Patch Detail
get:
Show a patch.
patch:
Update a patch.
put:
Update a patch.
GET /api/patches/87282/?format=api
{ "id": 87282, "url": "https://patches.dpdk.org/api/patches/87282/?format=api", "web_url": "https://patches.dpdk.org/project/dpdk/patch/20210126081746.1580627-4-feifei.wang2@arm.com/", "project": { "id": 1, "url": "https://patches.dpdk.org/api/projects/1/?format=api", "name": "DPDK", "link_name": "dpdk", "list_id": "dev.dpdk.org", "list_email": "dev@dpdk.org", "web_url": "http://core.dpdk.org", "scm_url": "git://dpdk.org/dpdk", "webscm_url": "http://git.dpdk.org/dpdk", "list_archive_url": "https://inbox.dpdk.org/dev", "list_archive_url_format": "https://inbox.dpdk.org/dev/{}", "commit_url_format": "" }, "msgid": "<20210126081746.1580627-4-feifei.wang2@arm.com>", "list_archive_url": "https://inbox.dpdk.org/dev/20210126081746.1580627-4-feifei.wang2@arm.com", "date": "2021-01-26T08:17:46", "name": "[v2,3/3] ring: refactor ring library", "commit_ref": null, "pull_url": null, "state": "superseded", "archived": true, "hash": "88cb831f678ad6cf6b31d36cb5cc3dcf26bda81c", "submitter": { "id": 1771, "url": "https://patches.dpdk.org/api/people/1771/?format=api", "name": "Feifei Wang", "email": "feifei.wang2@arm.com" }, "delegate": { "id": 24651, "url": "https://patches.dpdk.org/api/users/24651/?format=api", "username": "dmarchand", "first_name": "David", "last_name": "Marchand", "email": "david.marchand@redhat.com" }, "mbox": "https://patches.dpdk.org/project/dpdk/patch/20210126081746.1580627-4-feifei.wang2@arm.com/mbox/", "series": [ { "id": 14944, "url": "https://patches.dpdk.org/api/series/14944/?format=api", "web_url": "https://patches.dpdk.org/project/dpdk/list/?series=14944", "date": "2021-01-26T08:17:43", "name": "refactoring ring library", "version": 2, "mbox": "https://patches.dpdk.org/series/14944/mbox/" } ], "comments": "https://patches.dpdk.org/api/patches/87282/comments/", "check": "fail", "checks": "https://patches.dpdk.org/api/patches/87282/checks/", "tags": {}, "related": [], "headers": { "Return-Path": "<dev-bounces@dpdk.org>", "X-Original-To": "patchwork@inbox.dpdk.org", "Delivered-To": "patchwork@inbox.dpdk.org", "Received": [ "from mails.dpdk.org (mails.dpdk.org [217.70.189.124])\n\tby inbox.dpdk.org (Postfix) with ESMTP id 7A069A052A;\n\tTue, 26 Jan 2021 09:18:24 +0100 (CET)", "from [217.70.189.124] (localhost [127.0.0.1])\n\tby mails.dpdk.org (Postfix) with ESMTP id 87B971412A3;\n\tTue, 26 Jan 2021 09:18:07 +0100 (CET)", "from foss.arm.com (foss.arm.com [217.140.110.172])\n by mails.dpdk.org (Postfix) with ESMTP id A0D8D141286\n for <dev@dpdk.org>; Tue, 26 Jan 2021 09:18:05 +0100 (CET)", "from usa-sjc-imap-foss1.foss.arm.com (unknown [10.121.207.14])\n by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 085B8D6E;\n Tue, 26 Jan 2021 00:18:05 -0800 (PST)", "from net-x86-dell-8268.shanghai.arm.com\n (net-x86-dell-8268.shanghai.arm.com [10.169.210.127])\n by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPA id 15BA33F68F;\n Tue, 26 Jan 2021 00:18:01 -0800 (PST)" ], "From": "Feifei Wang <feifei.wang2@arm.com>", "To": "Honnappa Nagarahalli <honnappa.nagarahalli@arm.com>,\n Konstantin Ananyev <konstantin.ananyev@intel.com>", "Cc": "dev@dpdk.org, nd@arm.com, Feifei Wang <feifei.wang2@arm.com>,\n Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>,\n Ruifeng Wang <ruifeng.wang@arm.com>", "Date": "Tue, 26 Jan 2021 16:17:46 +0800", "Message-Id": "<20210126081746.1580627-4-feifei.wang2@arm.com>", "X-Mailer": "git-send-email 2.25.1", "In-Reply-To": "<20210126081746.1580627-1-feifei.wang2@arm.com>", "References": "<20201221111359.22013-1-feifei.wang2@arm.com>\n <20210126081746.1580627-1-feifei.wang2@arm.com>", "MIME-Version": "1.0", "Content-Transfer-Encoding": "8bit", "Subject": "[dpdk-dev] [PATCH v2 3/3] ring: refactor ring library", "X-BeenThere": "dev@dpdk.org", "X-Mailman-Version": "2.1.29", "Precedence": "list", "List-Id": "DPDK patches and discussions <dev.dpdk.org>", "List-Unsubscribe": "<https://mails.dpdk.org/options/dev>,\n <mailto:dev-request@dpdk.org?subject=unsubscribe>", "List-Archive": "<http://mails.dpdk.org/archives/dev/>", "List-Post": "<mailto:dev@dpdk.org>", "List-Help": "<mailto:dev-request@dpdk.org?subject=help>", "List-Subscribe": "<https://mails.dpdk.org/listinfo/dev>,\n <mailto:dev-request@dpdk.org?subject=subscribe>", "Errors-To": "dev-bounces@dpdk.org", "Sender": "\"dev\" <dev-bounces@dpdk.org>" }, "content": "For legacy modes, rename ring_generic/c11 to ring_generic/c11_pvt.\nFurthermore, add new file ring_elem_pvt.h which includes ring_do_eq/deq\nand ring element copy/delete APIs.\n\nFor other modes, rename xx_c11_mem to xx_elem_pvt. Move all private APIs\ninto these new header files.\n\nFinally, the external APIs and internal APIs will be separated from each\nother. This can remind users not to use internal APIs and make ring\nlibrary easier to maintain.\n\nSuggested-by: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>\nSigned-off-by: Feifei Wang <feifei.wang2@arm.com>\nReviewed-by: Honnappa Nagarahalli <Honnappa.Nagarahalli@arm.com>\nReviewed-by: Ruifeng Wang <ruifeng.wang@arm.com>\n---\n lib/librte_ring/meson.build | 11 +-\n ...{rte_ring_c11_mem.h => rte_ring_c11_pvt.h} | 6 +-\n lib/librte_ring/rte_ring_elem.h | 374 +----------------\n lib/librte_ring/rte_ring_elem_pvt.h | 385 ++++++++++++++++++\n ..._ring_generic.h => rte_ring_generic_pvt.h} | 6 +-\n lib/librte_ring/rte_ring_hts.h | 84 +---\n ..._hts_c11_mem.h => rte_ring_hts_elem_pvt.h} | 88 +++-\n lib/librte_ring/rte_ring_peek.h | 71 +---\n ...eek_c11_mem.h => rte_ring_peek_elem_pvt.h} | 75 +++-\n lib/librte_ring/rte_ring_peek_zc.h | 2 +-\n lib/librte_ring/rte_ring_rts.h | 84 +---\n ..._rts_c11_mem.h => rte_ring_rts_elem_pvt.h} | 88 +++-\n 12 files changed, 644 insertions(+), 630 deletions(-)\n rename lib/librte_ring/{rte_ring_c11_mem.h => rte_ring_c11_pvt.h} (98%)\n create mode 100644 lib/librte_ring/rte_ring_elem_pvt.h\n rename lib/librte_ring/{rte_ring_generic.h => rte_ring_generic_pvt.h} (97%)\n rename lib/librte_ring/{rte_ring_hts_c11_mem.h => rte_ring_hts_elem_pvt.h} (60%)\n rename lib/librte_ring/{rte_ring_peek_c11_mem.h => rte_ring_peek_elem_pvt.h} (61%)\n rename lib/librte_ring/{rte_ring_rts_c11_mem.h => rte_ring_rts_elem_pvt.h} (62%)", "diff": "diff --git a/lib/librte_ring/meson.build b/lib/librte_ring/meson.build\nindex 36fdcb6a5..f007893b5 100644\n--- a/lib/librte_ring/meson.build\n+++ b/lib/librte_ring/meson.build\n@@ -5,12 +5,13 @@ sources = files('rte_ring.c')\n headers = files('rte_ring.h',\n \t\t'rte_ring_core.h',\n \t\t'rte_ring_elem.h',\n-\t\t'rte_ring_c11_mem.h',\n-\t\t'rte_ring_generic.h',\n+\t\t'rte_ring_elem_pvt.h',\n+\t\t'rte_ring_c11_pvt.h',\n+\t\t'rte_ring_generic_pvt.h',\n \t\t'rte_ring_hts.h',\n-\t\t'rte_ring_hts_c11_mem.h',\n+\t\t'rte_ring_hts_elem_pvt.h',\n \t\t'rte_ring_peek.h',\n-\t\t'rte_ring_peek_c11_mem.h',\n+\t\t'rte_ring_peek_elem_pvt.h',\n \t\t'rte_ring_peek_zc.h',\n \t\t'rte_ring_rts.h',\n-\t\t'rte_ring_rts_c11_mem.h')\n+\t\t'rte_ring_rts_elem_pvt.h')\ndiff --git a/lib/librte_ring/rte_ring_c11_mem.h b/lib/librte_ring/rte_ring_c11_pvt.h\nsimilarity index 98%\nrename from lib/librte_ring/rte_ring_c11_mem.h\nrename to lib/librte_ring/rte_ring_c11_pvt.h\nindex 7f5eba262..759192f4c 100644\n--- a/lib/librte_ring/rte_ring_c11_mem.h\n+++ b/lib/librte_ring/rte_ring_c11_pvt.h\n@@ -7,8 +7,8 @@\n * Used as BSD-3 Licensed with permission from Kip Macy.\n */\n \n-#ifndef _RTE_RING_C11_MEM_H_\n-#define _RTE_RING_C11_MEM_H_\n+#ifndef _RTE_RING_C11_PVT_H_\n+#define _RTE_RING_C11_PVT_H_\n \n static __rte_always_inline void\n __rte_ring_update_tail(struct rte_ring_headtail *ht, uint32_t old_val,\n@@ -178,4 +178,4 @@ __rte_ring_move_cons_head(struct rte_ring *r, int is_sc,\n \treturn n;\n }\n \n-#endif /* _RTE_RING_C11_MEM_H_ */\n+#endif /* _RTE_RING_C11_PVT_H_ */\ndiff --git a/lib/librte_ring/rte_ring_elem.h b/lib/librte_ring/rte_ring_elem.h\nindex 57344d47d..98c5495e0 100644\n--- a/lib/librte_ring/rte_ring_elem.h\n+++ b/lib/librte_ring/rte_ring_elem.h\n@@ -21,6 +21,7 @@ extern \"C\" {\n #endif\n \n #include <rte_ring_core.h>\n+#include <rte_ring_elem_pvt.h>\n \n /**\n * Calculate the memory size needed for a ring with given element size\n@@ -105,379 +106,6 @@ ssize_t rte_ring_get_memsize_elem(unsigned int esize, unsigned int count);\n struct rte_ring *rte_ring_create_elem(const char *name, unsigned int esize,\n \t\t\tunsigned int count, int socket_id, unsigned int flags);\n \n-static __rte_always_inline void\n-__rte_ring_enqueue_elems_32(struct rte_ring *r, const uint32_t size,\n-\t\tuint32_t idx, const void *obj_table, uint32_t n)\n-{\n-\tunsigned int i;\n-\tuint32_t *ring = (uint32_t *)&r[1];\n-\tconst uint32_t *obj = (const uint32_t *)obj_table;\n-\tif (likely(idx + n < size)) {\n-\t\tfor (i = 0; i < (n & ~0x7); i += 8, idx += 8) {\n-\t\t\tring[idx] = obj[i];\n-\t\t\tring[idx + 1] = obj[i + 1];\n-\t\t\tring[idx + 2] = obj[i + 2];\n-\t\t\tring[idx + 3] = obj[i + 3];\n-\t\t\tring[idx + 4] = obj[i + 4];\n-\t\t\tring[idx + 5] = obj[i + 5];\n-\t\t\tring[idx + 6] = obj[i + 6];\n-\t\t\tring[idx + 7] = obj[i + 7];\n-\t\t}\n-\t\tswitch (n & 0x7) {\n-\t\tcase 7:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 6:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 5:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 4:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 3:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 2:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 1:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\t}\n-\t} else {\n-\t\tfor (i = 0; idx < size; i++, idx++)\n-\t\t\tring[idx] = obj[i];\n-\t\t/* Start at the beginning */\n-\t\tfor (idx = 0; i < n; i++, idx++)\n-\t\t\tring[idx] = obj[i];\n-\t}\n-}\n-\n-static __rte_always_inline void\n-__rte_ring_enqueue_elems_64(struct rte_ring *r, uint32_t prod_head,\n-\t\tconst void *obj_table, uint32_t n)\n-{\n-\tunsigned int i;\n-\tconst uint32_t size = r->size;\n-\tuint32_t idx = prod_head & r->mask;\n-\tuint64_t *ring = (uint64_t *)&r[1];\n-\tconst unaligned_uint64_t *obj = (const unaligned_uint64_t *)obj_table;\n-\tif (likely(idx + n < size)) {\n-\t\tfor (i = 0; i < (n & ~0x3); i += 4, idx += 4) {\n-\t\t\tring[idx] = obj[i];\n-\t\t\tring[idx + 1] = obj[i + 1];\n-\t\t\tring[idx + 2] = obj[i + 2];\n-\t\t\tring[idx + 3] = obj[i + 3];\n-\t\t}\n-\t\tswitch (n & 0x3) {\n-\t\tcase 3:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 2:\n-\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n-\t\tcase 1:\n-\t\t\tring[idx++] = obj[i++];\n-\t\t}\n-\t} else {\n-\t\tfor (i = 0; idx < size; i++, idx++)\n-\t\t\tring[idx] = obj[i];\n-\t\t/* Start at the beginning */\n-\t\tfor (idx = 0; i < n; i++, idx++)\n-\t\t\tring[idx] = obj[i];\n-\t}\n-}\n-\n-static __rte_always_inline void\n-__rte_ring_enqueue_elems_128(struct rte_ring *r, uint32_t prod_head,\n-\t\tconst void *obj_table, uint32_t n)\n-{\n-\tunsigned int i;\n-\tconst uint32_t size = r->size;\n-\tuint32_t idx = prod_head & r->mask;\n-\trte_int128_t *ring = (rte_int128_t *)&r[1];\n-\tconst rte_int128_t *obj = (const rte_int128_t *)obj_table;\n-\tif (likely(idx + n < size)) {\n-\t\tfor (i = 0; i < (n & ~0x1); i += 2, idx += 2)\n-\t\t\tmemcpy((void *)(ring + idx),\n-\t\t\t\t(const void *)(obj + i), 32);\n-\t\tswitch (n & 0x1) {\n-\t\tcase 1:\n-\t\t\tmemcpy((void *)(ring + idx),\n-\t\t\t\t(const void *)(obj + i), 16);\n-\t\t}\n-\t} else {\n-\t\tfor (i = 0; idx < size; i++, idx++)\n-\t\t\tmemcpy((void *)(ring + idx),\n-\t\t\t\t(const void *)(obj + i), 16);\n-\t\t/* Start at the beginning */\n-\t\tfor (idx = 0; i < n; i++, idx++)\n-\t\t\tmemcpy((void *)(ring + idx),\n-\t\t\t\t(const void *)(obj + i), 16);\n-\t}\n-}\n-\n-/* the actual enqueue of elements on the ring.\n- * Placed here since identical code needed in both\n- * single and multi producer enqueue functions.\n- */\n-static __rte_always_inline void\n-__rte_ring_enqueue_elems(struct rte_ring *r, uint32_t prod_head,\n-\t\tconst void *obj_table, uint32_t esize, uint32_t num)\n-{\n-\t/* 8B and 16B copies implemented individually to retain\n-\t * the current performance.