summaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorLinus Torvalds <torvalds@linux-foundation.org>2019-09-28 14:26:47 -0700
committerLinus Torvalds <torvalds@linux-foundation.org>2019-09-28 14:26:47 -0700
commitedf445ad7c8d58c2784a5b733790e80999093d8f (patch)
tree2ffeaee2454bf0b03530c22ac23b4eb5edb4d89d
parenta2953204b576ea3ba4afd07b917811d50fc49778 (diff)
parent76e654cc91bbe627aa6067916f02a4d3ac041620 (diff)
downloadlinux-edf445ad7c8d58c2784a5b733790e80999093d8f.tar.bz2
Merge branch 'hugepage-fallbacks' (hugepatch patches from David Rientjes)
Merge hugepage allocation updates from David Rientjes: "We (mostly Linus, Andrea, and myself) have been discussing offlist how to implement a sane default allocation strategy for hugepages on NUMA platforms. With these reverts in place, the page allocator will happily allocate a remote hugepage immediately rather than try to make a local hugepage available. This incurs a substantial performance degradation when memory compaction would have otherwise made a local hugepage available. This series reverts those reverts and attempts to propose a more sane default allocation strategy specifically for hugepages. Andrea acknowledges this is likely to fix the swap storms that he originally reported that resulted in the patches that removed __GFP_THISNODE from hugepage allocations. The immediate goal is to return 5.3 to the behavior the kernel has implemented over the past several years so that remote hugepages are not immediately allocated when local hugepages could have been made available because the increased access latency is untenable. The next goal is to introduce a sane default allocation strategy for hugepages allocations in general regardless of the configuration of the system so that we prevent thrashing of local memory when compaction is unlikely to succeed and can prefer remote hugepages over remote native pages when the local node is low on memory." Note on timing: this reverts the hugepage VM behavior changes that got introduced fairly late in the 5.3 cycle, and that fixed a huge performance regression for certain loads that had been around since 4.18. Andrea had this note: "The regression of 4.18 was that it was taking hours to start a VM where 3.10 was only taking a few seconds, I reported all the details on lkml when it was finally tracked down in August 2018. https://lore.kernel.org/linux-mm/20180820032640.9896-2-aarcange@redhat.com/ __GFP_THISNODE in MADV_HUGEPAGE made the above enterprise vfio workload degrade like in the "current upstream" above. And it still would have been that bad as above until 5.3-rc5" where the bad behavior ends up happening as you fill up a local node, and without that change, you'd get into the nasty swap storm behavior due to compaction working overtime to make room for more memory on the nodes. As a result 5.3 got the two performance fix reverts in rc5. However, David Rientjes then noted that those performance fixes in turn regressed performance for other loads - although not quite to the same degree. He suggested reverting the reverts and instead replacing them with two small changes to how hugepage allocations are done (patch descriptions rephrased by me): - "avoid expensive reclaim when compaction may not succeed": just admit that the allocation failed when you're trying to allocate a huge-page and compaction wasn't successful. - "allow hugepage fallback to remote nodes when madvised": when that node-local huge-page allocation failed, retry without forcing the local node. but by then I judged it too late to replace the fixes for a 5.3 release. So 5.3 was released with behavior that harked back to the pre-4.18 logic. But now we're in the merge window for 5.4, and we can see if this alternate model fixes not just the horrendous swap storm behavior, but also restores the performance regression that the late reverts caused. Fingers crossed. * emailed patches from David Rientjes <rientjes@google.com>: mm, page_alloc: allow hugepage fallback to remote nodes when madvised mm, page_alloc: avoid expensive reclaim when compaction may not succeed Revert "Revert "Revert "mm, thp: consolidate THP gfp handling into alloc_hugepage_direct_gfpmask"" Revert "Revert "mm, thp: restore node-local hugepage allocations""
