Donate to e Foundation | Murena handsets with /e/OS | Own a part of Murena! Learn more

Commit c88aea7a authored by Tejun Heo's avatar Tejun Heo
Browse files

x86-64, NUMA: Make emulation code build numa_meminfo and share the registration path 



NUMA emulation code built nodes[] array and had its own registration
path to set up the emulated nodes.  Update it such that it generates
emulated numa_meminfo and returns control to initmem_init() and shares
the same registration path with non-emulated cases.

Because {acpi|amd}_fake_nodes() expect nodes[] parameter,
fake_physnodes() now generates nodes[] from numa_meminfo.  This will
go away with further updates.

Signed-off-by: default avatarTejun Heo <tj@kernel.org>
Cc: Yinghai Lu <yinghai@kernel.org>
Cc: Brian Gerst <brgerst@gmail.com>
Cc: Cyrill Gorcunov <gorcunov@gmail.com>
Cc: Shaohui Zheng <shaohui.zheng@intel.com>
Cc: David Rientjes <rientjes@google.com>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: H. Peter Anvin <hpa@linux.intel.com>
parent 9d073cae

arch/x86/mm/numa_64.c

+86 −87
Original line number Original line Diff line number Diff line
@@ -541,7 +541,6 @@ static int __init numa_register_memblks(struct numa_meminfo *mi) 




#ifdef CONFIG_NUMA_EMU

#ifdef CONFIG_NUMA_EMU

/* Numa emulation */

/* Numa emulation */

static struct bootnode nodes[MAX_NUMNODES] __initdata;

static struct bootnode physnodes[MAX_NUMNODES] __initdata;

static struct bootnode physnodes[MAX_NUMNODES] __initdata;





static int emu_nid_to_phys[MAX_NUMNODES] __cpuinitdata;

static int emu_nid_to_phys[MAX_NUMNODES] __cpuinitdata;
@@ -626,9 +625,24 @@ static int __init setup_physnodes(unsigned long start, unsigned long end) 
	return ret;

	return ret;

}

}





static void __init fake_physnodes(int acpi, int amd, int nr_nodes)

static void __init fake_physnodes(int acpi, int amd,

				  const struct numa_meminfo *ei)

{

{

	int i;

	static struct bootnode nodes[MAX_NUMNODES] __initdata;

	int i, nr_nodes = 0;



	for (i = 0; i < ei->nr_blks; i++) {

		int nid = ei->blk[i].nid;



		if (nodes[nid].start == nodes[nid].end) {

			nodes[nid].start = ei->blk[i].start;

			nodes[nid].end = ei->blk[i].end;

			nr_nodes++;

		} else {

			nodes[nid].start = min(ei->blk[i].start, nodes[nid].start);

			nodes[nid].end = max(ei->blk[i].end, nodes[nid].end);

		}

	}





	BUG_ON(acpi && amd);

	BUG_ON(acpi && amd);

#ifdef CONFIG_ACPI_NUMA

#ifdef CONFIG_ACPI_NUMA
@@ -645,45 +659,44 @@ static void __init fake_physnodes(int acpi, int amd, int nr_nodes) 
}

}





/*

/*

 * Setups up nid to range from addr to addr + size.  If the end

 * Sets up nid to range from @start to @end.  The return value is -errno if

 * boundary is greater than max_addr, then max_addr is used instead.

 * something went wrong, 0 otherwise.

 * The return value is 0 if there is additional memory left for

 * allocation past addr and -1 otherwise.  addr is adjusted to be at

 * the end of the node.

 */

 */

static int __init setup_node_range(int nid, int physnid,

static int __init emu_setup_memblk(struct numa_meminfo *ei,

				   u64 *addr, u64 size, u64 max_addr)

				   int nid, int physnid, u64 start, u64 end)

{

{

	int ret = 0;

	struct numa_memblk *eb = &ei->blk[ei->nr_blks];

	nodes[nid].start = *addr;



	*addr += size;

	if (ei->nr_blks >= NR_NODE_MEMBLKS) {

	if (*addr >= max_addr) {

		pr_err("NUMA: Too many emulated memblks, failing emulation\n");

		*addr = max_addr;

		return -EINVAL;

		ret = -1;

	}

	}

	nodes[nid].end = *addr;



	node_set(nid, node_possible_map);

	ei->nr_blks++;

	eb->start = start;

	eb->end = end;

	eb->nid = nid;





	if (emu_nid_to_phys[nid] == NUMA_NO_NODE)

	if (emu_nid_to_phys[nid] == NUMA_NO_NODE)

		emu_nid_to_phys[nid] = physnid;

		emu_nid_to_phys[nid] = physnid;





	printk(KERN_INFO "Faking node %d at %016Lx-%016Lx (%LuMB)\n", nid,

	printk(KERN_INFO "Faking node %d at %016Lx-%016Lx (%LuMB)\n", nid,

	       nodes[nid].start, nodes[nid].end,

	       eb->start, eb->end, (eb->end - eb->start) >> 20);

	       (nodes[nid].end - nodes[nid].start) >> 20);

	return 0;

	return ret;

}

}





/*

/*

 * Sets up nr_nodes fake nodes interleaved over physical nodes ranging from addr

 * Sets up nr_nodes fake nodes interleaved over physical nodes ranging from addr

 * to max_addr.  The return value is the number of nodes allocated.

