tree: https://git.kernel.org/pub/scm/linux/kernel/git/vbabka/linux.git slub-local-lock-v4r1 head: cb9b19ebd40aa23420fd5f8889fb4cd81ff346bc commit: f141ef3ae9b1a6421e47136f93500ae74e1076e1 [29/35] mm: slub: Move flush_cpu_slab() invocations __free_slab() invocations out of IRQ context config: xtensa-randconfig-p002-20210810 (attached as .config) compiler: xtensa-linux-gcc (GCC) 10.3.0 reproduce (this is a W=1 build): wget https://raw.githubusercontent.com/intel/lkp-tests/master/sbin/make.cross -O ~/bin/make.cross chmod +x ~/bin/make.cross # https://git.kernel.org/pub/scm/linux/kernel/git/vbabka/linux.git/commit/?... git remote add vbabka https://git.kernel.org/pub/scm/linux/kernel/git/vbabka/linux.git git fetch --no-tags vbabka slub-local-lock-v4r1 git checkout f141ef3ae9b1a6421e47136f93500ae74e1076e1 # save the attached .config to linux build tree COMPILER_INSTALL_PATH=$HOME/0day COMPILER=gcc-10.3.0 make.cross ARCH=xtensa If you fix the issue, kindly add following tag as appropriate Reported-by: kernel test robot <lkp(a)intel.com&gt; All warnings (new ones prefixed by >>): 4396 | int __kmem_cache_do_shrink(struct kmem_cache *s) | ^~~~~~~~~~~~~~~~~~~~~~ vim +/__kmem_cache_do_shrink +4396 mm/slub.c 4386 4387 /* 4388 * kmem_cache_shrink discards empty slabs and promotes the slabs filled 4389 * up most to the head of the partial lists. New allocations will then 4390 * fill those up and thus they can be removed from the partial lists. 4391 * 4392 * The slabs with the least items are placed last. This results in them 4393 * being allocated from last increasing the chance that the last objects 4394 * are freed in them. 4395 */ 4397 { 4398 int node; 4399 int i; 4400 struct kmem_cache_node *n; 4401 struct page *page; 4402 struct page *t; 4403 struct list_head discard; 4404 struct list_head promote[SHRINK_PROMOTE_MAX]; 4405 unsigned long flags; 4406 int ret = 0; 4407 4408 for_each_kmem_cache_node(s, node, n) { 4409 INIT_LIST_HEAD(&discard); 4410 for (i = 0; i < SHRINK_PROMOTE_MAX; i++) 4411 INIT_LIST_HEAD(promote + i); 4412 4413 spin_lock_irqsave(&n->list_lock, flags); 4414 4415 /* 4416 * Build lists of slabs to discard or promote. 4417 * 4418 * Note that concurrent frees may occur while we hold the 4419 * list_lock. page->inuse here is the upper limit. 4420 */ 4421 list_for_each_entry_safe(page, t, &n->partial, slab_list) { 4422 int free = page->objects - page->inuse; 4423 4424 /* Do not reread page->inuse */ 4425 barrier(); 4426 4427 /* We do not keep full slabs on the list */ 4428 BUG_ON(free <= 0); 4429 4430 if (free == page->objects) { 4431 list_move(&page->slab_list, &discard); 4432 n->nr_partial--; 4433 } else if (free <= SHRINK_PROMOTE_MAX) 4434 list_move(&page->slab_list, promote + free - 1); 4435 } 4436 4437 /* 4438 * Promote the slabs filled up most to the head of the 4439 * partial list. 4440 */ 4441 for (i = SHRINK_PROMOTE_MAX - 1; i >= 0; i--) 4442 list_splice(promote + i, &n->partial); 4443 4444 spin_unlock_irqrestore(&n->list_lock, flags); 4445 4446 /* Release empty slabs */ 4447 list_for_each_entry_safe(page, t, &discard, slab_list) 4448 discard_slab(s, page); 4449 4450 if (slabs_node(s, node)) 4451 ret = 1; 4452 } 4453 4454 return ret; 4455 } 4456 --- 0-DAY CI Kernel Test Service, Intel Corporation https://lists.01.org/hyperkitty/list/kbuild-all@lists.01.org