[lttng-dev] [RFC lttng-modules v4] Add kmalloc failover to vmalloc
Mathieu Desnoyers
mathieu.desnoyers at efficios.com
Mon Sep 25 15:19:03 UTC 2017
----- On Sep 25, 2017, at 10:56 AM, Michael Jeanson mjeanson at efficios.com wrote:
> This patch is based on the kvmalloc helpers introduced in kernel 4.12.
>
> It will gracefully failover memory allocations of more than one page to
> vmalloc for systems under high memory pressure or fragmentation.
>
> Thoughts?
>
> V2:
> - Add vmalloc_sync_all
> - Use only for lttng_session
>
> V3:
> - Added to all allocations of potentially >PAGE_SIZE
> - Added the "node" variants
>
> V4
> - Added __lttng_vmalloc_node_fallback wrapper
>
> Signed-off-by: Michael Jeanson <mjeanson at efficios.com>
>
> See upstream commit:
> commit a7c3e901a46ff54c016d040847eda598a9e3e653
> Author: Michal Hocko <mhocko at suse.com>
> Date: Mon May 8 15:57:09 2017 -0700
>
> mm: introduce kv[mz]alloc helpers
>
> Patch series "kvmalloc", v5.
>
> There are many open coded kmalloc with vmalloc fallback instances in the
> tree. Most of them are not careful enough or simply do not care about
> the underlying semantic of the kmalloc/page allocator which means that
> a) some vmalloc fallbacks are basically unreachable because the kmalloc
> part will keep retrying until it succeeds b) the page allocator can
> invoke a really disruptive steps like the OOM killer to move forward
> which doesn't sound appropriate when we consider that the vmalloc
> fallback is available.
>
> As it can be seen implementing kvmalloc requires quite an intimate
> knowledge if the page allocator and the memory reclaim internals which
> strongly suggests that a helper should be implemented in the memory
> subsystem proper.
>
> Most callers, I could find, have been converted to use the helper
> instead. This is patch 6. There are some more relying on __GFP_REPEAT
> in the networking stack which I have converted as well and Eric Dumazet
> was not opposed [2] to convert them as well.
>
> [1] http://lkml.kernel.org/r/20170130094940.13546-1-mhocko@kernel.org
> [2]
> http://lkml.kernel.org/r/1485273626.16328.301.camel@edumazet-glaptop3.roam.corp.google.com
>
> This patch (of 9):
>
> Using kmalloc with the vmalloc fallback for larger allocations is a
> common pattern in the kernel code. Yet we do not have any common helper
> for that and so users have invented their own helpers. Some of them are
> really creative when doing so. Let's just add kv[mz]alloc and make sure
> it is implemented properly. This implementation makes sure to not make
> a large memory pressure for > PAGE_SZE requests (__GFP_NORETRY) and also
> to not warn about allocation failures. This also rules out the OOM
> killer as the vmalloc is a more approapriate fallback than a disruptive
> user visible action.
