Crashes in refresh_zone_stat_thresholds when some nodes have no memory

[Paul Mackerras]

It appears that commit 75ef71840539 ("mm, vmstat: add infrastructure
for per-node vmstats", 2016-07-28) has introduced a regression on
machines that have nodes which have no memory, such as the POWER8
server that I use for testing.  When I boot current upstream, I get a
splat like this:

[    1.713998] Unable to handle kernel paging request for data at address 0xff7a10000
[    1.714164] Faulting instruction address: 0xc000000000270cd0
[    1.714304] Oops: Kernel access of bad area, sig: 11 [#1]
[    1.714414] SMP NR_CPUS=2048 NUMA PowerNV
[    1.714530] Modules linked in:
[    1.714647] CPU: 0 PID: 1 Comm: swapper/0 Not tainted 4.7.0-kvm+ #118
[    1.714786] task: c000000ff0680010 task.stack: c000000ff0704000
[    1.714926] NIP: c000000000270cd0 LR: c000000000270ce8 CTR: 0000000000000000
[    1.715093] REGS: c000000ff0707900 TRAP: 0300   Not tainted  (4.7.0-kvm+)
[    1.715232] MSR: 9000000102009033 <SF,HV,VEC,EE,ME,IR,DR,RI,LE,TM[E]>  CR: 846b6824  XER: 20000000
[    1.715748] CFAR: c000000000008768 DAR: 0000000ff7a10000 DSISR: 42000000 SOFTE: 1 
GPR00: c000000000270d08 c000000ff0707b80 c0000000011fb200 0000000000000000 
GPR04: 0000000000000800 0000000000000000 0000000000000000 0000000000000000 
GPR08: ffffffffffffffff 0000000000000000 0000000ff7a10000 c00000000122aae0 
GPR12: c000000000a1e440 c00000000fb80000 c00000000000c188 0000000000000000 
GPR16: 0000000000000000 0000000000000000 0000000000000000 0000000000000000 
GPR20: 0000000000000000 0000000000000000 0000000000000000 c000000000cecad0 
GPR24: c000000000d035b8 c000000000d6cd18 c000000000d6cd18 c000001fffa86300 
GPR28: 0000000000000000 c000001fffa96300 c000000001230034 c00000000122eb18 
[    1.717484] NIP [c000000000270cd0] refresh_zone_stat_thresholds+0x80/0x240
[    1.717568] LR [c000000000270ce8] refresh_zone_stat_thresholds+0x98/0x240
[    1.717648] Call Trace:
[    1.717687] [c000000ff0707b80] [c000000000270d08] refresh_zone_stat_thresholds+0xb8/0x240 (unreliable)
[    1.717818] [c000000ff0707bd0] [c000000000a1e4d4] init_per_zone_wmark_min+0x94/0xb0
[    1.717932] [c000000ff0707c30] [c00000000000b90c] do_one_initcall+0x6c/0x1d0
[    1.718036] [c000000ff0707cf0] [c000000000d04244] kernel_init_freeable+0x294/0x384
[    1.718150] [c000000ff0707dc0] [c00000000000c1a8] kernel_init+0x28/0x160
[    1.718249] [c000000ff0707e30] [c000000000009968] ret_from_kernel_thread+0x5c/0x74
[    1.718358] Instruction dump:
[    1.718408] 3fc20003 3bde4e34 3b800000 60420000 3860ffff 3fbb0001 4800001c 60420000 
[    1.718575] 3d220003 3929f8e0 7d49502a e93d9c00 <7f8a49ae> 38a30001 38800800 7ca507b4 

It turns out that we can get a pgdat in the online pgdat list where
pgdat->per_cpu_nodestats is NULL.  On my machine the pgdats for nodes
1 and 17 are like this.  All the memory is in nodes 0 and 16.

With the patch below, the system boots normally.  I don't guarantee to
have found every place that needs a check, and it may be better to fix
this by allocating space for per-cpu statistics on nodes which have no
memory rather than checking at each use site.

Paul.
--------
mm: cope with memoryless nodes not having per-cpu statistics allocated

It seems that the pgdat for nodes which have no memory will also have
no per-cpu statistics space allocated, that is, pgdat->per_cpu_nodestats
is NULL.  Avoid crashing on machines which have memoryless nodes by
checking for non-NULL pgdat->per_cpu_nodestats.

Signed-off-by: Paul Mackerras <paulus@xxxxxxxxxx>
---
diff --git a/include/linux/vmstat.h b/include/linux/vmstat.h
index 6137719..48b2780 100644
--- a/include/linux/vmstat.h
+++ b/include/linux/vmstat.h
@@ -184,8 +184,9 @@ static inline unsigned long node_page_state_snapshot(pg_data_t *pgdat,
 
 #ifdef CONFIG_SMP
 	int cpu;
-	for_each_online_cpu(cpu)
-		x += per_cpu_ptr(pgdat->per_cpu_nodestats, cpu)->vm_node_stat_diff[item];
+	if (pgdat->per_cpu_nodestats)
+		for_each_online_cpu(cpu)
+			x += per_cpu_ptr(pgdat->per_cpu_nodestats, cpu)->vm_node_stat_diff[item];
 
 	if (x < 0)
 		x = 0;
diff --git a/mm/vmstat.c b/mm/vmstat.c
index 89cec42..d83881e 100644
--- a/mm/vmstat.c
+++ b/mm/vmstat.c
@@ -176,6 +176,10 @@ void refresh_zone_stat_thresholds(void)
 
 	/* Zero current pgdat thresholds */
 	for_each_online_pgdat(pgdat) {
+		if (!pgdat->per_cpu_nodestats) {
+			pr_err("No nodestats for node %d\n", pgdat->node_id);
+			continue;
+		}
 		for_each_online_cpu(cpu) {
 			per_cpu_ptr(pgdat->per_cpu_nodestats, cpu)->stat_threshold = 0;
 		}
@@ -184,6 +188,10 @@ void refresh_zone_stat_thresholds(void)
 	for_each_populated_zone(zone) {
 		struct pglist_data *pgdat = zone->zone_pgdat;
 		unsigned long max_drift, tolerate_drift;
+		if (!pgdat->per_cpu_nodestats) {
+			pr_err("No per cpu nodestats\n");
+			continue;
+		}
 
 		threshold = calculate_normal_threshold(zone);
 
@@ -701,6 +709,8 @@ static int refresh_cpu_vm_stats(bool do_pagesets)
 	for_each_online_pgdat(pgdat) {
 		struct per_cpu_nodestat __percpu *p = pgdat->per_cpu_nodestats;
 
+		if (!p)
+			continue;
 		for (i = 0; i < NR_VM_NODE_STAT_ITEMS; i++) {
 			int v;
 
@@ -748,6 +758,8 @@ void cpu_vm_stats_fold(int cpu)
 	for_each_online_pgdat(pgdat) {
 		struct per_cpu_nodestat *p;
 
+		if (!pgdat->per_cpu_nodestats)
+			continue;
 		p = per_cpu_ptr(pgdat->per_cpu_nodestats, cpu);
 
 		for (i = 0; i < NR_VM_NODE_STAT_ITEMS; i++)