Skip to contentSkip to conversation

Add support for reporting papr nvdimm health

PATCH Superseded Vaibhav Jain 2 participants 11 messages Jun 16, 2020 – Jun 18, 2020 libnvdimm: non-volatile memory device

v6 of 13 Superseded Latest is v13 →

lore
Share & tools
lore
#0 lore raw

v6 cover Add support for reporting papr nvdimm health

Changes since v5 [1]:
* Removed the patch introducing new dimm-ops 'dimm_init()' & 'dimm_uninit()'. Corrosponding code that used the dimm private initialization is also removed.
* Updated various dimm ops callback to rely on 'struct ndctl_cmd' arg instead of dimm-private.
* Added ndctl_bus_has_of_node() and ndctl_bus_is_papr_scm() to library ld version script.
* Simplified probing of new papr compatible nvdimm based on introduction of new exported library function ndctl_bus_is_papr_scm().
* Reworked various dimm-ops callbacks based on update uapi interface with papr_scm as defined at [5].
* Introduced a new header 'papr.h' that defines 'struct nd_pkg_papr' that holds 'struct nd_cmd_pkg gen' and 'struct nd_pkg_pdsm pdsm' together.

[1] https://lore.kernel.org/linux-nvdimm/[email protected]
---
This patch-set proposes changes to libndctl to add support for reporting health for nvdimms that support the PAPR standard[2]. The standard defines machenism (HCALL) through which a guest kernel can query and fetch health and performance stats of an nvdimm attached to the hypervisor[3]. Until now 'ndctl' was unable to report these stats for papr_scm dimms on PPC64 guests due to absence of ACPI/NFIT, a limitation which this patch-set tries to address.

The patch-set introduces support for the new PAPR PDSM family defined at [4] & [5] via a new dimm-ops named 'papr_dimm_ops'. Infrastructure to probe and distinguish papr-scm dimms from other dimm families that may support ACPI/NFIT is implemented by updating the 'struct ndctl_dimm' initialization routines to bifurcate based on the nvdimm type. We also introduce two new dimm-ops member for handling initialization of dimm specific data for specific DSM families.

These changes coupled with proposed kernel changes located at Ref[1] should provide a way for the user to retrieve NVDIMM health status using ndtcl for pseries guests. Below is a sample output using proposed kernel + ndctl changes:

 # ndctl list -DH
[
  {
    "dev":"nmem0",
    "flag_smart_event":true,
    "health":{
      "health_state":"fatal",
      "shutdown_state":"dirty"
    }
  }
]

Structure of the patchset
=========================

We start with a re-factoring patch that splits the 'add_dimm()' function into two functions one that take care of allocating and initializing 'struct ndctl_dimm' and another that takes care of initializing nfit specific dimm attributes.

Patch-2 introduces probe function of papr nvdimms and assigning 'papr_dimm_ops' defined in 'papr.c' to 'dimm->ops' if needed. The patch also code to parse the dimm flags specific to papr nvdimms

Patches-3,4 implements scaffolding to add support for PAPR PDSM requests and pull in their definitions from the kernel.

Finally Patch-6 add support for issuing and handling the result of 'struct ndctl_cmd' to request dimm health stats from papr_scm kernel module and returning appropriate health status to libndctl for reporting.

References
==========
[2] "Power Architecture Platform Reference"
https://en.wikipedia.org/wiki/Power_Architecture_Platform_Reference

[3] "Hypercall Op-codes (hcalls)"
https://github.com/torvalds/linux/blob/master/Documentation/powerpc/papr_hcalls.rst

[4] "powerpc/papr_scm: Add support for reporting nvdimm health"
https://lore.kernel.org/linux-nvdimm/[email protected]/

[5] "ndctl/papr_scm,uapi: Add support for PAPR nvdimm specific methods"
https://lore.kernel.org/linux-nvdimm/[email protected]/

Vaibhav Jain (5):
libndctl: Refactor out add_dimm() to handle NFIT specific init
libncdtl: Add initial support for NVDIMM_FAMILY_PAPR nvdimm family libndctl,papr_scm: Add definitions for PAPR nvdimm specific methods
papr: Add scaffolding to issue and handle PDSM requests libndctl,papr_scm: Implement support for PAPR_PDSM_HEALTH

 ndctl/lib/Makefile.am  |   1 +
 ndctl/lib/libndctl.c   | 264 +++++++++++++++++++++++++++++------------
 ndctl/lib/libndctl.sym |   5 +
 ndctl/lib/papr.c       | 218 ++++++++++++++++++++++++++++++++++
 ndctl/lib/papr.h       |  15 +++
 ndctl/lib/papr_pdsm.h  | 132 +++++++++++++++++++++
 ndctl/lib/private.h    |   4 +
 ndctl/libndctl.h       |   2 +
 ndctl/ndctl.h          |   1 +
 9 files changed, 569 insertions(+), 73 deletions(-)
 create mode 100644 ndctl/lib/papr.c
 create mode 100644 ndctl/lib/papr.h
 create mode 100644 ndctl/lib/papr_pdsm.h
Signature · 4 lines
2.26.2
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
#1 lore raw

PATCH v6 1/5 libndctl: Refactor out add_dimm() to handle NFIT specific init

Presently add_dimm() only probes dimms that support NFIT/ACPI. Hence this patch refactors this functionality into two functions namely add_dimm() and add_nfit_dimm(). Function add_dimm() performs allocation and common 'struct ndctl_dimm' initialization and depending on whether the dimm-bus supports NIFT, calls add_nfit_dimm(). Once the probe is completed based on the value of 'ndctl_dimm.cmd_family' appropriate dimm-ops are assigned to the dimm.

In case dimm-bus is of unknown type or doesn't support NFIT the initialization still continues, with no dimm-ops assigned to the 'struct ndctl_dimm' there-by limiting the functionality available.

This patch shouldn't introduce any behavioral change.

---
Changelog:

v5..v6:
* Changed a return code for add_nfit_dimm() in case a allocation failed. [ Vishal ]
* Updated an error message in error path of add_dimm() [ Vishal ]

v4..v5:
* None

v3..v4:
* None

v2..v3:
* None

v1..v2:
* None

---
 ndctl/lib/libndctl.c | 196 +++++++++++++++++++++++++------------------
 1 file changed, 115 insertions(+), 81 deletions(-)
ndctl/lib/libndctl.c +115 -81
index ee737cbbfe3e..85c4d5b34548 100644
@@ -1441,82 +1441,15 @@ static int ndctl_bind(struct ndctl_ctx *ctx, struct kmod_module *module,
 static int ndctl_unbind(struct ndctl_ctx *ctx, const char *devpath);
 static struct kmod_module *to_module(struct ndctl_ctx *ctx, const char *alias);
 