\n-\t */\n-\tif (esize == 8)\n-\t\t__rte_ring_enqueue_elems_64(r, prod_head, obj_table, num);\n-\telse if (esize == 16)\n-\t\t__rte_ring_enqueue_elems_128(r, prod_head, obj_table, num);\n-\telse {\n-\t\tuint32_t idx, scale, nr_idx, nr_num, nr_size;\n-\n-\t\t/* Normalize to uint32_t */\n-\t\tscale = esize / sizeof(uint32_t);\n-\t\tnr_num = num * scale;\n-\t\tidx = prod_head & r->mask;\n-\t\tnr_idx = idx * scale;\n-\t\tnr_size = r->size * scale;\n-\t\t__rte_ring_enqueue_elems_32(r, nr_size, nr_idx,\n-\t\t\t\tobj_table, nr_num);\n-\t}\n-}\n-\n-static __rte_always_inline void\n-__rte_ring_dequeue_elems_32(struct rte_ring *r, const uint32_t size,\n-\t\tuint32_t idx, void *obj_table, uint32_t n)\n-{\n-\tunsigned int i;\n-\tuint32_t *ring = (uint32_t *)&r[1];\n-\tuint32_t *obj = (uint32_t *)obj_table;\n-\tif (likely(idx + n < size)) {\n-\t\tfor (i = 0; i < (n & ~0x7); i += 8, idx += 8) {\n-\t\t\tobj[i] = ring[idx];\n-\t\t\tobj[i + 1] = ring[idx + 1];\n-\t\t\tobj[i + 2] = ring[idx + 2];\n-\t\t\tobj[i + 3] = ring[idx + 3];\n-\t\t\tobj[i + 4] = ring[idx + 4];\n-\t\t\tobj[i + 5] = ring[idx + 5];\n-\t\t\tobj[i + 6] = ring[idx + 6];\n-\t\t\tobj[i + 7] = ring[idx + 7];\n-\t\t}\n-\t\tswitch (n & 0x7) {\n-\t\tcase 7:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 6:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 5:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 4:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 3:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 2:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 1:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\t}\n-\t} else {\n-\t\tfor (i = 0; idx < size; i++, idx++)\n-\t\t\tobj[i] = ring[idx];\n-\t\t/* Start at the beginning */\n-\t\tfor (idx = 0; i < n; i++, idx++)\n-\t\t\tobj[i] = ring[idx];\n-\t}\n-}\n-\n-static __rte_always_inline void\n-__rte_ring_dequeue_elems_64(struct rte_ring *r, uint32_t prod_head,\n-\t\tvoid *obj_table, uint32_t n)\n-{\n-\tunsigned int i;\n-\tconst uint32_t size = r->size;\n-\tuint32_t idx = prod_head & r->mask;\n-\tuint64_t *ring = (uint64_t *)&r[1];\n-\tunaligned_uint64_t *obj = (unaligned_uint64_t *)obj_table;\n-\tif (likely(idx + n < size)) {\n-\t\tfor (i = 0; i < (n & ~0x3); i += 4, idx += 4) {\n-\t\t\tobj[i] = ring[idx];\n-\t\t\tobj[i + 1] = ring[idx + 1];\n-\t\t\tobj[i + 2] = ring[idx + 2];\n-\t\t\tobj[i + 3] = ring[idx + 3];\n-\t\t}\n-\t\tswitch (n & 0x3) {\n-\t\tcase 3:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 2:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\tcase 1:\n-\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n-\t\t}\n-\t} else {\n-\t\tfor (i = 0; idx < size; i++, idx++)\n-\t\t\tobj[i] = ring[idx];\n-\t\t/* Start at the beginning */\n-\t\tfor (idx = 0; i < n; i++, idx++)\n-\t\t\tobj[i] = ring[idx];\n-\t}\n-}\n-\n-static __rte_always_inline void\n-__rte_ring_dequeue_elems_128(struct rte_ring *r, uint32_t prod_head,\n-\t\tvoid *obj_table, uint32_t n)\n-{\n-\tunsigned int i;\n-\tconst uint32_t size = r->size;\n-\tuint32_t idx = prod_head & r->mask;\n-\trte_int128_t *ring = (rte_int128_t *)&r[1];\n-\trte_int128_t *obj = (rte_int128_t *)obj_table;\n-\tif (likely(idx + n < size)) {\n-\t\tfor (i = 0; i < (n & ~0x1); i += 2, idx += 2)\n-\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 32);\n-\t\tswitch (n & 0x1) {\n-\t\tcase 1:\n-\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 16);\n-\t\t}\n-\t} else {\n-\t\tfor (i = 0; idx < size; i++, idx++)\n-\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 16);\n-\t\t/* Start at the beginning */\n-\t\tfor (idx = 0; i < n; i++, idx++)\n-\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 16);\n-\t}\n-}\n-\n-/* the actual dequeue of elements from the ring.\n- * Placed here since identical code needed in both\n- * single and multi producer enqueue functions.\n- */\n-static __rte_always_inline void\n-__rte_ring_dequeue_elems(struct rte_ring *r, uint32_t cons_head,\n-\t\tvoid *obj_table, uint32_t esize, uint32_t num)\n-{\n-\t/* 8B and 16B copies implemented individually to retain\n-\t * the current performance.\n-\t */\n-\tif (esize == 8)\n-\t\t__rte_ring_dequeue_elems_64(r, cons_head, obj_table, num);\n-\telse if (esize == 16)\n-\t\t__rte_ring_dequeue_elems_128(r, cons_head, obj_table, num);\n-\telse {\n-\t\tuint32_t idx, scale, nr_idx, nr_num, nr_size;\n-\n-\t\t/* Normalize to uint32_t */\n-\t\tscale = esize / sizeof(uint32_t);\n-\t\tnr_num = num * scale;\n-\t\tidx = cons_head & r->mask;\n-\t\tnr_idx = idx * scale;\n-\t\tnr_size = r->size * scale;\n-\t\t__rte_ring_dequeue_elems_32(r, nr_size, nr_idx,\n-\t\t\t\tobj_table, nr_num);\n-\t}\n-}\n-\n-/* Between load and load. there might be cpu reorder in weak model\n- * (powerpc/arm).\n- * There are 2 choices for the users\n- * 1.use rmb() memory barrier\n- * 2.use one-direction load_acquire/store_release barrier\n- * It depends on performance test results.\n- * By default, move common functions to rte_ring_generic.h\n- */\n-#ifdef RTE_USE_C11_MEM_MODEL\n-#include \"rte_ring_c11_mem.h\"\n-#else\n-#include \"rte_ring_generic.h\"\n-#endif\n-\n-/**\n- * @internal Enqueue several objects on the ring\n- *\n- * @param r\n- * A pointer to the ring structure.\n- * @param obj_table\n- * A pointer to a table of objects.\n- * @param esize\n- * The size of ring element, in bytes. It must be a multiple of 4.\n- * This must be the same value used while creating the ring. Otherwise\n- * the results are undefined.\n- * @param n\n- * The number of objects to add in the ring from the obj_table.\n- * @param behavior\n- * RTE_RING_QUEUE_FIXED: Enqueue a fixed number of items from a ring\n- * RTE_RING_QUEUE_VARIABLE: Enqueue as many items as possible from ring\n- * @param is_sp\n- * Indicates whether to use single producer or multi-producer head update\n- * @param free_space\n- * returns the amount of space after the enqueue operation has finished\n- * @return\n- * Actual number of objects enqueued.\n- * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_enqueue_elem(struct rte_ring *r, const void *obj_table,\n-\t\tunsigned int esize, unsigned int n,\n-\t\tenum rte_ring_queue_behavior behavior, unsigned int is_sp,\n-\t\tunsigned int *free_space)\n-{\n-\tuint32_t prod_head, prod_next;\n-\tuint32_t free_entries;\n-\n-\tn = __rte_ring_move_prod_head(r, is_sp, n, behavior,\n-\t\t\t&prod_head, &prod_next, &free_entries);\n-\tif (n == 0)\n-\t\tgoto end;\n-\n-\t__rte_ring_enqueue_elems(r, prod_head, obj_table, esize, n);\n-\n-\t__rte_ring_update_tail(&r->prod, prod_head, prod_next, is_sp, 1);\n-end:\n-\tif (free_space != NULL)\n-\t\t*free_space = free_entries - n;\n-\treturn n;\n-}\n-\n-/**\n- * @internal Dequeue several objects from the ring\n- *\n- * @param r\n- * A pointer to the ring structure.\n- * @param obj_table\n- * A pointer to a table of objects.\n- * @param esize\n- * The size of ring element, in bytes. It must be a multiple of 4.\n- * This must be the same value used while creating the ring. Otherwise\n- * the results are undefined.\n- * @param n\n- * The number of objects to pull from the ring.