 * to max_addr.  The return value is the number of nodes allocated.

 */

 */

static int __init split_nodes_interleave(u64 addr, u64 max_addr, int nr_nodes)

static int __init split_nodes_interleave(struct numa_meminfo *ei,

					 u64 addr, u64 max_addr, int nr_nodes)

{

{

	nodemask_t physnode_mask = NODE_MASK_NONE;

	nodemask_t physnode_mask = NODE_MASK_NONE;

	u64 size;

	u64 size;

	int big;

	int big;

	int ret = 0;

	int nid = 0;

	int i;

	int i, ret;





	if (nr_nodes <= 0)

	if (nr_nodes <= 0)

		return -1;

		return -1;
@@ -721,7 +734,7 @@ static int __init split_nodes_interleave(u64 addr, u64 max_addr, int nr_nodes) 
			u64 end = physnodes[i].start + size;

			u64 end = physnodes[i].start + size;

			u64 dma32_end = PFN_PHYS(MAX_DMA32_PFN);

			u64 dma32_end = PFN_PHYS(MAX_DMA32_PFN);





			if (ret < big)

			if (nid < big)

				end += FAKE_NODE_MIN_SIZE;

				end += FAKE_NODE_MIN_SIZE;





			/*

			/*
@@ -760,16 +773,21 @@ static int __init split_nodes_interleave(u64 addr, u64 max_addr, int nr_nodes) 
			 * happen as a result of rounding down each node's size

			 * happen as a result of rounding down each node's size

			 * to FAKE_NODE_MIN_SIZE.

			 * to FAKE_NODE_MIN_SIZE.

			 */

			 */

			if (nodes_weight(physnode_mask) + ret >= nr_nodes)

			if (nodes_weight(physnode_mask) + nid >= nr_nodes)

				end = physnodes[i].end;

				end = physnodes[i].end;





			if (setup_node_range(ret++, i, &physnodes[i].start,

			ret = emu_setup_memblk(ei, nid++, i,

						end - physnodes[i].start,

					       physnodes[i].start,

						physnodes[i].end) < 0)

					       min(end, physnodes[i].end));

			if (ret < 0)

				return ret;



			physnodes[i].start = min(end, physnodes[i].end);

			if (physnodes[i].start == physnodes[i].end)

				node_clear(i, physnode_mask);

				node_clear(i, physnode_mask);

		}

		}

	}

	}

	return ret;

	return 0;

}

}





/*

/*
@@ -794,12 +812,13 @@ static u64 __init find_end_of_node(u64 start, u64 max_addr, u64 size) 
 * Sets up fake nodes of `size' interleaved over physical nodes ranging from

 * Sets up fake nodes of `size' interleaved over physical nodes ranging from

 * `addr' to `max_addr'.  The return value is the number of nodes allocated.

 * `addr' to `max_addr'.  The return value is the number of nodes allocated.

 */

 */

static int __init split_nodes_size_interleave(u64 addr, u64 max_addr, u64 size)

static int __init split_nodes_size_interleave(struct numa_meminfo *ei,

					      u64 addr, u64 max_addr, u64 size)

{

{

	nodemask_t physnode_mask = NODE_MASK_NONE;

	nodemask_t physnode_mask = NODE_MASK_NONE;

	u64 min_size;

	u64 min_size;

	int ret = 0;

	int nid = 0;

	int i;

	int i, ret;





	if (!size)

	if (!size)

		return -1;

		return -1;
@@ -854,30 +873,31 @@ static int __init split_nodes_size_interleave(u64 addr, u64 max_addr, u64 size) 
			    memblock_x86_hole_size(end, physnodes[i].end) < size)

			    memblock_x86_hole_size(end, physnodes[i].end) < size)

				end = physnodes[i].end;

				end = physnodes[i].end;





			/*

			ret = emu_setup_memblk(ei, nid++, i,

			 * Setup the fake node that will be allocated as bootmem

					       physnodes[i].start,

			 * later.  If setup_node_range() returns non-zero, there

					       min(end, physnodes[i].end));

			 * is no more memory available on this physical node.

			if (ret < 0)

			 */

				return ret;

			if (setup_node_range(ret++, i, &physnodes[i].start,



						end - physnodes[i].start,

			physnodes[i].start = min(end, physnodes[i].end);

						physnodes[i].end) < 0)

			if (physnodes[i].start == physnodes[i].end)

				node_clear(i, physnode_mask);

				node_clear(i, physnode_mask);

		}

		}

	}

	}

	return ret;

	return 0;

}

}





/*

/*

 * Sets up the system RAM area from start_pfn to last_pfn according to the

 * Sets up the system RAM area from start_pfn to last_pfn according to the

 * numa=fake command-line option.