>
> Signed-off-by: Michael Jeanson <mjeanson at efficios.com>
> ---
> lib/prio_heap/lttng_prio_heap.c | 7 +-
> lib/ringbuffer/ring_buffer_backend.c | 22 ++---
> lib/ringbuffer/ring_buffer_frontend.c | 13 +--
> lttng-context-perf-counters.c | 6 +-
> lttng-context.c | 6 +-
> lttng-events.c | 6 +-
> wrapper/vmalloc.h | 169 +++++++++++++++++++++++++++++++++-
> 7 files changed, 198 insertions(+), 31 deletions(-)
>
> diff --git a/lib/prio_heap/lttng_prio_heap.c b/lib/prio_heap/lttng_prio_heap.c
> index 6db7f52..01ed69f 100644
> --- a/lib/prio_heap/lttng_prio_heap.c
> +++ b/lib/prio_heap/lttng_prio_heap.c
> @@ -26,6 +26,7 @@
>
> #include <linux/slab.h>
> #include <lib/prio_heap/lttng_prio_heap.h>
> +#include <wrapper/vmalloc.h>
>
> #ifdef DEBUG_HEAP
> void lttng_check_heap(const struct lttng_ptr_heap *heap)
> @@ -70,12 +71,12 @@ int heap_grow(struct lttng_ptr_heap *heap, size_t new_len)
> return 0;
>
> heap->alloc_len = max_t(size_t, new_len, heap->alloc_len << 1);
> - new_ptrs = kmalloc(heap->alloc_len * sizeof(void *), heap->gfpmask);
> + new_ptrs = lttng_kvmalloc(heap->alloc_len * sizeof(void *), heap->gfpmask);
> if (!new_ptrs)
> return -ENOMEM;
> if (heap->ptrs)
> memcpy(new_ptrs, heap->ptrs, heap->len * sizeof(void *));
> - kfree(heap->ptrs);
> + lttng_kvfree(heap->ptrs);
> heap->ptrs = new_ptrs;
> return 0;
> }
> @@ -109,7 +110,7 @@ int lttng_heap_init(struct lttng_ptr_heap *heap, size_t
> alloc_len,
>
> void lttng_heap_free(struct lttng_ptr_heap *heap)
> {
> - kfree(heap->ptrs);
> + lttng_kvfree(heap->ptrs);
> }
>
> static void heapify(struct lttng_ptr_heap *heap, size_t i)
> diff --git a/lib/ringbuffer/ring_buffer_backend.c
> b/lib/ringbuffer/ring_buffer_backend.c
> index f760836..3efa1d1 100644
> --- a/lib/ringbuffer/ring_buffer_backend.c
> +++ b/lib/ringbuffer/ring_buffer_backend.c
> @@ -71,7 +71,7 @@ int lib_ring_buffer_backend_allocate(const struct
> lib_ring_buffer_config *config
> if (unlikely(!pages))
> goto pages_error;
>
> - bufb->array = kmalloc_node(ALIGN(sizeof(*bufb->array)
> + bufb->array = lttng_kvmalloc_node(ALIGN(sizeof(*bufb->array)
> * num_subbuf_alloc,
> 1 << INTERNODE_CACHE_SHIFT),
> GFP_KERNEL | __GFP_NOWARN,
> @@ -90,7 +90,7 @@ int lib_ring_buffer_backend_allocate(const struct
> lib_ring_buffer_config *config
> /* Allocate backend pages array elements */
> for (i = 0; i < num_subbuf_alloc; i++) {
> bufb->array[i] =
> - kzalloc_node(ALIGN(
> + lttng_kvzalloc_node(ALIGN(
> sizeof(struct lib_ring_buffer_backend_pages) +
> sizeof(struct lib_ring_buffer_backend_page)
> * num_pages_per_subbuf,
> @@ -102,7 +102,7 @@ int lib_ring_buffer_backend_allocate(const struct
> lib_ring_buffer_config *config
> }
>
> /* Allocate write-side subbuffer table */
> - bufb->buf_wsb = kzalloc_node(ALIGN(
> + bufb->buf_wsb = lttng_kvzalloc_node(ALIGN(
> sizeof(struct lib_ring_buffer_backend_subbuffer)
> * num_subbuf,
> 1 << INTERNODE_CACHE_SHIFT),
> @@ -122,7 +122,7 @@ int lib_ring_buffer_backend_allocate(const struct