-static void *add_dimm(void *parent, int id, const char *dimm_base)
+static int add_nfit_dimm(struct ndctl_dimm *dimm, const char *dimm_base)
 {
-	int formats, i;
-	struct ndctl_dimm *dimm;
+	int i, rc = -1;
 	char buf[SYSFS_ATTR_SIZE];
-	struct ndctl_bus *bus = parent;
-	struct ndctl_ctx *ctx = bus->ctx;
+	struct ndctl_ctx *ctx = dimm->bus->ctx;
 	char *path = calloc(1, strlen(dimm_base) + 100);
 
 	if (!path)
-		return NULL;
-
-	sprintf(path, "%s/nfit/formats", dimm_base);
-	if (sysfs_read_attr(ctx, path, buf) < 0)
-		formats = 1;
-	else
-		formats = clamp(strtoul(buf, NULL, 0), 1UL, 2UL);
-
-	dimm = calloc(1, sizeof(*dimm) + sizeof(int) * formats);
-	if (!dimm)
-		goto err_dimm;
-	dimm->bus = bus;
-	dimm->id = id;
-
-	sprintf(path, "%s/dev", dimm_base);
-	if (sysfs_read_attr(ctx, path, buf) < 0)
-		goto err_read;
-	if (sscanf(buf, "%d:%d", &dimm->major, &dimm->minor) != 2)
-		goto err_read;
-
-	sprintf(path, "%s/commands", dimm_base);
-	if (sysfs_read_attr(ctx, path, buf) < 0)
-		goto err_read;
-	dimm->cmd_mask = parse_commands(buf, 1);
-
-	dimm->dimm_buf = calloc(1, strlen(dimm_base) + 50);
-	if (!dimm->dimm_buf)
-		goto err_read;
-	dimm->buf_len = strlen(dimm_base) + 50;
-
-	dimm->dimm_path = strdup(dimm_base);
-	if (!dimm->dimm_path)
-		goto err_read;
-
-	sprintf(path, "%s/modalias", dimm_base);
-	if (sysfs_read_attr(ctx, path, buf) < 0)
-		goto err_read;
-	dimm->module = to_module(ctx, buf);
-
-	dimm->handle = -1;
-	dimm->phys_id = -1;
-	dimm->serial = -1;
-	dimm->vendor_id = -1;
-	dimm->device_id = -1;
-	dimm->revision_id = -1;
-	dimm->health_eventfd = -1;
-	dimm->dirty_shutdown = -ENOENT;
-	dimm->subsystem_vendor_id = -1;
-	dimm->subsystem_device_id = -1;
-	dimm->subsystem_revision_id = -1;
-	dimm->manufacturing_date = -1;
-	dimm->manufacturing_location = -1;
-	dimm->cmd_family = -1;
-	dimm->nfit_dsm_mask = ULONG_MAX;
-	for (i = 0; i < formats; i++)
-		dimm->format[i] = -1;
-
-	sprintf(path, "%s/flags", dimm_base);
-	if (sysfs_read_attr(ctx, path, buf) < 0) {
-		dimm->locked = -1;
-		dimm->aliased = -1;
-	} else
-		parse_dimm_flags(dimm, buf);
-
-	if (!ndctl_bus_has_nfit(bus))
-		goto out;
+		return -ENOMEM;
 
 	/*
 	 * 'unique_id' may not be available on older kernels, so don't
@@ -1582,24 +1515,15 @@ static void *add_dimm(void *parent, int id, const char *dimm_base)
 	sprintf(path, "%s/nfit/family", dimm_base);
 	if (sysfs_read_attr(ctx, path, buf) == 0)
 		dimm->cmd_family = strtoul(buf, NULL, 0);
-	if (dimm->cmd_family == NVDIMM_FAMILY_INTEL)
-		dimm->ops = intel_dimm_ops;
-	if (dimm->cmd_family == NVDIMM_FAMILY_HPE1)
-		dimm->ops = hpe1_dimm_ops;
-	if (dimm->cmd_family == NVDIMM_FAMILY_MSFT)
-		dimm->ops = msft_dimm_ops;
-	if (dimm->cmd_family == NVDIMM_FAMILY_HYPERV)
-		dimm->ops = hyperv_dimm_ops;
 
 	sprintf(path, "%s/nfit/dsm_mask", dimm_base);
 	if (sysfs_read_attr(ctx, path, buf) == 0)
 		dimm->nfit_dsm_mask = strtoul(buf, NULL, 0);
 
-	dimm->formats = formats;
 	sprintf(path, "%s/nfit/format", dimm_base);
 	if (sysfs_read_attr(ctx, path, buf) == 0)
 		dimm->format[0] = strtoul(buf, NULL, 0);
-	for (i = 1; i < formats; i++) {
+	for (i = 1; i < dimm->formats; i++) {
 		sprintf(path, "%s/nfit/format%d", dimm_base, i);
 		if (sysfs_read_attr(ctx, path, buf) == 0)
 			dimm->format[i] = strtoul(buf, NULL, 0);
@@ -1610,7 +1534,117 @@ static void *add_dimm(void *parent, int id, const char *dimm_base)
 		parse_nfit_mem_flags(dimm, buf);
 
 	dimm->health_eventfd = open(path, O_RDONLY|O_CLOEXEC);
+	rc = 0;
+ err_read:
+
+	free(path);
+	return rc;
+}
+
+static void *add_dimm(void *parent, int id, const char *dimm_base)
+{
+	int formats, i, rc = -ENODEV;
+	struct ndctl_dimm *dimm = NULL;
+	char buf[SYSFS_ATTR_SIZE];
+	struct ndctl_bus *bus = parent;
+	struct ndctl_ctx *ctx = bus->ctx;
+	char *path = calloc(1, strlen(dimm_base) + 100);
+
+	if (!path)
+		return NULL;
+
+	sprintf(path, "%s/nfit/formats", dimm_base);
+	if (sysfs_read_attr(ctx, path, buf) < 0)
+		formats = 1;
+	else
+		formats = clamp(strtoul(buf, NULL, 0), 1UL, 2UL);
+
+	dimm = calloc(1, sizeof(*dimm) + sizeof(int) * formats);
+	if (!dimm)
+		goto err_dimm;
+	dimm->bus = bus;
+	dimm->id = id;
+
+	sprintf(path, "%s/dev", dimm_base);
+	if (sysfs_read_attr(ctx, path, buf) < 0)
+		goto err_read;
+	if (sscanf(buf, "%d:%d", &dimm->major, &dimm->minor) != 2)
+		goto err_read;
+
+	sprintf(path, "%s/commands", dimm_base);
+	if (sysfs_read_attr(ctx, path, buf) < 0)
+		goto err_read;
+	dimm->cmd_mask = parse_commands(buf, 1);
+
+	dimm->dimm_buf = calloc(1, strlen(dimm_base) + 50);
+	if (!dimm->dimm_buf)
+		goto err_read;
+	dimm->buf_len = strlen(dimm_base) + 50;
+
+	dimm->dimm_path = strdup(dimm_base);
+	if (!dimm->dimm_path)
+		goto err_read;
+
+	sprintf(path, "%s/modalias", dimm_base);
+	if (sysfs_read_attr(ctx, path, buf) < 0)
+		goto err_read;
+	dimm->module = to_module(ctx, buf);
+
+	dimm->handle = -1;
+	dimm->phys_id = -1;
+	dimm->serial = -1;
+	dimm->vendor_id = -1;
+	dimm->device_id = -1;
+	dimm->revision_id = -1;
+	dimm->health_eventfd = -1;
+	dimm->dirty_shutdown = -ENOENT;
+	dimm->subsystem_vendor_id = -1;
+	dimm->subsystem_device_id = -1;
+	dimm->subsystem_revision_id = -1;
+	dimm->manufacturing_date = -1;
+	dimm->manufacturing_location = -1;
+	dimm->cmd_family = -1;
+	dimm->nfit_dsm_mask = ULONG_MAX;
+	for (i = 0; i < formats; i++)
+		dimm->format[i] = -1;
+
+	sprintf(path, "%s/flags", dimm_base);
+	if (sysfs_read_attr(ctx, path, buf) < 0) {
+		dimm->locked = -1;
+		dimm->aliased = -1;
+	} else
+		parse_dimm_flags(dimm, buf);
+
+	/* Check if the given dimm supports nfit */
+	if (ndctl_bus_has_nfit(bus)) {
+		dimm->formats = formats;
+		rc = add_nfit_dimm(dimm, dimm_base);
+	}
+
+	if (rc == -ENODEV) {
+		/* Unprobed dimm with no family */
+		rc = 0;
+		goto out;
+	}
+
+	/* Assign dimm-ops based on command family */
+	if (dimm->cmd_family == NVDIMM_FAMILY_INTEL)
+		dimm->ops = intel_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_HPE1)
+		dimm->ops = hpe1_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_MSFT)
+		dimm->ops = msft_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_HYPERV)
+		dimm->ops = hyperv_dimm_ops;
  out:
+	if (rc) {
+		/* Ensure dimm_uninit() is not called during free_dimm() */
+		dimm->ops = NULL;
+		err(ctx, "%s: probe failed: %s\n", ndctl_dimm_get_devname(dimm),
+		    strerror(-rc));
+		goto err_read;
+	}
+
 	list_add(&bus->dimms, &dimm->list);
 	free(path);
 
