec = pci_find_dvsec_capability(to_pci_dev(dev), PCI_VENDOR_ID_CXL,
> +			is_port ? CXL_DVSEC_PORT_GPF : CXL_DVSEC_DEVICE_GPF);
> +	if (!dvsec)
> +		dev_warn(dev, "%s GPF DVSEC not present\n",
> +			 is_port ? "Port" : "Device");
> +	return dvsec;
> +}
> +EXPORT_SYMBOL_NS_GPL(cxl_gpf_get_dvsec, "CXL");
> +
>  static int update_gpf_port_dvsec(struct pci_dev *pdev, int dvsec, int phase)
>  {
>  	u64 base, scale;
> @@ -1116,26 +1132,20 @@ int cxl_gpf_port_setup(struct device *dport_dev, struct cxl_port *port)
>  {
>  	struct pci_dev *pdev;
>  
> -	if (!dev_is_pci(dport_dev))
> -		return 0;
> -
> -	pdev = to_pci_dev(dport_dev);
> -	if (!pdev || !port)
> +	if (!port)
>  		return -EINVAL;
>  
>  	if (!port->gpf_dvsec) {
>  		int dvsec;
>  
> -		dvsec = pci_find_dvsec_capability(pdev, PCI_VENDOR_ID_CXL,
> -						  CXL_DVSEC_PORT_GPF);
> -		if (!dvsec) {
> -			pci_warn(pdev, "Port GPF DVSEC not present\n");
> +		dvsec = cxl_gpf_get_dvsec(dport_dev, true);
> +		if (!dvsec)
>  			return -EINVAL;
> -		}
>  
>  		port->gpf_dvsec = dvsec;
>  	}
>  
> +	pdev = to_pci_dev(dport_dev);
>  	update_gpf_port_dvsec(pdev, port->gpf_dvsec, 1);
>  	update_gpf_port_dvsec(pdev, port->gpf_dvsec, 2);
>  
> diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
> index 6baec4ba9141..29f2ab0d5bf6 100644
> --- a/drivers/cxl/cxl.h
> +++ b/drivers/cxl/cxl.h
> @@ -901,4 +901,6 @@ bool cxl_endpoint_decoder_reset_detected(struct cxl_port *port);
>  #define __mock static
>  #endif
>  
> +u16 cxl_gpf_get_dvsec(struct device *dev, bool is_port);
> +
>  #endif /* __CXL_H__ */
> -- 
> 2.39.5
> 

[PATCH 2/4] cxl/pmem: Rename cxl_dirty_shutdown_state()

[Davidlohr Bueso]

... to a better suited 'cxl_arm_dirty_shutdown()'.

Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Li Ming <ming.li@zohomail.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 drivers/cxl/core/mbox.c | 4 ++--
 drivers/cxl/cxlmem.h    | 2 +-
 drivers/cxl/pmem.c      | 2 +-
 3 files changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
index c5eedcae3b02..86d13f4a1c18 100644
--- a/drivers/cxl/core/mbox.c
+++ b/drivers/cxl/core/mbox.c
@@ -1281,7 +1281,7 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
 }
 EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
 
-int cxl_dirty_shutdown_state(struct cxl_memdev_state *mds)
+int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
 {
 	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
 	struct cxl_mbox_cmd mbox_cmd;
@@ -1297,7 +1297,7 @@ int cxl_dirty_shutdown_state(struct cxl_memdev_state *mds)
 
 	return cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
 }
-EXPORT_SYMBOL_NS_GPL(cxl_dirty_shutdown_state, "CXL");
+EXPORT_SYMBOL_NS_GPL(cxl_arm_dirty_shutdown, "CXL");
 
 int cxl_set_timestamp(struct cxl_memdev_state *mds)
 {
diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
index 8e1e46c348f5..6d60030139df 100644
--- a/drivers/cxl/cxlmem.h
+++ b/drivers/cxl/cxlmem.h
@@ -822,7 +822,7 @@ void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
 			    enum cxl_event_log_type type,
 			    enum cxl_event_type event_type,
 			    const uuid_t *uuid, union cxl_event *evt);
-int cxl_dirty_shutdown_state(struct cxl_memdev_state *mds);
+int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds);
 int cxl_set_timestamp(struct cxl_memdev_state *mds);
 int cxl_poison_state_init(struct cxl_memdev_state *mds);
 int cxl_mem_get_poison(struct cxl_memdev *cxlmd, u64 offset, u64 len,
diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
index a39e2c52d7ab..6b284962592f 100644
--- a/drivers/cxl/pmem.c
+++ b/drivers/cxl/pmem.c
@@ -90,7 +90,7 @@ static int cxl_nvdimm_probe(struct device *dev)
 	 * clear it upon a successful GPF flow. The exception to this
 	 * is upon Viral detection, per CXL 3.2 section 12.4.2.
 	 */
-	if (cxl_dirty_shutdown_state(mds))
+	if (cxl_arm_dirty_shutdown(mds))
 		dev_warn(dev, "GPF: could not dirty shutdown state\n");
 
 	dev_set_drvdata(dev, nvdimm);
-- 
2.39.5

Re: [PATCH 2/4] cxl/pmem: Rename cxl_dirty_shutdown_state()

[Ira Weiny]

Davidlohr Bueso wrote:
> ... to a better suited 'cxl_arm_dirty_shutdown()'.
> 
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> Reviewed-by: Li Ming <ming.li@zohomail.com>

Reviewed-by: Ira Weiny <ira.weiny@intel.com>

> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

[snip]

Re: [PATCH 2/4] cxl/pmem: Rename cxl_dirty_shutdown_state()

[Jonathan Cameron]

On Wed, 19 Feb 2025 17:36:02 -0800
Davidlohr Bueso <dave@stgolabs.net> wrote:

> ... to a better suited 'cxl_arm_dirty_shutdown()'.

But it works on x86 as well! I'll cope I suppose.

Mind you why not let it take the state as a parameter
and always pass 1?  From a mailbox command point of
view 0 is valid. The text in table 8-152 has me a little confused
but I think it says you can set it to 1 and back to 0 and
then reset at which point the dirty shutdown count will
not increment.

No idea why you'd do that, but just passing a 1 into
this function would make a call of
cxl_set_dirty_shutdown_state(mds, 1)
seem reasonable, or pass a boolean.
 



> 
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> Reviewed-by: Li Ming <ming.li@zohomail.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
> ---
>  drivers/cxl/core/mbox.c | 4 ++--
>  drivers/cxl/cxlmem.h    | 2 +-
>  drivers/cxl/pmem.c      | 2 +-
>  3 files changed, 4 insertions(+), 4 deletions(-)
> 
> diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
> index c5eedcae3b02..86d13f4a1c18 100644
> --- a/drivers/cxl/core/mbox.c
> +++ b/drivers/cxl/core/mbox.c
> @@ -1281,7 +1281,7 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
>  }
>  EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
>  
> -int cxl_dirty_shutdown_state(struct cxl_memdev_state *mds)
> +int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
>  {
>  	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
>  	struct cxl_mbox_cmd mbox_cmd;
> @@ -1297,7 +1297,7 @@ int cxl_dirty_shutdown_state(struct cxl_memdev_state *mds)
>  
>  	return cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
>  }
> -EXPORT_SYMBOL_NS_GPL(cxl_dirty_shutdown_state, "CXL");
> +EXPORT_SYMBOL_NS_GPL(cxl_arm_dirty_shutdown, "CXL");
>  
>  int cxl_set_timestamp(struct cxl_memdev_state *mds)
>  {

Re: [PATCH 2/4] cxl/pmem: Rename cxl_dirty_shutdown_state()

[Davidlohr Bueso]

On Thu, 20 Feb 2025, Jonathan Cameron wrote:

>On Wed, 19 Feb 2025 17:36:02 -0800
>Davidlohr Bueso <dave@stgolabs.net> wrote:
>
>> ... to a better suited 'cxl_arm_dirty_shutdown()'.
>
>But it works on x86 as well! I'll cope I suppose.
>
>Mind you why not let it take the state as a parameter
>and always pass 1?  From a mailbox command point of
>view 0 is valid. The text in table 8-152 has me a little confused
>but I think it says you can set it to 1 and back to 0 and
>then reset at which point the dirty shutdown count will
>not increment.
>
>No idea why you'd do that, but just passing a 1 into
>this function would make a call of
>cxl_set_dirty_shutdown_state(mds, 1)
>seem reasonable, or pass a boolean.

I prefer how it is now, really.

No point passing the arg which will always be 1, and that
"arm/setup" is in the name, so passing 0 would be counter
intuitive as well.

Thanks,
Davidlohr

[PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Davidlohr Bueso]

Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
introduce:

  /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown

Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
DVSEC exists, and 3) the count itself can be retrieved.