-rw-r--r--include/linux/gfp.h12
-rw-r--r--include/linux/mempolicy.h2
-rw-r--r--mm/huge_memory.c51
-rw-r--r--mm/mempolicy.c45
-rw-r--r--mm/page_alloc.c22
-rw-r--r--mm/shmem.c2
6 files changed, 92 insertions, 42 deletions
diff --git a/include/linux/gfp.h b/include/linux/gfp.h
index f33881688f42..fb07b503dc45 100644
--- a/include/linux/gfp.h
+++ b/include/linux/gfp.h
@@ -510,18 +510,22 @@ alloc_pages(gfp_t gfp_mask, unsigned int order)
}
extern struct page *alloc_pages_vma(gfp_t gfp_mask, int order,
struct vm_area_struct *vma, unsigned long addr,
- int node);
+ int node, bool hugepage);
+#define alloc_hugepage_vma(gfp_mask, vma, addr, order) \
+ alloc_pages_vma(gfp_mask, order, vma, addr, numa_node_id(), true)
#else
#define alloc_pages(gfp_mask, order) \
alloc_pages_node(numa_node_id(), gfp_mask, order)
-#define alloc_pages_vma(gfp_mask, order, vma, addr, node)\
+#define alloc_pages_vma(gfp_mask, order, vma, addr, node, false)\
+ alloc_pages(gfp_mask, order)
+#define alloc_hugepage_vma(gfp_mask, vma, addr, order) \
alloc_pages(gfp_mask, order)
#endif
#define alloc_page(gfp_mask) alloc_pages(gfp_mask, 0)
#define alloc_page_vma(gfp_mask, vma, addr) \
- alloc_pages_vma(gfp_mask, 0, vma, addr, numa_node_id())
+ alloc_pages_vma(gfp_mask, 0, vma, addr, numa_node_id(), false)
#define alloc_page_vma_node(gfp_mask, vma, addr, node) \
- alloc_pages_vma(gfp_mask, 0, vma, addr, node)
+ alloc_pages_vma(gfp_mask, 0, vma, addr, node, false)
extern unsigned long __get_free_pages(gfp_t gfp_mask, unsigned int order);
extern unsigned long get_zeroed_page(gfp_t gfp_mask);
diff --git a/include/linux/mempolicy.h b/include/linux/mempolicy.h
index bac395f1d00a..5228c62af416 100644
--- a/include/linux/mempolicy.h
+++ b/include/linux/mempolicy.h
@@ -139,8 +139,6 @@ struct mempolicy *mpol_shared_policy_lookup(struct shared_policy *sp,
struct mempolicy *get_task_policy(struct task_struct *p);
struct mempolicy *__get_vma_policy(struct vm_area_struct *vma,
unsigned long addr);
-struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
- unsigned long addr);
bool vma_policy_mof(struct vm_area_struct *vma);
extern void numa_default_policy(void);
diff --git a/mm/huge_memory.c b/mm/huge_memory.c
index 73fc517c08d2..c5cb6dcd6c69 100644
--- a/mm/huge_memory.c
+++ b/mm/huge_memory.c
@@ -659,40 +659,30 @@ release:
* available
* never: never stall for any thp allocation
*/
-static inline gfp_t alloc_hugepage_direct_gfpmask(struct vm_area_struct *vma, unsigned long addr)
+static inline gfp_t alloc_hugepage_direct_gfpmask(struct vm_area_struct *vma)
{
const bool vma_madvised = !!(vma->vm_flags & VM_HUGEPAGE);
- gfp_t this_node = 0;
-
-#ifdef CONFIG_NUMA
- struct mempolicy *pol;
- /*
- * __GFP_THISNODE is used only when __GFP_DIRECT_RECLAIM is not
- * specified, to express a general desire to stay on the current
- * node for optimistic allocation attempts. If the defrag mode
- * and/or madvise hint requires the direct reclaim then we prefer
- * to fallback to other node rather than node reclaim because that
- * can lead to excessive reclaim even though there is free memory
- * on other nodes. We expect that NUMA preferences are specified
- * by memory policies.
