This patch will find the max contiguous area to determine the largest namespace size that can be created. If the requested size exceeds the largest available, ENOSPC error will be returned.
This fixes the allocation underrun error and wrong error return code that have otherwise been observed as the following kernel warning:
WARNING: CPU: <CPU> PID: <PID> at drivers/nvdimm/namespace_devs.c:913 size_store
Fixes: a1f3e4d6a0c3 ("libnvdimm, region: update nd_region_available_dpa() for multi-pmem support") Cc: stable@vger.kernel.org Signed-off-by: Keith Busch keith.busch@intel.com --- v1 -> v2:
Updated changelog to indicate the warning this patch fixes and copy stable.
Fixed code comments with the correct name of a function
drivers/nvdimm/dimm_devs.c | 29 +++++++++++++++++++++++++++++ drivers/nvdimm/namespace_devs.c | 2 +- drivers/nvdimm/nd-core.h | 3 +++ drivers/nvdimm/region_devs.c | 23 +++++++++++++++++++++++ 4 files changed, 56 insertions(+), 1 deletion(-)
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..791a93cd8eb1 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,35 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/** + * nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max + * contiguous unallocated dpa range. + * @nd_region: constrain available space check to this reference region + * @nd_mapping: container of dpa-resource-root + labels + */ +resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region, + struct nd_mapping *nd_mapping) +{ + struct nvdimm_drvdata *ndd = to_ndd(nd_mapping); + resource_size_t start, end, max = 0; + struct resource *res; + + if (!ndd) + return 0; + + start = nd_mapping->start; + end = start + nd_mapping->size; + + for_each_dpa_resource(ndd, res) { + if ((res->start - start) > max) + max = res->start - start; + start = res->start + resource_size(res); + } + if (start < end && end - start > max) + max = end - start; + return max; +} + /** * nd_pmem_available_dpa - for the given dimm+region account unallocated dpa * @nd_mapping: container of dpa-resource-root + labels diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c index 28afdd668905..c00d1d2d48f9 100644 --- a/drivers/nvdimm/namespace_devs.c +++ b/drivers/nvdimm/namespace_devs.c @@ -1032,7 +1032,7 @@ static ssize_t __size_store(struct device *dev, unsigned long long val)
allocated += nvdimm_allocated_dpa(ndd, &label_id); } - available = nd_region_available_dpa(nd_region); + available = nd_region_contiguous_max(nd_region);
if (val > available + allocated) return -ENOSPC; diff --git a/drivers/nvdimm/nd-core.h b/drivers/nvdimm/nd-core.h index 79274ead54fb..8e4acd075b0f 100644 --- a/drivers/nvdimm/nd-core.h +++ b/drivers/nvdimm/nd-core.h @@ -100,6 +100,9 @@ struct nd_region; struct nvdimm_drvdata; struct nd_mapping; void nd_mapping_free_labels(struct nd_mapping *nd_mapping); +resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region, + struct nd_mapping *nd_mapping); +resource_size_t nd_region_contiguous_max(struct nd_region *nd_region); resource_size_t nd_pmem_available_dpa(struct nd_region *nd_region, struct nd_mapping *nd_mapping, resource_size_t *overlap); resource_size_t nd_blk_available_dpa(struct nd_region *nd_region); diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c index ec3543b83330..8af483d7ef57 100644 --- a/drivers/nvdimm/region_devs.c +++ b/drivers/nvdimm/region_devs.c @@ -357,6 +357,29 @@ static ssize_t set_cookie_show(struct device *dev, } static DEVICE_ATTR_RO(set_cookie);
+resource_size_t nd_region_contiguous_max(struct nd_region *nd_region) +{ + resource_size_t available = 0; + int i; + + WARN_ON(!is_nvdimm_bus_locked(&nd_region->dev)); + for (i = 0; i < nd_region->ndr_mappings; i++) { + struct nd_mapping *nd_mapping = &nd_region->mapping[i]; + struct nvdimm_drvdata *ndd = to_ndd(nd_mapping); + + /* if a dimm is disabled the available capacity is zero */ + if (!ndd) + return 0; + + if (is_memory(&nd_region->dev)) + available += nd_pmem_max_contiguous_dpa(nd_region, + nd_mapping); + else if (is_nd_blk(&nd_region->dev)) + available += nd_blk_available_dpa(nd_region); + } + return available; +} + resource_size_t nd_region_available_dpa(struct nd_region *nd_region) { resource_size_t blk_max_overlap = 0, available, overlap;
On Thu, Jul 5, 2018 at 1:17 PM, Keith Busch keith.busch@intel.com wrote:
This patch will find the max contiguous area to determine the largest namespace size that can be created. If the requested size exceeds the largest available, ENOSPC error will be returned.
