Re: [dpdk-dev] [RFC 08/11] malloc: allow adding memory to named heaps

[Alejandro Lucero <alejandro.lucero@netronome.com>]

On Fri, Jul 6, 2018 at 2:17 PM, Anatoly Burakov <anatoly.burakov@intel.com>
wrote:

> Add an API to add externally allocated memory to malloc heap. The
> memory will be stored in memseg lists like regular DPDK memory.
> Multiple segments are allowed within a heap. If IOVA table is
> not provided, IOVA addresses are filled in with RTE_BAD_IOVA.
>
> Signed-off-by: Anatoly Burakov <anatoly.burakov@intel.com>
> ---
>  lib/librte_eal/common/include/rte_malloc.h | 44 ++++++++++++++
>  lib/librte_eal/common/malloc_heap.c        | 70 ++++++++++++++++++++++
>  lib/librte_eal/common/malloc_heap.h        |  4 ++
>  lib/librte_eal/common/rte_malloc.c         | 39 ++++++++++++
>  lib/librte_eal/rte_eal_version.map         |  1 +
>  5 files changed, 158 insertions(+)
>
> diff --git a/lib/librte_eal/common/include/rte_malloc.h
> b/lib/librte_eal/common/include/rte_malloc.h
> index fa6de073a..5f933993b 100644
> --- a/lib/librte_eal/common/include/rte_malloc.h
> +++ b/lib/librte_eal/common/include/rte_malloc.h
> @@ -274,6 +274,50 @@ rte_free(void *ptr);
>  int __rte_experimental
>  rte_malloc_heap_create(const char *heap_name);
>
> +/**
> + * Add more memory to heap with specified name.
> + *
> + * @note Concurrently adding memory to or removing memory from different
> heaps
> + *   is not safe.
> + *
> + * @note This function does not need to be called in multiple processes,
> as
> + *   multiprocess synchronization will happen automatically as far as
> heap data
> + *   is concerned. However, before accessing pointers to memory in this
> heap, it
> + *   is responsibility of the user to ensure that the heap memory is
> accessible
> + *   in all processes.
> + *
> + * @note Memory must be previously allocated for DPDK to be able to use
> it as a
> + *   malloc heap. Failing to do so will result in undefined behavior, up
> to and
> + *   including crashes.
> + *
> + * @note Adding memory to heap may fail in multiple processes scenario, as
> + *   attaching to ``rte_fbarray`` structures may not always be successful
> in
> + *   secondary processes.
> + *
> + * @param heap_name
> + *   Name of the heap to create.
>

Name of the heap to add memory to.