- */
- pol = get_vma_policy(vma, addr);
- if (pol->mode != MPOL_BIND)
- this_node = __GFP_THISNODE;
- mpol_cond_put(pol);
-#endif
+ /* Always do synchronous compaction */
if (test_bit(TRANSPARENT_HUGEPAGE_DEFRAG_DIRECT_FLAG, &transparent_hugepage_flags))
return GFP_TRANSHUGE | (vma_madvised ? 0 : __GFP_NORETRY);
+
+ /* Kick kcompactd and fail quickly */
if (test_bit(TRANSPARENT_HUGEPAGE_DEFRAG_KSWAPD_FLAG, &transparent_hugepage_flags))
- return GFP_TRANSHUGE_LIGHT | __GFP_KSWAPD_RECLAIM | this_node;
+ return GFP_TRANSHUGE_LIGHT | __GFP_KSWAPD_RECLAIM;
+
+ /* Synchronous compaction if madvised, otherwise kick kcompactd */
if (test_bit(TRANSPARENT_HUGEPAGE_DEFRAG_KSWAPD_OR_MADV_FLAG, &transparent_hugepage_flags))
- return GFP_TRANSHUGE_LIGHT | (vma_madvised ? __GFP_DIRECT_RECLAIM :
- __GFP_KSWAPD_RECLAIM | this_node);
+ return GFP_TRANSHUGE_LIGHT |
+ (vma_madvised ? __GFP_DIRECT_RECLAIM :
+ __GFP_KSWAPD_RECLAIM);
+
+ /* Only do synchronous compaction if madvised */
if (test_bit(TRANSPARENT_HUGEPAGE_DEFRAG_REQ_MADV_FLAG, &transparent_hugepage_flags))
- return GFP_TRANSHUGE_LIGHT | (vma_madvised ? __GFP_DIRECT_RECLAIM :
- this_node);
- return GFP_TRANSHUGE_LIGHT | this_node;
+ return GFP_TRANSHUGE_LIGHT |
+ (vma_madvised ? __GFP_DIRECT_RECLAIM : 0);
+
+ return GFP_TRANSHUGE_LIGHT;
}
/* Caller must hold page table lock. */
@@ -764,8 +754,8 @@ vm_fault_t do_huge_pmd_anonymous_page(struct vm_fault *vmf)
pte_free(vma->vm_mm, pgtable);
return ret;
}
- gfp = alloc_hugepage_direct_gfpmask(vma, haddr);
- page = alloc_pages_vma(gfp, HPAGE_PMD_ORDER, vma, haddr, numa_node_id());
+ gfp = alloc_hugepage_direct_gfpmask(vma);
+ page = alloc_hugepage_vma(gfp, vma, haddr, HPAGE_PMD_ORDER);
if (unlikely(!page)) {
count_vm_event(THP_FAULT_FALLBACK);
return VM_FAULT_FALLBACK;
@@ -1372,9 +1362,8 @@ vm_fault_t do_huge_pmd_wp_page(struct vm_fault *vmf, pmd_t orig_pmd)
alloc:
if (__transparent_hugepage_enabled(vma) &&
!transparent_hugepage_debug_cow()) {
- huge_gfp = alloc_hugepage_direct_gfpmask(vma, haddr);
- new_page = alloc_pages_vma(huge_gfp, HPAGE_PMD_ORDER, vma,
- haddr, numa_node_id());
+ huge_gfp = alloc_hugepage_direct_gfpmask(vma);
+ new_page = alloc_hugepage_vma(huge_gfp, vma, haddr, HPAGE_PMD_ORDER);
} else
new_page = NULL;
diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index de27d08b1ff8..4ae967bcf954 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -1179,8 +1179,8 @@ static struct page *new_page(struct page *page, unsigned long start)
} else if (PageTransHuge(page)) {
struct page *thp;
- thp = alloc_pages_vma(GFP_TRANSHUGE, HPAGE_PMD_ORDER, vma,
- address, numa_node_id());
+ thp = alloc_hugepage_vma(GFP_TRANSHUGE, vma, address,
+ HPAGE_PMD_ORDER);
if (!thp)
return NULL;
prep_transhuge_page(thp);
@@ -1732,7 +1732,7 @@ struct mempolicy *__get_vma_policy(struct vm_area_struct *vma,
* freeing by another task. It is the caller's responsibility to free the
* extra reference for shared policies.
*/
-struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
+static struct mempolicy *get_vma_policy(struct vm_area_struct *vma,
unsigned long addr)
{
struct mempolicy *pol = __get_vma_policy(vma, addr);
@@ -2081,6 +2081,7 @@ static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
* @vma: Pointer to VMA or NULL if not available.