This fixes the allocation underrun error and wrong error return code that have otherwise been observed as the following kernel warning:
WARNING: CPU: <CPU> PID: <PID> at drivers/nvdimm/namespace_devs.c:913 size_store
Fixes: a1f3e4d6a0c3 ("libnvdimm, region: update nd_region_available_dpa() for multi-pmem support") Cc: stable@vger.kernel.org Signed-off-by: Keith Busch keith.busch@intel.com
v1 -> v2:
Updated changelog to indicate the warning this patch fixes and copy stable.
Fixed code comments with the correct name of a function
Looks good to me, applied to my for-4.18/libnvdimm tracking branch.
On Thu, Jul 5, 2018 at 2:13 PM, Dan Williams dan.j.williams@intel.com wrote:
On Thu, Jul 5, 2018 at 1:17 PM, Keith Busch keith.busch@intel.com wrote:
This patch will find the max contiguous area to determine the largest namespace size that can be created. If the requested size exceeds the largest available, ENOSPC error will be returned.
This fixes the allocation underrun error and wrong error return code that have otherwise been observed as the following kernel warning:
WARNING: CPU: <CPU> PID: <PID> at drivers/nvdimm/namespace_devs.c:913 size_store
Fixes: a1f3e4d6a0c3 ("libnvdimm, region: update nd_region_available_dpa() for multi-pmem support") Cc: stable@vger.kernel.org Signed-off-by: Keith Busch keith.busch@intel.com
v1 -> v2:
Updated changelog to indicate the warning this patch fixes and copy stable.
Fixed code comments with the correct name of a function
Looks good to me, applied to my for-4.18/libnvdimm tracking branch.
I take it back, this is causing the libndctl unit test to fail. debugging...
On Thu, Jul 5, 2018 at 1:17 PM, Keith Busch keith.busch@intel.com wrote:
This patch will find the max contiguous area to determine the largest namespace size that can be created. If the requested size exceeds the largest available, ENOSPC error will be returned.
This fixes the allocation underrun error and wrong error return code that have otherwise been observed as the following kernel warning:
WARNING: CPU: <CPU> PID: <PID> at drivers/nvdimm/namespace_devs.c:913 size_store
Fixes: a1f3e4d6a0c3 ("libnvdimm, region: update nd_region_available_dpa() for multi-pmem support") Cc: stable@vger.kernel.org Signed-off-by: Keith Busch keith.busch@intel.com
v1 -> v2:
Updated changelog to indicate the warning this patch fixes and copy stable.
Fixed code comments with the correct name of a function
drivers/nvdimm/dimm_devs.c | 29 +++++++++++++++++++++++++++++ drivers/nvdimm/namespace_devs.c | 2 +- drivers/nvdimm/nd-core.h | 3 +++ drivers/nvdimm/region_devs.c | 23 +++++++++++++++++++++++ 4 files changed, 56 insertions(+), 1 deletion(-)
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..791a93cd8eb1 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,35 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/** x + * nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max
contiguous unallocated dpa range.