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 21 +++++
 drivers/cxl/cxl.h                             |  1 +
 drivers/cxl/cxlmem.h                          | 13 +++
 drivers/cxl/pmem.c                            | 79 ++++++++++++++++---
 6 files changed, 118 insertions(+), 10 deletions(-)

diff --git a/Documentation/ABI/testing/sysfs-bus-cxl b/Documentation/ABI/testing/sysfs-bus-cxl
index 3f5627a1210a..a7491d214098 100644
--- a/Documentation/ABI/testing/sysfs-bus-cxl
+++ b/Documentation/ABI/testing/sysfs-bus-cxl
@@ -586,3 +586,15 @@ Description:
 		See Documentation/ABI/stable/sysfs-devices-node. access0 provides
 		the number to the closest initiator and access1 provides the
 		number to the closest CPU.
+
+
+What:		/sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
+Date:		Feb, 2025
+KernelVersion:	v6.15
+Contact:	linux-cxl@vger.kernel.org
+Description:
+		(RO) The device dirty shutdown count value, which is the number
+		of times the device could have incurred in potential data loss.
+		The count is persistent across power loss and wraps back to 0
+		upon overflow. If this file is not present, the device does not
+		have the necessary support for dirty tracking.
diff --git a/Documentation/driver-api/cxl/maturity-map.rst b/Documentation/driver-api/cxl/maturity-map.rst
index 99dd2c841e69..a2288f9df658 100644
--- a/Documentation/driver-api/cxl/maturity-map.rst
+++ b/Documentation/driver-api/cxl/maturity-map.rst
@@ -130,7 +130,7 @@ Mailbox commands
 * [0] Switch CCI
 * [3] Timestamp
 * [1] PMEM labels
-* [1] PMEM GPF / Dirty Shutdown
+* [3] PMEM GPF / Dirty Shutdown
 * [0] Scan Media
 
 PMU
diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
index 86d13f4a1c18..6bc398182a5d 100644
--- a/drivers/cxl/core/mbox.c
+++ b/drivers/cxl/core/mbox.c
@@ -1281,6 +1281,27 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
 }
 EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
 
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count)
+{
+	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
+	struct cxl_mbox_get_health_info_out hi;
+	struct cxl_mbox_cmd mbox_cmd;
+	int rc;
+
+	mbox_cmd = (struct cxl_mbox_cmd) {
+		.opcode = CXL_MBOX_OP_GET_HEALTH_INFO,
+		.size_out = sizeof(hi),
+		.payload_out = &hi,
+	};
+
+	rc = cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
+	if (!rc)
+		*count = le32_to_cpu(hi.dirty_shutdown_cnt);
+
+	return rc;
+}
+EXPORT_SYMBOL_NS_GPL(cxl_get_dirty_count, "CXL");
+
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
 {
 	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
index 29f2ab0d5bf6..8bdfa536262e 100644
--- a/drivers/cxl/cxl.h
+++ b/drivers/cxl/cxl.h
@@ -542,6 +542,7 @@ struct cxl_nvdimm {
 	struct device dev;
 	struct cxl_memdev *cxlmd;
 	u8 dev_id[CXL_DEV_ID_LEN]; /* for nvdimm, string of 'serial' */
+	u64 dirty_shutdowns;
 };
 
 struct cxl_pmem_region_mapping {
diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
index 6d60030139df..3b6ef9e936c3 100644
--- a/drivers/cxl/cxlmem.h
+++ b/drivers/cxl/cxlmem.h
@@ -681,6 +681,18 @@ struct cxl_mbox_set_partition_info {
 
 #define  CXL_SET_PARTITION_IMMEDIATE_FLAG	BIT(0)
 
+/* Get Health Info Output Payload CXL 3.2 Spec 8.2.10.9.3.1 Table 8-148 */
+struct cxl_mbox_get_health_info_out {
+	u8 health_status;
+	u8 media_status;
+	u8 additional_status;
+	u8 life_used;
+	__le16 device_temperature;
+	__le32 dirty_shutdown_cnt;
+	__le32 corrected_volatile_error_cnt;
+	__le32 corrected_persistent_error_cnt;
+} __packed;
+
 /* Set Shutdown State Input Payload CXL 3.2 Spec 8.2.10.9.3.5 Table 8-152 */
 struct cxl_mbox_set_shutdown_state_in {
 	u8 state;
@@ -822,6 +834,7 @@ void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
 			    enum cxl_event_log_type type,
 			    enum cxl_event_type event_type,
 			    const uuid_t *uuid, union cxl_event *evt);
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count);
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds);
 int cxl_set_timestamp(struct cxl_memdev_state *mds);
 int cxl_poison_state_init(struct cxl_memdev_state *mds);
diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
index 6b284962592f..cb039cfc62cb 100644
--- a/drivers/cxl/pmem.c
+++ b/drivers/cxl/pmem.c
@@ -38,19 +38,48 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
 	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
 	struct cxl_dev_state *cxlds = cxl_nvd->cxlmd->cxlds;
 
-	return sysfs_emit(buf, "%lld\n", cxlds->serial);
+	return sysfs_emit(buf, "%llu\n", cxlds->serial);
 }
 static DEVICE_ATTR_RO(id);
 
+static ssize_t dirty_shutdown_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	struct nvdimm *nvdimm = to_nvdimm(dev);
+	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+	return sysfs_emit(buf, "%lld\n", cxl_nvd->dirty_shutdowns);
+}
+static DEVICE_ATTR_RO(dirty_shutdown);
+
 static struct attribute *cxl_dimm_attributes[] = {
 	&dev_attr_id.attr,
 	&dev_attr_provider.attr,
+	&dev_attr_dirty_shutdown.attr,
 	NULL
 };
 
+#define CXL_INVALID_DIRTY_SHUTDOWN_COUNT ULLONG_MAX
+static umode_t cxl_dimm_visible(struct kobject *kobj,
+				struct attribute *a, int n)
+{
+	if (a == &dev_attr_dirty_shutdown.attr) {
+		struct device *dev = kobj_to_dev(kobj);
+		struct nvdimm *nvdimm = to_nvdimm(dev);
+		struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+		if (cxl_nvd->dirty_shutdowns ==
+		    CXL_INVALID_DIRTY_SHUTDOWN_COUNT)
+			return 0;
+	}
+
+	return a->mode;
+}
+
 static const struct attribute_group cxl_dimm_attribute_group = {
 	.name = "cxl",
 	.attrs = cxl_dimm_attributes,
+	.is_visible = cxl_dimm_visible
 };
 
 static const struct attribute_group *cxl_dimm_attribute_groups[] = {
@@ -58,6 +87,38 @@ static const struct attribute_group *cxl_dimm_attribute_groups[] = {
 	NULL
 };
 
+static void cxl_nvdimm_setup_dirty_tracking(struct cxl_nvdimm *cxl_nvd)
+{
+	struct cxl_memdev *cxlmd = cxl_nvd->cxlmd;
+	struct cxl_dev_state *cxlds = cxlmd->cxlds;
+	struct cxl_memdev_state *mds = to_cxl_memdev_state(cxlds);
+	struct device *dev = &cxl_nvd->dev;
+	u32 count;
+
+	/*
+	 * Dirty tracking is enabled and exposed to the user, only when:
+	 *   - dirty shutdown on the device can be set, and,
+	 *   - the device has a Device GPF DVSEC (albeit unused), and,
+	 *   - the Get Health Info cmd can retrieve the device's dirty count.
+	 */
+	cxl_nvd->dirty_shutdowns = CXL_INVALID_DIRTY_SHUTDOWN_COUNT;
+
+	if (cxl_arm_dirty_shutdown(mds)) {
+		dev_warn(dev, "GPF: could not set dirty shutdown state\n");
+		return;
+	}
+
+	if (!cxl_gpf_get_dvsec(cxlds->dev, false))
+		return;
+
+	if (cxl_get_dirty_count(mds, &count)) {
+		dev_warn(dev, "GPF: could not retrieve dirty count\n");
+		return;
+	}
+
+	cxl_nvd->dirty_shutdowns = count;
+}
+
 static int cxl_nvdimm_probe(struct device *dev)
 {
 	struct cxl_nvdimm *cxl_nvd = to_cxl_nvdimm(dev);
@@ -78,20 +139,20 @@ static int cxl_nvdimm_probe(struct device *dev)
 	set_bit(ND_CMD_GET_CONFIG_SIZE, &cmd_mask);
 	set_bit(ND_CMD_GET_CONFIG_DATA, &cmd_mask);
 	set_bit(ND_CMD_SET_CONFIG_DATA, &cmd_mask);
-	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
-				 cxl_dimm_attribute_groups, flags,
-				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
-				 cxl_security_ops, NULL);
-	if (!nvdimm)
-		return -ENOMEM;
 
 	/*
 	 * Set dirty shutdown now, with the expectation that the device
 	 * clear it upon a successful GPF flow. The exception to this
 	 * is upon Viral detection, per CXL 3.2 section 12.4.2.
 	 */
-	if (cxl_arm_dirty_shutdown(mds))
-		dev_warn(dev, "GPF: could not dirty shutdown state\n");
+	cxl_nvdimm_setup_dirty_tracking(cxl_nvd);
+
+	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
+				 cxl_dimm_attribute_groups, flags,
+				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
+				 cxl_security_ops, NULL);
+	if (!nvdimm)
+		return -ENOMEM;
 
 	dev_set_drvdata(dev, nvdimm);
 	return devm_add_action_or_reset(dev, unregister_nvdimm, nvdimm);
-- 
2.39.5

Re: [PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Ira Weiny]

Davidlohr Bueso wrote:
> Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
> introduce:
> 
>   /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
> 
> Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
> DVSEC exists, and 3) the count itself can be retrieved.
> 
> Suggested-by: Dan Williams <dan.j.williams@intel.com>
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>

Reviewed-by: Ira Weiny <ira.weiny@intel.com>

> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

[snip]

Re: [PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Jonathan Cameron]

On Wed, 19 Feb 2025 17:36:03 -0800
Davidlohr Bueso <dave@stgolabs.net> wrote:

> Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
> introduce:
> 
>   /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
> 
> Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
> DVSEC exists, and 3) the count itself can be retrieved.
> 
> Suggested-by: Dan Williams <dan.j.williams@intel.com>
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
One trivial thing otherwise
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>


> diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
> index 6b284962592f..cb039cfc62cb 100644
> --- a/drivers/cxl/pmem.c
> +++ b/drivers/cxl/pmem.c
> @@ -38,19 +38,48 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
>  	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
>  	struct cxl_dev_state *cxlds = cxl_nvd->cxlmd->cxlds;
>  
> -	return sysfs_emit(buf, "%lld\n", cxlds->serial);
> +	return sysfs_emit(buf, "%llu\n", cxlds->serial);

I guess you 'fixed' the wrong one?

