On 3/27/19 7:44 AM, Michal Hocko wrote> What? Normal spin lock implementation doesn't disable interrupts. So > either I misunderstand what you are saying or you seem to be confused. > the thing is that in_atomic relies on preempt_count to work properly and > if you have CONFIG_PREEMPT_COUNT=n then you simply never know whether > preemption is disabled so you do not know that a spin_lock is held. > irqs_disabled on the other hand checks whether arch specific flag for > IRQs handling is set (or cleared). So you would only catch irq safe spin > locks with the above check. Exactly, because kmemleak_alloc() is only called in a few call sites, slab allocation, neigh_hash_alloc(), alloc_page_ext(), sg_kmalloc(), early_amd_iommu_init() and blk_mq_alloc_rqs(), my review does not yield any of those holding irq unsafe spinlocks. Could future code changes suddenly call kmemleak_alloc() with a irq unsafe spinlock held? Always possible, but it is unlikely to happen. I could put some comments on kmemleak_alloc() about this though.
