diff mbox

[v5,4/5] mm/sparse: add new sparse_init_nid() and sparse_init()

Message ID 20180712203730.8703-5-pasha.tatashin@oracle.com (mailing list archive)
State New, archived
Headers show

Commit Message

Pavel Tatashin July 12, 2018, 8:37 p.m. UTC
sparse_init() requires to temporary allocate two large buffers:
usemap_map and map_map. Baoquan He has identified that these buffers are so
large that Linux is not bootable on small memory machines, such as a kdump
boot. The buffers are especially large when CONFIG_X86_5LEVEL is set, as
they are scaled to the maximum physical memory size.

Baoquan provided a fix, which reduces these sizes of these buffers, but it
is much better to get rid of them entirely.

Add a new way to initialize sparse memory: sparse_init_nid(), which only
operates within one memory node, and thus allocates memory either in large
contiguous block or allocates section by section. This eliminates the need
for use of temporary buffers.

For simplified bisecting and review temporarly call sparse_init()
new_sparse_init(), the new interface is going to be enabled as well as old
code removed in the next patch.

Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
---
 mm/sparse.c | 85 +++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 85 insertions(+)

Comments

Oscar Salvador July 13, 2018, 12:03 p.m. UTC | #1
On Thu, Jul 12, 2018 at 04:37:29PM -0400, Pavel Tatashin wrote:
> sparse_init() requires to temporary allocate two large buffers:
> usemap_map and map_map. Baoquan He has identified that these buffers are so
> large that Linux is not bootable on small memory machines, such as a kdump
> boot. The buffers are especially large when CONFIG_X86_5LEVEL is set, as
> they are scaled to the maximum physical memory size.
> 
> Baoquan provided a fix, which reduces these sizes of these buffers, but it
> is much better to get rid of them entirely.
> 
> Add a new way to initialize sparse memory: sparse_init_nid(), which only
> operates within one memory node, and thus allocates memory either in large
> contiguous block or allocates section by section. This eliminates the need
> for use of temporary buffers.
> 
> For simplified bisecting and review temporarly call sparse_init()
> new_sparse_init(), the new interface is going to be enabled as well as old
> code removed in the next patch.
> 
> Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>

Looks good to me, and it will make the code much shorter/easier.

Reviewed-by: Oscar Salvador <osalvador@suse.de>

Thanks
Pavel Tatashin July 13, 2018, 12:37 p.m. UTC | #2
> > Signed-off-by: Pavel Tatashin <pasha.tatashin@oracle.com>
>
> Looks good to me, and it will make the code much shorter/easier.
>
> Reviewed-by: Oscar Salvador <osalvador@suse.de>
>

Thank you!

Pave
diff mbox

Patch

diff --git a/mm/sparse.c b/mm/sparse.c
index 01c616342909..4087b94afddf 100644
--- a/mm/sparse.c
+++ b/mm/sparse.c
@@ -200,6 +200,11 @@  static inline int next_present_section_nr(int section_nr)
 	      (section_nr <= __highest_present_section_nr));	\
 	     section_nr = next_present_section_nr(section_nr))
 
+static inline unsigned long first_present_section_nr(void)
+{
+	return next_present_section_nr(-1);
+}
+
 /*
  * Record how many memory sections are marked as present
  * during system bootup.
@@ -668,6 +673,86 @@  void __init sparse_init(void)
 	memblock_free_early(__pa(usemap_map), size);
 }
 
+/*
+ * Initialize sparse on a specific node. The node spans [pnum_begin, pnum_end)
+ * And number of present sections in this node is map_count.
+ */
+static void __init sparse_init_nid(int nid, unsigned long pnum_begin,
+				   unsigned long pnum_end,
+				   unsigned long map_count)
+{
+	unsigned long pnum, usemap_longs, *usemap;
+	struct page *map;
+
+	usemap_longs = BITS_TO_LONGS(SECTION_BLOCKFLAGS_BITS);
+	usemap = sparse_early_usemaps_alloc_pgdat_section(NODE_DATA(nid),
+							  usemap_size() *
+							  map_count);
+	if (!usemap) {
+		pr_err("%s: node[%d] usemap allocation failed", __func__, nid);
+		goto failed;
+	}
+	sparse_buffer_init(map_count * section_map_size(), nid);
+	for_each_present_section_nr(pnum_begin, pnum) {
+		if (pnum >= pnum_end)
+			break;
+
+		map = sparse_mem_map_populate(pnum, nid, NULL);
+		if (!map) {
+			pr_err("%s: node[%d] memory map backing failed. Some memory will not be available.",
+			       __func__, nid);
+			pnum_begin = pnum;
+			goto failed;
+		}
+		check_usemap_section_nr(nid, usemap);
+		sparse_init_one_section(__nr_to_section(pnum), pnum, map, usemap);
+		usemap += usemap_longs;
+	}
+	sparse_buffer_fini();
+	return;
+failed:
+	/* We failed to allocate, mark all the following pnums as not present */
+	for_each_present_section_nr(pnum_begin, pnum) {
+		struct mem_section *ms;
+
+		if (pnum >= pnum_end)
+			break;
+		ms = __nr_to_section(pnum);
+		ms->section_mem_map = 0;
+	}
+}
+
+/*
+ * Allocate the accumulated non-linear sections, allocate a mem_map
+ * for each and record the physical to section mapping.
+ */
+void __init new_sparse_init(void)
+{
+	unsigned long pnum_begin = first_present_section_nr();
+	int nid_begin = sparse_early_nid(__nr_to_section(pnum_begin));
+	unsigned long pnum_end, map_count = 1;
+
+	/* Setup pageblock_order for HUGETLB_PAGE_SIZE_VARIABLE */
+	set_pageblock_order();
+
+	for_each_present_section_nr(pnum_begin + 1, pnum_end) {
+		int nid = sparse_early_nid(__nr_to_section(pnum_end));
+
+		if (nid == nid_begin) {
+			map_count++;
+			continue;
+		}
+		/* Init node with sections in range [pnum_begin, pnum_end) */
+		sparse_init_nid(nid_begin, pnum_begin, pnum_end, map_count);
+		nid_begin = nid;
+		pnum_begin = pnum_end;
+		map_count = 1;
+	}
+	/* cover the last node */
+	sparse_init_nid(nid_begin, pnum_begin, pnum_end, map_count);
+	vmemmap_populate_print_last();
+}
+
 #ifdef CONFIG_MEMORY_HOTPLUG
 
 /* Mark all memory sections within the pfn range as online */