- @nd_region: constrain available space check to this reference region
- @nd_mapping: container of dpa-resource-root + labels
- */
+resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region,
struct nd_mapping *nd_mapping)+{
struct nvdimm_drvdata *ndd = to_ndd(nd_mapping);resource_size_t start, end, max = 0;struct resource *res;if (!ndd)return 0;start = nd_mapping->start;end = start + nd_mapping->size;for_each_dpa_resource(ndd, res) {if ((res->start - start) > max)max = res->start - start;start = res->start + resource_size(res);}if (start < end && end - start > max)max = end - start;return max;+}
So, I liked how simple and straightforward this was, but the unit test reminded me how complicated the free space allocation could be given that we still might need to account for aliased BLK capacity.
A few things I now notice:
* It is possible for a dpa resource to start before nd_mapping->start and/or end after nd_mapping->size. See the diagram in tools/testing/nvdimm/test/nfit.c
* We need to be summing the max extents across all DIMMs in the set while also accounting for this BLK overlap problem that nd_pmem_available_dpa() handles.
The original bug arose from the fact that the nd_region_available_dpa() accounts for total busy capacity and the allocation path looks for extents. It would be nice to unify those two functions... or maybe standardize on scan_allocate() everywhere.
On Thu, Jul 5, 2018 at 6:25 PM, Dan Williams dan.j.williams@intel.com wrote:
On Thu, Jul 5, 2018 at 1:17 PM, Keith Busch keith.busch@intel.com wrote:
This patch will find the max contiguous area to determine the largest namespace size that can be created. If the requested size exceeds the largest available, ENOSPC error will be returned.
This fixes the allocation underrun error and wrong error return code that have otherwise been observed as the following kernel warning:
WARNING: CPU: <CPU> PID: <PID> at drivers/nvdimm/namespace_devs.c:913 size_store
Fixes: a1f3e4d6a0c3 ("libnvdimm, region: update nd_region_available_dpa() for multi-pmem support") Cc: stable@vger.kernel.org Signed-off-by: Keith Busch keith.busch@intel.com
v1 -> v2:
Updated changelog to indicate the warning this patch fixes and copy stable.
Fixed code comments with the correct name of a function
drivers/nvdimm/dimm_devs.c | 29 +++++++++++++++++++++++++++++ drivers/nvdimm/namespace_devs.c | 2 +- drivers/nvdimm/nd-core.h | 3 +++ drivers/nvdimm/region_devs.c | 23 +++++++++++++++++++++++ 4 files changed, 56 insertions(+), 1 deletion(-)
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..791a93cd8eb1 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,35 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/** x + * nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max
contiguous unallocated dpa range.
- @nd_region: constrain available space check to this reference region
- @nd_mapping: container of dpa-resource-root + labels
- */
+resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region,
struct nd_mapping *nd_mapping)+{
struct nvdimm_drvdata *ndd = to_ndd(nd_mapping);resource_size_t start, end, max = 0;struct resource *res;if (!ndd)return 0;start = nd_mapping->start;end = start + nd_mapping->size;for_each_dpa_resource(ndd, res) {if ((res->start - start) > max)max = res->start - start;start = res->start + resource_size(res);}if (start < end && end - start > max)max = end - start;return max;+}
So, I liked how simple and straightforward this was, but the unit test reminded me how complicated the free space allocation could be given that we still might need to account for aliased BLK capacity.
A few things I now notice:
- It is possible for a dpa resource to start before nd_mapping->start
and/or end after nd_mapping->size. See the diagram in tools/testing/nvdimm/test/nfit.c
- We need to be summing the max extents across all DIMMs in the set
while also accounting for this BLK overlap problem that nd_pmem_available_dpa() handles.
The original bug arose from the fact that the nd_region_available_dpa() accounts for total busy capacity and the allocation path looks for extents. It would be nice to unify those two functions... or maybe standardize on scan_allocate() everywhere.