> + * @param va_addr
> + *   Start of virtual area to add to the heap.
> + * @param len
> + *   Length of virtual area to add to the heap.
> + * @param iova_addrs
> + *   Array of page IOVA addresses corresponding to each page in this
> memory
> + *   area. Can be NULL, in which case page IOVA addresses will be set to
> + *   RTE_BAD_IOVA.
> + * @param n_pages
> + *   Number of elements in the iova_addrs array. Must be zero of
> ``iova_addrs``
> + *   is NULL.
> + * @param page_sz
> + *   Page size of the underlying memory.
> + *
> + * @return
> + *   - 0 on successful creation.
> + *   - -1 on error.
> + */
> +int __rte_experimental
> +rte_malloc_heap_add_memory(const char *heap_name, void *va_addr, size_t
> len,
> +               rte_iova_t iova_addrs[], unsigned int n_pages, size_t
> page_sz);
> +
>  /**
>   * If malloc debug is enabled, check a memory block for header
>   * and trailer markers to indicate that all is well with the block.
> diff --git a/lib/librte_eal/common/malloc_heap.c
> b/lib/librte_eal/common/malloc_heap.c
> index f5d103626..29446cac9 100644
> --- a/lib/librte_eal/common/malloc_heap.c
> +++ b/lib/librte_eal/common/malloc_heap.c
> @@ -892,6 +892,76 @@ malloc_heap_dump(struct malloc_heap *heap, FILE *f)
>         rte_spinlock_unlock(&heap->lock);
>  }
>
> +int
> +malloc_heap_add_external_memory(struct malloc_heap *heap, void *va_addr,
> +               rte_iova_t iova_addrs[], unsigned int n_pages, size_t
> page_sz)
> +{
> +       struct rte_mem_config *mcfg = rte_eal_get_configuration()->m
> em_config;
> +       char fbarray_name[RTE_FBARRAY_NAME_LEN];
> +       struct rte_memseg_list *msl = NULL;
> +       struct rte_fbarray *arr;
> +       size_t seg_len = n_pages * page_sz;
> +       unsigned int i;
> +
> +       /* first, find a free memseg list */
> +       for (i = 0; i < RTE_MAX_MEMSEG_LISTS; i++) {
> +               struct rte_memseg_list *tmp = &mcfg->memsegs[i];
> +               if (tmp->base_va == NULL) {
> +                       msl = tmp;
> +                       break;
> +               }
> +       }
> +       if (msl == NULL) {
> +               RTE_LOG(ERR, EAL, "Couldn't find empty memseg list\n");
> +               rte_errno = ENOSPC;
> +               return -1;
> +       }
> +
> +       snprintf(fbarray_name, sizeof(fbarray_name) - 1, "%s_%p",
> +                       heap->name, va_addr);
> +
> +       /* create the backing fbarray */
> +       if (rte_fbarray_init(&msl->memseg_arr, fbarray_name, n_pages,
> +                       sizeof(struct rte_memseg)) < 0) {
> +               RTE_LOG(ERR, EAL, "Couldn't create fbarray backing the
> memseg list\n");
> +               return -1;
> +       }
> +       arr = &msl->memseg_arr;
> +
> +       /* fbarray created, fill it up */
> +       for (i = 0; i < n_pages; i++) {
> +               struct rte_memseg *ms;
> +
> +               rte_fbarray_set_used(arr, i);
> +               ms = rte_fbarray_get(arr, i);
> +               ms->addr = RTE_PTR_ADD(va_addr, n_pages * page_sz);
> +               ms->iova = iova_addrs == NULL ? RTE_BAD_IOVA :
> iova_addrs[i];
> +               ms->hugepage_sz = page_sz;
> +               ms->len = page_sz;
> +               ms->nchannel = rte_memory_get_nchannel();
> +               ms->nrank = rte_memory_get_nrank();
> +               ms->socket_id = -1; /* invalid socket ID */
> +       }
> +
> +       /* set up the memseg list */
> +       msl->base_va = va_addr;
> +       msl->page_sz = page_sz;
> +       msl->socket_id = -1; /* invalid socket ID */
> +       msl->version = 0;
> +       msl->external = true;
> +
> +       /* now, add newly minted memory to the malloc heap */
> +       malloc_heap_add_memory(heap, msl, va_addr, seg_len);
> +
> +       heap->total_size += seg_len;
> +
> +       /* all done! */
> +       RTE_LOG(DEBUG, EAL, "Added segment for heap %s starting at %p\n",
> +                       heap->name, va_addr);
> +
> +       return 0;
> +}
> +
>  int
>  malloc_heap_create(struct malloc_heap *heap, const char *heap_name)
>  {
> diff --git a/lib/librte_eal/common/malloc_heap.h
> b/lib/librte_eal/common/malloc_heap.h
> index aa819ef65..3be4656d0 100644
> --- a/lib/librte_eal/common/malloc_heap.h
> +++ b/lib/librte_eal/common/malloc_heap.h
> @@ -38,6 +38,10 @@ malloc_heap_alloc_on_heap_id(const char *type, size_t
> size,
>  int
>  malloc_heap_create(struct malloc_heap *heap, const char *heap_name);
>
> +int
> +malloc_heap_add_external_memory(struct malloc_heap *heap, void *va_addr,
> +               rte_iova_t iova_addrs[], unsigned int n_pages, size_t
> page_sz);
> +
>  int
>  malloc_heap_find_named_heap_idx(const char *name);
>
> diff --git a/lib/librte_eal/common/rte_malloc.c
> b/lib/librte_eal/common/rte_malloc.c
> index e000dc5b7..db0f604ad 100644
> --- a/lib/librte_eal/common/rte_malloc.c
> +++ b/lib/librte_eal/common/rte_malloc.c
> @@ -274,6 +274,45 @@ rte_malloc_virt2iova(const void *addr)
>         return ms->iova + RTE_PTR_DIFF(addr, ms->addr);
>  }
>
> +int
> +rte_malloc_heap_add_memory(const char *heap_name, void *va_addr, size_t
> len,
> +               rte_iova_t iova_addrs[], unsigned int n_pages, size_t
> page_sz)
> +{
> +       struct malloc_heap *heap = NULL;
> +       unsigned int n;
> +       int ret;
> +
> +       /* iova_addrs is allowed to be NULL */
> +       if (heap_name == NULL || va_addr == NULL ||
> +                       /* n_pages can be 0 if iova_addrs is NULL */
> +                       ((iova_addrs != NULL) == (n_pages == 0)) ||
> +                       page_sz == 0 || !rte_is_power_of_2(page_sz) ||
> +                       strnlen(heap_name, RTE_HEAP_NAME_MAX_LEN) == 0 ||
> +                       strnlen(heap_name, RTE_HEAP_NAME_MAX_LEN) ==
> +                               RTE_HEAP_NAME_MAX_LEN) {
> +               rte_errno = EINVAL;
> +               return -1;
> +       }
> +       /* find our heap */
> +       heap = malloc_heap_find_named_heap(heap_name);
> +       if (heap == NULL) {
> +               rte_errno = EINVAL;
> +               return -1;
> +       }
> +       n = len / page_sz;
> +       if (n != n_pages && iova_addrs != NULL) {
> +               rte_errno = EINVAL;
> +               return -1;
> +       }
> +
> +       rte_spinlock_lock(&heap->lock);
> +       ret = malloc_heap_add_external_memory(heap, va_addr, iova_addrs,
> n,
> +                       page_sz);
> +       rte_spinlock_unlock(&heap->lock);
> +
> +       return ret;
> +}
> +
>  int
>  rte_malloc_heap_create(const char *heap_name)
>  {
> diff --git a/lib/librte_eal/rte_eal_version.map
> b/lib/librte_eal/rte_eal_version.map
> index f3c375156..6290cc910 100644
> --- a/lib/librte_eal/rte_eal_version.map
> +++ b/lib/librte_eal/rte_eal_version.map
> @@ -280,6 +280,7 @@ EXPERIMENTAL {
>         rte_malloc_dump_heaps;
>         rte_malloc_from_heap;
>         rte_malloc_get_stats_from_heap;
> +       rte_malloc_heap_add_memory;
>         rte_malloc_heap_create;
>         rte_mem_alloc_validator_register;
>         rte_mem_alloc_validator_unregister;
> --
> 2.17.1
>