>  }
>  static DEVICE_ATTR_RO(id);
>  
> +static ssize_t dirty_shutdown_show(struct device *dev,
> +				   struct device_attribute *attr, char *buf)
> +{
> +	struct nvdimm *nvdimm = to_nvdimm(dev);
> +	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
> +
> +	return sysfs_emit(buf, "%lld\n", cxl_nvd->dirty_shutdowns);

It's unsigned so %llu though I hope no one ever tests that by
doing that many dirty shutdowns.

> +}
> +static DEVICE_ATTR_RO(dirty_shutdown);

Re: [PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Davidlohr Bueso]

On Thu, 20 Feb 2025, Jonathan Cameron wrote:

>On Wed, 19 Feb 2025 17:36:03 -0800
>Davidlohr Bueso <dave@stgolabs.net> wrote:
>
>> Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
>> introduce:
>>
>>   /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
>>
>> Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
>> DVSEC exists, and 3) the count itself can be retrieved.
>>
>> Suggested-by: Dan Williams <dan.j.williams@intel.com>
>> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
>> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
>One trivial thing otherwise
>Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
>
>
>> diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
>> index 6b284962592f..cb039cfc62cb 100644
>> --- a/drivers/cxl/pmem.c
>> +++ b/drivers/cxl/pmem.c
>> @@ -38,19 +38,48 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
>>  	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
>>  	struct cxl_dev_state *cxlds = cxl_nvd->cxlmd->cxlds;
>>
>> -	return sysfs_emit(buf, "%lld\n", cxlds->serial);
>> +	return sysfs_emit(buf, "%llu\n", cxlds->serial);
>
>I guess you 'fixed' the wrong one?

Yes, I'm an idiot. Will send a hopefully final v5 with this fix.

[PATCH 4/4] tools/testing/cxl: Set Shutdown State support

[Davidlohr Bueso]

Add support to emulate the CXL Set Shutdown State operation.

Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Li Ming <ming.li@zohomail.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 tools/testing/cxl/test/mem.c | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/tools/testing/cxl/test/mem.c b/tools/testing/cxl/test/mem.c
index 495199238335..9ca210b80e27 100644
--- a/tools/testing/cxl/test/mem.c
+++ b/tools/testing/cxl/test/mem.c
@@ -65,6 +65,10 @@ static struct cxl_cel_entry mock_cel[] = {
 		.opcode = cpu_to_le16(CXL_MBOX_OP_GET_HEALTH_INFO),
 		.effect = CXL_CMD_EFFECT_NONE,
 	},
+	{
+		.opcode = cpu_to_le16(CXL_MBOX_OP_SET_SHUTDOWN_STATE),
+		.effect = POLICY_CHANGE_IMMEDIATE,
+	},
 	{
 		.opcode = cpu_to_le16(CXL_MBOX_OP_GET_POISON),
 		.effect = CXL_CMD_EFFECT_NONE,
@@ -161,6 +165,7 @@ struct cxl_mockmem_data {
 	u8 event_buf[SZ_4K];
 	u64 timestamp;
 	unsigned long sanitize_timeout;
+	u8 shutdown_state;
 };
 
 static struct mock_event_log *event_find_log(struct device *dev, int log_type)
@@ -1088,6 +1093,21 @@ static int mock_health_info(struct cxl_mbox_cmd *cmd)
 	return 0;
 }
 
+static int mock_set_shutdown_state(struct cxl_mockmem_data *mdata,
+				   struct cxl_mbox_cmd *cmd)
+{
+	struct cxl_mbox_set_shutdown_state_in *ss = cmd->payload_in;
+
+	if (cmd->size_in != sizeof(*ss))
+		return -EINVAL;
+
+	if (cmd->size_out != 0)
+		return -EINVAL;
+
+	mdata->shutdown_state = ss->state;
+	return 0;
+}
+
 static struct mock_poison {
 	struct cxl_dev_state *cxlds;
 	u64 dpa;
@@ -1421,6 +1441,9 @@ static int cxl_mock_mbox_send(struct cxl_mailbox *cxl_mbox,
 	case CXL_MBOX_OP_PASSPHRASE_SECURE_ERASE:
 		rc = mock_passphrase_secure_erase(mdata, cmd);
 		break;
+	case CXL_MBOX_OP_SET_SHUTDOWN_STATE:
+		rc = mock_set_shutdown_state(mdata, cmd);
+		break;
 	case CXL_MBOX_OP_GET_POISON:
 		rc = mock_get_poison(cxlds, cmd);
 		break;
-- 
2.39.5

Re: [PATCH 4/4] tools/testing/cxl: Set Shutdown State support

[Ira Weiny]

Davidlohr Bueso wrote:
> Add support to emulate the CXL Set Shutdown State operation.
> 
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> Reviewed-by: Li Ming <ming.li@zohomail.com>

Reviewed-by: Ira Weiny <ira.weiny@intel.com>

> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

[snip]

Re: [PATCH 4/4] tools/testing/cxl: Set Shutdown State support

[Jonathan Cameron]

On Wed, 19 Feb 2025 17:36:04 -0800
Davidlohr Bueso <dave@stgolabs.net> wrote:

> Add support to emulate the CXL Set Shutdown State operation.
> 
> Reviewed-by: Dave Jiang <dave.jiang@intel.com>
> Reviewed-by: Li Ming <ming.li@zohomail.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
FWIW looks fine to me as well.
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>

[PATCH v5 0/4] cxl: Dirty shutdown followups

[Davidlohr Bueso]

Changes from v4:
  - use %llu on the correct sysfs file in patch 3 (Jonathan)
  - s/cxl_nvdimm_setup_dirty_tracking()/cxl_nvdimm_arm_dirty_shutdown_tracking()
    in patch 3 which was feedback from v1.

Hi,

Some followup patches to the GPF work. First two patches are from feedback
provided by DaveJ. The third patch adds a $platform/dirty_shutdown sysfs
attribute to expose the count to userspace. Fourth patch adds support
emulating the set shutdown state command for the mock device.

Applies against the -next branch of cxl.git.

Thanks!

Davidlohr Bueso (4):
  cxl/pci: Introduce cxl_gpf_get_dvsec()
  cxl/pmem: Rename cxl_dirty_shutdown_state()
  cxl/pmem: Export dirty shutdown count via sysfs
  tools/testing/cxl: Set Shutdown State support

 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 25 +++++-
 drivers/cxl/core/pci.c                        | 30 +++++---
 drivers/cxl/cxl.h                             |  3 +
 drivers/cxl/cxlmem.h                          | 15 +++-
 drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
 tools/testing/cxl/test/mem.c                  | 23 ++++++
 8 files changed, 165 insertions(+), 22 deletions(-)

-- 
2.39.5

[PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Davidlohr Bueso]

Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
introduce:

  /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown

Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
DVSEC exists, and 3) the count itself can be retrieved.