Signature · 4 lines
2.26.2
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
maintainer
#2 lore raw

Re: [ndctl PATCH v6 1/5] libndctl: Refactor out add_dimm() to handle NFIT specific init

Quoted 1 line · Vaibhav Jain

Presently add_dimm() only probes dimms that support NFIT/ACPI. Hence

"...probes NVDIMMs for platforms that support the ACPI NFIT"

The NFIT is a platform firmware thing, not directly related to the DIMMs themselves.

Quoted 1 line

this patch refactors this functionality into two functions namely

s/Hence this patch refactors/Refactor/

Quoted 9 lines

add_dimm() and add_nfit_dimm(). Function add_dimm() performs allocation and common 'struct ndctl_dimm' initialization and depending on whether the dimm-bus supports NIFT, calls add_nfit_dimm(). Once the probe is completed based on the value of 'ndctl_dimm.cmd_family' appropriate dimm-ops are assigned to the dimm.

In case dimm-bus is of unknown type or doesn't support NFIT the initialization still continues, with no dimm-ops assigned to the 'struct ndctl_dimm' there-by limiting the functionality available.

No need to hyphenate 'thereby'

Quoted 27 lines

This patch shouldn't introduce any behavioral change.

---
Changelog:

v5..v6:
* Changed a return code for add_nfit_dimm() in case a allocation failed. [ Vishal ]
* Updated an error message in error path of add_dimm() [ Vishal ]

v4..v5:
* None

v3..v4:
* None

v2..v3:
* None

v1..v2:
* None

---
 ndctl/lib/libndctl.c | 196 +++++++++++++++++++++++++------------------
 1 file changed, 115 insertions(+), 81 deletions(-)

[..]

Quoted 4 lines
ndctl/lib/libndctl.c +4 -0
+
+	/* Assign dimm-ops based on command family */
+	if (dimm->cmd_family == NVDIMM_FAMILY_INTEL)
+		dimm->ops = intel_dimm_ops;

ndctl/lib/libndctl.c static void *add_dimm(void *parent, int id, const char *dimm_base) · line 1642

Quoted 10 lines
ndctl/lib/libndctl.c +9 -0
+	if (dimm->cmd_family == NVDIMM_FAMILY_HPE1)
+		dimm->ops = hpe1_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_MSFT)
+		dimm->ops = msft_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_HYPERV)
+		dimm->ops = hyperv_dimm_ops;
  out:
+	if (rc) {
+		/* Ensure dimm_uninit() is not called during free_dimm() */
+		dimm->ops = NULL;

I think the above assignment and comment are now stale..

ndctl/lib/libndctl.c static void *add_dimm(void *parent, int id, const char *dimm_base) · line 1649

Quoted 8 lines
ndctl/lib/libndctl.c +5 -0
+		err(ctx, "%s: probe failed: %s\n", ndctl_dimm_get_devname(dimm),
+		    strerror(-rc));
+		goto err_read;
+	}
+
 	list_add(&bus->dimms, &dimm->list);
 	free(path);

_______________________________________________
Linux-nvdimm mailing list -- [email protected] To unsubscribe send an email to [email protected]

#3 lore raw

Re: [ndctl PATCH v6 1/5] libndctl: Refactor out add_dimm() to handle NFIT specific init

Hi Vishal,

Thanks for reviewing this patch. I will be addressing your review comments in v7 of this patch series.

~ Vaibhav

Quoted 78 lines · Vishal L
Quoted 1 line · Vaibhav Jain

Presently add_dimm() only probes dimms that support NFIT/ACPI. Hence

"...probes NVDIMMs for platforms that support the ACPI NFIT"

The NFIT is a platform firmware thing, not directly related to the DIMMs themselves.

Quoted 1 line

this patch refactors this functionality into two functions namely

s/Hence this patch refactors/Refactor/

Quoted 9 lines

add_dimm() and add_nfit_dimm(). Function add_dimm() performs allocation and common 'struct ndctl_dimm' initialization and depending on whether the dimm-bus supports NIFT, calls add_nfit_dimm(). Once the probe is completed based on the value of 'ndctl_dimm.cmd_family' appropriate dimm-ops are assigned to the dimm.

In case dimm-bus is of unknown type or doesn't support NFIT the initialization still continues, with no dimm-ops assigned to the 'struct ndctl_dimm' there-by limiting the functionality available.

No need to hyphenate 'thereby'

Quoted 27 lines

This patch shouldn't introduce any behavioral change.

---
Changelog:

v5..v6:
* Changed a return code for add_nfit_dimm() in case a allocation failed. [ Vishal ]
* Updated an error message in error path of add_dimm() [ Vishal ]

v4..v5:
* None

v3..v4:
* None

v2..v3:
* None

v1..v2:
* None

---
 ndctl/lib/libndctl.c | 196 +++++++++++++++++++++++++------------------
 1 file changed, 115 insertions(+), 81 deletions(-)

[..]