> lib_ring_buffer_config *config
> bufb->buf_rsb.id = subbuffer_id(config, 0, 1, 0);
>
> /* Allocate subbuffer packet counter table */
> - bufb->buf_cnt = kzalloc_node(ALIGN(
> + bufb->buf_cnt = lttng_kvzalloc_node(ALIGN(
> sizeof(struct lib_ring_buffer_backend_counts)
> * num_subbuf,
> 1 << INTERNODE_CACHE_SHIFT),
> @@ -154,15 +154,15 @@ int lib_ring_buffer_backend_allocate(const struct
> lib_ring_buffer_config *config
> return 0;
>
> free_wsb:
> - kfree(bufb->buf_wsb);
> + lttng_kvfree(bufb->buf_wsb);
> free_array:
> for (i = 0; (i < num_subbuf_alloc && bufb->array[i]); i++)
> - kfree(bufb->array[i]);
> + lttng_kvfree(bufb->array[i]);
> depopulate:
> /* Free all allocated pages */
> for (i = 0; (i < num_pages && pages[i]); i++)
> __free_page(pages[i]);
> - kfree(bufb->array);
> + lttng_kvfree(bufb->array);
> array_error:
> vfree(pages);
> pages_error:
> @@ -191,14 +191,14 @@ void lib_ring_buffer_backend_free(struct
> lib_ring_buffer_backend *bufb)
> if (chanb->extra_reader_sb)
> num_subbuf_alloc++;
>
> - kfree(bufb->buf_wsb);
> - kfree(bufb->buf_cnt);
> + lttng_kvfree(bufb->buf_wsb);
> + lttng_kvfree(bufb->buf_cnt);
> for (i = 0; i < num_subbuf_alloc; i++) {
> for (j = 0; j < bufb->num_pages_per_subbuf; j++)
> __free_page(pfn_to_page(bufb->array[i]->p[j].pfn));
> - kfree(bufb->array[i]);
> + lttng_kvfree(bufb->array[i]);
> }
> - kfree(bufb->array);
> + lttng_kvfree(bufb->array);
> bufb->allocated = 0;
> }
>
> diff --git a/lib/ringbuffer/ring_buffer_frontend.c
> b/lib/ringbuffer/ring_buffer_frontend.c
> index 3d60c14..c9ab43f 100644
> --- a/lib/ringbuffer/ring_buffer_frontend.c
> +++ b/lib/ringbuffer/ring_buffer_frontend.c
> @@ -65,6 +65,7 @@
> #include <wrapper/kref.h>
> #include <wrapper/percpu-defs.h>
> #include <wrapper/timer.h>
> +#include <wrapper/vmalloc.h>
>
> /*
> * Internal structure representing offsets to use at a sub-buffer switch.
> @@ -147,8 +148,8 @@ void lib_ring_buffer_free(struct lib_ring_buffer *buf)
> struct channel *chan = buf->backend.chan;
>
> lib_ring_buffer_print_errors(chan, buf, buf->backend.cpu);
> - kfree(buf->commit_hot);
> - kfree(buf->commit_cold);
> + lttng_kvfree(buf->commit_hot);
> + lttng_kvfree(buf->commit_cold);
>
> lib_ring_buffer_backend_free(&buf->backend);
> }
> @@ -245,7 +246,7 @@ int lib_ring_buffer_create(struct lib_ring_buffer *buf,
> return ret;
>
> buf->commit_hot =
> - kzalloc_node(ALIGN(sizeof(*buf->commit_hot)
> + lttng_kvzalloc_node(ALIGN(sizeof(*buf->commit_hot)
> * chan->backend.num_subbuf,
> 1 << INTERNODE_CACHE_SHIFT),
> GFP_KERNEL | __GFP_NOWARN,
> @@ -256,7 +257,7 @@ int lib_ring_buffer_create(struct lib_ring_buffer *buf,
> }
>
> buf->commit_cold =
> - kzalloc_node(ALIGN(sizeof(*buf->commit_cold)
> + lttng_kvzalloc_node(ALIGN(sizeof(*buf->commit_cold)
> * chan->backend.num_subbuf,
> 1 << INTERNODE_CACHE_SHIFT),
> GFP_KERNEL | __GFP_NOWARN,
> @@ -305,9 +306,9 @@ int lib_ring_buffer_create(struct lib_ring_buffer *buf,
>
> /* Error handling */
