Show a cover letter.

GET /api/covers/45125/?format=api
Content-Type: application/json
Vary: Accept

    "id": 45125,
    "url": "",
    "web_url": "",
    "project": {
        "id": 1,
        "url": "",
        "name": "DPDK",
        "link_name": "dpdk",
        "list_id": "",
        "list_email": "",
        "web_url": "",
        "scm_url": "git://",
        "webscm_url": ""
    "msgid": "<>",
    "date": "2018-09-21T16:13:49",
    "name": "[v4,00/20] Support externally allocated memory in DPDK",
    "submitter": {
        "id": 4,
        "url": "",
        "name": "Burakov, Anatoly",
        "email": ""
    "mbox": "",
    "series": [
            "id": 1446,
            "url": "",
            "web_url": "",
            "date": "2018-09-21T16:13:49",
            "name": "Support externally allocated memory in DPDK",
            "version": 4,
            "mbox": ""
    "comments": "",
    "headers": {
        "Return-Path": "<>",
        "References": "<>",
        "X-Mailman-Version": "2.1.15",
        "X-IronPort-AV": "E=Sophos;i=\"5.54,285,1534834800\"; d=\"scan'208\";a=\"75173450\"",
        "From": "Anatoly Burakov <>",
        "List-Post": "<>",
        "List-Help": "<>",
        "X-ExtLoop1": "1",
        "X-BeenThere": "",
        "Message-Id": "<>",
        "X-Original-To": "",
        "Received": [
            "from [] (localhost [])\n\tby (Postfix) with ESMTP id E37A25F1C;\n\tFri, 21 Sep 2018 18:14:15 +0200 (CEST)",
            "from ( [])\n\tby (Postfix) with ESMTP id CE55B5F14\n\tfor <>; Fri, 21 Sep 2018 18:14:14 +0200 (CEST)",
            "from ([])\n\tby with ESMTP/TLS/DHE-RSA-AES256-GCM-SHA384;\n\t21 Sep 2018 09:14:13 -0700",
            "from ([])\n\tby with ESMTP; 21 Sep 2018 09:14:10 -0700",
            "from (\n\t[])\n\tby (8.14.3/8.13.6/MailSET/Hub) with ESMTP id\n\tw8LGE9Ow029177; Fri, 21 Sep 2018 17:14:09 +0100",
            "from (localhost [])\n\tby with ESMTP id w8LGE9Sd002758;\n\tFri, 21 Sep 2018 17:14:09 +0100",
            "(from aburakov@localhost)\n\tby with LOCAL id w8LGE9ok002752;\n\tFri, 21 Sep 2018 17:14:09 +0100"
        "X-Mailer": "git-send-email",
        "Sender": "\"dev\" <>",
        "Subject": "[dpdk-dev] [PATCH v4 00/20] Support externally allocated memory in\n\tDPDK",
        "Precedence": "list",
        "Date": "Fri, 21 Sep 2018 17:13:49 +0100",
        "List-Archive": "<>",
        "X-Amp-File-Uploaded": "False",
        "List-Subscribe": "<>,\n\t<>",
        "Cc": ",,\n\,,\n\,,\n\,,\n\,,\n\,,,\n\,,",
        "List-Id": "DPDK patches and discussions <>",
        "Delivered-To": "",
        "In-Reply-To": "<>",
        "List-Unsubscribe": "<>,\n\t<>",
        "To": "",
        "X-Amp-Result": "SKIPPED(no attachment in message)",
        "Errors-To": ""
    "content": "This is a proposal to enable using externally allocated memory\nin DPDK.\n\nIn a nutshell, here is what is being done here:\n\n- Index internal malloc heaps by NUMA node index, rather than NUMA\n  node itself (external heaps will have ID's in order of creation)\n- Add identifier string to malloc heap, to uniquely identify it\n  - Each new heap will receive a unique socket ID that will be used by\n    allocator to decide from which heap (internal or external) to\n    allocate requested amount of memory\n- Allow creating named heaps and add/remove memory to/from those heaps\n- Allocate memseg lists at runtime, to keep track of IOVA addresses\n  of externally allocated memory\n  - If IOVA addresses aren't provided, use RTE_BAD_IOVA\n- Allow malloc and memzones to allocate from external heaps\n- Allow other data structures to allocate from externall heaps\n\nThe responsibility to ensure memory is accessible before using it is\non the shoulders of the user - there is no checking done with regards\nto validity of the memory (nor could there be...).\n\nThe general approach is to create heap and add memory into it. For any\nother process wishing to use the same memory, said memory must first\nbe attached (otherwise some things will not work).\n\nA design decision was made to make multiprocess synchronization a\nmanual process. Due to underlying issues with attaching to fbarrays in\nsecondary processes, this design was deemed to be better because we\ndon't want to fail to create external heap in the primary because\nsomething in the secondary has failed when in fact we may not eve have\nwanted this memory to be accessible in the secondary in the first\nplace.\n\nUsing external memory in multiprocess is *hard*, because not only\nmemory space needs to be preallocated, but it also needs to be attached\nin each process to allow other processes to access the page table. The\nattach API call may or may not succeed, depending on memory layout, for\nreasons similar to other multiprocess failures. This is treated as a\n\"known issue\" for this release.