| Message ID | 20210507113948.38950-1-vaibhav@linux.ibm.com (mailing list archive) |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v2] powerpc/papr_scm: Make 'perf_stats' invisible if perf-stats unavailable | expand |
On Fri, May 7, 2021 at 4:40 AM Vaibhav Jain <vaibhav@linux.ibm.com> wrote: > > In case performance stats for an nvdimm are not available, reading the > 'perf_stats' sysfs file returns an -ENOENT error. A better approach is > to make the 'perf_stats' file entirely invisible to indicate that > performance stats for an nvdimm are unavailable. > > So this patch updates 'papr_nd_attribute_group' to add a 'is_visible' > callback implemented as newly introduced 'papr_nd_attribute_visible()' > that returns an appropriate mode in case performance stats aren't > supported in a given nvdimm. > > Also the initialization of 'papr_scm_priv.stat_buffer_len' is moved > from papr_scm_nvdimm_init() to papr_scm_probe() so that it value is > available when 'papr_nd_attribute_visible()' is called during nvdimm > initialization. > > Fixes: 2d02bf835e57('powerpc/papr_scm: Fetch nvdimm performance stats from PHYP') Since this has been the exposed ABI since v5.9 perhaps a note / analysis is needed here that the disappearance of this file will not break any existing scripts/tools. > Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com> > --- > Changelog: > > v2: > * Removed a redundant dev_info() from pap_scm_nvdimm_init() [ Aneesh ] > * Marked papr_nd_attribute_visible() as static which also fixed the > build warning reported by kernel build robot > --- > arch/powerpc/platforms/pseries/papr_scm.c | 35 ++++++++++++++++------- > 1 file changed, 24 insertions(+), 11 deletions(-) > > diff --git a/arch/powerpc/platforms/pseries/papr_scm.c b/arch/powerpc/platforms/pseries/papr_scm.c > index e2b69cc3beaf..11e7b90a3360 100644 > --- a/arch/powerpc/platforms/pseries/papr_scm.c > +++ b/arch/powerpc/platforms/pseries/papr_scm.c > @@ -907,6 +907,20 @@ static ssize_t flags_show(struct device *dev, > } > DEVICE_ATTR_RO(flags); > > +static umode_t papr_nd_attribute_visible(struct kobject *kobj, > + struct attribute *attr, int n) > +{ > + struct device *dev = container_of(kobj, typeof(*dev), kobj); This can use the kobj_to_dev() helper. > + struct nvdimm *nvdimm = to_nvdimm(dev); > + struct papr_scm_priv *p = nvdimm_provider_data(nvdimm); > + > + /* For if perf-stats not available remove perf_stats sysfs */ > + if (attr == &dev_attr_perf_stats.attr && p->stat_buffer_len == 0) > + return 0; > + > + return attr->mode; > +} > + > /* papr_scm specific dimm attributes */ > static struct attribute *papr_nd_attributes[] = { > &dev_attr_flags.attr, > @@ -916,6 +930,7 @@ static struct attribute *papr_nd_attributes[] = { > > static struct attribute_group papr_nd_attribute_group = { > .name = "papr", > + .is_visible = papr_nd_attribute_visible, > .attrs = papr_nd_attributes, > }; > > @@ -931,7 +946,6 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p) > struct nd_region_desc ndr_desc; > unsigned long dimm_flags; > int target_nid, online_nid; > - ssize_t stat_size; > > p->bus_desc.ndctl = papr_scm_ndctl; > p->bus_desc.module = THIS_MODULE; > @@ -1016,16 +1030,6 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p) > list_add_tail(&p->region_list, &papr_nd_regions); > mutex_unlock(&papr_ndr_lock); > > - /* Try retriving the stat buffer and see if its supported */ > - stat_size = drc_pmem_query_stats(p, NULL, 0); > - if (stat_size > 0) { > - p->stat_buffer_len = stat_size; > - dev_dbg(&p->pdev->dev, "Max perf-stat size %lu-bytes\n", > - p->stat_buffer_len); > - } else { > - dev_info(&p->pdev->dev, "Dimm performance stats unavailable\n"); > - } > - > return 0; > > err: nvdimm_bus_unregister(p->bus); > @@ -1102,6 +1106,7 @@ static int papr_scm_probe(struct platform_device *pdev) > u64 blocks, block_size; > struct papr_scm_priv *p; > const char *uuid_str; > + ssize_t stat_size; > u64 uuid[2]; > int rc; > > @@ -1179,6 +1184,14 @@ static int papr_scm_probe(struct platform_device *pdev) > p->res.name = pdev->name; > p->res.flags = IORESOURCE_MEM; > > + /* Try retriving the stat buffer and see if its supported */ s/retriving/retrieving/ > + stat_size = drc_pmem_query_stats(p, NULL, 0); > + if (stat_size > 0) { > + p->stat_buffer_len = stat_size; > + dev_dbg(&p->pdev->dev, "Max perf-stat size %lu-bytes\n", > + p->stat_buffer_len); > + } > + > rc = papr_scm_nvdimm_init(p); > if (rc) > goto err2; > -- > 2.31.1 > After the minor fixups above you can add: Reviewed-by: Dan Williams <dan.j.williams@intel.com> ...I assume this will go through the PPC tree.