\n- * @param behavior\n- * RTE_RING_QUEUE_FIXED: Dequeue a fixed number of items from a ring\n- * RTE_RING_QUEUE_VARIABLE: Dequeue as many items as possible from ring\n- * @param is_sc\n- * Indicates whether to use single consumer or multi-consumer head update\n- * @param available\n- * returns the number of remaining ring entries after the dequeue has finished\n- * @return\n- * - Actual number of objects dequeued.\n- * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_dequeue_elem(struct rte_ring *r, void *obj_table,\n-\t\tunsigned int esize, unsigned int n,\n-\t\tenum rte_ring_queue_behavior behavior, unsigned int is_sc,\n-\t\tunsigned int *available)\n-{\n-\tuint32_t cons_head, cons_next;\n-\tuint32_t entries;\n-\n-\tn = __rte_ring_move_cons_head(r, (int)is_sc, n, behavior,\n-\t\t\t&cons_head, &cons_next, &entries);\n-\tif (n == 0)\n-\t\tgoto end;\n-\n-\t__rte_ring_dequeue_elems(r, cons_head, obj_table, esize, n);\n-\n-\t__rte_ring_update_tail(&r->cons, cons_head, cons_next, is_sc, 0);\n-\n-end:\n-\tif (available != NULL)\n-\t\t*available = entries - n;\n-\treturn n;\n-}\n-\n /**\n * Enqueue several objects on the ring (multi-producers safe).\n *\ndiff --git a/lib/librte_ring/rte_ring_elem_pvt.h b/lib/librte_ring/rte_ring_elem_pvt.h\nnew file mode 100644\nindex 000000000..275ec5539\n--- /dev/null\n+++ b/lib/librte_ring/rte_ring_elem_pvt.h\n@@ -0,0 +1,385 @@\n+/* SPDX-License-Identifier: BSD-3-Clause\n+ *\n+ * Copyright (c) 2017,2018 HXT-semitech Corporation.\n+ * Copyright (c) 2007-2009 Kip Macy kmacy@freebsd.org\n+ * All rights reserved.\n+ * Derived from FreeBSD's bufring.h\n+ * Used as BSD-3 Licensed with permission from Kip Macy.\n+ */\n+\n+#ifndef _RTE_RING_ELEM_PVT_H_\n+#define _RTE_RING_ELEM_PVT_H_\n+\n+static __rte_always_inline void\n+__rte_ring_enqueue_elems_32(struct rte_ring *r, const uint32_t size,\n+\t\tuint32_t idx, const void *obj_table, uint32_t n)\n+{\n+\tunsigned int i;\n+\tuint32_t *ring = (uint32_t *)&r[1];\n+\tconst uint32_t *obj = (const uint32_t *)obj_table;\n+\tif (likely(idx + n < size)) {\n+\t\tfor (i = 0; i < (n & ~0x7); i += 8, idx += 8) {\n+\t\t\tring[idx] = obj[i];\n+\t\t\tring[idx + 1] = obj[i + 1];\n+\t\t\tring[idx + 2] = obj[i + 2];\n+\t\t\tring[idx + 3] = obj[i + 3];\n+\t\t\tring[idx + 4] = obj[i + 4];\n+\t\t\tring[idx + 5] = obj[i + 5];\n+\t\t\tring[idx + 6] = obj[i + 6];\n+\t\t\tring[idx + 7] = obj[i + 7];\n+\t\t}\n+\t\tswitch (n & 0x7) {\n+\t\tcase 7:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 6:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 5:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 4:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 3:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 2:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 1:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\t}\n+\t} else {\n+\t\tfor (i = 0; idx < size; i++, idx++)\n+\t\t\tring[idx] = obj[i];\n+\t\t/* Start at the beginning */\n+\t\tfor (idx = 0; i < n; i++, idx++)\n+\t\t\tring[idx] = obj[i];\n+\t}\n+}\n+\n+static __rte_always_inline void\n+__rte_ring_enqueue_elems_64(struct rte_ring *r, uint32_t prod_head,\n+\t\tconst void *obj_table, uint32_t n)\n+{\n+\tunsigned int i;\n+\tconst uint32_t size = r->size;\n+\tuint32_t idx = prod_head & r->mask;\n+\tuint64_t *ring = (uint64_t *)&r[1];\n+\tconst unaligned_uint64_t *obj = (const unaligned_uint64_t *)obj_table;\n+\tif (likely(idx + n < size)) {\n+\t\tfor (i = 0; i < (n & ~0x3); i += 4, idx += 4) {\n+\t\t\tring[idx] = obj[i];\n+\t\t\tring[idx + 1] = obj[i + 1];\n+\t\t\tring[idx + 2] = obj[i + 2];\n+\t\t\tring[idx + 3] = obj[i + 3];\n+\t\t}\n+\t\tswitch (n & 0x3) {\n+\t\tcase 3:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 2:\n+\t\t\tring[idx++] = obj[i++]; /* fallthrough */\n+\t\tcase 1:\n+\t\t\tring[idx++] = obj[i++];\n+\t\t}\n+\t} else {\n+\t\tfor (i = 0; idx < size; i++, idx++)\n+\t\t\tring[idx] = obj[i];\n+\t\t/* Start at the beginning */\n+\t\tfor (idx = 0; i < n; i++, idx++)\n+\t\t\tring[idx] = obj[i];\n+\t}\n+}\n+\n+static __rte_always_inline void\n+__rte_ring_enqueue_elems_128(struct rte_ring *r, uint32_t prod_head,\n+\t\tconst void *obj_table, uint32_t n)\n+{\n+\tunsigned int i;\n+\tconst uint32_t size = r->size;\n+\tuint32_t idx = prod_head & r->mask;\n+\trte_int128_t *ring = (rte_int128_t *)&r[1];\n+\tconst rte_int128_t *obj = (const rte_int128_t *)obj_table;\n+\tif (likely(idx + n < size)) {\n+\t\tfor (i = 0; i < (n & ~0x1); i += 2, idx += 2)\n+\t\t\tmemcpy((void *)(ring + idx),\n+\t\t\t\t(const void *)(obj + i), 32);\n+\t\tswitch (n & 0x1) {\n+\t\tcase 1:\n+\t\t\tmemcpy((void *)(ring + idx),\n+\t\t\t\t(const void *)(obj + i), 16);\n+\t\t}\n+\t} else {\n+\t\tfor (i = 0; idx < size; i++, idx++)\n+\t\t\tmemcpy((void *)(ring + idx),\n+\t\t\t\t(const void *)(obj + i), 16);\n+\t\t/* Start at the beginning */\n+\t\tfor (idx = 0; i < n; i++, idx++)\n+\t\t\tmemcpy((void *)(ring + idx),\n+\t\t\t\t(const void *)(obj + i), 16);\n+\t}\n+}\n+\n+/* the actual enqueue of elements on the ring.\n+ * Placed here since identical code needed in both\n+ * single and multi producer enqueue functions.\n+ */\n+static __rte_always_inline void\n+__rte_ring_enqueue_elems(struct rte_ring *r, uint32_t prod_head,\n+\t\tconst void *obj_table, uint32_t esize, uint32_t num)\n+{\n+\t/* 8B and 16B copies implemented individually to retain\n+\t * the current performance.\n+\t */\n+\tif (esize == 8)\n+\t\t__rte_ring_enqueue_elems_64(r, prod_head, obj_table, num);\n+\telse if (esize == 16)\n+\t\t__rte_ring_enqueue_elems_128(r, prod_head, obj_table, num);\n+\telse {\n+\t\tuint32_t idx, scale, nr_idx, nr_num, nr_size;\n+\n+\t\t/* Normalize to uint32_t */\n+\t\tscale = esize / sizeof(uint32_t);\n+\t\tnr_num = num * scale;\n+\t\tidx = prod_head & r->mask;\n+\t\tnr_idx = idx * scale;\n+\t\tnr_size = r->size * scale;\n+\t\t__rte_ring_enqueue_elems_32(r, nr_size, nr_idx,\n+\t\t\t\tobj_table, nr_num);\n+\t}\n+}\n+\n+static __rte_always_inline void\n+__rte_ring_dequeue_elems_32(struct rte_ring *r, const uint32_t size,\n+\t\tuint32_t idx, void *obj_table, uint32_t n)\n+{\n+\tunsigned int i;\n+\tuint32_t *ring = (uint32_t *)&r[1];\n+\tuint32_t *obj = (uint32_t *)obj_table;\n+\tif (likely(idx + n < size)) {\n+\t\tfor (i = 0; i < (n & ~0x7); i += 8, idx += 8) {\n+\t\t\tobj[i] = ring[idx];\n+\t\t\tobj[i + 1] = ring[idx + 1];\n+\t\t\tobj[i + 2] = ring[idx + 2];\n+\t\t\tobj[i + 3] = ring[idx + 3];\n+\t\t\tobj[i + 4] = ring[idx + 4];\n+\t\t\tobj[i + 5] = ring[idx + 5];\n+\t\t\tobj[i + 6] = ring[idx + 6];\n+\t\t\tobj[i + 7] = ring[idx + 7];\n+\t\t}\n+\t\tswitch (n & 0x7) {\n+\t\tcase 7:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 6:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 5:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 4:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 3:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 2:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 1:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\t}\n+\t} else {\n+\t\tfor (i = 0; idx < size; i++, idx++)\n+\t\t\tobj[i] = ring[idx];\n+\t\t/* Start at the beginning */\n+\t\tfor (idx = 0; i < n; i++, idx++)\n+\t\t\tobj[i] = ring[idx];\n+\t}\n+}\n+\n+static __rte_always_inline void\n+__rte_ring_dequeue_elems_64(struct rte_ring *r, uint32_t prod_head,\n+\t\tvoid *obj_table, uint32_t n)\n+{\n+\tunsigned int i;\n+\tconst uint32_t size = r->size;\n+\tuint32_t idx = prod_head & r->mask;\n+\tuint64_t *ring = (uint64_t *)&r[1];\n+\tunaligned_uint64_t *obj = (unaligned_uint64_t *)obj_table;\n+\tif (likely(idx + n < size)) {\n+\t\tfor (i = 0; i < (n & ~0x3); i += 4, idx += 4) {\n+\t\t\tobj[i] = ring[idx];\n+\t\t\tobj[i + 1] = ring[idx + 1];\n+\t\t\tobj[i + 2] = ring[idx + 2];\n+\t\t\tobj[i + 3] = ring[idx + 3];\n+\t\t}\n+\t\tswitch (n & 0x3) {\n+\t\tcase 3:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 2:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\tcase 1:\n+\t\t\tobj[i++] = ring[idx++]; /* fallthrough */\n+\t\t}\n+\t} else {\n+\t\tfor (i = 0; idx < size; i++, idx++)\n+\t\t\tobj[i] = ring[idx];\n+\t\t/* Start at the beginning */\n+\t\tfor (idx = 0; i < n; i++, idx++)\n+\t\t\tobj[i] = ring[idx];\n+\t}\n+}\n+\n+static __rte_always_inline void\n+__rte_ring_dequeue_elems_128(struct rte_ring *r, uint32_t prod_head,\n+\t\tvoid *obj_table, uint32_t n)\n+{\n+\tunsigned int i;\n+\tconst uint32_t size = r->size;\n+\tuint32_t idx = prod_head & r->mask;\n+\trte_int128_t *ring = (rte_int128_t *)&r[1];\n+\trte_int128_t *obj = (rte_int128_t *)obj_table;\n+\tif (likely(idx + n < size)) {\n+\t\tfor (i = 0; i < (n & ~0x1); i += 2, idx += 2)\n+\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 32);\n+\t\tswitch (n & 0x1) {\n+\t\tcase 1:\n+\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 16);\n+\t\t}\n+\t} else {\n+\t\tfor (i = 0; idx < size; i++, idx++)\n+\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 16);\n+\t\t/* Start at the beginning */\n+\t\tfor (idx = 0; i < n; i++, idx++)\n+\t\t\tmemcpy((void *)(obj + i), (void *)(ring + idx), 16);\n+\t}\n+}\n+\n+/* the actual dequeue of elements from the ring.\n+ * Placed here since identical code needed in both\n+ * single and multi producer enqueue functions.\n+ */\n+static __rte_always_inline void\n+__rte_ring_dequeue_elems(struct rte_ring *r, uint32_t cons_head,\n+\t\tvoid *obj_table, uint32_t esize, uint32_t num)\n+{\n+\t/* 8B and 16B copies implemented individually to retain\n+\t * the current performance.\n+\t */\n+\tif (esize == 8)\n+\t\t__rte_ring_dequeue_elems_64(r, cons_head, obj_table, num);\n+\telse if (esize == 16)\n+\t\t__rte_ring_dequeue_elems_128(r, cons_head, obj_table, num);\n+\telse {\n+\t\tuint32_t idx, scale, nr_idx, nr_num, nr_size;\n+\n+\t\t/* Normalize to uint32_t */\n+\t\tscale = esize / sizeof(uint32_t);\n+\t\tnr_num = num * scale;\n+\t\tidx = cons_head & r->mask;\n+\t\tnr_idx = idx * scale;\n+\t\tnr_size = r->size * scale;\n+\t\t__rte_ring_dequeue_elems_32(r, nr_size, nr_idx,\n+\t\t\t\tobj_table, nr_num);\n+\t}\n+}\n+\n+/* Between load and load. there might be cpu reorder in weak model\n+ * (powerpc/arm).\n+ * There are 2 choices for the users\n+ * 1.use rmb() memory barrier\n+ * 2.use one-direction load_acquire/store_release barrier\n+ * It depends on performance test results.\n+ */\n+#ifdef RTE_USE_C11_MEM_MODEL\n+#include \"rte_ring_c11_pvt.h\"\n+#else\n+#include \"rte_ring_generic_pvt.h\"\n+#endif\n+\n+/**\n+ * @internal Enqueue several objects on the ring\n+ *\n+ * @param r\n+ * A pointer to the ring structure.\n+ * @param obj_table\n+ * A pointer to a table of objects.\n+ * @param esize\n+ * The size of ring element, in bytes. It must be a multiple of 4.\n+ * This must be the same value used while creating the ring. Otherwise\n+ * the results are undefined.\n+ * @param n\n+ * The number of objects to add in the ring from the obj_table.\n+ * @param behavior\n+ * RTE_RING_QUEUE_FIXED: Enqueue a fixed number of items from a ring\n+ * RTE_RING_QUEUE_VARIABLE: Enqueue as many items as possible from ring\n+ * @param is_sp\n+ * Indicates whether to use single producer or multi-producer head update\n+ * @param free_space\n+ * returns the amount of space after the enqueue operation has finished\n+ * @return\n+ * Actual number of objects enqueued.\n+ * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_enqueue_elem(struct rte_ring *r, const void *obj_table,\n+\t\tunsigned int esize, unsigned int n,\n+\t\tenum rte_ring_queue_behavior behavior, unsigned int is_sp,\n+\t\tunsigned int *free_space)\n+{\n+\tuint32_t prod_head, prod_next;\n+\tuint32_t free_entries;\n+\n+\tn = __rte_ring_move_prod_head(r, is_sp, n, behavior,\n+\t\t\t&prod_head, &prod_next, &free_entries);\n+\tif (n == 0)\n+\t\tgoto end;\n+\n+\t__rte_ring_enqueue_elems(r, prod_head, obj_table, esize, n);\n+\n+\t__rte_ring_update_tail(&r->prod, prod_head, prod_next, is_sp, 1);\n+end:\n+\tif (free_space != NULL)\n+\t\t*free_space = free_entries - n;\n+\treturn n;\n+}\n+\n+/**\n+ * @internal Dequeue several objects from the ring\n+ *\n+ * @param r\n+ * A pointer to the ring structure.\n+ * @param obj_table\n+ * A pointer to a table of objects.\n+ * @param esize\n+ * The size of ring element, in bytes. It must be a multiple of 4.\n+ * This must be the same value used while creating the ring. Otherwise\n+ * the results are undefined.\n+ * @param n\n+ * The number of objects to pull from the ring.\n+ * @param behavior\n+ * RTE_RING_QUEUE_FIXED: Dequeue a fixed number of items from a ring\n+ * RTE_RING_QUEUE_VARIABLE: Dequeue as many items as possible from ring\n+ * @param is_sc\n+ * Indicates whether to use single consumer or multi-consumer head update\n+ * @param available\n+ * returns the number of remaining ring entries after the dequeue has finished\n+ * @return\n+ * - Actual number of objects dequeued.