Quoted 4 lines
ndctl/lib/libndctl.c +4 -0
+
+	/* Assign dimm-ops based on command family */
+	if (dimm->cmd_family == NVDIMM_FAMILY_INTEL)
+		dimm->ops = intel_dimm_ops;

ndctl/lib/libndctl.c static void *add_dimm(void *parent, int id, const char *dimm_base) · line 1642

Quoted 10 lines
ndctl/lib/libndctl.c +9 -0
+	if (dimm->cmd_family == NVDIMM_FAMILY_HPE1)
+		dimm->ops = hpe1_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_MSFT)
+		dimm->ops = msft_dimm_ops;
+	if (dimm->cmd_family == NVDIMM_FAMILY_HYPERV)
+		dimm->ops = hyperv_dimm_ops;
  out:
+	if (rc) {
+		/* Ensure dimm_uninit() is not called during free_dimm() */
+		dimm->ops = NULL;

I think the above assignment and comment are now stale..

Quoted 8 lines
ndctl/lib/libndctl.c +5 -0
+		err(ctx, "%s: probe failed: %s\n", ndctl_dimm_get_devname(dimm),
+		    strerror(-rc));
+		goto err_read;
+	}
+
 	list_add(&bus->dimms, &dimm->list);
 	free(path);
Signature · 5 lines
Cheers
~ Vaibhav
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
#4 lore raw

PATCH v6 2/5 libncdtl: Add initial support for NVDIMM_FAMILY_PAPR nvdimm family

In reply to Vaibhav Jain

Add necessary scaffolding in libndctl for dimms that support papr_scm specification[1]. Since there can be platforms that support Open-Firmware[2] but not the papr_scm specification, hence the changes proposed first add support for probing if the dimm bus supports Open-Firmware. This is done via querying for sysfs attribute 'of_node' in dimm device sysfs directory. If available newly introduced member 'struct ndctl_bus.has_of_node' is set. During the probe of the dimm and execution of add_dimm(), the newly introduced add_papr_dimm() is called if dimm bus reports supports Open-Firmware.

Function add_papr_dimm() queries the 'compatible' device tree attribute via newly introduced ndctl_bus_is_papr_scm() and based on its value assign NVDIMM_FAMILY_PAPR to the dimm command family. In future, based on the contents of 'compatible' attribute more of_pmem dimm families can be queried.

We also add support for parsing the dimm flags for NVDIMM_FAMILY_PAPR supporting nvdimms as described at [3]. A newly introduced function parse_papr_flags() reads the contents of this flag file and sets appropriate flag bits in 'struct ndctl_dimm.flags'.

Also we advertise support for monitor mode by allocating a file descriptor to the dimm 'flags' file and assigning it to 'struct ndctl_dimm.health_event_fd'.

The dimm-ops implementation for NVDIMM_FAMILY_PAPR is available in global variable 'papr_dimm_ops' which points to skeleton implementation in newly introduced file 'lib/papr.c'.

References:
[1] Documentation/powerpc/papr_hcalls.rst
https://lore.kernel.org/linux-nvdimm/[email protected]

[2] https://en.wikipedia.org/wiki/Open_Firmware

[3] Documentation/ABI/testing/sysfs-bus-papr-pmem
https://lore.kernel.org/linux-nvdimm/[email protected]

---
Changelog:

v5..v6:
* Updated various dbg messages to use consistent dimm device name rather than dimm-id. [Vishal]
* Minor changes to log messages in add_papr_dimm() [ Vishal ]
* Minor change in parse_papr_flags() to aid in code readability [Vishal]
* Added ndctl_bus_has_of_node() and ndctl_bus_is_papr_scm to library ld version script. [Vishal]
* Added new library function ndctl_bus_is_papr_scm() that checks the device tree compatible field [Vishal]
* Updated add_papr_dimm() to use ndctl_bus_is_papr_scm() [Vishal]
* Updated the license header of 'papr.c' to LGPL 2.1

v4..v5:
* Renamed file 'papr_scm.c' to 'papr.c'
* s/NVDIMM_FAMILY_PAPR_SCM/NVDIMM_FAMILY_PAPR/g
* s/papr_scm_dimm_ops/papr_dimm_ops/g
* s/parse_papr_scm_flags()/parse_papr_flags()/g
* Updated patch description & title to reflect new command family name and file-names.

v3..v4:
* None

v2..v3:
* Renamed add_of_pmem() to add_papr_dimm() [ Aneesh ]

v1..v2:
* Squashed the patch to parse dimm flags
* Updated the dimm flags parsing to add case for 'restore_fail' and 'flush_fail'.
* Renamed parse_of_pmem_flags() to parse_papr_scm_flags().
* Updated the path to dimm flags file to 'papr/flags'.
* Introduced 'papr_scm.c' file in this patch rather than later in the patch series.
* Update add_of_pmem_dimm() to parse dimm flags and enable monitoring only if 'ibm,pmemory' compatible nvdimm is found.

---
 ndctl/lib/Makefile.am  |  1 +
 ndctl/lib/libndctl.c   | 84 ++++++++++++++++++++++++++++++++++++++++++
 ndctl/lib/libndctl.sym |  5 +++
 ndctl/lib/papr.c       | 22 +++++++++++
 ndctl/lib/private.h    |  2 +
 ndctl/libndctl.h       |  2 +
 ndctl/ndctl.h          |  1 +
 7 files changed, 117 insertions(+)
 create mode 100644 ndctl/lib/papr.c
ndctl/lib/Makefile.am +1 -0
index d6be5c3acd26..e15bb2288575 100644
@@ -23,6 +23,7 @@ libndctl_la_SOURCES =\
 	hpe1.c \
 	msft.c \
 	hyperv.c \
+	papr.c \
 	ars.c \
 	firmware.c \
 	libndctl.c \
ndctl/lib/libndctl.c +84 -0

120 lines not on this page · raw message

ndctl/lib/libndctl.sym +5 -0
index ac575a23d035..9cfaccf5d3b7 100644
@@ -431,3 +431,8 @@ LIBNDCTL_23 {
 	ndctl_region_get_align;
 	ndctl_region_set_align;
 } LIBNDCTL_22;
+
+LIBNDCTL_24 {
+	ndctl_bus_has_of_node;
+	ndctl_bus_is_papr_scm;
+} LIBNDCTL_23;
ndctl/lib/papr.c new file +22 -0
new file mode 100644
index 000000000000..4b6ce8beccab
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: LGPL-2.1
+
+#include <stdint.h>
+#include <stdlib.h>
+#include <limits.h>
+#include <util/log.h>
+#include <ndctl.h>
+#include <ndctl/libndctl.h>
+#include <lib/private.h>
+
+static bool papr_cmd_is_supported(struct ndctl_dimm *dimm, int cmd)
+{
+	/* Handle this separately to support monitor mode */
+	if (cmd == ND_CMD_SMART)
+		return true;
+
+	return !!(dimm->cmd_mask & (1ULL << cmd));
+}
+
+struct ndctl_dimm_ops * const papr_dimm_ops = &(struct ndctl_dimm_ops) {
+	.cmd_is_supported = papr_cmd_is_supported,
+};
ndctl/lib/private.h +2 -0
index 2e537f0a8649..d90236b1f98b 100644
@@ -167,6 +167,7 @@ struct ndctl_bus {
 	int dimms_init;
 	int regions_init;
 	int has_nfit;
+	int has_of_node;
 	char *bus_path;
 	char *bus_buf;
 	size_t buf_len;
@@ -352,6 +353,7 @@ extern struct ndctl_dimm_ops * const intel_dimm_ops;
 extern struct ndctl_dimm_ops * const hpe1_dimm_ops;
 extern struct ndctl_dimm_ops * const msft_dimm_ops;
 extern struct ndctl_dimm_ops * const hyperv_dimm_ops;
+extern struct ndctl_dimm_ops * const papr_dimm_ops;
 
