[PATCH RFC 08/10] dax/pmem: Add device-dax support for PFN_MODE_NONE

[Joao Martins]

Allowing dax pmem driver to work without struct pages means
that user will request to not create any PFN metadata by writing
seed's device mode to PFN_MODE_NONE.

When the underlying nd_pfn->mode is PFN_MODE_NONE, most dax_pmem
initialization steps can be skipped because we won't have/need a
pfn superblock for the pagemap/struct-pages. We only allocate
an opaque zeroed object with the chosen align requested, and
finally add PFN_SPECIAL to the region pfn_flags.

Signed-off-by: Joao Martins <joao.m.martins@xxxxxxxxxx>
---
 drivers/dax/pmem/core.c | 36 ++++++++++++++++++++++++++++++------
 1 file changed, 30 insertions(+), 6 deletions(-)

diff --git a/drivers/dax/pmem/core.c b/drivers/dax/pmem/core.c
index 2bedf8414fff..67f5604a8291 100644
--- a/drivers/dax/pmem/core.c
+++ b/drivers/dax/pmem/core.c
@@ -17,15 +17,38 @@ struct dev_dax *__dax_pmem_probe(struct device *dev, enum dev_dax_subsys subsys)
 	struct nd_namespace_io *nsio;
 	struct dax_region *dax_region;
 	struct dev_pagemap pgmap = { };
+	struct dev_pagemap *devmap = NULL;
 	struct nd_namespace_common *ndns;
 	struct nd_dax *nd_dax = to_nd_dax(dev);
 	struct nd_pfn *nd_pfn = &nd_dax->nd_pfn;
 	struct nd_region *nd_region = to_nd_region(dev->parent);
+	unsigned long long pfn_flags = PFN_DEV;
 
 	ndns = nvdimm_namespace_common_probe(dev);
 	if (IS_ERR(ndns))
 		return ERR_CAST(ndns);
 
+	rc = sscanf(dev_name(&ndns->dev), "namespace%d.%d", &region_id, &id);
+	if (rc != 2)
+		return ERR_PTR(-EINVAL);
+
+	if (is_nd_dax(&nd_pfn->dev) && nd_pfn->mode == PFN_MODE_NONE) {
+		/* allocate a dummy super block */
+		pfn_sb = devm_kzalloc(&nd_pfn->dev, sizeof(*pfn_sb),
+				      GFP_KERNEL);
+		if (!pfn_sb)
+			return ERR_PTR(-ENOMEM);
+
+		memset(pfn_sb, 0, sizeof(*pfn_sb));
+		pfn_sb->align = nd_pfn->align;
+		nd_pfn->pfn_sb = pfn_sb;
+		pfn_flags |= PFN_SPECIAL;
+
+		nsio = to_nd_namespace_io(&ndns->dev);
+		memcpy(&res, &nsio->res, sizeof(res));
+		goto no_pfn_sb;
+	}
+
 	/* parse the 'pfn' info block via ->rw_bytes */
 	rc = devm_namespace_enable(dev, ndns, nd_info_block_reserve());
 	if (rc)
@@ -45,20 +68,21 @@ struct dev_dax *__dax_pmem_probe(struct device *dev, enum dev_dax_subsys subsys)
 		return ERR_PTR(-EBUSY);
 	}
 
-	rc = sscanf(dev_name(&ndns->dev), "namespace%d.%d", &region_id, &id);
-	if (rc != 2)
-		return ERR_PTR(-EINVAL);
-
 	/* adjust the dax_region resource to the start of data */
 	memcpy(&res, &pgmap.res, sizeof(res));
 	res.start += offset;
+	devmap = &pgmap;
+	pfn_flags |= PFN_MAP;
+
+no_pfn_sb:
 	dax_region = alloc_dax_region(dev, region_id, &res,
 			nd_region->target_node, le32_to_cpu(pfn_sb->align),
-			PFN_DEV|PFN_MAP);
+			pfn_flags);
 	if (!dax_region)
 		return ERR_PTR(-ENOMEM);
 
-	dev_dax = __devm_create_dev_dax(dax_region, id, &pgmap, subsys);
+
+	dev_dax = __devm_create_dev_dax(dax_region, id, devmap, subsys);
 
 	/* child dev_dax instances now own the lifetime of the dax_region */
 	dax_region_put(dax_region);
-- 
2.17.1