Thanks for looking into this patch Dan, Dan Williams <dan.j.williams@intel.com> writes: > On Fri, May 7, 2021 at 4:40 AM Vaibhav Jain <vaibhav@linux.ibm.com> wrote: >> >> In case performance stats for an nvdimm are not available, reading the >> 'perf_stats' sysfs file returns an -ENOENT error. A better approach is >> to make the 'perf_stats' file entirely invisible to indicate that >> performance stats for an nvdimm are unavailable. >> >> So this patch updates 'papr_nd_attribute_group' to add a 'is_visible' >> callback implemented as newly introduced 'papr_nd_attribute_visible()' >> that returns an appropriate mode in case performance stats aren't >> supported in a given nvdimm. >> >> Also the initialization of 'papr_scm_priv.stat_buffer_len' is moved >> from papr_scm_nvdimm_init() to papr_scm_probe() so that it value is >> available when 'papr_nd_attribute_visible()' is called during nvdimm >> initialization. >> >> Fixes: 2d02bf835e57('powerpc/papr_scm: Fetch nvdimm performance stats from PHYP') > > Since this has been the exposed ABI since v5.9 perhaps a note / > analysis is needed here that the disappearance of this file will not > break any existing scripts/tools. > Yes, I have added a note in patch description and also updated Documentation/ABI/testing/sysfs-bus-papr-pmem with a note on when this sysfs attr would be available. Thanks for this suggestion. >> Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com> >> --- >> Changelog: >> >> v2: >> * Removed a redundant dev_info() from pap_scm_nvdimm_init() [ Aneesh ] >> * Marked papr_nd_attribute_visible() as static which also fixed the >> build warning reported by kernel build robot >> --- >> arch/powerpc/platforms/pseries/papr_scm.c | 35 ++++++++++++++++------- >> 1 file changed, 24 insertions(+), 11 deletions(-) >> >> diff --git a/arch/powerpc/platforms/pseries/papr_scm.c b/arch/powerpc/platforms/pseries/papr_scm.c >> index e2b69cc3beaf..11e7b90a3360 100644 >> --- a/arch/powerpc/platforms/pseries/papr_scm.c >> +++ b/arch/powerpc/platforms/pseries/papr_scm.c >> @@ -907,6 +907,20 @@ static ssize_t flags_show(struct device *dev, >> } >> DEVICE_ATTR_RO(flags); >> >> +static umode_t papr_nd_attribute_visible(struct kobject *kobj, >> + struct attribute *attr, int n) >> +{ >> + struct device *dev = container_of(kobj, typeof(*dev), kobj); > > This can use the kobj_to_dev() helper. > Fixed in v3 >> + struct nvdimm *nvdimm = to_nvdimm(dev); >> + struct papr_scm_priv *p = nvdimm_provider_data(nvdimm); >> + >> + /* For if perf-stats not available remove perf_stats sysfs */ >> + if (attr == &dev_attr_perf_stats.attr && p->stat_buffer_len == 0) >> + return 0; >> + >> + return attr->mode; >> +} >> + >> /* papr_scm specific dimm attributes */ >> static struct attribute *papr_nd_attributes[] = { >> &dev_attr_flags.attr, >> @@ -916,6 +930,7 @@ static struct attribute *papr_nd_attributes[] = { >> >> static struct attribute_group papr_nd_attribute_group = { >> .name = "papr", >> + .is_visible = papr_nd_attribute_visible, >> .attrs = papr_nd_attributes, >> }; >> >> @@ -931,7 +946,6 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p) >> struct nd_region_desc ndr_desc; >> unsigned long dimm_flags; >> int target_nid, online_nid; >> - ssize_t stat_size; >> >> p->bus_desc.ndctl = papr_scm_ndctl; >> p->bus_desc.module = THIS_MODULE; >> @@ -1016,16 +1030,6 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p) >> list_add_tail(&p->region_list, &papr_nd_regions); >> mutex_unlock(&papr_ndr_lock); >> >> - /* Try retriving the stat buffer and see if its supported */ >> - stat_size = drc_pmem_query_stats(p, NULL, 0); >> - if (stat_size > 0) { >> - p->stat_buffer_len = stat_size; >> - dev_dbg(&p->pdev->dev, "Max perf-stat size %lu-bytes\n", >> - p->stat_buffer_len); >> - } else { >> - dev_info(&p->pdev->dev, "Dimm performance stats unavailable\n"); >> - } >> - >> return 0; >> >> err: nvdimm_bus_unregister(p->bus); >> @@ -1102,6 +1106,7 @@ static int papr_scm_probe(struct platform_device *pdev) >> u64 blocks, block_size; >> struct papr_scm_priv *p; >> const char *uuid_str; >> + ssize_t stat_size; >> u64 uuid[2]; >> int rc; >> >> @@ -1179,6 +1184,14 @@ static int papr_scm_probe(struct platform_device *pdev) >> p->res.name = pdev->name; >> p->res.flags = IORESOURCE_MEM; >> >> + /* Try retriving the stat buffer and see if its supported */ > > s/retriving/retrieving/ > Fixed in v3. >> + stat_size = drc_pmem_query_stats(p, NULL, 0); >> + if (stat_size > 0) { >> + p->stat_buffer_len = stat_size; >> + dev_dbg(&p->pdev->dev, "Max perf-stat size %lu-bytes\n", >> + p->stat_buffer_len); >> + } >> + >> rc = papr_scm_nvdimm_init(p); >> if (rc) >> goto err2; >> -- >> 2.31.1 >> > > After the minor fixups above you can add: > > Reviewed-by: Dan Williams <dan.j.williams@intel.com> Thanks > > ...I assume this will go through the PPC tree. Yes