 * numa=fake command-line option.

 */

 */

static int __init numa_emulation(int acpi, int amd)

static bool __init numa_emulation(int acpi, int amd)

{

{

	static struct numa_meminfo ei __initdata;

	static struct numa_meminfo ei __initdata;

	const u64 max_addr = max_pfn << PAGE_SHIFT;

	const u64 max_addr = max_pfn << PAGE_SHIFT;

	int num_nodes;

	int i, ret;

	int i;



	memset(&ei, 0, sizeof(ei));





	for (i = 0; i < MAX_NUMNODES; i++)

	for (i = 0; i < MAX_NUMNODES; i++)

		emu_nid_to_phys[i] = NUMA_NO_NODE;

		emu_nid_to_phys[i] = NUMA_NO_NODE;
@@ -891,52 +911,33 @@ static int __init numa_emulation(int acpi, int amd) 
		u64 size;

		u64 size;





		size = memparse(emu_cmdline, &emu_cmdline);

		size = memparse(emu_cmdline, &emu_cmdline);

		num_nodes = split_nodes_size_interleave(0, max_addr, size);

		ret = split_nodes_size_interleave(&ei, 0, max_addr, size);

	} else {

	} else {

		unsigned long n;

		unsigned long n;





		n = simple_strtoul(emu_cmdline, NULL, 0);

		n = simple_strtoul(emu_cmdline, NULL, 0);

		num_nodes = split_nodes_interleave(0, max_addr, n);

		ret = split_nodes_interleave(&ei, 0, max_addr, n);

	}



	if (ret < 0)

		return false;



	if (numa_cleanup_meminfo(&ei) < 0) {

		pr_warning("NUMA: Warning: constructed meminfo invalid, disabling emulation\n");

		return false;

	}

	}





	if (num_nodes < 0)

	/* commit */

		return num_nodes;

	numa_meminfo = ei;





	/* make sure all emulated nodes are mapped to a physical node */

	/* make sure all emulated nodes are mapped to a physical node */

	for (i = 0; i < ARRAY_SIZE(emu_nid_to_phys); i++)

	for (i = 0; i < ARRAY_SIZE(emu_nid_to_phys); i++)

		if (emu_nid_to_phys[i] == NUMA_NO_NODE)

		if (emu_nid_to_phys[i] == NUMA_NO_NODE)

			emu_nid_to_phys[i] = 0;

			emu_nid_to_phys[i] = 0;





	ei.nr_blks = num_nodes;

	fake_physnodes(acpi, amd, &ei);

	for (i = 0; i < ei.nr_blks; i++) {

		ei.blk[i].start = nodes[i].start;

		ei.blk[i].end = nodes[i].end;

		ei.blk[i].nid = i;

	}



	memnode_shift = compute_hash_shift(&ei);

	if (memnode_shift < 0) {

		memnode_shift = 0;

		printk(KERN_ERR "No NUMA hash function found.  NUMA emulation "

		       "disabled.\n");

		return -1;

	}



	/*

	 * We need to vacate all active ranges that may have been registered for

	 * the e820 memory map.

	 */

	remove_all_active_ranges();

	for_each_node_mask(i, node_possible_map)

		memblock_x86_register_active_regions(i, nodes[i].start >> PAGE_SHIFT,

						nodes[i].end >> PAGE_SHIFT);

	init_memory_mapping_high();

	for_each_node_mask(i, node_possible_map)

		setup_node_bootmem(i, nodes[i].start, nodes[i].end);

	fake_physnodes(acpi, amd, num_nodes);

	numa_init_array();

	numa_emu_dist = true;

	numa_emu_dist = true;

	return 0;

	return true;

}

}

#endif /* CONFIG_NUMA_EMU */

#endif /* CONFIG_NUMA_EMU */
@@ -988,15 +989,13 @@ void __init initmem_init(void) 
			continue;

			continue;

#ifdef CONFIG_NUMA_EMU

#ifdef CONFIG_NUMA_EMU

		setup_physnodes(0, max_pfn << PAGE_SHIFT);

		setup_physnodes(0, max_pfn << PAGE_SHIFT);

		if (emu_cmdline && !numa_emulation(i == 0, i == 1))

		/*

			return;

		 * If requested, try emulation.  If emulation is not used,



		 * build identity emu_nid_to_phys[] for numa_add_cpu()

		/* not emulating, build identity mapping for numa_add_cpu() */

		 */

		if (!emu_cmdline || !numa_emulation(i == 0, i == 1))

			for (j = 0; j < ARRAY_SIZE(emu_nid_to_phys); j++)

			for (j = 0; j < ARRAY_SIZE(emu_nid_to_phys); j++)

				emu_nid_to_phys[j] = j;

				emu_nid_to_phys[j] = j;



		nodes_clear(node_possible_map);

		nodes_clear(node_online_map);

#endif

#endif

		if (numa_register_memblks(&numa_meminfo) < 0)

		if (numa_register_memblks(&numa_meminfo) < 0)

			continue;

			continue;