I think you can redo this using reserve_free_pmem() and then find the largest free space reservation after accounting for blk_max_overlap as calculated by nd_region_available_dpa().
reserve_free_pmem() inserts resources into the dpa resource tree with the name "pmem-reserve" then you need to filter those against blk_max_overlap to find which ones might appear free, but shadowed by a BLK allocation on one of the DIMMs in the set. Yes, it allocates memory and is heavy handed, but I'm having a hard time thinking of something better at the moment.
On Thu, Jul 05, 2018 at 06:48:25PM -0700, Dan Williams wrote:
On Thu, Jul 5, 2018 at 6:25 PM, Dan Williams dan.j.williams@intel.com wrote:
So, I liked how simple and straightforward this was, but the unit test reminded me how complicated the free space allocation could be given that we still might need to account for aliased BLK capacity.
A few things I now notice:
- It is possible for a dpa resource to start before nd_mapping->start
and/or end after nd_mapping->size. See the diagram in tools/testing/nvdimm/test/nfit.c
- We need to be summing the max extents across all DIMMs in the set
while also accounting for this BLK overlap problem that nd_pmem_available_dpa() handles.
The original bug arose from the fact that the nd_region_available_dpa() accounts for total busy capacity and the allocation path looks for extents. It would be nice to unify those two functions... or maybe standardize on scan_allocate() everywhere.
I think you can redo this using reserve_free_pmem() and then find the largest free space reservation after accounting for blk_max_overlap as calculated by nd_region_available_dpa().
reserve_free_pmem() inserts resources into the dpa resource tree with the name "pmem-reserve" then you need to filter those against blk_max_overlap to find which ones might appear free, but shadowed by a BLK allocation on one of the DIMMs in the set. Yes, it allocates memory and is heavy handed, but I'm having a hard time thinking of something better at the moment.
Okay, I think I follow what you're saying, but let me just see if the following is what you had in mind before sending a v3:
--- diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..f30e0c3b0282 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,31 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/** + * nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max + * contiguous unallocated dpa range. + * @nd_region: constrain available space check to this reference region + * @nd_mapping: container of dpa-resource-root + labels + */ +resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region, + struct nd_mapping *nd_mapping) +{ + struct nvdimm_drvdata *ndd = to_ndd(nd_mapping); + resource_size_t max = 0; + struct resource *res; + + if (!ndd) + return 0; + + for_each_dpa_resource(ndd, res) { + if (strcmp(res->name, "pmem-reserve") != 0) + continue; + if (resource_size(res) > max) + max = resource_size(res); + } + return max; +} + /** * nd_pmem_available_dpa - for the given dimm+region account unallocated dpa * @nd_mapping: container of dpa-resource-root + labels diff --git a/drivers/nvdimm/namespace_devs.c b/drivers/nvdimm/namespace_devs.c index 28afdd668905..ba5120aba9bf 100644 --- a/drivers/nvdimm/namespace_devs.c +++ b/drivers/nvdimm/namespace_devs.c @@ -836,7 +836,7 @@ static int __reserve_free_pmem(struct device *dev, void *data) return 0; }
-static void release_free_pmem(struct nvdimm_bus *nvdimm_bus, +void release_free_pmem(struct nvdimm_bus *nvdimm_bus, struct nd_mapping *nd_mapping) { struct nvdimm_drvdata *ndd = to_ndd(nd_mapping); @@ -847,7 +847,7 @@ static void release_free_pmem(struct nvdimm_bus *nvdimm_bus, nvdimm_free_dpa(ndd, res); }