diff --git a/arch/powerpc/platforms/pseries/papr_scm.c b/arch/powerpc/platforms/pseries/papr_scm.c index e2b69cc3beaf..11e7b90a3360 100644 --- a/arch/powerpc/platforms/pseries/papr_scm.c +++ b/arch/powerpc/platforms/pseries/papr_scm.c @@ -907,6 +907,20 @@ static ssize_t flags_show(struct device *dev, } DEVICE_ATTR_RO(flags); +static umode_t papr_nd_attribute_visible(struct kobject *kobj, + struct attribute *attr, int n) +{ + struct device *dev = container_of(kobj, typeof(*dev), kobj); + struct nvdimm *nvdimm = to_nvdimm(dev); + struct papr_scm_priv *p = nvdimm_provider_data(nvdimm); + + /* For if perf-stats not available remove perf_stats sysfs */ + if (attr == &dev_attr_perf_stats.attr && p->stat_buffer_len == 0) + return 0; + + return attr->mode; +} + /* papr_scm specific dimm attributes */ static struct attribute *papr_nd_attributes[] = { &dev_attr_flags.attr, @@ -916,6 +930,7 @@ static struct attribute *papr_nd_attributes[] = { static struct attribute_group papr_nd_attribute_group = { .name = "papr", + .is_visible = papr_nd_attribute_visible, .attrs = papr_nd_attributes, }; @@ -931,7 +946,6 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p) struct nd_region_desc ndr_desc; unsigned long dimm_flags; int target_nid, online_nid; - ssize_t stat_size; p->bus_desc.ndctl = papr_scm_ndctl; p->bus_desc.module = THIS_MODULE; @@ -1016,16 +1030,6 @@ static int papr_scm_nvdimm_init(struct papr_scm_priv *p) list_add_tail(&p->region_list, &papr_nd_regions); mutex_unlock(&papr_ndr_lock); - /* Try retriving the stat buffer and see if its supported */ - stat_size = drc_pmem_query_stats(p, NULL, 0); - if (stat_size > 0) { - p->stat_buffer_len = stat_size; - dev_dbg(&p->pdev->dev, "Max perf-stat size %lu-bytes\n", - p->stat_buffer_len); - } else { - dev_info(&p->pdev->dev, "Dimm performance stats unavailable\n"); - } - return 0; err: nvdimm_bus_unregister(p->bus); @@ -1102,6 +1106,7 @@ static int papr_scm_probe(struct platform_device *pdev) u64 blocks, block_size; struct papr_scm_priv *p; const char *uuid_str; + ssize_t stat_size; u64 uuid[2]; int rc; @@ -1179,6 +1184,14 @@ static int papr_scm_probe(struct platform_device *pdev) p->res.name = pdev->name; p->res.flags = IORESOURCE_MEM; + /* Try retriving the stat buffer and see if its supported */ + stat_size = drc_pmem_query_stats(p, NULL, 0); + if (stat_size > 0) { + p->stat_buffer_len = stat_size; + dev_dbg(&p->pdev->dev, "Max perf-stat size %lu-bytes\n", + p->stat_buffer_len); + } + rc = papr_scm_nvdimm_init(p); if (rc) goto err2;
In case performance stats for an nvdimm are not available, reading the 'perf_stats' sysfs file returns an -ENOENT error. A better approach is to make the 'perf_stats' file entirely invisible to indicate that performance stats for an nvdimm are unavailable. So this patch updates 'papr_nd_attribute_group' to add a 'is_visible' callback implemented as newly introduced 'papr_nd_attribute_visible()' that returns an appropriate mode in case performance stats aren't supported in a given nvdimm. Also the initialization of 'papr_scm_priv.stat_buffer_len' is moved from papr_scm_nvdimm_init() to papr_scm_probe() so that it value is available when 'papr_nd_attribute_visible()' is called during nvdimm initialization. Fixes: 2d02bf835e57('powerpc/papr_scm: Fetch nvdimm performance stats from PHYP') Signed-off-by: Vaibhav Jain <vaibhav@linux.ibm.com> --- Changelog: v2: * Removed a redundant dev_info() from pap_scm_nvdimm_init() [ Aneesh ] * Marked papr_nd_attribute_visible() as static which also fixed the build warning reported by kernel build robot --- arch/powerpc/platforms/pseries/papr_scm.c | 35 ++++++++++++++++------- 1 file changed, 24 insertions(+), 11 deletions(-)