> free_init:
> - kfree(buf->commit_cold);
> + lttng_kvfree(buf->commit_cold);
> free_commit:
> - kfree(buf->commit_hot);
> + lttng_kvfree(buf->commit_hot);
> free_chanbuf:
> lib_ring_buffer_backend_free(&buf->backend);
> return ret;
> diff --git a/lttng-context-perf-counters.c b/lttng-context-perf-counters.c
> index 8afc11f..260e5d0 100644
> --- a/lttng-context-perf-counters.c
> +++ b/lttng-context-perf-counters.c
> @@ -119,7 +119,7 @@ void lttng_destroy_perf_counter_field(struct lttng_ctx_field
> *field)
> #endif /* #else #if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,10,0)) */
> kfree(field->event_field.name);
> kfree(field->u.perf_counter->attr);
> - kfree(events);
> + lttng_kvfree(events);
> kfree(field->u.perf_counter);
> }
>
> @@ -237,7 +237,7 @@ int lttng_add_perf_counter_to_ctx(uint32_t type,
> int ret;
> char *name_alloc;
>
> - events = kzalloc(num_possible_cpus() * sizeof(*events), GFP_KERNEL);
> + events = lttng_kvzalloc(num_possible_cpus() * sizeof(*events), GFP_KERNEL);
> if (!events)
> return -ENOMEM;
>
> @@ -372,6 +372,6 @@ name_alloc_error:
> error_alloc_perf_field:
> kfree(attr);
> error_attr:
> - kfree(events);
> + lttng_kvfree(events);
> return ret;
> }
> diff --git a/lttng-context.c b/lttng-context.c
> index 406f479..544e95f 100644
> --- a/lttng-context.c
> +++ b/lttng-context.c
> @@ -95,12 +95,12 @@ struct lttng_ctx_field *lttng_append_context(struct
> lttng_ctx **ctx_p)
> struct lttng_ctx_field *new_fields;
>
> ctx->allocated_fields = max_t(size_t, 1, 2 * ctx->allocated_fields);
> - new_fields = kzalloc(ctx->allocated_fields * sizeof(struct lttng_ctx_field),
> GFP_KERNEL);
> + new_fields = lttng_kvzalloc(ctx->allocated_fields * sizeof(struct
> lttng_ctx_field), GFP_KERNEL);
> if (!new_fields)
> return NULL;
> if (ctx->fields)
> memcpy(new_fields, ctx->fields, sizeof(*ctx->fields) * ctx->nr_fields);
> - kfree(ctx->fields);
> + lttng_kvfree(ctx->fields);
> ctx->fields = new_fields;
> }
> field = &ctx->fields[ctx->nr_fields];
> @@ -240,7 +240,7 @@ void lttng_destroy_context(struct lttng_ctx *ctx)
> if (ctx->fields[i].destroy)
> ctx->fields[i].destroy(&ctx->fields[i]);
> }
> - kfree(ctx->fields);
> + lttng_kvfree(ctx->fields);
> kfree(ctx);
> }
>
> diff --git a/lttng-events.c b/lttng-events.c
> index 6aa994c..21c4113 100644
> --- a/lttng-events.c
> +++ b/lttng-events.c
> @@ -132,7 +132,7 @@ struct lttng_session *lttng_session_create(void)
> int i;
>
> mutex_lock(&sessions_mutex);
> - session = kzalloc(sizeof(struct lttng_session), GFP_KERNEL);
> + session = lttng_kvzalloc(sizeof(struct lttng_session), GFP_KERNEL);
> if (!session)
> goto err;
> INIT_LIST_HEAD(&session->chan);
> @@ -163,7 +163,7 @@ struct lttng_session *lttng_session_create(void)
> err_free_cache:
> kfree(metadata_cache);
> err_free_session:
> - kfree(session);
> + lttng_kvfree(session);
> err:
> mutex_unlock(&sessions_mutex);
> return NULL;
> @@ -212,7 +212,7 @@ void lttng_session_destroy(struct lttng_session *session)