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
Reviewed-by: Dave Jiang <dave.jiang@intel.com>
Reviewed-by: Ira Weiny <ira.weiny@intel.com>
Reviewed-by: Jonathan Cameron <Jonathan.Cameron@huawei.com>
---
 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 21 +++++
 drivers/cxl/cxl.h                             |  1 +
 drivers/cxl/cxlmem.h                          | 13 ++++
 drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
 6 files changed, 117 insertions(+), 9 deletions(-)

diff --git a/Documentation/ABI/testing/sysfs-bus-cxl b/Documentation/ABI/testing/sysfs-bus-cxl
index 04a880bd1dde..6d911f046a78 100644
--- a/Documentation/ABI/testing/sysfs-bus-cxl
+++ b/Documentation/ABI/testing/sysfs-bus-cxl
@@ -604,3 +604,15 @@ Description:
 		See Documentation/ABI/stable/sysfs-devices-node. access0 provides
 		the number to the closest initiator and access1 provides the
 		number to the closest CPU.
+
+
+What:		/sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
+Date:		Feb, 2025
+KernelVersion:	v6.15
+Contact:	linux-cxl@vger.kernel.org
+Description:
+		(RO) The device dirty shutdown count value, which is the number
+		of times the device could have incurred in potential data loss.
+		The count is persistent across power loss and wraps back to 0
+		upon overflow. If this file is not present, the device does not
+		have the necessary support for dirty tracking.
diff --git a/Documentation/driver-api/cxl/maturity-map.rst b/Documentation/driver-api/cxl/maturity-map.rst
index 99dd2c841e69..a2288f9df658 100644
--- a/Documentation/driver-api/cxl/maturity-map.rst
+++ b/Documentation/driver-api/cxl/maturity-map.rst
@@ -130,7 +130,7 @@ Mailbox commands
 * [0] Switch CCI
 * [3] Timestamp
 * [1] PMEM labels
-* [1] PMEM GPF / Dirty Shutdown
+* [3] PMEM GPF / Dirty Shutdown
 * [0] Scan Media
 
 PMU
diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
index 86d13f4a1c18..6bc398182a5d 100644
--- a/drivers/cxl/core/mbox.c
+++ b/drivers/cxl/core/mbox.c
@@ -1281,6 +1281,27 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
 }
 EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
 
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count)
+{
+	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
+	struct cxl_mbox_get_health_info_out hi;
+	struct cxl_mbox_cmd mbox_cmd;
+	int rc;
+
+	mbox_cmd = (struct cxl_mbox_cmd) {
+		.opcode = CXL_MBOX_OP_GET_HEALTH_INFO,
+		.size_out = sizeof(hi),
+		.payload_out = &hi,
+	};
+
+	rc = cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
+	if (!rc)
+		*count = le32_to_cpu(hi.dirty_shutdown_cnt);
+
+	return rc;
+}
+EXPORT_SYMBOL_NS_GPL(cxl_get_dirty_count, "CXL");
+
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
 {
 	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
index 29f2ab0d5bf6..8bdfa536262e 100644
--- a/drivers/cxl/cxl.h
+++ b/drivers/cxl/cxl.h
@@ -542,6 +542,7 @@ struct cxl_nvdimm {
 	struct device dev;
 	struct cxl_memdev *cxlmd;
 	u8 dev_id[CXL_DEV_ID_LEN]; /* for nvdimm, string of 'serial' */
+	u64 dirty_shutdowns;
 };
 
 struct cxl_pmem_region_mapping {
diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
index 6d60030139df..3b6ef9e936c3 100644
--- a/drivers/cxl/cxlmem.h
+++ b/drivers/cxl/cxlmem.h
@@ -681,6 +681,18 @@ struct cxl_mbox_set_partition_info {
 
 #define  CXL_SET_PARTITION_IMMEDIATE_FLAG	BIT(0)
 
+/* Get Health Info Output Payload CXL 3.2 Spec 8.2.10.9.3.1 Table 8-148 */
+struct cxl_mbox_get_health_info_out {
+	u8 health_status;
+	u8 media_status;
+	u8 additional_status;
+	u8 life_used;
+	__le16 device_temperature;
+	__le32 dirty_shutdown_cnt;
+	__le32 corrected_volatile_error_cnt;
+	__le32 corrected_persistent_error_cnt;
+} __packed;
+
 /* Set Shutdown State Input Payload CXL 3.2 Spec 8.2.10.9.3.5 Table 8-152 */
 struct cxl_mbox_set_shutdown_state_in {
 	u8 state;
@@ -822,6 +834,7 @@ void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
 			    enum cxl_event_log_type type,
 			    enum cxl_event_type event_type,
 			    const uuid_t *uuid, union cxl_event *evt);
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count);
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds);
 int cxl_set_timestamp(struct cxl_memdev_state *mds);
 int cxl_poison_state_init(struct cxl_memdev_state *mds);
diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
index 5afd9ca9a944..d061fe3d2b86 100644
--- a/drivers/cxl/pmem.c
+++ b/drivers/cxl/pmem.c
@@ -42,15 +42,44 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
 }
 static DEVICE_ATTR_RO(id);
 
+static ssize_t dirty_shutdown_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	struct nvdimm *nvdimm = to_nvdimm(dev);
+	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+	return sysfs_emit(buf, "%llu\n", cxl_nvd->dirty_shutdowns);
+}
+static DEVICE_ATTR_RO(dirty_shutdown);
+
 static struct attribute *cxl_dimm_attributes[] = {
 	&dev_attr_id.attr,
 	&dev_attr_provider.attr,
+	&dev_attr_dirty_shutdown.attr,
 	NULL
 };
 
+#define CXL_INVALID_DIRTY_SHUTDOWN_COUNT ULLONG_MAX
+static umode_t cxl_dimm_visible(struct kobject *kobj,
+				struct attribute *a, int n)
+{
+	if (a == &dev_attr_dirty_shutdown.attr) {
+		struct device *dev = kobj_to_dev(kobj);
+		struct nvdimm *nvdimm = to_nvdimm(dev);
+		struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+		if (cxl_nvd->dirty_shutdowns ==
+		    CXL_INVALID_DIRTY_SHUTDOWN_COUNT)
+			return 0;
+	}
+
+	return a->mode;
+}
+
 static const struct attribute_group cxl_dimm_attribute_group = {
 	.name = "cxl",
 	.attrs = cxl_dimm_attributes,
+	.is_visible = cxl_dimm_visible
 };
 
 static const struct attribute_group *cxl_dimm_attribute_groups[] = {
@@ -58,6 +87,38 @@ static const struct attribute_group *cxl_dimm_attribute_groups[] = {
 	NULL
 };
 
+static void cxl_nvdimm_arm_dirty_shutdown_tracking(struct cxl_nvdimm *cxl_nvd)
+{
+	struct cxl_memdev *cxlmd = cxl_nvd->cxlmd;
+	struct cxl_dev_state *cxlds = cxlmd->cxlds;
+	struct cxl_memdev_state *mds = to_cxl_memdev_state(cxlds);
+	struct device *dev = &cxl_nvd->dev;
+	u32 count;
+
+	/*
+	 * Dirty tracking is enabled and exposed to the user, only when:
+	 *   - dirty shutdown on the device can be set, and,
+	 *   - the device has a Device GPF DVSEC (albeit unused), and,
+	 *   - the Get Health Info cmd can retrieve the device's dirty count.
+	 */
+	cxl_nvd->dirty_shutdowns = CXL_INVALID_DIRTY_SHUTDOWN_COUNT;
+
+	if (cxl_arm_dirty_shutdown(mds)) {
+		dev_warn(dev, "GPF: could not set dirty shutdown state\n");
+		return;
+	}
+
+	if (!cxl_gpf_get_dvsec(cxlds->dev, false))
+		return;
+
+	if (cxl_get_dirty_count(mds, &count)) {
+		dev_warn(dev, "GPF: could not retrieve dirty count\n");
+		return;
+	}
+
+	cxl_nvd->dirty_shutdowns = count;
+}
+
 static int cxl_nvdimm_probe(struct device *dev)
 {
 	struct cxl_nvdimm *cxl_nvd = to_cxl_nvdimm(dev);
@@ -78,20 +139,20 @@ static int cxl_nvdimm_probe(struct device *dev)
 	set_bit(ND_CMD_GET_CONFIG_SIZE, &cmd_mask);
 	set_bit(ND_CMD_GET_CONFIG_DATA, &cmd_mask);
 	set_bit(ND_CMD_SET_CONFIG_DATA, &cmd_mask);
-	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
-				 cxl_dimm_attribute_groups, flags,
-				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
-				 cxl_security_ops, NULL);
-	if (!nvdimm)
-		return -ENOMEM;
 
 	/*
 	 * Set dirty shutdown now, with the expectation that the device
 	 * clear it upon a successful GPF flow. The exception to this
 	 * is upon Viral detection, per CXL 3.2 section 12.4.2.
 	 */
-	if (cxl_arm_dirty_shutdown(mds))
-		dev_warn(dev, "GPF: could not dirty shutdown state\n");
+	cxl_nvdimm_arm_dirty_shutdown_tracking(cxl_nvd);
+
+	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
+				 cxl_dimm_attribute_groups, flags,
+				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
+				 cxl_security_ops, NULL);
+	if (!nvdimm)
+		return -ENOMEM;
 
 	dev_set_drvdata(dev, nvdimm);
 	return devm_add_action_or_reset(dev, unregister_nvdimm, nvdimm);
-- 
2.39.5

[PATCH v3 0/4] cxl: Dirty shutdown followups

[Davidlohr Bueso]

Changes from v2:
  - remove leftover cxl_gpf_device() declaration
  
Changes from v1 (https://lore.kernel.org/linux-cxl/20250205040842.1253616-1-dave@stgolabs.net/):
  - crated a new cxl_gpf_get_dvsec() helper to share for port and dev gpf (DaveJ)
  - renamed cxl_dirty_shutdown_state() to cxl_arm_dirty_shutdown() (DaveJ)
  - exported the cxl_gpf_get_dvsec() symbol used outside of core (Yi, DaveJ)
  - introduced CXL_INVALID_DIRTY_SHUTDOWN_COUNT (DaveJ)
  - use u64 for cxl_nvd->dirty_shutdowns,
  - rename to cxl_nvdimm_setup_dirty_tracking() and use return statements (DaveJ)
  - picked up review tag in patch 4 (DaveJ)

Hi,

Some followup patches to the GPF work. First two patches are from feedback
provided by DaveJ. The third patch adds a $platform/dirty_shutdown sysfs
attribute to expose the count to userspace. Fourth patch adds support
emulating the set shutdown state command for the mock device.