-static int reserve_free_pmem(struct nvdimm_bus *nvdimm_bus, +int reserve_free_pmem(struct nvdimm_bus *nvdimm_bus, struct nd_mapping *nd_mapping) { struct nvdimm *nvdimm = nd_mapping->nvdimm; @@ -1032,7 +1032,7 @@ static ssize_t __size_store(struct device *dev, unsigned long long val)
allocated += nvdimm_allocated_dpa(ndd, &label_id); } - available = nd_region_available_dpa(nd_region); + available = nd_region_contiguous_max(nd_region);
if (val > available + allocated) return -ENOSPC; diff --git a/drivers/nvdimm/nd-core.h b/drivers/nvdimm/nd-core.h index 79274ead54fb..391b97e5c820 100644 --- a/drivers/nvdimm/nd-core.h +++ b/drivers/nvdimm/nd-core.h @@ -100,6 +100,15 @@ struct nd_region; struct nvdimm_drvdata; struct nd_mapping; void nd_mapping_free_labels(struct nd_mapping *nd_mapping); + +int reserve_free_pmem(struct nvdimm_bus *nvdimm_bus, + struct nd_mapping *nd_mapping); +void release_free_pmem(struct nvdimm_bus *nvdimm_bus, + struct nd_mapping *nd_mapping); + +resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region, + struct nd_mapping *nd_mapping); +resource_size_t nd_region_contiguous_max(struct nd_region *nd_region); resource_size_t nd_pmem_available_dpa(struct nd_region *nd_region, struct nd_mapping *nd_mapping, resource_size_t *overlap); resource_size_t nd_blk_available_dpa(struct nd_region *nd_region); diff --git a/drivers/nvdimm/region_devs.c b/drivers/nvdimm/region_devs.c index ec3543b83330..48b3bd5219d2 100644 --- a/drivers/nvdimm/region_devs.c +++ b/drivers/nvdimm/region_devs.c @@ -357,6 +357,35 @@ static ssize_t set_cookie_show(struct device *dev, } static DEVICE_ATTR_RO(set_cookie);
+resource_size_t nd_region_contiguous_max(struct nd_region *nd_region) +{ + struct nvdimm_bus *nvdimm_bus = walk_to_nvdimm_bus(&nd_region->dev); + resource_size_t available = 0; + int i, rc; + + WARN_ON(!is_nvdimm_bus_locked(&nd_region->dev)); + for (i = 0; i < nd_region->ndr_mappings; i++) { + struct nd_mapping *nd_mapping = &nd_region->mapping[i]; + struct nvdimm_drvdata *ndd = to_ndd(nd_mapping); + + /* if a dimm is disabled the available capacity is zero */ + if (!ndd) + return 0; + + + if (is_memory(&nd_region->dev)) { + rc = reserve_free_pmem(nvdimm_bus, nd_mapping); + if (rc) + return 0; + available += nd_pmem_max_contiguous_dpa(nd_region, + nd_mapping); + release_free_pmem(nvdimm_bus, nd_mapping); + } else if (is_nd_blk(&nd_region->dev)) + available += nd_blk_available_dpa(nd_region); + } + return available; +} + resource_size_t nd_region_available_dpa(struct nd_region *nd_region) { resource_size_t blk_max_overlap = 0, available, overlap; --
On Fri, Jul 6, 2018 at 3:06 PM, Keith Busch keith.busch@intel.com wrote:
On Thu, Jul 05, 2018 at 06:48:25PM -0700, Dan Williams wrote:
On Thu, Jul 5, 2018 at 6:25 PM, Dan Williams dan.j.williams@intel.com wrote:
So, I liked how simple and straightforward this was, but the unit test reminded me how complicated the free space allocation could be given that we still might need to account for aliased BLK capacity.
A few things I now notice:
- It is possible for a dpa resource to start before nd_mapping->start
and/or end after nd_mapping->size. See the diagram in tools/testing/nvdimm/test/nfit.c
- We need to be summing the max extents across all DIMMs in the set
while also accounting for this BLK overlap problem that nd_pmem_available_dpa() handles.
The original bug arose from the fact that the nd_region_available_dpa() accounts for total busy capacity and the allocation path looks for extents. It would be nice to unify those two functions... or maybe standardize on scan_allocate() everywhere.