 static inline struct ndctl_bus *cmd_to_bus(struct ndctl_cmd *cmd)
 {
ndctl/libndctl.h +2 -0
index 2580f433ade8..e16fb7eaf34b 100644
@@ -119,6 +119,8 @@ struct ndctl_bus *ndctl_bus_get_next(struct ndctl_bus *bus);
              bus = ndctl_bus_get_next(bus))
 struct ndctl_ctx *ndctl_bus_get_ctx(struct ndctl_bus *bus);
 int ndctl_bus_has_nfit(struct ndctl_bus *bus);
+int ndctl_bus_has_of_node(struct ndctl_bus *bus);
+int ndctl_bus_is_papr_scm(struct ndctl_bus *bus);
 unsigned int ndctl_bus_get_major(struct ndctl_bus *bus);
 unsigned int ndctl_bus_get_minor(struct ndctl_bus *bus);
 const char *ndctl_bus_get_devname(struct ndctl_bus *bus);
ndctl/ndctl.h +1 -0
index 008f81cdeb9f..3b64f66d58cc 100644
@@ -263,6 +263,7 @@ struct nd_cmd_pkg {
 #define NVDIMM_FAMILY_HPE2 2
 #define NVDIMM_FAMILY_MSFT 3
 #define NVDIMM_FAMILY_HYPERV 4
+#define NVDIMM_FAMILY_PAPR 5
 
 #define ND_IOCTL_CALL			_IOWR(ND_IOCTL, ND_CMD_CALL,\
 					struct nd_cmd_pkg)
Signature · 4 lines
2.26.2
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
maintainer
#5 lore raw

Re: [ndctl PATCH v6 2/5] libncdtl: Add initial support for NVDIMM_FAMILY_PAPR nvdimm family

Quoted 1 line · Vaibhav Jain

Add necessary scaffolding in libndctl for dimms that support papr_scm

support /the/ papr_scm specification

Quoted 2 lines

specification[1]. Since there can be platforms that support Open-Firmware[2] but not the papr_scm specification, hence the changes

s/hence//

Quoted 2 lines

proposed first add support for probing if the dimm bus supports Open-Firmware. This is done via querying for sysfs attribute 'of_node'

This is done /by/ querying for /the/ sysfs attribute

Quoted 3 lines

in dimm device sysfs directory. If available newly introduced member 'struct ndctl_bus.has_of_node' is set. During the probe of the dimm and execution of add_dimm(), the newly introduced add_papr_dimm()

During 'add_dimm()', the newly introduced..

Quoted 5 lines

is called if dimm bus reports supports Open-Firmware.

Function add_papr_dimm() queries the 'compatible' device tree attribute via newly introduced ndctl_bus_is_papr_scm() and based on its value assign NVDIMM_FAMILY_PAPR to the dimm command family. In

based on its value, assigns NVDIM_..

Quoted 1 line

future, based on the contents of 'compatible' attribute more of_pmem

In /the/ future

Quoted 3 lines

dimm families can be queried.

We also add support for parsing the dimm flags for

'We' can be ambiguous. Say something like: "Additionally, add support.."

Quoted 6 lines

NVDIMM_FAMILY_PAPR supporting nvdimms as described at [3]. A newly introduced function parse_papr_flags() reads the contents of this flag file and sets appropriate flag bits in 'struct ndctl_dimm.flags'.

Also we advertise support for monitor mode by allocating a file

"Advertise support for monitor mode.."

Quoted 6 lines

descriptor to the dimm 'flags' file and assigning it to 'struct ndctl_dimm.health_event_fd'.

The dimm-ops implementation for NVDIMM_FAMILY_PAPR is available in global variable 'papr_dimm_ops' which points to skeleton implementation in newly introduced file 'lib/papr.c'.

This paragraph can just be dropped - it's a minor implementation detail, and doesn't add much to the commit message. The same actually goes for the part above that talks about setting flags.

Quoted 9 lines

References:
[1] Documentation/powerpc/papr_hcalls.rst
https://lore.kernel.org/linux-nvdimm/[email protected]

[2] https://en.wikipedia.org/wiki/Open_Firmware

[3] Documentation/ABI/testing/sysfs-bus-papr-pmem
https://lore.kernel.org/linux-nvdimm/[email protected]

Not a huge deal, but the lore links can probably be updated to the latest posting.

Quoted 4 lines

---

[..]

ndctl/lib/papr.c line 1

Quoted 7 lines
ndctl/lib/papr.c new file +1 -0
new file mode 100644
index 000000000000..4b6ce8beccab
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: LGPL-2.1

I'm not sure if you intended to drop the copyright line here :)

Quoted 12 lines
ndctl/lib/papr.c new file +11 -0
+
+#include <stdint.h>
+#include <stdlib.h>
+#include <limits.h>
+#include <util/log.h>
+#include <ndctl.h>
+#include <ndctl/libndctl.h>
+#include <lib/private.h>
+
+static bool papr_cmd_is_supported(struct ndctl_dimm *dimm, int cmd)
+{

ndctl/lib/papr.c line 22

Quoted 10 lines
ndctl/lib/papr.c new file +10 -0
+	/* Handle this separately to support monitor mode */
+	if (cmd == ND_CMD_SMART)
+		return true;
+
+	return !!(dimm->cmd_mask & (1ULL << cmd));
+}
+
+struct ndctl_dimm_ops * const papr_dimm_ops = &(struct ndctl_dimm_ops) {
+	.cmd_is_supported = papr_cmd_is_supported,
+};

_______________________________________________
Linux-nvdimm mailing list -- [email protected] To unsubscribe send an email to [email protected]

#6 lore raw

Re: [ndctl PATCH v6 2/5] libncdtl: Add initial support for NVDIMM_FAMILY_PAPR nvdimm family

Hi Vishal,

Thanks for reviewing this patch. I will be addressing your review comments in v7 of this patch series.

~ Vaibhav

Quoted 110 lines · Vishal L
Quoted 1 line · Vaibhav Jain

Add necessary scaffolding in libndctl for dimms that support papr_scm

support /the/ papr_scm specification

Quoted 2 lines

specification[1]. Since there can be platforms that support Open-Firmware[2] but not the papr_scm specification, hence the changes

s/hence//

Quoted 2 lines

proposed first add support for probing if the dimm bus supports Open-Firmware. This is done via querying for sysfs attribute 'of_node'

This is done /by/ querying for /the/ sysfs attribute

Quoted 3 lines

in dimm device sysfs directory. If available newly introduced member 'struct ndctl_bus.has_of_node' is set. During the probe of the dimm and execution of add_dimm(), the newly introduced add_papr_dimm()

During 'add_dimm()', the newly introduced..

Quoted 5 lines

is called if dimm bus reports supports Open-Firmware.