Applies against the -next branch of cxl.git.

Thanks!

Davidlohr Bueso (4):
  cxl/pci: Introduce cxl_gpf_get_dvsec()
  cxl/pmem: Rename cxl_dirty_shutdown_state()
  cxl/pmem: Export dirty shutdown count via sysfs
  tools/testing/cxl: Set Shutdown State support

 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 25 +++++-
 drivers/cxl/core/pci.c                        | 38 ++++++---
 drivers/cxl/cxl.h                             |  3 +
 drivers/cxl/cxlmem.h                          | 15 +++-
 drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
 tools/testing/cxl/test/mem.c                  | 23 ++++++
 8 files changed, 173 insertions(+), 22 deletions(-)

--
2.39.5

[PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Davidlohr Bueso]

Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
introduce:

  /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown

Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
DVSEC exists, and 3) the count itself can be retrieved.

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 21 +++++
 drivers/cxl/cxl.h                             |  1 +
 drivers/cxl/cxlmem.h                          | 13 ++++
 drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
 6 files changed, 117 insertions(+), 9 deletions(-)

diff --git a/Documentation/ABI/testing/sysfs-bus-cxl b/Documentation/ABI/testing/sysfs-bus-cxl
index 3f5627a1210a..a7491d214098 100644
--- a/Documentation/ABI/testing/sysfs-bus-cxl
+++ b/Documentation/ABI/testing/sysfs-bus-cxl
@@ -586,3 +586,15 @@ Description:
 		See Documentation/ABI/stable/sysfs-devices-node. access0 provides
 		the number to the closest initiator and access1 provides the
 		number to the closest CPU.
+
+
+What:		/sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
+Date:		Feb, 2025
+KernelVersion:	v6.15
+Contact:	linux-cxl@vger.kernel.org
+Description:
+		(RO) The device dirty shutdown count value, which is the number
+		of times the device could have incurred in potential data loss.
+		The count is persistent across power loss and wraps back to 0
+		upon overflow. If this file is not present, the device does not
+		have the necessary support for dirty tracking.
diff --git a/Documentation/driver-api/cxl/maturity-map.rst b/Documentation/driver-api/cxl/maturity-map.rst
index 99dd2c841e69..a2288f9df658 100644
--- a/Documentation/driver-api/cxl/maturity-map.rst
+++ b/Documentation/driver-api/cxl/maturity-map.rst
@@ -130,7 +130,7 @@ Mailbox commands
 * [0] Switch CCI
 * [3] Timestamp
 * [1] PMEM labels
-* [1] PMEM GPF / Dirty Shutdown
+* [3] PMEM GPF / Dirty Shutdown
 * [0] Scan Media
 
 PMU
diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
index 86d13f4a1c18..f1009a265f9d 100644
--- a/drivers/cxl/core/mbox.c
+++ b/drivers/cxl/core/mbox.c
@@ -1281,6 +1281,27 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
 }
 EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
 
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count)
+{
+	int rc;
+	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
+	struct cxl_mbox_cmd mbox_cmd;
+	struct cxl_mbox_get_health_info_out hi;
+
+	mbox_cmd = (struct cxl_mbox_cmd) {
+		.opcode = CXL_MBOX_OP_GET_HEALTH_INFO,
+		.size_out = sizeof(hi),
+		.payload_out = &hi,
+	};
+
+	rc = cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
+	if (!rc)
+		*count = le32_to_cpu(hi.dirty_shutdown_cnt);
+
+	return rc;
+}
+EXPORT_SYMBOL_NS_GPL(cxl_get_dirty_count, "CXL");
+
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
 {
 	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
index acbbba41356d..4dbf1cc60047 100644
--- a/drivers/cxl/cxl.h
+++ b/drivers/cxl/cxl.h
@@ -542,6 +542,7 @@ struct cxl_nvdimm {
 	struct device dev;
 	struct cxl_memdev *cxlmd;
 	u8 dev_id[CXL_DEV_ID_LEN]; /* for nvdimm, string of 'serial' */
+	u64 dirty_shutdowns;
 };
 
 struct cxl_pmem_region_mapping {
diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
index 6d60030139df..3b6ef9e936c3 100644
--- a/drivers/cxl/cxlmem.h
+++ b/drivers/cxl/cxlmem.h
@@ -681,6 +681,18 @@ struct cxl_mbox_set_partition_info {
 
 #define  CXL_SET_PARTITION_IMMEDIATE_FLAG	BIT(0)
 
+/* Get Health Info Output Payload CXL 3.2 Spec 8.2.10.9.3.1 Table 8-148 */
+struct cxl_mbox_get_health_info_out {
+	u8 health_status;
+	u8 media_status;
+	u8 additional_status;
+	u8 life_used;
+	__le16 device_temperature;
+	__le32 dirty_shutdown_cnt;
+	__le32 corrected_volatile_error_cnt;
+	__le32 corrected_persistent_error_cnt;
+} __packed;
+
 /* Set Shutdown State Input Payload CXL 3.2 Spec 8.2.10.9.3.5 Table 8-152 */
 struct cxl_mbox_set_shutdown_state_in {
 	u8 state;
@@ -822,6 +834,7 @@ void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
 			    enum cxl_event_log_type type,
 			    enum cxl_event_type event_type,
 			    const uuid_t *uuid, union cxl_event *evt);
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count);
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds);
 int cxl_set_timestamp(struct cxl_memdev_state *mds);
 int cxl_poison_state_init(struct cxl_memdev_state *mds);
diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
index 6b284962592f..aee1afe9d287 100644
--- a/drivers/cxl/pmem.c
+++ b/drivers/cxl/pmem.c
@@ -42,15 +42,44 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
 }
 static DEVICE_ATTR_RO(id);
 
+static ssize_t dirty_shutdown_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	struct nvdimm *nvdimm = to_nvdimm(dev);
+	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+	return sysfs_emit(buf, "%lld\n", cxl_nvd->dirty_shutdowns);
+}
+static DEVICE_ATTR_RO(dirty_shutdown);
+
 static struct attribute *cxl_dimm_attributes[] = {
 	&dev_attr_id.attr,
 	&dev_attr_provider.attr,
+	&dev_attr_dirty_shutdown.attr,
 	NULL
 };
 
+#define CXL_INVALID_DIRTY_SHUTDOWN_COUNT -1
+static umode_t cxl_dimm_visible(struct kobject *kobj,
+				struct attribute *a, int n)
+{
+	if (a == &dev_attr_dirty_shutdown.attr) {
+		struct device *dev = kobj_to_dev(kobj);
+		struct nvdimm *nvdimm = to_nvdimm(dev);
+		struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+		if (cxl_nvd->dirty_shutdowns ==
+		    CXL_INVALID_DIRTY_SHUTDOWN_COUNT)
+			return 0;
+	}
+
+	return a->mode;
+}
+
 static const struct attribute_group cxl_dimm_attribute_group = {
 	.name = "cxl",
 	.attrs = cxl_dimm_attributes,
+	.is_visible = cxl_dimm_visible
 };
 
 static const struct attribute_group *cxl_dimm_attribute_groups[] = {
@@ -58,6 +87,38 @@ static const struct attribute_group *cxl_dimm_attribute_groups[] = {
 	NULL
 };
 
+static void cxl_nvdimm_setup_dirty_tracking(struct cxl_nvdimm *cxl_nvd)
+{
+	u32 count;
+	struct cxl_memdev *cxlmd = cxl_nvd->cxlmd;
+	struct cxl_dev_state *cxlds = cxlmd->cxlds;
+	struct cxl_memdev_state *mds = to_cxl_memdev_state(cxlds);
+	struct device *dev = &cxl_nvd->dev;
+
+	/*
+	 * Dirty tracking is enabled and exposed to the user, only when:
+	 *   - dirty shutdown on the device can be set, and,
+	 *   - the device has a Device GPF DVSEC (albeit unused), and,
+	 *   - the Get Health Info cmd can retrieve the device's dirty count.
+	 */
+	cxl_nvd->dirty_shutdowns = CXL_INVALID_DIRTY_SHUTDOWN_COUNT;
+
+	if (cxl_arm_dirty_shutdown(mds)) {
+		dev_warn(dev, "GPF: could not set dirty shutdown state\n");
+		return;
+	}
+
+	if (cxl_gpf_get_dvsec(cxlds->dev, false) <= 0)
+		return;
+
+	if (cxl_get_dirty_count(mds, &count)) {
+		dev_warn(dev, "GPF: could not retrieve dirty count\n");
+		return;
+	}
+
+	cxl_nvd->dirty_shutdowns = count;
+}
+
 static int cxl_nvdimm_probe(struct device *dev)
 {
 	struct cxl_nvdimm *cxl_nvd = to_cxl_nvdimm(dev);
@@ -78,20 +139,20 @@ static int cxl_nvdimm_probe(struct device *dev)
 	set_bit(ND_CMD_GET_CONFIG_SIZE, &cmd_mask);
 	set_bit(ND_CMD_GET_CONFIG_DATA, &cmd_mask);
 	set_bit(ND_CMD_SET_CONFIG_DATA, &cmd_mask);
-	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
-				 cxl_dimm_attribute_groups, flags,
-				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
-				 cxl_security_ops, NULL);
-	if (!nvdimm)
-		return -ENOMEM;
 