\n+ * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_dequeue_elem(struct rte_ring *r, void *obj_table,\n+\t\tunsigned int esize, unsigned int n,\n+\t\tenum rte_ring_queue_behavior behavior, unsigned int is_sc,\n+\t\tunsigned int *available)\n+{\n+\tuint32_t cons_head, cons_next;\n+\tuint32_t entries;\n+\n+\tn = __rte_ring_move_cons_head(r, (int)is_sc, n, behavior,\n+\t\t\t&cons_head, &cons_next, &entries);\n+\tif (n == 0)\n+\t\tgoto end;\n+\n+\t__rte_ring_dequeue_elems(r, cons_head, obj_table, esize, n);\n+\n+\t__rte_ring_update_tail(&r->cons, cons_head, cons_next, is_sc, 0);\n+\n+end:\n+\tif (available != NULL)\n+\t\t*available = entries - n;\n+\treturn n;\n+}\n+\n+#endif /* _RTE_RING_ELEM_PVT_H_ */\ndiff --git a/lib/librte_ring/rte_ring_generic.h b/lib/librte_ring/rte_ring_generic_pvt.h\nsimilarity index 97%\nrename from lib/librte_ring/rte_ring_generic.h\nrename to lib/librte_ring/rte_ring_generic_pvt.h\nindex 37c62b8d6..532deb5e7 100644\n--- a/lib/librte_ring/rte_ring_generic.h\n+++ b/lib/librte_ring/rte_ring_generic_pvt.h\n@@ -7,8 +7,8 @@\n * Used as BSD-3 Licensed with permission from Kip Macy.\n */\n \n-#ifndef _RTE_RING_GENERIC_H_\n-#define _RTE_RING_GENERIC_H_\n+#ifndef _RTE_RING_GENERIC_PVT_H_\n+#define _RTE_RING_GENERIC_PVT_H_\n \n static __rte_always_inline void\n __rte_ring_update_tail(struct rte_ring_headtail *ht, uint32_t old_val,\n@@ -170,4 +170,4 @@ __rte_ring_move_cons_head(struct rte_ring *r, unsigned int is_sc,\n \treturn n;\n }\n \n-#endif /* _RTE_RING_GENERIC_H_ */\n+#endif /* _RTE_RING_GENERIC_PVT_H_ */\ndiff --git a/lib/librte_ring/rte_ring_hts.h b/lib/librte_ring/rte_ring_hts.h\nindex 359b15771..a9342083f 100644\n--- a/lib/librte_ring/rte_ring_hts.h\n+++ b/lib/librte_ring/rte_ring_hts.h\n@@ -29,89 +29,7 @@\n extern \"C\" {\n #endif\n \n-#include <rte_ring_hts_c11_mem.h>\n-\n-/**\n- * @internal Enqueue several objects on the HTS ring.\n- *\n- * @param r\n- * A pointer to the ring structure.\n- * @param obj_table\n- * A pointer to a table of objects.\n- * @param esize\n- * The size of ring element, in bytes. It must be a multiple of 4.\n- * This must be the same value used while creating the ring. Otherwise\n- * the results are undefined.\n- * @param n\n- * The number of objects to add in the ring from the obj_table.\n- * @param behavior\n- * RTE_RING_QUEUE_FIXED: Enqueue a fixed number of items from a ring\n- * RTE_RING_QUEUE_VARIABLE: Enqueue as many items as possible from ring\n- * @param free_space\n- * returns the amount of space after the enqueue operation has finished\n- * @return\n- * Actual number of objects enqueued.\n- * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_hts_enqueue_elem(struct rte_ring *r, const void *obj_table,\n-\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n-\tuint32_t *free_space)\n-{\n-\tuint32_t free, head;\n-\n-\tn = __rte_ring_hts_move_prod_head(r, n, behavior, &head, &free);\n-\n-\tif (n != 0) {\n-\t\t__rte_ring_enqueue_elems(r, head, obj_table, esize, n);\n-\t\t__rte_ring_hts_update_tail(&r->hts_prod, head, n, 1);\n-\t}\n-\n-\tif (free_space != NULL)\n-\t\t*free_space = free - n;\n-\treturn n;\n-}\n-\n-/**\n- * @internal Dequeue several objects from the HTS ring.\n- *\n- * @param r\n- * A pointer to the ring structure.\n- * @param obj_table\n- * A pointer to a table of objects.\n- * @param esize\n- * The size of ring element, in bytes. It must be a multiple of 4.\n- * This must be the same value used while creating the ring. Otherwise\n- * the results are undefined.\n- * @param n\n- * The number of objects to pull from the ring.\n- * @param behavior\n- * RTE_RING_QUEUE_FIXED: Dequeue a fixed number of items from a ring\n- * RTE_RING_QUEUE_VARIABLE: Dequeue as many items as possible from ring\n- * @param available\n- * returns the number of remaining ring entries after the dequeue has finished\n- * @return\n- * - Actual number of objects dequeued.\n- * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_hts_dequeue_elem(struct rte_ring *r, void *obj_table,\n-\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n-\tuint32_t *available)\n-{\n-\tuint32_t entries, head;\n-\n-\tn = __rte_ring_hts_move_cons_head(r, n, behavior, &head, &entries);\n-\n-\tif (n != 0) {\n-\t\t__rte_ring_dequeue_elems(r, head, obj_table, esize, n);\n-\t\t__rte_ring_hts_update_tail(&r->hts_cons, head, n, 0);\n-\t}\n-\n-\tif (available != NULL)\n-\t\t*available = entries - n;\n-\treturn n;\n-}\n+#include <rte_ring_hts_elem_pvt.h>\n \n /**\n * Enqueue several objects on the HTS ring (multi-producers safe).\ndiff --git a/lib/librte_ring/rte_ring_hts_c11_mem.h b/lib/librte_ring/rte_ring_hts_elem_pvt.h\nsimilarity index 60%\nrename from lib/librte_ring/rte_ring_hts_c11_mem.h\nrename to lib/librte_ring/rte_ring_hts_elem_pvt.h\nindex 16e54b6ff..a1c0683c2 100644\n--- a/lib/librte_ring/rte_ring_hts_c11_mem.h\n+++ b/lib/librte_ring/rte_ring_hts_elem_pvt.h\n@@ -7,8 +7,8 @@\n * Used as BSD-3 Licensed with permission from Kip Macy.\n */\n \n-#ifndef _RTE_RING_HTS_C11_MEM_H_\n-#define _RTE_RING_HTS_C11_MEM_H_\n+#ifndef _RTE_RING_HTS_ELEM_PVT_H_\n+#define _RTE_RING_HTS_ELEM_PVT_H_\n \n /**\n * @file rte_ring_hts_c11_mem.h\n@@ -161,4 +161,86 @@ __rte_ring_hts_move_cons_head(struct rte_ring *r, unsigned int num,\n \treturn n;\n }\n \n-#endif /* _RTE_RING_HTS_C11_MEM_H_ */\n+/**\n+ * @internal Enqueue several objects on the HTS ring.\n+ *\n+ * @param r\n+ * A pointer to the ring structure.\n+ * @param obj_table\n+ * A pointer to a table of objects.\n+ * @param esize\n+ * The size of ring element, in bytes. It must be a multiple of 4.\n+ * This must be the same value used while creating the ring. Otherwise\n+ * the results are undefined.\n+ * @param n\n+ * The number of objects to add in the ring from the obj_table.\n+ * @param behavior\n+ * RTE_RING_QUEUE_FIXED: Enqueue a fixed number of items from a ring\n+ * RTE_RING_QUEUE_VARIABLE: Enqueue as many items as possible from ring\n+ * @param free_space\n+ * returns the amount of space after the enqueue operation has finished\n+ * @return\n+ * Actual number of objects enqueued.\n+ * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_hts_enqueue_elem(struct rte_ring *r, const void *obj_table,\n+\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n+\tuint32_t *free_space)\n+{\n+\tuint32_t free, head;\n+\n+\tn = __rte_ring_hts_move_prod_head(r, n, behavior, &head, &free);\n+\n+\tif (n != 0) {\n+\t\t__rte_ring_enqueue_elems(r, head, obj_table, esize, n);\n+\t\t__rte_ring_hts_update_tail(&r->hts_prod, head, n, 1);\n+\t}\n+\n+\tif (free_space != NULL)\n+\t\t*free_space = free - n;\n+\treturn n;\n+}\n+\n+/**\n+ * @internal Dequeue several objects from the HTS ring.