I think you can redo this using reserve_free_pmem() and then find the largest free space reservation after accounting for blk_max_overlap as calculated by nd_region_available_dpa().
reserve_free_pmem() inserts resources into the dpa resource tree with the name "pmem-reserve" then you need to filter those against blk_max_overlap to find which ones might appear free, but shadowed by a BLK allocation on one of the DIMMs in the set. Yes, it allocates memory and is heavy handed, but I'm having a hard time thinking of something better at the moment.
Okay, I think I follow what you're saying, but let me just see if the following is what you had in mind before sending a v3:
This is going in the right direction... but still needs to account for the blk_overlap.
So, on a given DIMM BLK capacity is allocated from the top of DPA space going down and PMEM capacity is allocated from the bottom of the DPA space going up.
Since BLK capacity is single DIMM, and PMEM capacity is striped you could get into the situation where one DIMM is fully allocated for BLK usage and that would shade / remove the possibility to use the PMEM capacity on the other DIMMs in the PMEM set. PMEM needs all the same DPAs in all the DIMMs to be free.
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..f30e0c3b0282 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,31 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/**
- nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max
contiguous unallocated dpa range.
- @nd_region: constrain available space check to this reference region
- @nd_mapping: container of dpa-resource-root + labels
- */
+resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region,
struct nd_mapping *nd_mapping)+{
struct nvdimm_drvdata *ndd = to_ndd(nd_mapping);resource_size_t max = 0;struct resource *res;if (!ndd)return 0;for_each_dpa_resource(ndd, res) {if (strcmp(res->name, "pmem-reserve") != 0)continue;if (resource_size(res) > max)
...so instead straight resource_size() here you need trim the end of this "pmem-reserve" resource to the start of the first BLK allocation in any of the DIMMs in the set.
See blk_start calculation in nd_pmem_available_dpa().
On Fri, Jul 06, 2018 at 03:25:15PM -0700, Dan Williams wrote:
This is going in the right direction... but still needs to account for the blk_overlap.
So, on a given DIMM BLK capacity is allocated from the top of DPA space going down and PMEM capacity is allocated from the bottom of the DPA space going up.
Since BLK capacity is single DIMM, and PMEM capacity is striped you could get into the situation where one DIMM is fully allocated for BLK usage and that would shade / remove the possibility to use the PMEM capacity on the other DIMMs in the PMEM set. PMEM needs all the same DPAs in all the DIMMs to be free.
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..f30e0c3b0282 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,31 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/**
- nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max
contiguous unallocated dpa range.
- @nd_region: constrain available space check to this reference region
- @nd_mapping: container of dpa-resource-root + labels
- */
+resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region,
struct nd_mapping *nd_mapping)+{
struct nvdimm_drvdata *ndd = to_ndd(nd_mapping);resource_size_t max = 0;struct resource *res;if (!ndd)return 0;for_each_dpa_resource(ndd, res) {if (strcmp(res->name, "pmem-reserve") != 0)continue;if (resource_size(res) > max)...so instead straight resource_size() here you need trim the end of this "pmem-reserve" resource to the start of the first BLK allocation in any of the DIMMs in the set.
See blk_start calculation in nd_pmem_available_dpa().
Hmm, the resources defining this are a bit inconvenient given these constraints. If an unallocated portion of a DIMM may only be used for BLK because an overlapping range in another DIMM is allocated that way, would it make since to insert something like a "blk-reserve" resource in all the other DIMMs so we don't need multiple iterations to calculate which DPAs can be used for PMEM?
On Mon, Jul 9, 2018 at 8:44 AM, Keith Busch keith.busch@intel.com wrote:
On Fri, Jul 06, 2018 at 03:25:15PM -0700, Dan Williams wrote:
This is going in the right direction... but still needs to account for the blk_overlap.
So, on a given DIMM BLK capacity is allocated from the top of DPA space going down and PMEM capacity is allocated from the bottom of the DPA space going up.