Function add_papr_dimm() queries the 'compatible' device tree attribute via newly introduced ndctl_bus_is_papr_scm() and based on its value assign NVDIMM_FAMILY_PAPR to the dimm command family. In

based on its value, assigns NVDIM_..

Quoted 1 line

future, based on the contents of 'compatible' attribute more of_pmem

In /the/ future

Quoted 3 lines

dimm families can be queried.

We also add support for parsing the dimm flags for

'We' can be ambiguous. Say something like: "Additionally, add support.."

Quoted 6 lines

NVDIMM_FAMILY_PAPR supporting nvdimms as described at [3]. A newly introduced function parse_papr_flags() reads the contents of this flag file and sets appropriate flag bits in 'struct ndctl_dimm.flags'.

Also we advertise support for monitor mode by allocating a file

"Advertise support for monitor mode.."

Quoted 6 lines

descriptor to the dimm 'flags' file and assigning it to 'struct ndctl_dimm.health_event_fd'.

The dimm-ops implementation for NVDIMM_FAMILY_PAPR is available in global variable 'papr_dimm_ops' which points to skeleton implementation in newly introduced file 'lib/papr.c'.

This paragraph can just be dropped - it's a minor implementation detail, and doesn't add much to the commit message. The same actually goes for the part above that talks about setting flags.

Quoted 9 lines

References:
[1] Documentation/powerpc/papr_hcalls.rst
https://lore.kernel.org/linux-nvdimm/[email protected]

[2] https://en.wikipedia.org/wiki/Open_Firmware

[3] Documentation/ABI/testing/sysfs-bus-papr-pmem
https://lore.kernel.org/linux-nvdimm/[email protected]

Not a huge deal, but the lore links can probably be updated to the latest posting.

Quoted 4 lines

---

[..]

ndctl/lib/papr.c line 1

Quoted 7 lines
ndctl/lib/papr.c new file +1 -0
new file mode 100644
index 000000000000..4b6ce8beccab
@@ -0,0 +1,22 @@
+// SPDX-License-Identifier: LGPL-2.1

I'm not sure if you intended to drop the copyright line here :)

Quoted 22 lines
ndctl/lib/papr.c new file +21 -0
+
+#include <stdint.h>
+#include <stdlib.h>
+#include <limits.h>
+#include <util/log.h>
+#include <ndctl.h>
+#include <ndctl/libndctl.h>
+#include <lib/private.h>
+
+static bool papr_cmd_is_supported(struct ndctl_dimm *dimm, int cmd)
+{
+	/* Handle this separately to support monitor mode */
+	if (cmd == ND_CMD_SMART)
+		return true;
+
+	return !!(dimm->cmd_mask & (1ULL << cmd));
+}
+
+struct ndctl_dimm_ops * const papr_dimm_ops = &(struct ndctl_dimm_ops) {
+	.cmd_is_supported = papr_cmd_is_supported,
+};
Signature · 5 lines
Cheers
~ Vaibhav
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
maintainer
#7 lore raw

Re: [ndctl PATCH v6 2/5] libncdtl: Add initial support for NVDIMM_FAMILY_PAPR nvdimm family

Quoted 7 lines · Vaibhav Jain

Hi Vishal,

Thanks for reviewing this patch. I will be addressing your review comments in v7 of this patch series.

~ Vaibhav

Thanks Vaibhav. The rest of the series looks good to me, so once you send the update, I'll queue it up for v69.

Thanks,
-Vishal
_______________________________________________
Linux-nvdimm mailing list -- [email protected] To unsubscribe send an email to [email protected]

#8 lore raw

PATCH v6 3/5 libndctl,papr_scm: Add definitions for PAPR nvdimm specific methods

In reply to Vaibhav Jain

Pull the kernel definition of PAPR nvdimm specific methods which is located in the proposed kernel tree patches at Ref[1] & [2].

Add a new header file 'papr.h' that introduces defines 'struct nd_pkg_papr' that holds an instance of 'struct nd_cmd_pkg' and 'struct nd_pkg_pdsm' together. Also add an instance of 'struct nd_pkg_pdsm' to 'struct ndctl_cmd' named as 'papr'.

References:
[1] "powerpc/papr_scm: Add support for reporting nvdimm health"
https://lore.kernel.org/linux-nvdimm/[email protected]/

[2] "ndctl/papr_scm,uapi: Add support for PAPR nvdimm specific methods"
https://lore.kernel.org/linux-nvdimm/[email protected]/

---
Changelog:

v5..v6:
* Updated 'papr_pdsm.h' to recent kernel-uapi version.
* Introduced a new header 'papr.h' and 'struct nd_pkg_papr'.
* Updated patch description.

v4..v5:
* Renamed 'papr_scm_pdsm.h' to 'papr_pdsm.h'
* Updated 'papr_pdsm.h' to recent kernel-uapi version.

v3..v4:
* Updated the definition of 'struct nd_pdsm_cmd_pkg' and pdsm_cmd_to_payload() to remove 'payload_offset' field. [ Aneesh ]

v2..v3:
* Added instance of 'struct nd_pdsm_cmd_pkg' to 'struct ndctl_cmd'
* Updated commit description.
* Updated the papr_scm_pdsm.h header to recently proposed kernel version.

v1..v2:
* Switched from obsolete papr_scm_dsm.h to recent papr_scm_pdsm.h that describes the ND_CMD_CALL interface between libndctl and papr_scm module.

---
 ndctl/lib/papr.c      |   1 +
 ndctl/lib/papr.h      |  15 +++++
 ndctl/lib/papr_pdsm.h | 132 ++++++++++++++++++++++++++++++++++++++++++
 ndctl/lib/private.h   |   2 +
 4 files changed, 150 insertions(+)
 create mode 100644 ndctl/lib/papr.h
 create mode 100644 ndctl/lib/papr_pdsm.h
ndctl/lib/papr.c +1 -0
index 4b6ce8beccab..37b4c4c78709 100644
@@ -7,6 +7,7 @@
 #include <ndctl.h>
 #include <ndctl/libndctl.h>
 #include <lib/private.h>
+#include "papr.h"
 
 static bool papr_cmd_is_supported(struct ndctl_dimm *dimm, int cmd)
 {
ndctl/lib/papr.h new file +15 -0
new file mode 100644
index 000000000000..77579396a7bd
@@ -0,0 +1,15 @@
+/* SPDX-License-Identifier: LGPL-2.1 */
+/* (C) Copyright IBM 2020 */
+
+#ifndef __PAPR_H__
+#define __PAPR_H__
+
+#include <papr_pdsm.h>
+
+/* Wraps a nd_cmd generic header with pdsm header */
+struct nd_pkg_papr {
+	struct nd_cmd_pkg gen;
+	struct nd_pkg_pdsm pdsm;
+};
+
+#endif /* __PAPR_H__ */
ndctl/lib/papr_pdsm.h new file +132 -0

132 lines not on this page · raw message

ndctl/lib/private.h +2 -0
index d90236b1f98b..c3d5fd7ac9a9 100644
@@ -32,6 +32,7 @@
 #include "hpe1.h"
 #include "msft.h"
 #include "hyperv.h"
+#include "papr.h"
 #include "libndctl-nfit.h"
 
 struct nvdimm_data {
@@ -277,6 +278,7 @@ struct ndctl_cmd {
 		struct ndn_pkg_msft msft[0];
 		struct nd_pkg_hyperv hyperv[0];
 		struct nd_pkg_intel intel[0];
+		struct nd_pkg_papr papr[0];
 		struct nd_cmd_get_config_size get_size[0];
 		struct nd_cmd_get_config_data_hdr get_data[0];
 		struct nd_cmd_set_config_hdr set_data[0];
Signature · 4 lines
2.26.2
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
#9 lore raw

PATCH v6 4/5 papr: Add scaffolding to issue and handle PDSM requests

In reply to Vaibhav Jain

This patch implement necessary infrastructure inside 'papr.c' to issue and handle PDSM requests. Changes implemented are:

* New helper function allocate_cmd() to allocate command packages for a specific PDSM command and payload size.

