NoaSoftware
/
TencentOS-kernel
forked from TencentOS/TencentOS-kernel
0
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity
TencentOS-kernel/mm
History
Dongli Zhang 891d438291 page_frag: Recover from memory pressure 
[ Upstream commit d8c19014bba8f565d8a2f1f46b4e38d1d97bf1a7 ]

The ethernet driver may allocate skb (and skb->data) via napi_alloc_skb().
This ends up to page_frag_alloc() to allocate skb->data from
page_frag_cache->va.

During the memory pressure, page_frag_cache->va may be allocated as
pfmemalloc page. As a result, the skb->pfmemalloc is always true as
skb->data is from page_frag_cache->va. The skb will be dropped if the
sock (receiver) does not have SOCK_MEMALLOC. This is expected behaviour
under memory pressure.

However, once kernel is not under memory pressure any longer (suppose large
amount of memory pages are just reclaimed), the page_frag_alloc() may still
re-use the prior pfmemalloc page_frag_cache->va to allocate skb->data. As a
result, the skb->pfmemalloc is always true unless page_frag_cache->va is
re-allocated, even if the kernel is not under memory pressure any longer.

Here is how kernel runs into issue.

1. The kernel is under memory pressure and allocation of
PAGE_FRAG_CACHE_MAX_ORDER in __page_frag_cache_refill() will fail. Instead,
the pfmemalloc page is allocated for page_frag_cache->va.

2: All skb->data from page_frag_cache->va (pfmemalloc) will have
skb->pfmemalloc=true. The skb will always be dropped by sock without
SOCK_MEMALLOC. This is an expected behaviour.

3. Suppose a large amount of pages are reclaimed and kernel is not under
memory pressure any longer. We expect skb->pfmemalloc drop will not happen.

4. Unfortunately, page_frag_alloc() does not proactively re-allocate
page_frag_alloc->va and will always re-use the prior pfmemalloc page. The
skb->pfmemalloc is always true even kernel is not under memory pressure any
longer.

Fix this by freeing and re-allocating the page instead of recycling it.