Since BLK capacity is single DIMM, and PMEM capacity is striped you could get into the situation where one DIMM is fully allocated for BLK usage and that would shade / remove the possibility to use the PMEM capacity on the other DIMMs in the PMEM set. PMEM needs all the same DPAs in all the DIMMs to be free.
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..f30e0c3b0282 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,31 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/**
- nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max
contiguous unallocated dpa range.
- @nd_region: constrain available space check to this reference region
- @nd_mapping: container of dpa-resource-root + labels
- */
+resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region,
struct nd_mapping *nd_mapping)+{
struct nvdimm_drvdata *ndd = to_ndd(nd_mapping);resource_size_t max = 0;struct resource *res;if (!ndd)return 0;for_each_dpa_resource(ndd, res) {if (strcmp(res->name, "pmem-reserve") != 0)continue;if (resource_size(res) > max)...so instead straight resource_size() here you need trim the end of this "pmem-reserve" resource to the start of the first BLK allocation in any of the DIMMs in the set.
See blk_start calculation in nd_pmem_available_dpa().
Hmm, the resources defining this are a bit inconvenient given these constraints. If an unallocated portion of a DIMM may only be used for BLK because an overlapping range in another DIMM is allocated that way, would it make since to insert something like a "blk-reserve" resource in all the other DIMMs so we don't need multiple iterations to calculate which DPAs can be used for PMEM?
Do you mean temporarily allocate the blk-reserve? You could have a different sized / offset BLK allocation on each DIMM in the PMEM set.
On Mon, Jul 09, 2018 at 02:49:54PM -0700, Dan Williams wrote:
On Mon, Jul 9, 2018 at 8:44 AM, Keith Busch keith.busch@intel.com wrote:
On Fri, Jul 06, 2018 at 03:25:15PM -0700, Dan Williams wrote:
This is going in the right direction... but still needs to account for the blk_overlap.
So, on a given DIMM BLK capacity is allocated from the top of DPA space going down and PMEM capacity is allocated from the bottom of the DPA space going up.
Since BLK capacity is single DIMM, and PMEM capacity is striped you could get into the situation where one DIMM is fully allocated for BLK usage and that would shade / remove the possibility to use the PMEM capacity on the other DIMMs in the PMEM set. PMEM needs all the same DPAs in all the DIMMs to be free.
diff --git a/drivers/nvdimm/dimm_devs.c b/drivers/nvdimm/dimm_devs.c index 8d348b22ba45..f30e0c3b0282 100644 --- a/drivers/nvdimm/dimm_devs.c +++ b/drivers/nvdimm/dimm_devs.c @@ -536,6 +536,31 @@ resource_size_t nd_blk_available_dpa(struct nd_region *nd_region) return info.available; }
+/**
- nd_pmem_max_contiguous_dpa - For the given dimm+region, return the max
contiguous unallocated dpa range.
- @nd_region: constrain available space check to this reference region
- @nd_mapping: container of dpa-resource-root + labels
- */
+resource_size_t nd_pmem_max_contiguous_dpa(struct nd_region *nd_region,
struct nd_mapping *nd_mapping)+{
struct nvdimm_drvdata *ndd = to_ndd(nd_mapping);resource_size_t max = 0;struct resource *res;if (!ndd)return 0;for_each_dpa_resource(ndd, res) {if (strcmp(res->name, "pmem-reserve") != 0)continue;if (resource_size(res) > max)...so instead straight resource_size() here you need trim the end of this "pmem-reserve" resource to the start of the first BLK allocation in any of the DIMMs in the set.
See blk_start calculation in nd_pmem_available_dpa().
Okay, I think I've got it this time. __reserve_free_pmem won't reserve a resource above the first blk resource, so we can't get some fragmented alternating mix of blk and pmem within a dimm. With that in mind, if we reserve pmem across all dimms in a region, take the largest reserved pmem within each dimm, then take the smallest of those across all dimms, that should be the largest allocatable pmem extent.
linux-stable-mirror@lists.linaro.org
-
Dan Williams -
Keith Busch