* @addr: Virtual Address of the allocation. Must be inside the VMA.
* @node: Which node to prefer for allocation (modulo policy).
+ * @hugepage: for hugepages try only the preferred node if possible
*
* This function allocates a page from the kernel page pool and applies
* a NUMA policy associated with the VMA or the current process.
@@ -2091,7 +2092,7 @@ static struct page *alloc_page_interleave(gfp_t gfp, unsigned order,
*/
struct page *
alloc_pages_vma(gfp_t gfp, int order, struct vm_area_struct *vma,
- unsigned long addr, int node)
+ unsigned long addr, int node, bool hugepage)
{
struct mempolicy *pol;
struct page *page;
@@ -2109,6 +2110,42 @@ alloc_pages_vma(gfp_t gfp, int order, struct vm_area_struct *vma,
goto out;
}
+ if (unlikely(IS_ENABLED(CONFIG_TRANSPARENT_HUGEPAGE) && hugepage)) {
+ int hpage_node = node;
+
+ /*
+ * For hugepage allocation and non-interleave policy which
+ * allows the current node (or other explicitly preferred
+ * node) we only try to allocate from the current/preferred
+ * node and don't fall back to other nodes, as the cost of
+ * remote accesses would likely offset THP benefits.
+ *
+ * If the policy is interleave, or does not allow the current
+ * node in its nodemask, we allocate the standard way.
+ */
+ if (pol->mode == MPOL_PREFERRED && !(pol->flags & MPOL_F_LOCAL))
+ hpage_node = pol->v.preferred_node;
+
+ nmask = policy_nodemask(gfp, pol);
+ if (!nmask || node_isset(hpage_node, *nmask)) {
+ mpol_cond_put(pol);
+ page = __alloc_pages_node(hpage_node,
+ gfp | __GFP_THISNODE, order);
+
+ /*
+ * If hugepage allocations are configured to always
+ * synchronous compact or the vma has been madvised
+ * to prefer hugepage backing, retry allowing remote
+ * memory as well.
+ */
+ if (!page && (gfp & __GFP_DIRECT_RECLAIM))
+ page = __alloc_pages_node(hpage_node,
+ gfp | __GFP_NORETRY, order);
+
+ goto out;
+ }
+ }
+
nmask = policy_nodemask(gfp, pol);
preferred_nid = policy_node(gfp, pol, node);
page = __alloc_pages_nodemask(gfp, order, preferred_nid, nmask);
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index 3334a769eb91..15c2050c629b 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -4467,6 +4467,28 @@ retry_cpuset:
if (page)
goto got_pg;
+ if (order >= pageblock_order && (gfp_mask & __GFP_IO)) {
+ /*
+ * If allocating entire pageblock(s) and compaction
+ * failed because all zones are below low watermarks
+ * or is prohibited because it recently failed at this
+ * order, fail immediately.
+ *
+ * Reclaim is
+ * - potentially very expensive because zones are far
+ * below their low watermarks or this is part of very
+ * bursty high order allocations,
+ * - not guaranteed to help because isolate_freepages()
+ * may not iterate over freed pages as part of its
+ * linear scan, and
+ * - unlikely to make entire pageblocks free on its
+ * own.
+ */
+ if (compact_result == COMPACT_SKIPPED ||
+ compact_result == COMPACT_DEFERRED)
+ goto nopage;
+ }
+
/*
* Checks for costly allocations with __GFP_NORETRY, which
* includes THP page fault allocations
diff --git a/mm/shmem.c b/mm/shmem.c
index 30ce722c23fa..cd570cc79c76 100644
--- a/mm/shmem.c
+++ b/mm/shmem.c
@@ -1481,7 +1481,7 @@ static struct page *shmem_alloc_hugepage(gfp_t gfp,
shmem_pseudo_vma_init(&pvma, info, hindex);
page = alloc_pages_vma(gfp | __GFP_COMP | __GFP_NORETRY | __GFP_NOWARN,
- HPAGE_PMD_ORDER, &pvma, 0, numa_node_id());
+ HPAGE_PMD_ORDER, &pvma, 0, numa_node_id(), true);
shmem_pseudo_vma_destroy(&pvma);
if (page)
prep_transhuge_page(page);