The kasan_populate_vmalloc() will cost a lot of time when populate large size, it will cause soft lockup, watchdog: BUG: soft lockup - CPU#3 stuck for 26s! [insmod:458] _raw_spin_unlock_irqrestore+0x50/0xb8 rmqueue_bulk+0x434/0x6b8 get_page_from_freelist+0xdd4/0x1680 __alloc_pages+0x244/0x508 alloc_pages+0xf0/0x218 __get_free_pages+0x1c/0x50 kasan_populate_vmalloc_pte+0x30/0x188 __apply_to_page_range+0x3ec/0x650 apply_to_page_range+0x1c/0x30 kasan_populate_vmalloc+0x60/0x70 alloc_vmap_area.part.67+0x328/0xe50 alloc_vmap_area+0x4c/0x78 __get_vm_area_node.constprop.76+0x130/0x240 __vmalloc_node_range+0x12c/0x340 __vmalloc_node+0x8c/0xb0 vmalloc+0x2c/0x40 Fix it by adding a cond_resched(). Signed-off-by: Kefeng Wang <wangkefeng.wang@xxxxxxxxxx> --- mm/kasan/shadow.c | 2 ++ 1 file changed, 2 insertions(+) diff --git a/mm/kasan/shadow.c b/mm/kasan/shadow.c index dd772f9d0f08..fd15e38ff80e 100644 --- a/mm/kasan/shadow.c +++ b/mm/kasan/shadow.c @@ -317,6 +317,8 @@ static int kasan_populate_vmalloc_pte(pte_t *ptep, unsigned long addr, unsigned long page; pte_t pte; + cond_resched(); + if (likely(!pte_none(ptep_get(ptep)))) return 0; -- 2.41.0