\n+ *\n+ * @param r\n+ * A pointer to the ring structure.\n+ * @param obj_table\n+ * A pointer to a table of objects.\n+ * @param esize\n+ * The size of ring element, in bytes. It must be a multiple of 4.\n+ * This must be the same value used while creating the ring. Otherwise\n+ * the results are undefined.\n+ * @param n\n+ * The number of objects to pull from the ring.\n+ * @param behavior\n+ * RTE_RING_QUEUE_FIXED: Dequeue a fixed number of items from a ring\n+ * RTE_RING_QUEUE_VARIABLE: Dequeue as many items as possible from ring\n+ * @param available\n+ * returns the number of remaining ring entries after the dequeue has finished\n+ * @return\n+ * - Actual number of objects dequeued.\n+ * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_hts_dequeue_elem(struct rte_ring *r, void *obj_table,\n+\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n+\tuint32_t *available)\n+{\n+\tuint32_t entries, head;\n+\n+\tn = __rte_ring_hts_move_cons_head(r, n, behavior, &head, &entries);\n+\n+\tif (n != 0) {\n+\t\t__rte_ring_dequeue_elems(r, head, obj_table, esize, n);\n+\t\t__rte_ring_hts_update_tail(&r->hts_cons, head, n, 0);\n+\t}\n+\n+\tif (available != NULL)\n+\t\t*available = entries - n;\n+\treturn n;\n+}\n+\n+#endif /* _RTE_RING_HTS_ELEM_PVT_H_ */\ndiff --git a/lib/librte_ring/rte_ring_peek.h b/lib/librte_ring/rte_ring_peek.h\nindex 45f707dc7..3f8f6fc1c 100644\n--- a/lib/librte_ring/rte_ring_peek.h\n+++ b/lib/librte_ring/rte_ring_peek.h\n@@ -48,39 +48,7 @@\n extern \"C\" {\n #endif\n \n-#include <rte_ring_peek_c11_mem.h>\n-\n-/**\n- * @internal This function moves prod head value.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_enqueue_start(struct rte_ring *r, uint32_t n,\n-\t\tenum rte_ring_queue_behavior behavior, uint32_t *free_space)\n-{\n-\tuint32_t free, head, next;\n-\n-\tswitch (r->prod.sync_type) {\n-\tcase RTE_RING_SYNC_ST:\n-\t\tn = __rte_ring_move_prod_head(r, RTE_RING_SYNC_ST, n,\n-\t\t\tbehavior, &head, &next, &free);\n-\t\tbreak;\n-\tcase RTE_RING_SYNC_MT_HTS:\n-\t\tn = __rte_ring_hts_move_prod_head(r, n, behavior,\n-\t\t\t&head, &free);\n-\t\tbreak;\n-\tcase RTE_RING_SYNC_MT:\n-\tcase RTE_RING_SYNC_MT_RTS:\n-\tdefault:\n-\t\t/* unsupported mode, shouldn't be here */\n-\t\tRTE_ASSERT(0);\n-\t\tn = 0;\n-\t\tfree = 0;\n-\t}\n-\n-\tif (free_space != NULL)\n-\t\t*free_space = free - n;\n-\treturn n;\n-}\n+#include <rte_ring_peek_elem_pvt.h>\n \n /**\n * Start to enqueue several objects on the ring.\n@@ -248,43 +216,6 @@ rte_ring_enqueue_finish(struct rte_ring *r, void * const *obj_table,\n \trte_ring_enqueue_elem_finish(r, obj_table, sizeof(uintptr_t), n);\n }\n \n-/**\n- * @internal This function moves cons head value and copies up to *n*\n- * objects from the ring to the user provided obj_table.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_dequeue_start(struct rte_ring *r, void *obj_table,\n-\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n-\tuint32_t *available)\n-{\n-\tuint32_t avail, head, next;\n-\n-\tswitch (r->cons.sync_type) {\n-\tcase RTE_RING_SYNC_ST:\n-\t\tn = __rte_ring_move_cons_head(r, RTE_RING_SYNC_ST, n,\n-\t\t\tbehavior, &head, &next, &avail);\n-\t\tbreak;\n-\tcase RTE_RING_SYNC_MT_HTS:\n-\t\tn = __rte_ring_hts_move_cons_head(r, n, behavior,\n-\t\t\t&head, &avail);\n-\t\tbreak;\n-\tcase RTE_RING_SYNC_MT:\n-\tcase RTE_RING_SYNC_MT_RTS:\n-\tdefault:\n-\t\t/* unsupported mode, shouldn't be here */\n-\t\tRTE_ASSERT(0);\n-\t\tn = 0;\n-\t\tavail = 0;\n-\t}\n-\n-\tif (n != 0)\n-\t\t__rte_ring_dequeue_elems(r, head, obj_table, esize, n);\n-\n-\tif (available != NULL)\n-\t\t*available = avail - n;\n-\treturn n;\n-}\n-\n /**\n * Start to dequeue several objects from the ring.\n * Note that user has to call appropriate dequeue_finish()\ndiff --git a/lib/librte_ring/rte_ring_peek_c11_mem.h b/lib/librte_ring/rte_ring_peek_elem_pvt.h\nsimilarity index 61%\nrename from lib/librte_ring/rte_ring_peek_c11_mem.h\nrename to lib/librte_ring/rte_ring_peek_elem_pvt.h\nindex 283c7e70b..3d1a487a3 100644\n--- a/lib/librte_ring/rte_ring_peek_c11_mem.h\n+++ b/lib/librte_ring/rte_ring_peek_elem_pvt.h\n@@ -7,8 +7,8 @@\n * Used as BSD-3 Licensed with permission from Kip Macy.\n */\n \n-#ifndef _RTE_RING_PEEK_C11_MEM_H_\n-#define _RTE_RING_PEEK_C11_MEM_H_\n+#ifndef _RTE_RING_PEEK_ELEM_PVT_H_\n+#define _RTE_RING_PEEK_ELEM_PVT_H_\n \n /**\n * @file rte_ring_peek_c11_mem.h\n@@ -107,4 +107,73 @@ __rte_ring_hts_set_head_tail(struct rte_ring_hts_headtail *ht, uint32_t tail,\n \t__atomic_store_n(&ht->ht.raw, p.raw, __ATOMIC_RELEASE);\n }\n \n-#endif /* _RTE_RING_PEEK_C11_MEM_H_ */\n+/**\n+ * @internal This function moves prod head value.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_enqueue_start(struct rte_ring *r, uint32_t n,\n+\t\tenum rte_ring_queue_behavior behavior, uint32_t *free_space)\n+{\n+\tuint32_t free, head, next;\n+\n+\tswitch (r->prod.sync_type) {\n+\tcase RTE_RING_SYNC_ST:\n+\t\tn = __rte_ring_move_prod_head(r, RTE_RING_SYNC_ST, n,\n+\t\t\tbehavior, &head, &next, &free);\n+\t\tbreak;\n+\tcase RTE_RING_SYNC_MT_HTS:\n+\t\tn = __rte_ring_hts_move_prod_head(r, n, behavior,\n+\t\t\t&head, &free);\n+\t\tbreak;\n+\tcase RTE_RING_SYNC_MT:\n+\tcase RTE_RING_SYNC_MT_RTS:\n+\tdefault:\n+\t\t/* unsupported mode, shouldn't be here */\n+\t\tRTE_ASSERT(0);\n+\t\tn = 0;\n+\t\tfree = 0;\n+\t}\n+\n+\tif (free_space != NULL)\n+\t\t*free_space = free - n;\n+\treturn n;\n+}\n+\n+/**\n+ * @internal This function moves cons head value and copies up to *n*\n+ * objects from the ring to the user provided obj_table.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_dequeue_start(struct rte_ring *r, void *obj_table,\n+\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n+\tuint32_t *available)\n+{\n+\tuint32_t avail, head, next;\n+\n+\tswitch (r->cons.sync_type) {\n+\tcase RTE_RING_SYNC_ST:\n+\t\tn = __rte_ring_move_cons_head(r, RTE_RING_SYNC_ST, n,\n+\t\t\tbehavior, &head, &next, &avail);\n+\t\tbreak;\n+\tcase RTE_RING_SYNC_MT_HTS:\n+\t\tn = __rte_ring_hts_move_cons_head(r, n, behavior,\n+\t\t\t&head, &avail);\n+\t\tbreak;\n+\tcase RTE_RING_SYNC_MT:\n+\tcase RTE_RING_SYNC_MT_RTS:\n+\tdefault:\n+\t\t/* unsupported mode, shouldn't be here */\n+\t\tRTE_ASSERT(0);\n+\t\tn = 0;\n+\t\tavail = 0;\n+\t}\n+\n+\tif (n != 0)\n+\t\t__rte_ring_dequeue_elems(r, head, obj_table, esize, n);\n+\n+\tif (available != NULL)\n+\t\t*available = avail - n;\n+\treturn n;\n+}\n+\n+#endif /* _RTE_RING_PEEK_ELEM_PVT_H_ */\ndiff --git a/lib/librte_ring/rte_ring_peek_zc.h b/lib/librte_ring/rte_ring_peek_zc.h\nindex cb3bbd067..be677a3e1 100644\n--- a/lib/librte_ring/rte_ring_peek_zc.h\n+++ b/lib/librte_ring/rte_ring_peek_zc.h\n@@ -72,7 +72,7 @@\n extern \"C\" {\n #endif\n \n-#include <rte_ring_peek_c11_mem.h>\n+#include <rte_ring_peek_elem_pvt.h>\n \n /**\n * Ring zero-copy information structure.