 	/*
 	 * Set dirty shutdown now, with the expectation that the device
 	 * clear it upon a successful GPF flow. The exception to this
 	 * is upon Viral detection, per CXL 3.2 section 12.4.2.
 	 */
-	if (cxl_arm_dirty_shutdown(mds))
-		dev_warn(dev, "GPF: could not dirty shutdown state\n");
+	cxl_nvdimm_setup_dirty_tracking(cxl_nvd);
+
+	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
+				 cxl_dimm_attribute_groups, flags,
+				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
+				 cxl_security_ops, NULL);
+	if (!nvdimm)
+		return -ENOMEM;
 
 	dev_set_drvdata(dev, nvdimm);
 	return devm_add_action_or_reset(dev, unregister_nvdimm, nvdimm);
-- 
2.39.5

Re: [PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Dave Jiang]

On 2/18/25 11:28 PM, Davidlohr Bueso wrote:
> Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
> introduce:
> 
>   /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
> 
> Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
> DVSEC exists, and 3) the count itself can be retrieved.
> 
> Suggested-by: Dan Williams <dan.j.williams@intel.com>
> Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>

Reviewed-by: Dave Jiang <dave.jiang@intel.com>

just a couple nits below

> ---
>  Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
>  Documentation/driver-api/cxl/maturity-map.rst |  2 +-
>  drivers/cxl/core/mbox.c                       | 21 +++++
>  drivers/cxl/cxl.h                             |  1 +
>  drivers/cxl/cxlmem.h                          | 13 ++++
>  drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
>  6 files changed, 117 insertions(+), 9 deletions(-)
> 
> diff --git a/Documentation/ABI/testing/sysfs-bus-cxl b/Documentation/ABI/testing/sysfs-bus-cxl
> index 3f5627a1210a..a7491d214098 100644
> --- a/Documentation/ABI/testing/sysfs-bus-cxl
> +++ b/Documentation/ABI/testing/sysfs-bus-cxl
> @@ -586,3 +586,15 @@ Description:
>  		See Documentation/ABI/stable/sysfs-devices-node. access0 provides
>  		the number to the closest initiator and access1 provides the
>  		number to the closest CPU.
> +
> +
> +What:		/sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
> +Date:		Feb, 2025
> +KernelVersion:	v6.15
> +Contact:	linux-cxl@vger.kernel.org
> +Description:
> +		(RO) The device dirty shutdown count value, which is the number
> +		of times the device could have incurred in potential data loss.
> +		The count is persistent across power loss and wraps back to 0
> +		upon overflow. If this file is not present, the device does not
> +		have the necessary support for dirty tracking.
> diff --git a/Documentation/driver-api/cxl/maturity-map.rst b/Documentation/driver-api/cxl/maturity-map.rst
> index 99dd2c841e69..a2288f9df658 100644
> --- a/Documentation/driver-api/cxl/maturity-map.rst
> +++ b/Documentation/driver-api/cxl/maturity-map.rst
> @@ -130,7 +130,7 @@ Mailbox commands
>  * [0] Switch CCI
>  * [3] Timestamp
>  * [1] PMEM labels
> -* [1] PMEM GPF / Dirty Shutdown
> +* [3] PMEM GPF / Dirty Shutdown
>  * [0] Scan Media
>  
>  PMU
> diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
> index 86d13f4a1c18..f1009a265f9d 100644
> --- a/drivers/cxl/core/mbox.c
> +++ b/drivers/cxl/core/mbox.c
> @@ -1281,6 +1281,27 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
>  }
>  EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
>  
> +int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count)
> +{
> +	int rc;
> +	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
> +	struct cxl_mbox_cmd mbox_cmd;
> +	struct cxl_mbox_get_health_info_out hi;

reverse xmas tree pls

> +
> +	mbox_cmd = (struct cxl_mbox_cmd) {
> +		.opcode = CXL_MBOX_OP_GET_HEALTH_INFO,
> +		.size_out = sizeof(hi),
> +		.payload_out = &hi,
> +	};
> +
> +	rc = cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
> +	if (!rc)
> +		*count = le32_to_cpu(hi.dirty_shutdown_cnt);
> +
> +	return rc;
> +}
> +EXPORT_SYMBOL_NS_GPL(cxl_get_dirty_count, "CXL");
> +
>  int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
>  {
>  	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
> diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
> index acbbba41356d..4dbf1cc60047 100644
> --- a/drivers/cxl/cxl.h
> +++ b/drivers/cxl/cxl.h
> @@ -542,6 +542,7 @@ struct cxl_nvdimm {
>  	struct device dev;
>  	struct cxl_memdev *cxlmd;
>  	u8 dev_id[CXL_DEV_ID_LEN]; /* for nvdimm, string of 'serial' */
> +	u64 dirty_shutdowns;
>  };
>  
>  struct cxl_pmem_region_mapping {
> diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
> index 6d60030139df..3b6ef9e936c3 100644
> --- a/drivers/cxl/cxlmem.h
> +++ b/drivers/cxl/cxlmem.h
> @@ -681,6 +681,18 @@ struct cxl_mbox_set_partition_info {
>  
>  #define  CXL_SET_PARTITION_IMMEDIATE_FLAG	BIT(0)
>  
> +/* Get Health Info Output Payload CXL 3.2 Spec 8.2.10.9.3.1 Table 8-148 */
> +struct cxl_mbox_get_health_info_out {
> +	u8 health_status;
> +	u8 media_status;
> +	u8 additional_status;
> +	u8 life_used;
> +	__le16 device_temperature;
> +	__le32 dirty_shutdown_cnt;
> +	__le32 corrected_volatile_error_cnt;
> +	__le32 corrected_persistent_error_cnt;
> +} __packed;
> +
>  /* Set Shutdown State Input Payload CXL 3.2 Spec 8.2.10.9.3.5 Table 8-152 */
>  struct cxl_mbox_set_shutdown_state_in {
>  	u8 state;
> @@ -822,6 +834,7 @@ void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
>  			    enum cxl_event_log_type type,
>  			    enum cxl_event_type event_type,
>  			    const uuid_t *uuid, union cxl_event *evt);
> +int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count);
>  int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds);
>  int cxl_set_timestamp(struct cxl_memdev_state *mds);
>  int cxl_poison_state_init(struct cxl_memdev_state *mds);
> diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
> index 6b284962592f..aee1afe9d287 100644
> --- a/drivers/cxl/pmem.c
> +++ b/drivers/cxl/pmem.c
> @@ -42,15 +42,44 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
>  }
>  static DEVICE_ATTR_RO(id);
>  
> +static ssize_t dirty_shutdown_show(struct device *dev,
> +				   struct device_attribute *attr, char *buf)
> +{
> +	struct nvdimm *nvdimm = to_nvdimm(dev);
> +	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
> +
> +	return sysfs_emit(buf, "%lld\n", cxl_nvd->dirty_shutdowns);
> +}
> +static DEVICE_ATTR_RO(dirty_shutdown);
> +
>  static struct attribute *cxl_dimm_attributes[] = {
>  	&dev_attr_id.attr,
>  	&dev_attr_provider.attr,
> +	&dev_attr_dirty_shutdown.attr,
>  	NULL
>  };
>  
> +#define CXL_INVALID_DIRTY_SHUTDOWN_COUNT -1
> +static umode_t cxl_dimm_visible(struct kobject *kobj,
> +				struct attribute *a, int n)
> +{
> +	if (a == &dev_attr_dirty_shutdown.attr) {
> +		struct device *dev = kobj_to_dev(kobj);
> +		struct nvdimm *nvdimm = to_nvdimm(dev);
> +		struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
> +
> +		if (cxl_nvd->dirty_shutdowns ==
> +		    CXL_INVALID_DIRTY_SHUTDOWN_COUNT)
> +			return 0;
> +	}
> +
> +	return a->mode;
> +}
> +
>  static const struct attribute_group cxl_dimm_attribute_group = {
>  	.name = "cxl",
>  	.attrs = cxl_dimm_attributes,
> +	.is_visible = cxl_dimm_visible
>  };
>  
>  static const struct attribute_group *cxl_dimm_attribute_groups[] = {
> @@ -58,6 +87,38 @@ static const struct attribute_group *cxl_dimm_attribute_groups[] = {
>  	NULL
>  };
>  
> +static void cxl_nvdimm_setup_dirty_tracking(struct cxl_nvdimm *cxl_nvd)
> +{
> +	u32 count;
> +	struct cxl_memdev *cxlmd = cxl_nvd->cxlmd;
> +	struct cxl_dev_state *cxlds = cxlmd->cxlds;
> +	struct cxl_memdev_state *mds = to_cxl_memdev_state(cxlds);
> +	struct device *dev = &cxl_nvd->dev;