\n\nCreating and destroying heaps is currently restricted to primary\nprocesses, because we need to keep track of all socket ID's we've ever\nused to prevent their reuse, and obviously different processes would\nhave kept different socket ID counters, and it isn't important enough\nto put into shared memory. This means that secondary processes will\nnot be able to create new heaps. If this use case is important\nenough, we can put the max socket ID into shared memory, or allow\nsocket ID reuse (which i do not think is a good idea because it has\nthe potential to make things harder to debug).\n\nv4 -> v3 changes:\n- Dropped sample application in favor of new testpmd flag\n- Added new flag to testpmd, with four options of mempool allocation\n- Added new API to check if a socket ID belongs to an external heap\n- Adjusted malloc and mempool code to not make any assumptions about\n  IOVA-contiguousness when dealing with externally allocated memory\n\nv3 -> v2 changes:\n- Rebase on top of latest master\n- Clarifications added to mempool code as per Andrew Rynchenko's\n  comments\n\nv2 -> v1 changes:\n- Fixed NULL dereference on heap socket ID lookup\n- Fixed memseg offset calculation on adding memory to heap\n- Improved unit test to test for above bugfixes\n- Restricted heap creation to primary processes only\n- Added sample application\n- Added documentation\n\nRFC -> v1 changes:\n- Removed the \"named heaps\" API, allocate using fake socket ID instead\n- Added multiprocess support\n- Everything is now thread-safe\n- Numerous bugfixes and API improvements\n\nAnatoly Burakov (20):\n  mem: add length to memseg list\n  mem: allow memseg lists to be marked as external\n  malloc: index heaps using heap ID rather than NUMA node\n  mem: do not check for invalid socket ID\n  flow_classify: do not check for invalid socket ID\n  pipeline: do not check for invalid socket ID\n  sched: do not check for invalid socket ID\n  malloc: add name to malloc heaps\n  malloc: add function to query socket ID of named heap\n  malloc: add function to check if socket is external\n  malloc: allow creating malloc heaps\n  malloc: allow destroying heaps\n  malloc: allow adding memory to named heaps\n  malloc: allow removing memory from named heaps\n  malloc: allow attaching to external memory chunks\n  malloc: allow detaching from external memory\n  test: add unit tests for external memory support\n  app/testpmd: add support for external memory\n  doc: add external memory feature to the release notes\n  doc: add external memory feature to programmer's guide\n\n app/test-pmd/config.c                         |  21 +-\n app/test-pmd/parameters.c                     |  23 +-\n app/test-pmd/testpmd.c                        | 337 +++++++++++++-\n app/test-pmd/testpmd.h                        |  13 +-\n config/common_base                            |   1 +\n config/rte_config.h                           |   1 +\n .../prog_guide/env_abstraction_layer.rst      |  38 ++\n doc/guides/rel_notes/deprecation.rst          |  15 -\n doc/guides/rel_notes/release_18_11.rst        |  24 +-\n doc/guides/testpmd_app_ug/run_app.rst         |  12 +\n drivers/bus/fslmc/fslmc_vfio.c                |   7 +-\n drivers/bus/pci/linux/pci.c                   |   2 +-\n drivers/net/mlx4/mlx4_mr.c                    |   3 +\n drivers/net/mlx5/mlx5.c                       |   5 +-\n drivers/net/mlx5/mlx5_mr.c                    |   3 +\n drivers/net/virtio/virtio_user/vhost_kernel.c |   5 +-\n lib/librte_eal/bsdapp/eal/Makefile            |   2 +-\n lib/librte_eal/bsdapp/eal/eal.c               |   3 +\n lib/librte_eal/bsdapp/eal/eal_memory.c        |   9 +-\n lib/librte_eal/common/eal_common_memory.c     |   8 +-\n lib/librte_eal/common/eal_common_memzone.c    |   8 +-\n .../common/include/rte_eal_memconfig.h        |   6 +-\n lib/librte_eal/common/include/rte_malloc.h    | 198 +++++++++\n .../common/include/rte_malloc_heap.h          |   3 +\n lib/librte_eal/common/include/rte_memory.h    |   9 +\n lib/librte_eal/common/malloc_elem.c           |  10 +-\n lib/librte_eal/common/malloc_heap.c           | 300 +++++++++++--\n lib/librte_eal/common/malloc_heap.h           |  17 +\n lib/librte_eal/common/rte_malloc.c            | 420 +++++++++++++++++-\n lib/librte_eal/linuxapp/eal/Makefile          |   2 +-\n lib/librte_eal/linuxapp/eal/eal.c             |  10 +-\n lib/librte_eal/linuxapp/eal/eal_memalloc.c    |  12 +-\n lib/librte_eal/linuxapp/eal/eal_memory.c      |   4 +-\n lib/librte_eal/linuxapp/eal/eal_vfio.c        |  17 +-\n lib/librte_eal/                    |   2 +-\n lib/librte_eal/            |   8 +\n lib/librte_flow_classify/rte_flow_classify.c  |   3 +-\n lib/librte_mempool/rte_mempool.c              |  57 ++-\n lib/librte_pipeline/rte_pipeline.c            |   3 +-\n lib/librte_sched/rte_sched.c                  |   2 +-\n test/test/Makefile                            |   1 +\n test/test/                    |  14 +-\n test/test/                         |   1 +\n test/test/test_external_mem.c                 | 389 ++++++++++++++++\n test/test/test_malloc.c                       |   3 +\n test/test/test_memzone.c                      |   3 +\n 46 files changed, 1897 insertions(+), 137 deletions(-)\n create mode 100644 test/test/test_external_mem.c"