> kref_put(&session->metadata_cache->refcount, metadata_cache_destroy);
> list_del(&session->list);
> mutex_unlock(&sessions_mutex);
> - kfree(session);
> + lttng_kvfree(session);
> }
>
> int lttng_session_statedump(struct lttng_session *session)
> diff --git a/wrapper/vmalloc.h b/wrapper/vmalloc.h
> index 2332439..2dd06cb 100644
> --- a/wrapper/vmalloc.h
> +++ b/wrapper/vmalloc.h
> @@ -25,6 +25,9 @@
> * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
> */
>
> +#include <linux/version.h>
> +#include <linux/vmalloc.h>
> +
> #ifdef CONFIG_KALLSYMS
>
> #include <linux/kallsyms.h>
> @@ -51,8 +54,6 @@ void wrapper_vmalloc_sync_all(void)
> }
> #else
>
> -#include <linux/vmalloc.h>
> -
> static inline
> void wrapper_vmalloc_sync_all(void)
> {
> @@ -60,4 +61,168 @@ void wrapper_vmalloc_sync_all(void)
> }
> #endif
>
> +#if (LINUX_VERSION_CODE >= KERNEL_VERSION(4,12,0))
> +static inline
> +void *lttng_kvmalloc_node(unsigned long size, gfp_t flags, int node)
> +{
> + void *ret;
> +
> + ret = kvmalloc_node(size, flags, node);
> + if (is_vmalloc_addr(ret)) {
> + /*
> + * Make sure we don't trigger recursive page faults in the
> + * tracing fast path.
> + */
> + wrapper_vmalloc_sync_all();
> + }
> + return ret;
> +}
> +
> +static inline
> +void *lttng_kvzalloc_node(unsigned long size, gfp_t flags, int node)
> +{
> + return lttng_kvmalloc_node(size, flags | __GFP_ZERO, node);
> +}
> +
> +static inline
> +void *lttng_kvmalloc(unsigned long size, gfp_t flags)
> +{
> + return lttng_kvmalloc_node(size, flags, NUMA_NO_NODE);
> +}
> +
> +static inline
> +void *lttng_kvzalloc(unsigned long size, gfp_t flags)
> +{
> + return lttng_kvzalloc_node(size, flags, NUMA_NO_NODE);
> +}
> +
> +static inline
> +void lttng_kvfree(const void *addr)
> +{
> + kvfree(addr);
> +}
> +
> +#else
> +
> +#include <linux/slab.h>
> +#include <linux/mm.h>
> +
> +/*
> + * kallsyms wrapper of __vmalloc_node with a fallback to kmalloc_node.
> + */
> +static inline
> +void *__lttng_vmalloc_node_fallback(unsigned long size, unsigned long align,
> + gfp_t gfp_mask, pgprot_t prot, int node, void *caller)
> +{
> + void *ret;
> +
> +#ifdef CONFIG_KALLSYMS
> + /*
> + * If we have KALLSYMS, get * __vmalloc_node which is not exported.
> + */
> + void *(*lttng__vmalloc_node)(unsigned long size, unsigned long align,
> + gfp_t gfp_mask, pgprot_t prot, int node, void *caller);
> +
> + lttng__vmalloc_node = (void *) kallsyms_lookup_funcptr("__vmalloc_node");
> + ret = lttng__vmalloc_node(size, align, gfp_mask, prot, node, caller);
> +#else
> + /*
> + * If we don't have KALLSYMS, fallback to kmalloc_node.
> + */
> + ret = kmalloc_node(size, flags, node);
> +#endif
> +
> + return ret;
> +}
> +
> +/**
> + * lttng_kvmalloc_node - attempt to allocate physically contiguous memory, but
> upon
> + * failure, fall back to non-contiguous (vmalloc) allocation.
> + * @size: size of the request.
> + * @flags: gfp mask for the allocation - must be compatible with GFP_KERNEL.