reverse xmas tree pls

> +
> +	/*
> +	 * Dirty tracking is enabled and exposed to the user, only when:
> +	 *   - dirty shutdown on the device can be set, and,
> +	 *   - the device has a Device GPF DVSEC (albeit unused), and,
> +	 *   - the Get Health Info cmd can retrieve the device's dirty count.
> +	 */
> +	cxl_nvd->dirty_shutdowns = CXL_INVALID_DIRTY_SHUTDOWN_COUNT;
> +
> +	if (cxl_arm_dirty_shutdown(mds)) {
> +		dev_warn(dev, "GPF: could not set dirty shutdown state\n");
> +		return;
> +	}
> +
> +	if (cxl_gpf_get_dvsec(cxlds->dev, false) <= 0)
> +		return;
> +
> +	if (cxl_get_dirty_count(mds, &count)) {
> +		dev_warn(dev, "GPF: could not retrieve dirty count\n");
> +		return;
> +	}
> +
> +	cxl_nvd->dirty_shutdowns = count;
> +}
> +
>  static int cxl_nvdimm_probe(struct device *dev)
>  {
>  	struct cxl_nvdimm *cxl_nvd = to_cxl_nvdimm(dev);
> @@ -78,20 +139,20 @@ static int cxl_nvdimm_probe(struct device *dev)
>  	set_bit(ND_CMD_GET_CONFIG_SIZE, &cmd_mask);
>  	set_bit(ND_CMD_GET_CONFIG_DATA, &cmd_mask);
>  	set_bit(ND_CMD_SET_CONFIG_DATA, &cmd_mask);
> -	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
> -				 cxl_dimm_attribute_groups, flags,
> -				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
> -				 cxl_security_ops, NULL);
> -	if (!nvdimm)
> -		return -ENOMEM;
>  
>  	/*
>  	 * Set dirty shutdown now, with the expectation that the device
>  	 * clear it upon a successful GPF flow. The exception to this
>  	 * is upon Viral detection, per CXL 3.2 section 12.4.2.
>  	 */
> -	if (cxl_arm_dirty_shutdown(mds))
> -		dev_warn(dev, "GPF: could not dirty shutdown state\n");
> +	cxl_nvdimm_setup_dirty_tracking(cxl_nvd);
> +
> +	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
> +				 cxl_dimm_attribute_groups, flags,
> +				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
> +				 cxl_security_ops, NULL);
> +	if (!nvdimm)
> +		return -ENOMEM;
>  
>  	dev_set_drvdata(dev, nvdimm);
>  	return devm_add_action_or_reset(dev, unregister_nvdimm, nvdimm);

Re: [PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Ira Weiny]

Davidlohr Bueso wrote:

[snip]

> +static ssize_t dirty_shutdown_show(struct device *dev,
> +				   struct device_attribute *attr, char *buf)
> +{
> +	struct nvdimm *nvdimm = to_nvdimm(dev);
> +	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
> +
> +	return sysfs_emit(buf, "%lld\n", cxl_nvd->dirty_shutdowns);
> +}
> +static DEVICE_ATTR_RO(dirty_shutdown);
> +
>  static struct attribute *cxl_dimm_attributes[] = {
>  	&dev_attr_id.attr,
>  	&dev_attr_provider.attr,
> +	&dev_attr_dirty_shutdown.attr,
>  	NULL
>  };
>  
> +#define CXL_INVALID_DIRTY_SHUTDOWN_COUNT -1

I think this should be defined as ULLONG_MAX.  Just for clarity since
dirty_shutdown is u64.

Ira

> +static umode_t cxl_dimm_visible(struct kobject *kobj,
> +				struct attribute *a, int n)
> +{
> +	if (a == &dev_attr_dirty_shutdown.attr) {
> +		struct device *dev = kobj_to_dev(kobj);
> +		struct nvdimm *nvdimm = to_nvdimm(dev);
> +		struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
> +
> +		if (cxl_nvd->dirty_shutdowns ==
> +		    CXL_INVALID_DIRTY_SHUTDOWN_COUNT)
> +			return 0;
> +	}
> +
> +	return a->mode;
> +}
> +

[snip]

[PATCH v2 0/4] cxl: Dirty shutdown followups

[Davidlohr Bueso]

Changes from v1 (https://lore.kernel.org/linux-cxl/20250205040842.1253616-1-dave@stgolabs.net/):
  - crated a new cxl_gpf_get_dvsec() helper to share for port and dev gpf (DaveJ)
  - renamed cxl_dirty_shutdown_state() to cxl_arm_dirty_shutdown() (DaveJ)
  - exported the cxl_gpf_get_dvsec() symbol used outside of core (Yi, DaveJ)
  - introduced CXL_INVALID_DIRTY_SHUTDOWN_COUNT (DaveJ)
  - rename to cxl_nvdimm_setup_dirty_tracking() and use return statements (DaveJ)
  - picked up review tag in patch 4 (DaveJ)

Hi,

Some followup patches to the GPF work. First two patches are from feedback
provided by DaveJ. The third patch adds a $platform/dirty_shutdown sysfs
attribute to expose the count to userspace. Fourth patch adds support
emulating the set shutdown state command for the mock device.

Applies against the -next branch of cxl.git.

Thanks!

Davidlohr Bueso (4):
  cxl/pci: Introduce cxl_gpf_get_dvsec()
  cxl/pmem: Rename cxl_dirty_shutdown_state()
  cxl/pmem: Export dirty shutdown count via sysfs
  tools/testing/cxl: Set Shutdown State support

 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 25 +++++-
 drivers/cxl/core/pci.c                        | 38 ++++++---
 drivers/cxl/cxl.h                             |  3 +
 drivers/cxl/cxlmem.h                          | 17 +++-
 drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
 tools/testing/cxl/test/mem.c                  | 23 ++++++
 8 files changed, 175 insertions(+), 22 deletions(-)

--
2.39.5

[PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Davidlohr Bueso]

Similar to how the acpi_nfit driver exports Optane dirty shutdown count,
introduce:

  /sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown

Under the conditions that 1) dirty shutdown can be set, 2) Device GPF
DVSEC exists, and 3) the count itself can be retrieved.

Suggested-by: Dan Williams <dan.j.williams@intel.com>
Signed-off-by: Davidlohr Bueso <dave@stgolabs.net>
---
 Documentation/ABI/testing/sysfs-bus-cxl       | 12 +++
 Documentation/driver-api/cxl/maturity-map.rst |  2 +-
 drivers/cxl/core/mbox.c                       | 21 +++++
 drivers/cxl/cxl.h                             |  1 +
 drivers/cxl/cxlmem.h                          | 15 ++++
 drivers/cxl/pmem.c                            | 77 +++++++++++++++++--
 6 files changed, 119 insertions(+), 9 deletions(-)

diff --git a/Documentation/ABI/testing/sysfs-bus-cxl b/Documentation/ABI/testing/sysfs-bus-cxl
index 3f5627a1210a..a7491d214098 100644
--- a/Documentation/ABI/testing/sysfs-bus-cxl
+++ b/Documentation/ABI/testing/sysfs-bus-cxl
@@ -586,3 +586,15 @@ Description:
 		See Documentation/ABI/stable/sysfs-devices-node. access0 provides
 		the number to the closest initiator and access1 provides the
 		number to the closest CPU.
+
+
+What:		/sys/bus/cxl/devices/nvdimm-bridge0/ndbusX/nmemY/cxl/dirty_shutdown
+Date:		Feb, 2025
+KernelVersion:	v6.15
+Contact:	linux-cxl@vger.kernel.org
+Description:
+		(RO) The device dirty shutdown count value, which is the number
+		of times the device could have incurred in potential data loss.
+		The count is persistent across power loss and wraps back to 0
+		upon overflow. If this file is not present, the device does not
+		have the necessary support for dirty tracking.
diff --git a/Documentation/driver-api/cxl/maturity-map.rst b/Documentation/driver-api/cxl/maturity-map.rst
index 99dd2c841e69..a2288f9df658 100644
--- a/Documentation/driver-api/cxl/maturity-map.rst
+++ b/Documentation/driver-api/cxl/maturity-map.rst
@@ -130,7 +130,7 @@ Mailbox commands
 * [0] Switch CCI
 * [3] Timestamp
 * [1] PMEM labels
-* [1] PMEM GPF / Dirty Shutdown
+* [3] PMEM GPF / Dirty Shutdown
 * [0] Scan Media
 
 PMU
diff --git a/drivers/cxl/core/mbox.c b/drivers/cxl/core/mbox.c
index 86d13f4a1c18..f1009a265f9d 100644
--- a/drivers/cxl/core/mbox.c
+++ b/drivers/cxl/core/mbox.c
@@ -1281,6 +1281,27 @@ int cxl_mem_dpa_fetch(struct cxl_memdev_state *mds, struct cxl_dpa_info *info)
 }
 EXPORT_SYMBOL_NS_GPL(cxl_mem_dpa_fetch, "CXL");
 