References: https://lore.kernel.org/lkml/20201103193239.1807-1-dongli.zhang@oracle.com/
References: https://lore.kernel.org/linux-mm/20201105042140.5253-1-willy@infradead.org/
Suggested-by: Matthew Wilcox (Oracle) <willy@infradead.org>
Cc: Aruna Ramakrishna <aruna.ramakrishna@oracle.com>
Cc: Bert Barbe <bert.barbe@oracle.com>
Cc: Rama Nichanamatlu <rama.nichanamatlu@oracle.com>
Cc: Venkat Venkatsubra <venkat.x.venkatsubra@oracle.com>
Cc: Manjunath Patil <manjunath.b.patil@oracle.com>
Cc: Joe Jin <joe.jin@oracle.com>
Cc: SRINIVAS <srinivas.eeda@oracle.com>
Fixes: 79930f5892e1 ("net: do not deplete pfmemalloc reserve")
Signed-off-by: Dongli Zhang <dongli.zhang@oracle.com>
Acked-by: Vlastimil Babka <vbabka@suse.cz>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Link: https://lore.kernel.org/r/20201115201029.11903-1-dongli.zhang@oracle.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
 		2021-03-16 16:41:08 +08:00
..
kasan kasan: disable branch tracing for core runtime 2021-03-16 16:28:20 +08:00
Kconfig Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
Kconfig.debug Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
Makefile Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
backing-dev.c bdi: add a ->dev_name field to struct backing_dev_info 2021-03-16 16:27:44 +08:00
balloon_compaction.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
cleancache.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
cma.c cma: don't quit at first error when activating reserved areas 2021-03-16 16:34:57 +08:00
cma.h Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
cma_debug.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
compaction.c mm, compaction: make capture control handling safe wrt interrupts 2021-03-16 16:31:23 +08:00
debug.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
debug_page_ref.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
dmapool.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
early_ioremap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
fadvise.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
failslab.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
filemap.c mm/error_inject: Fix allow_error_inject function signatures. 2021-03-16 16:38:34 +08:00
frame_vector.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
frontswap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
gup.c mm/gup: fix gup_fast with dynamic page table folding 2021-03-16 16:37:40 +08:00
gup_benchmark.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
highmem.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
hmm.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
huge_memory.c mm/page_owner: change split_page_owner to take a count 2021-03-16 16:38:58 +08:00
hugetlb.c mm/hugetlb: fix a race between hugetlb sysctl handlers 2021-03-16 16:35:45 +08:00
hugetlb_cgroup.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
hwpoison-inject.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
init-mm.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
internal.h Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
interval_tree.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
khugepaged.c mm: khugepaged: recalculate min_free_kbytes after memory hotplug as expected by khugepaged 2021-03-16 16:38:06 +08:00
kmemleak-test.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
kmemleak.c mm/kmemleak.c: use address-of operator on section symbols 2021-03-16 16:37:10 +08:00
ksm.c mm/ksm: fix NULL pointer dereference when KSM zero page is enabled 2021-03-16 16:26:56 +08:00
list_lru.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
maccess.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
madvise.c mm: validate pmd after splitting 2021-03-16 16:37:37 +08:00
memblock.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
memcontrol.c mm/memcg: fix device private memcg accounting 2021-03-16 16:38:46 +08:00
memfd.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
memory-failure.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
memory.c mm: avoid data corruption on CoW fault into PFN-mapped VMA 2021-03-16 16:36:58 +08:00
memory_hotplug.c mm: don't rely on system state to detect hot-plug operations 2021-03-16 16:37:50 +08:00
mempolicy.c mm: mempolicy: fix potential pte_unmap_unlock pte error 2021-03-16 16:40:21 +08:00
mempool.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
memremap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
memtest.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
migrate.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mincore.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mlock.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mm_init.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mmap.c mm/mmap.c: initialize align_offset explicitly for vm_unmapped_area 2021-03-16 16:37:12 +08:00
mmu_context.c mm: fix kthread_use_mm() vs TLB invalidate 2021-03-16 16:34:57 +08:00
mmu_gather.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mmu_notifier.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mmzone.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mprotect.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
mremap.c mm: Fix mremap not considering huge pmd devmap 2021-03-16 16:28:43 +08:00
msync.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
nommu.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
oom_kill.c mm, oom_adj: don't loop through tasks in __set_oom_adj when not necessary 2021-03-16 16:38:47 +08:00
page-writeback.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
page_alloc.c page_frag: Recover from memory pressure 2021-03-16 16:41:08 +08:00
page_counter.c mm/page_counter.c: fix protection usage propagation 2021-03-16 16:34:07 +08:00
page_ext.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
page_idle.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
page_io.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
page_isolation.c mm/memory_hotplug: drain per-cpu pages again during memory offline 2021-03-16 16:36:22 +08:00
page_owner.c mm/page_owner: change split_page_owner to take a count 2021-03-16 16:38:58 +08:00
page_poison.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
page_vma_mapped.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
pagewalk.c mm: pagewalk: fix termination condition in walk_pte_range() 2021-03-16 16:36:50 +08:00
percpu-internal.h Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
percpu-km.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
percpu-stats.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
percpu-vm.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
percpu.c percpu: fix first chunk size calculation for populated bitmap 2021-03-16 16:36:21 +08:00
pgtable-generic.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
process_vm_access.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
readahead.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
rmap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
rodata_test.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
shmem.c shmem: fix possible deadlocks on shmlock_user_lock 2021-03-16 16:27:44 +08:00
shuffle.c mm/shuffle: don't move pages between zones and don't read garbage memmaps 2021-03-16 16:34:57 +08:00
shuffle.h Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
slab.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
slab.h Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
slab_common.c mm: memcg/slab: fix memory leak at non-root kmem_cache destroy 2021-03-16 16:32:38 +08:00
slob.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
slub.c mm/slub: fix panic in slab_alloc_node() 2021-03-16 16:40:54 +08:00
sparse-vmemmap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
sparse.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
swap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
swap_cgroup.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
swap_slots.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
swap_state.c mm/swap_state: fix a data race in swapin_nr_pages 2021-03-16 16:37:24 +08:00
swapfile.c mm/swapfile.c: fix potential memory leak in sys_swapon 2021-03-16 16:38:46 +08:00
truncate.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
usercopy.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
userfaultfd.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
util.c mm: add kvfree_sensitive() for freeing sensitive data objects 2021-03-16 16:28:58 +08:00
vmacache.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
vmalloc.c mm/vunmap: add cond_resched() in vunmap_pmd_range 2021-03-16 16:34:58 +08:00
vmpressure.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
vmscan.c mm/vmscan.c: fix data races using kswapd_classzone_idx 2021-03-16 16:37:11 +08:00
vmstat.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
workingset.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
z3fold.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
zbud.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
zpool.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
zsmalloc.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00
zswap.c Init Repo base on linux 5.4.32 long term, and add base tlinux kernel interfaces. 2021-03-16 11:01:34 +08:00