* Provide a skeleton implementation of 'dimm_ops->smart_get_flags' to check the submitted command for any errors. Presently no flags are returned from this callback and future patch will update this callback function to check submitted ndctl_cmd instance and return appropriate flags.

* cmd_is_valid() to validate submitted 'struct ndctl_cmd' instance for correct command family and PDSM command id.

* Implement callbacks for 'dimm_ops->get_firmware_status' and 'dimm_ops->get_xlat_firmware_status' to return translated firmware command status from 'struct ndctl_cmd'.

* Logging helpers for papr_scm that use the underlying libndctl provided logging.

---
Changelog:
v5..v6:
* Reworked the patch based on recent changes introduced for layout changes to 'struct nd_cmd_pkg'.
* Removed function update_dimm_stats() and papr_dimm_init/uninit() as they depended on an now a defunct patch "libndctl: Introduce new dimm-ops dimm_init() & dimm_uninit()"
* Introduced various helper macros to_pdsm(), to_ndcmd(), to_payload() and to_pdsm_cmd().

v4..v5:
* Updated papr_dbg/err macros to remove the 'scm' suffix.
* s/NVDIMM_FAMILY_PAPR_SCM/NVDIMM_FAMILY_PAPR/g
* s/PAPR_SCM_PDSM_*/PAPR_PDSM_*/g
* Minor changes to various comments and patch-description to remove usage of term 'scm'.

v3..v4:
* Remove the initialization of 'nd_pdsm_cmd_pkg.payload_offset' field from allocate_cmd(). [ Aneesh ]

v2..v3:
* None

v1..v2:
* Added new dimm callback 'papr_get_firmware_status'
* Switched to new papr_scm interface as described by papr_scm_dsm.h
* Changed the case of logging functions [ Santosh Sivaraj ]
* Removed redundant logging functions.
* Minor updates to patch description to s/DSM/PDSM/

---
 ndctl/lib/papr.c | 124 +++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 124 insertions(+)
ndctl/lib/papr.c +124 -0
index 37b4c4c78709..aff294823e41 100644
@@ -9,6 +9,29 @@
 #include <lib/private.h>
 #include "papr.h"
 
+/* Utility logging maros for simplify logging */
+#define papr_dbg(_dimm, _format_str, ...) dbg(_dimm->bus->ctx,		\
+					      "%s:" _format_str,	\
+					      ndctl_dimm_get_devname(_dimm), \
+					      ##__VA_ARGS__)
+
+#define papr_err(_dimm, _format_str, ...) err(_dimm->bus->ctx,		\
+					      "%s:" _format_str,	\
+					      ndctl_dimm_get_devname(_dimm), \
+					      ##__VA_ARGS__)
+
+/* Convert a ndctl_cmd to pdsm package */
+#define to_pdsm(C)  (&(C)->papr[0].pdsm)
+
+/* Convert a ndctl_cmd to nd_cmd_pkg */
+#define to_ndcmd(C)  (&(C)->papr[0].gen)
+
+/* Return payload from a ndctl_cmd */
+#define to_payload(C) (&(C)->papr[0].pdsm.payload)
+
+/* return the pdsm command */
+#define to_pdsm_cmd(C) ((enum papr_pdsm)to_ndcmd(C)->nd_command)
+
 static bool papr_cmd_is_supported(struct ndctl_dimm *dimm, int cmd)
 {
 	/* Handle this separately to support monitor mode */
@@ -18,6 +41,107 @@ static bool papr_cmd_is_supported(struct ndctl_dimm *dimm, int cmd)
 	return !!(dimm->cmd_mask & (1ULL << cmd));
 }
 
+static u32 papr_get_firmware_status(struct ndctl_cmd *cmd)
+{
+	const struct nd_pkg_pdsm *pcmd = to_pdsm(cmd);
+
+	return (u32) pcmd->cmd_status;
+}
+
+static int papr_xlat_firmware_status(struct ndctl_cmd *cmd)
+{
+	const struct nd_pkg_pdsm *pcmd = to_pdsm(cmd);
+
+	return pcmd->cmd_status;
+}
+
+/* Verify if the given command is supported and valid */
+static bool cmd_is_valid(struct ndctl_cmd *cmd)
+{
+	const struct nd_cmd_pkg  *ncmd = NULL;
+
+	if (cmd == NULL)
+		return false;
+
+	ncmd = to_ndcmd(cmd);
+
+	/* Verify the command family */
+	if (ncmd->nd_family != NVDIMM_FAMILY_PAPR) {
+		papr_err(cmd->dimm, "Invalid command family:0x%016llx\n",
+			 ncmd->nd_family);
+		return false;
+	}
+
+	/* Verify the PDSM */
+	if (ncmd->nd_command <= PAPR_PDSM_MIN ||
+	    ncmd->nd_command >= PAPR_PDSM_MAX) {
+		papr_err(cmd->dimm, "Invalid command :0x%016llx\n",
+			 ncmd->nd_command);
+		return false;
+	}
+
+	return true;
+}
+
+/* Allocate a struct ndctl_cmd for given pdsm request with payload size */
+static struct ndctl_cmd *allocate_cmd(struct ndctl_dimm *dimm,
+				      enum papr_pdsm pdsm_cmd,
+				      size_t payload_size)
+{
+	struct ndctl_cmd *cmd;
+
+	/* Verify that payload size is within acceptable range */
+	if (payload_size > ND_PDSM_PAYLOAD_MAX_SIZE) {
+		papr_err(dimm, "Requested payload size too large %lu bytes\n",
+			 payload_size);
+		return NULL;
+	}
+
+	cmd = calloc(1, sizeof(struct ndctl_cmd) + sizeof(struct nd_pkg_papr));
+	if (!cmd)
+		return NULL;
+
+	ndctl_cmd_ref(cmd);
+	cmd->dimm = dimm;
+	cmd->type = ND_CMD_CALL;
+	cmd->status = 0;
+	cmd->get_firmware_status = &papr_get_firmware_status;
+
+	/* Populate the nd_cmd_pkg contained in nd_pkg_pdsm */
+	*to_ndcmd(cmd) =  (struct nd_cmd_pkg) {
+		.nd_family = NVDIMM_FAMILY_PAPR,
+		.nd_command = pdsm_cmd,
+		.nd_size_in = 0,
+		.nd_size_out = ND_PDSM_HDR_SIZE + payload_size,
+		.nd_fw_size = 0,
+	};
+	return cmd;
+}
+
+/* Validate the ndctl_cmd and return applicable flags */
+static unsigned int papr_smart_get_flags(struct ndctl_cmd *cmd)
+{
+	struct nd_pkg_pdsm *pcmd;
+
+	if (!cmd_is_valid(cmd))
+		return 0;
+
+	pcmd = to_pdsm(cmd);
+	/* If error reported then return empty flags */
+	if (pcmd->cmd_status) {
+		papr_err(cmd->dimm, "PDSM(0x%x) reported error:%d\n",
+			 to_pdsm_cmd(cmd), pcmd->cmd_status);
+		return 0;
+	}
+
+	/* return empty flags for now */
+	return 0;
+}
+
+
 struct ndctl_dimm_ops * const papr_dimm_ops = &(struct ndctl_dimm_ops) {
 	.cmd_is_supported = papr_cmd_is_supported,
+	.smart_get_flags = papr_smart_get_flags,
+	.get_firmware_status =  papr_get_firmware_status,
+	.xlat_firmware_status = papr_xlat_firmware_status,
 };
Signature · 4 lines
2.26.2
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]
#10 lore raw