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count)
+{
+	int rc;
+	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
+	struct cxl_mbox_cmd mbox_cmd;
+	struct cxl_mbox_get_health_info_out hi;
+
+	mbox_cmd = (struct cxl_mbox_cmd) {
+		.opcode = CXL_MBOX_OP_GET_HEALTH_INFO,
+		.size_out = sizeof(hi),
+		.payload_out = &hi,
+	};
+
+	rc = cxl_internal_send_cmd(cxl_mbox, &mbox_cmd);
+	if (!rc)
+		*count = le32_to_cpu(hi.dirty_shutdown_cnt);
+
+	return rc;
+}
+EXPORT_SYMBOL_NS_GPL(cxl_get_dirty_count, "CXL");
+
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds)
 {
 	struct cxl_mailbox *cxl_mbox = &mds->cxlds.cxl_mbox;
diff --git a/drivers/cxl/cxl.h b/drivers/cxl/cxl.h
index acbbba41356d..4dbf1cc60047 100644
--- a/drivers/cxl/cxl.h
+++ b/drivers/cxl/cxl.h
@@ -542,6 +542,7 @@ struct cxl_nvdimm {
 	struct device dev;
 	struct cxl_memdev *cxlmd;
 	u8 dev_id[CXL_DEV_ID_LEN]; /* for nvdimm, string of 'serial' */
+	u64 dirty_shutdowns;
 };
 
 struct cxl_pmem_region_mapping {
diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
index 6d60030139df..03ad3c8ba88d 100644
--- a/drivers/cxl/cxlmem.h
+++ b/drivers/cxl/cxlmem.h
@@ -681,6 +681,18 @@ struct cxl_mbox_set_partition_info {
 
 #define  CXL_SET_PARTITION_IMMEDIATE_FLAG	BIT(0)
 
+/* Get Health Info Output Payload CXL 3.2 Spec 8.2.10.9.3.1 Table 8-148 */
+struct cxl_mbox_get_health_info_out {
+	u8 health_status;
+	u8 media_status;
+	u8 additional_status;
+	u8 life_used;
+	__le16 device_temperature;
+	__le32 dirty_shutdown_cnt;
+	__le32 corrected_volatile_error_cnt;
+	__le32 corrected_persistent_error_cnt;
+} __packed;
+
 /* Set Shutdown State Input Payload CXL 3.2 Spec 8.2.10.9.3.5 Table 8-152 */
 struct cxl_mbox_set_shutdown_state_in {
 	u8 state;
@@ -822,6 +834,7 @@ void cxl_event_trace_record(const struct cxl_memdev *cxlmd,
 			    enum cxl_event_log_type type,
 			    enum cxl_event_type event_type,
 			    const uuid_t *uuid, union cxl_event *evt);
+int cxl_get_dirty_count(struct cxl_memdev_state *mds, u32 *count);
 int cxl_arm_dirty_shutdown(struct cxl_memdev_state *mds);
 int cxl_set_timestamp(struct cxl_memdev_state *mds);
 int cxl_poison_state_init(struct cxl_memdev_state *mds);
@@ -866,4 +879,6 @@ struct cxl_hdm {
 struct seq_file;
 struct dentry *cxl_debugfs_create_dir(const char *dir);
 void cxl_dpa_debug(struct seq_file *file, struct cxl_dev_state *cxlds);
+
+int cxl_gpf_device(struct cxl_dev_state *cxlds);
 #endif /* __CXL_MEM_H__ */
diff --git a/drivers/cxl/pmem.c b/drivers/cxl/pmem.c
index 6b284962592f..aee1afe9d287 100644
--- a/drivers/cxl/pmem.c
+++ b/drivers/cxl/pmem.c
@@ -42,15 +42,44 @@ static ssize_t id_show(struct device *dev, struct device_attribute *attr, char *
 }
 static DEVICE_ATTR_RO(id);
 
+static ssize_t dirty_shutdown_show(struct device *dev,
+				   struct device_attribute *attr, char *buf)
+{
+	struct nvdimm *nvdimm = to_nvdimm(dev);
+	struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+	return sysfs_emit(buf, "%lld\n", cxl_nvd->dirty_shutdowns);
+}
+static DEVICE_ATTR_RO(dirty_shutdown);
+
 static struct attribute *cxl_dimm_attributes[] = {
 	&dev_attr_id.attr,
 	&dev_attr_provider.attr,
+	&dev_attr_dirty_shutdown.attr,
 	NULL
 };
 
+#define CXL_INVALID_DIRTY_SHUTDOWN_COUNT -1
+static umode_t cxl_dimm_visible(struct kobject *kobj,
+				struct attribute *a, int n)
+{
+	if (a == &dev_attr_dirty_shutdown.attr) {
+		struct device *dev = kobj_to_dev(kobj);
+		struct nvdimm *nvdimm = to_nvdimm(dev);
+		struct cxl_nvdimm *cxl_nvd = nvdimm_provider_data(nvdimm);
+
+		if (cxl_nvd->dirty_shutdowns ==
+		    CXL_INVALID_DIRTY_SHUTDOWN_COUNT)
+			return 0;
+	}
+
+	return a->mode;
+}
+
 static const struct attribute_group cxl_dimm_attribute_group = {
 	.name = "cxl",
 	.attrs = cxl_dimm_attributes,
+	.is_visible = cxl_dimm_visible
 };
 
 static const struct attribute_group *cxl_dimm_attribute_groups[] = {
@@ -58,6 +87,38 @@ static const struct attribute_group *cxl_dimm_attribute_groups[] = {
 	NULL
 };
 
+static void cxl_nvdimm_setup_dirty_tracking(struct cxl_nvdimm *cxl_nvd)
+{
+	u32 count;
+	struct cxl_memdev *cxlmd = cxl_nvd->cxlmd;
+	struct cxl_dev_state *cxlds = cxlmd->cxlds;
+	struct cxl_memdev_state *mds = to_cxl_memdev_state(cxlds);
+	struct device *dev = &cxl_nvd->dev;
+
+	/*
+	 * Dirty tracking is enabled and exposed to the user, only when:
+	 *   - dirty shutdown on the device can be set, and,
+	 *   - the device has a Device GPF DVSEC (albeit unused), and,
+	 *   - the Get Health Info cmd can retrieve the device's dirty count.
+	 */
+	cxl_nvd->dirty_shutdowns = CXL_INVALID_DIRTY_SHUTDOWN_COUNT;
+
+	if (cxl_arm_dirty_shutdown(mds)) {
+		dev_warn(dev, "GPF: could not set dirty shutdown state\n");
+		return;
+	}
+
+	if (cxl_gpf_get_dvsec(cxlds->dev, false) <= 0)
+		return;
+
+	if (cxl_get_dirty_count(mds, &count)) {
+		dev_warn(dev, "GPF: could not retrieve dirty count\n");
+		return;
+	}
+
+	cxl_nvd->dirty_shutdowns = count;
+}
+
 static int cxl_nvdimm_probe(struct device *dev)
 {
 	struct cxl_nvdimm *cxl_nvd = to_cxl_nvdimm(dev);
@@ -78,20 +139,20 @@ static int cxl_nvdimm_probe(struct device *dev)
 	set_bit(ND_CMD_GET_CONFIG_SIZE, &cmd_mask);
 	set_bit(ND_CMD_GET_CONFIG_DATA, &cmd_mask);
 	set_bit(ND_CMD_SET_CONFIG_DATA, &cmd_mask);
-	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
-				 cxl_dimm_attribute_groups, flags,
-				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
-				 cxl_security_ops, NULL);
-	if (!nvdimm)
-		return -ENOMEM;
 
 	/*
 	 * Set dirty shutdown now, with the expectation that the device
 	 * clear it upon a successful GPF flow. The exception to this
 	 * is upon Viral detection, per CXL 3.2 section 12.4.2.
 	 */
-	if (cxl_arm_dirty_shutdown(mds))
-		dev_warn(dev, "GPF: could not dirty shutdown state\n");
+	cxl_nvdimm_setup_dirty_tracking(cxl_nvd);
+
+	nvdimm = __nvdimm_create(cxl_nvb->nvdimm_bus, cxl_nvd,
+				 cxl_dimm_attribute_groups, flags,
+				 cmd_mask, 0, NULL, cxl_nvd->dev_id,
+				 cxl_security_ops, NULL);
+	if (!nvdimm)
+		return -ENOMEM;
 
 	dev_set_drvdata(dev, nvdimm);
 	return devm_add_action_or_reset(dev, unregister_nvdimm, nvdimm);
-- 
2.39.5

Re: [PATCH 3/4] cxl/pmem: Export dirty shutdown count via sysfs

[Davidlohr Bueso]

On Tue, 18 Feb 2025, Davidlohr Bueso wrote:

>diff --git a/drivers/cxl/cxlmem.h b/drivers/cxl/cxlmem.h
>index 6d60030139df..03ad3c8ba88d 100644
>--- a/drivers/cxl/cxlmem.h
>+++ b/drivers/cxl/cxlmem.h
>@@ -866,4 +879,6 @@ struct cxl_hdm {
> struct seq_file;
> struct dentry *cxl_debugfs_create_dir(const char *dir);
> void cxl_dpa_debug(struct seq_file *file, struct cxl_dev_state *cxlds);
>+
>+int cxl_gpf_device(struct cxl_dev_state *cxlds);

Bleh this is a leftover of v1, it slipped through. I will send a v3.

Thanks,
Davidlohr