> + *
> + * Uses kmalloc to get the memory but if the allocation fails then falls back
> + * to the vmalloc allocator. Use lttng_kvfree to free the memory.
> + *
> + * Reclaim modifiers - __GFP_NORETRY, __GFP_REPEAT and __GFP_NOFAIL are not
> supported
> + */
> +static inline
> +void *lttng_kvmalloc_node(unsigned long size, gfp_t flags, int node)
> +{
> + void *ret;
> +
> + /*
> + * vmalloc uses GFP_KERNEL for some internal allocations (e.g page tables)
> + * so the given set of flags has to be compatible.
> + */
> + WARN_ON_ONCE((flags & GFP_KERNEL) != GFP_KERNEL);
> +
> + /*
> + * If the allocation fits in a single page, do not fallback.
> + */
> + if (size <= PAGE_SIZE) {
> + return kmalloc_node(size, flags, node);
> + }
> +
> + /*
> + * Make sure that larger requests are not too disruptive - no OOM
> + * killer and no allocation failure warnings as we have a fallback
> + */
> + ret = kmalloc_node(size, flags | __GFP_NOWARN | __GFP_NORETRY, node);
> + if (!ret) {
> + if (node == NUMA_NO_NODE) {
> + /*
> + * If no node was specified, use __vmalloc which is
> + * always exported.
> + */
> + ret = __vmalloc(size, flags | __GFP_HIGHMEM, PAGE_KERNEL);
> + } else {
> + /*
> + * Otherwise, we need to select a node but __vmalloc_node
> + * is not exported, use this fallback wrapper which uses
> + * kallsyms if available or falls back to kmalloc_node.
> + */
> + ret = __lttng_vmalloc_node_fallback(size, 1,
> + flags | __GFP_HIGHMEM, PAGE_KERNEL, node,
> + __builtin_return_address(0));
I try to never use __builtin_return_address(0) directly. It's buggy on powerpc32,
and causes stack corruption.
The kernel exposes _RET_IP_ nowadays. Can we use it instead ?
And I'm tempted to do a trick similar to lttng-ust there, e.g.:
/*
* Use of __builtin_return_address(0) sometimes seems to cause stack
* corruption on 32-bit PowerPC. Disable this feature on that
* architecture for now by always using the NULL value for the ip
* context.
*/
#if defined(__PPC__) && !defined(__PPC64__)
#define LTTNG_UST_CALLER_IP() NULL
#else /* #if defined(__PPC__) && !defined(__PPC64__) */
#define LTTNG_UST_CALLER_IP() __builtin_return_address(0)
#endif /* #else #if defined(__PPC__) && !defined(__PPC64__) */
Thanks,
Mathieu
> + }
> +
> + /*
> + * Make sure we don't trigger recursive page faults in the
> + * tracing fast path.
> + */
> + wrapper_vmalloc_sync_all();
> + }
> + return ret;
> +}
> +
> +static inline
> +void *lttng_kvzalloc_node(unsigned long size, gfp_t flags, int node)
> +{
> + return lttng_kvmalloc_node(size, flags | __GFP_ZERO, node);
> +}
> +
> +static inline
> +void *lttng_kvmalloc(unsigned long size, gfp_t flags)
> +{
> + return lttng_kvmalloc_node(size, flags, NUMA_NO_NODE);
> +}
> +
> +static inline
> +void *lttng_kvzalloc(unsigned long size, gfp_t flags)
> +{
> + return lttng_kvzalloc_node(size, flags, NUMA_NO_NODE);
> +}
> +
> +static inline
> +void lttng_kvfree(const void *addr)
> +{
> + if (is_vmalloc_addr(addr)) {
> + vfree(addr);
> + } else {
> + kfree(addr);
> + }
> +}
> +#endif
> +
> #endif /* _LTTNG_WRAPPER_VMALLOC_H */
> --
> 2.7.4
--
Mathieu Desnoyers
EfficiOS Inc.
http://www.efficios.com
More information about the lttng-dev
mailing list