PATCH v6 5/5 libndctl,papr_scm: Implement support for PAPR_PDSM_HEALTH

In reply to Vaibhav Jain

Add support for reporting DIMM health and shutdown state by issuing PAPR_PDSM_HEALTH request to papr_scm module. It returns an instance of 'struct nd_papr_pdsm_health' as defined in 'papr_pdsm.h'. The patch provides support for dimm-ops 'new_smart', 'smart_get_health' & 'smart_get_shutdown_state' as newly introduced functions papr_new_smart_health(), papr_smart_get_health() & papr_smart_get_shutdown_state() respectively. These callbacks should enable ndctl to report DIMM health.

---
Changelog:
v5..v6:
* Updated patch description to reflect removal of update_dimm_health().
* Update papr_smart_get_flags() to parse the 'struct ndctl_cmd' instance and return appropriate ND_SMART_XXX flags.
* Callbacks papr_smart_get_health() , papr_smart_get_shutdown_state() callbacks to use 'struct ndctl_cmd' instead of 'dimm_priv'.
* Added update_dimm_flags() to update 'struct ndctl_dimm' with the flags returned in 'struct nd_papr_psdm_health'.

v4..v5:
* Updated patch description to reflect updated names of struct and defines that have the term 'scm' removed.

v3..v4:
* None

v2..v3:
* None

v1..v2:
* Squashed patch to report nvdimm bad shutdown state with this patch.
* Switched to new structs/enums as defined in papr_scm_pdsm.h

---
 ndctl/lib/papr.c | 73 +++++++++++++++++++++++++++++++++++++++++++++++-
 1 file changed, 72 insertions(+), 1 deletion(-)
ndctl/lib/papr.c +72 -1
index aff294823e41..e1acc3c44295 100644
@@ -118,10 +118,23 @@ static struct ndctl_cmd *allocate_cmd(struct ndctl_dimm *dimm,
 	return cmd;
 }
 
+/* Parse the nd_papr_pdsm_health and update dimm flags */
+static int update_dimm_flags(struct ndctl_dimm *dimm, struct nd_papr_pdsm_health *health)
+{
+	/* Update the dimm flags */
+	dimm->flags.f_arm = health->dimm_unarmed;
+	dimm->flags.f_flush = health->dimm_bad_shutdown;
+	dimm->flags.f_restore = health->dimm_bad_restore;
+	dimm->flags.f_smart = (health->dimm_health != 0);
+
+	return 0;
+}
+
 /* Validate the ndctl_cmd and return applicable flags */
 static unsigned int papr_smart_get_flags(struct ndctl_cmd *cmd)
 {
 	struct nd_pkg_pdsm *pcmd;
+	struct nd_papr_pdsm_health health;
 
 	if (!cmd_is_valid(cmd))
 		return 0;
@@ -134,14 +147,72 @@ static unsigned int papr_smart_get_flags(struct ndctl_cmd *cmd)
 		return 0;
 	}
 
-	/* return empty flags for now */
+	/*
+	 * In case of nvdimm health PDSM, update dimm flags
+	 * and  return possible flags.
+	 */
+	if (to_pdsm_cmd(cmd) == PAPR_PDSM_HEALTH) {
+		health = pcmd->payload.health;
+		update_dimm_flags(cmd->dimm, &health);
+		return ND_SMART_HEALTH_VALID | ND_SMART_SHUTDOWN_VALID;
+	}
+
+	/* Else return empty flags */
 	return 0;
 }
 
+static struct ndctl_cmd *papr_new_smart_health(struct ndctl_dimm *dimm)
+{
+	struct ndctl_cmd *cmd;
+
+	cmd = allocate_cmd(dimm, PAPR_PDSM_HEALTH,
+			       sizeof(struct nd_papr_pdsm_health));
+	if (!cmd)
+		papr_err(dimm, "Unable to allocate smart_health command\n");
+
+	return cmd;
+}
+
+static unsigned int papr_smart_get_health(struct ndctl_cmd *cmd)
+{
+	struct nd_papr_pdsm_health health;
+
+	/* Ignore in case of error or invalid pdsm */
+	if (!cmd_is_valid(cmd) ||
+	    to_pdsm(cmd)->cmd_status != 0 ||
+	    to_pdsm_cmd(cmd) != PAPR_PDSM_HEALTH)
+		return 0;
+
+	/* get the payload from command */
+	health = to_payload(cmd)->health;
+
+	/* Use some math to return one of defined ND_SMART_*_HEALTH values */
+	return  !health.dimm_health ? 0 : 1 << (health.dimm_health - 1);
+}
+
+static unsigned int papr_smart_get_shutdown_state(struct ndctl_cmd *cmd)
+{
+	struct nd_papr_pdsm_health health;
+
+	/* Ignore in case of error or invalid pdsm */
+	if (!cmd_is_valid(cmd) ||
+	    to_pdsm(cmd)->cmd_status != 0 ||
+	    to_pdsm_cmd(cmd) != PAPR_PDSM_HEALTH)
+		return 0;
+
+	/* get the payload from command */
+	health = to_payload(cmd)->health;
+
+	/* return the bad shutdown flag returned from papr_scm */
+	return health.dimm_bad_shutdown;
+}
 
 struct ndctl_dimm_ops * const papr_dimm_ops = &(struct ndctl_dimm_ops) {
 	.cmd_is_supported = papr_cmd_is_supported,
 	.smart_get_flags = papr_smart_get_flags,
 	.get_firmware_status =  papr_get_firmware_status,
 	.xlat_firmware_status = papr_xlat_firmware_status,
+	.new_smart = papr_new_smart_health,
+	.smart_get_health = papr_smart_get_health,
+	.smart_get_shutdown_state = papr_smart_get_shutdown_state,
 };
Signature · 4 lines
2.26.2
_______________________________________________
Linux-nvdimm mailing list -- [email protected]
To unsubscribe send an email to [email protected]