x86, mm: introduce vmem_altmap to augment vmemmap_populate()
In support of providing struct page for large persistent memory
capacities, use struct vmem_altmap to change the default policy for
allocating memory for the memmap array. The default vmemmap_populate()
allocates page table storage area from the page allocator. Given
persistent memory capacities relative to DRAM it may not be feasible to
store the memmap in 'System Memory'. Instead vmem_altmap represents
pre-allocated "device pages" to satisfy vmemmap_alloc_block_buf()
requests.
Signed-off-by: Dan Williams <dan.j.williams@intel.com>
Reported-by: kbuild test robot <lkp@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: Ingo Molnar <mingo@redhat.com>
Cc: "H. Peter Anvin" <hpa@zytor.com>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
diff --git a/kernel/memremap.c b/kernel/memremap.c
index 61cfbf4..562f647 100644
--- a/kernel/memremap.c
+++ b/kernel/memremap.c
@@ -166,6 +166,7 @@
struct resource res;
struct percpu_ref *ref;
struct dev_pagemap pgmap;
+ struct vmem_altmap altmap;
};
static void pgmap_radix_release(struct resource *res)
@@ -183,6 +184,7 @@
struct page_map *page_map = data;
struct resource *res = &page_map->res;
resource_size_t align_start, align_size;
+ struct dev_pagemap *pgmap = &page_map->pgmap;
pgmap_radix_release(res);
@@ -190,6 +192,8 @@
align_start = res->start & ~(SECTION_SIZE - 1);
align_size = ALIGN(resource_size(res), SECTION_SIZE);
arch_remove_memory(align_start, align_size);
+ dev_WARN_ONCE(dev, pgmap->altmap && pgmap->altmap->alloc,
+ "%s: failed to free all reserved pages\n", __func__);
}
/* assumes rcu_read_lock() held at entry */
@@ -203,11 +207,23 @@
return page_map ? &page_map->pgmap : NULL;
}
-void *devm_memremap_pages(struct device *dev, struct resource *res)
+/**
+ * devm_memremap_pages - remap and provide memmap backing for the given resource
+ * @dev: hosting device for @res
+ * @res: "host memory" address range
+ * @altmap: optional descriptor for allocating the memmap from @res
+ *
+ * Note, the expectation is that @res is a host memory range that could
+ * feasibly be treated as a "System RAM" range, i.e. not a device mmio
+ * range, but this is not enforced.
+ */
+void *devm_memremap_pages(struct device *dev, struct resource *res,
+ struct vmem_altmap *altmap)
{
int is_ram = region_intersects(res->start, resource_size(res),
"System RAM");
resource_size_t key, align_start, align_size;
+ struct dev_pagemap *pgmap;
struct page_map *page_map;
int error, nid;
@@ -220,14 +236,27 @@
if (is_ram == REGION_INTERSECTS)
return __va(res->start);
+ if (altmap && !IS_ENABLED(CONFIG_SPARSEMEM_VMEMMAP)) {
+ dev_err(dev, "%s: altmap requires CONFIG_SPARSEMEM_VMEMMAP=y\n",
+ __func__);
+ return ERR_PTR(-ENXIO);
+ }
+
page_map = devres_alloc_node(devm_memremap_pages_release,
sizeof(*page_map), GFP_KERNEL, dev_to_node(dev));
if (!page_map)
return ERR_PTR(-ENOMEM);
+ pgmap = &page_map->pgmap;
memcpy(&page_map->res, res, sizeof(*res));
- page_map->pgmap.dev = dev;
+ pgmap->dev = dev;
+ if (altmap) {
+ memcpy(&page_map->altmap, altmap, sizeof(*altmap));
+ pgmap->altmap = &page_map->altmap;
+ }
+ pgmap->res = &page_map->res;
+
mutex_lock(&pgmap_lock);
error = 0;
for (key = res->start; key <= res->end; key += SECTION_SIZE) {
@@ -273,4 +302,43 @@
return ERR_PTR(error);
}
EXPORT_SYMBOL(devm_memremap_pages);
+
+unsigned long vmem_altmap_offset(struct vmem_altmap *altmap)
+{
+ /* number of pfns from base where pfn_to_page() is valid */
+ return altmap->reserve + altmap->free;
+}
+
+void vmem_altmap_free(struct vmem_altmap *altmap, unsigned long nr_pfns)
+{
+ altmap->alloc -= nr_pfns;
+}
+
+#ifdef CONFIG_SPARSEMEM_VMEMMAP
+struct vmem_altmap *to_vmem_altmap(unsigned long memmap_start)
+{
+ /*
+ * 'memmap_start' is the virtual address for the first "struct
+ * page" in this range of the vmemmap array. In the case of
+ * CONFIG_SPARSE_VMEMMAP a page_to_pfn conversion is simple
+ * pointer arithmetic, so we can perform this to_vmem_altmap()
+ * conversion without concern for the initialization state of
+ * the struct page fields.
+ */
+ struct page *page = (struct page *) memmap_start;
+ struct dev_pagemap *pgmap;
+
+ /*
+ * Uncoditionally retrieve a dev_pagemap associated with the
+ * given physical address, this is only for use in the
+ * arch_{add|remove}_memory() for setting up and tearing down
+ * the memmap.
+ */
+ rcu_read_lock();
+ pgmap = find_dev_pagemap(__pfn_to_phys(page_to_pfn(page)));
+ rcu_read_unlock();
+
+ return pgmap ? pgmap->altmap : NULL;
+}
+#endif /* CONFIG_SPARSEMEM_VMEMMAP */
#endif /* CONFIG_ZONE_DEVICE */