Commit b74e6278fd6db5848163ccdc6e9d8eb6efdee9bd

Authored by Alex Thorlton
Committed by Ingo Molnar
1 parent aa39477b56

sched: Fix KMALLOC_MAX_SIZE overflow during cpumask allocation

When allocating space for load_balance_mask, in sched_init, when
CPUMASK_OFFSTACK is set, we've managed to spill over
KMALLOC_MAX_SIZE on our 6144 core machine.  The patch below
breaks up the allocations so that they don't overflow the max
alloc size.  It also allocates the masks on the the node from
which they'll most commonly be accessed, to minimize remote
accesses on NUMA machines.

Suggested-by: George Beshers <gbeshers@sgi.com>
Signed-off-by: Alex Thorlton <athorlton@sgi.com>
Cc: George Beshers <gbeshers@sgi.com>
Cc: Russ Anderson <rja@sgi.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Link: http://lkml.kernel.org/r/1418928270-148543-1-git-send-email-athorlton@sgi.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>

Showing 1 changed file with 5 additions and 8 deletions Side-by-side Diff

... ... @@ -7113,9 +7113,6 @@
7113 7113 #ifdef CONFIG_RT_GROUP_SCHED
7114 7114 alloc_size += 2 * nr_cpu_ids * sizeof(void **);
7115 7115 #endif
7116   -#ifdef CONFIG_CPUMASK_OFFSTACK
7117   - alloc_size += num_possible_cpus() * cpumask_size();
7118   -#endif
7119 7116 if (alloc_size) {
7120 7117 ptr = (unsigned long)kzalloc(alloc_size, GFP_NOWAIT);
7121 7118  
7122 7119  
7123 7120  
... ... @@ -7135,13 +7132,13 @@
7135 7132 ptr += nr_cpu_ids * sizeof(void **);
7136 7133  
7137 7134 #endif /* CONFIG_RT_GROUP_SCHED */
  7135 + }
7138 7136 #ifdef CONFIG_CPUMASK_OFFSTACK
7139   - for_each_possible_cpu(i) {
7140   - per_cpu(load_balance_mask, i) = (void *)ptr;
7141   - ptr += cpumask_size();
7142   - }
7143   -#endif /* CONFIG_CPUMASK_OFFSTACK */
  7137 + for_each_possible_cpu(i) {
  7138 + per_cpu(load_balance_mask, i) = (cpumask_var_t)kzalloc_node(
  7139 + cpumask_size(), GFP_KERNEL, cpu_to_node(i));
7144 7140 }
  7141 +#endif /* CONFIG_CPUMASK_OFFSTACK */
7145 7142  
7146 7143 init_rt_bandwidth(&def_rt_bandwidth,
7147 7144 global_rt_period(), global_rt_runtime());