\ndiff --git a/lib/librte_ring/rte_ring_rts.h b/lib/librte_ring/rte_ring_rts.h\nindex afc12abe2..9570aec8f 100644\n--- a/lib/librte_ring/rte_ring_rts.h\n+++ b/lib/librte_ring/rte_ring_rts.h\n@@ -56,89 +56,7 @@\n extern \"C\" {\n #endif\n \n-#include <rte_ring_rts_c11_mem.h>\n-\n-/**\n- * @internal Enqueue several objects on the RTS ring.\n- *\n- * @param r\n- * A pointer to the ring structure.\n- * @param obj_table\n- * A pointer to a table of objects.\n- * @param esize\n- * The size of ring element, in bytes. It must be a multiple of 4.\n- * This must be the same value used while creating the ring. Otherwise\n- * the results are undefined.\n- * @param n\n- * The number of objects to add in the ring from the obj_table.\n- * @param behavior\n- * RTE_RING_QUEUE_FIXED: Enqueue a fixed number of items from a ring\n- * RTE_RING_QUEUE_VARIABLE: Enqueue as many items as possible from ring\n- * @param free_space\n- * returns the amount of space after the enqueue operation has finished\n- * @return\n- * Actual number of objects enqueued.\n- * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_rts_enqueue_elem(struct rte_ring *r, const void *obj_table,\n-\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n-\tuint32_t *free_space)\n-{\n-\tuint32_t free, head;\n-\n-\tn = __rte_ring_rts_move_prod_head(r, n, behavior, &head, &free);\n-\n-\tif (n != 0) {\n-\t\t__rte_ring_enqueue_elems(r, head, obj_table, esize, n);\n-\t\t__rte_ring_rts_update_tail(&r->rts_prod);\n-\t}\n-\n-\tif (free_space != NULL)\n-\t\t*free_space = free - n;\n-\treturn n;\n-}\n-\n-/**\n- * @internal Dequeue several objects from the RTS ring.\n- *\n- * @param r\n- * A pointer to the ring structure.\n- * @param obj_table\n- * A pointer to a table of objects.\n- * @param esize\n- * The size of ring element, in bytes. It must be a multiple of 4.\n- * This must be the same value used while creating the ring. Otherwise\n- * the results are undefined.\n- * @param n\n- * The number of objects to pull from the ring.\n- * @param behavior\n- * RTE_RING_QUEUE_FIXED: Dequeue a fixed number of items from a ring\n- * RTE_RING_QUEUE_VARIABLE: Dequeue as many items as possible from ring\n- * @param available\n- * returns the number of remaining ring entries after the dequeue has finished\n- * @return\n- * - Actual number of objects dequeued.\n- * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n- */\n-static __rte_always_inline unsigned int\n-__rte_ring_do_rts_dequeue_elem(struct rte_ring *r, void *obj_table,\n-\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n-\tuint32_t *available)\n-{\n-\tuint32_t entries, head;\n-\n-\tn = __rte_ring_rts_move_cons_head(r, n, behavior, &head, &entries);\n-\n-\tif (n != 0) {\n-\t\t__rte_ring_dequeue_elems(r, head, obj_table, esize, n);\n-\t\t__rte_ring_rts_update_tail(&r->rts_cons);\n-\t}\n-\n-\tif (available != NULL)\n-\t\t*available = entries - n;\n-\treturn n;\n-}\n+#include <rte_ring_rts_elem_pvt.h>\n \n /**\n * Enqueue several objects on the RTS ring (multi-producers safe).\ndiff --git a/lib/librte_ring/rte_ring_rts_c11_mem.h b/lib/librte_ring/rte_ring_rts_elem_pvt.h\nsimilarity index 62%\nrename from lib/librte_ring/rte_ring_rts_c11_mem.h\nrename to lib/librte_ring/rte_ring_rts_elem_pvt.h\nindex 327f22796..cfc963820 100644\n--- a/lib/librte_ring/rte_ring_rts_c11_mem.h\n+++ b/lib/librte_ring/rte_ring_rts_elem_pvt.h\n@@ -7,8 +7,8 @@\n * Used as BSD-3 Licensed with permission from Kip Macy.\n */\n \n-#ifndef _RTE_RING_RTS_C11_MEM_H_\n-#define _RTE_RING_RTS_C11_MEM_H_\n+#ifndef _RTE_RING_RTS_ELEM_PVT_H_\n+#define _RTE_RING_RTS_ELEM_PVT_H_\n \n /**\n * @file rte_ring_rts_c11_mem.h\n@@ -176,4 +176,86 @@ __rte_ring_rts_move_cons_head(struct rte_ring *r, uint32_t num,\n \treturn n;\n }\n \n-#endif /* _RTE_RING_RTS_C11_MEM_H_ */\n+/**\n+ * @internal Enqueue several objects on the RTS ring.\n+ *\n+ * @param r\n+ * A pointer to the ring structure.\n+ * @param obj_table\n+ * A pointer to a table of objects.\n+ * @param esize\n+ * The size of ring element, in bytes. It must be a multiple of 4.\n+ * This must be the same value used while creating the ring. Otherwise\n+ * the results are undefined.\n+ * @param n\n+ * The number of objects to add in the ring from the obj_table.\n+ * @param behavior\n+ * RTE_RING_QUEUE_FIXED: Enqueue a fixed number of items from a ring\n+ * RTE_RING_QUEUE_VARIABLE: Enqueue as many items as possible from ring\n+ * @param free_space\n+ * returns the amount of space after the enqueue operation has finished\n+ * @return\n+ * Actual number of objects enqueued.\n+ * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_rts_enqueue_elem(struct rte_ring *r, const void *obj_table,\n+\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n+\tuint32_t *free_space)\n+{\n+\tuint32_t free, head;\n+\n+\tn = __rte_ring_rts_move_prod_head(r, n, behavior, &head, &free);\n+\n+\tif (n != 0) {\n+\t\t__rte_ring_enqueue_elems(r, head, obj_table, esize, n);\n+\t\t__rte_ring_rts_update_tail(&r->rts_prod);\n+\t}\n+\n+\tif (free_space != NULL)\n+\t\t*free_space = free - n;\n+\treturn n;\n+}\n+\n+/**\n+ * @internal Dequeue several objects from the RTS ring.\n+ *\n+ * @param r\n+ * A pointer to the ring structure.\n+ * @param obj_table\n+ * A pointer to a table of objects.\n+ * @param esize\n+ * The size of ring element, in bytes. It must be a multiple of 4.\n+ * This must be the same value used while creating the ring. Otherwise\n+ * the results are undefined.\n+ * @param n\n+ * The number of objects to pull from the ring.\n+ * @param behavior\n+ * RTE_RING_QUEUE_FIXED: Dequeue a fixed number of items from a ring\n+ * RTE_RING_QUEUE_VARIABLE: Dequeue as many items as possible from ring\n+ * @param available\n+ * returns the number of remaining ring entries after the dequeue has finished\n+ * @return\n+ * - Actual number of objects dequeued.\n+ * If behavior == RTE_RING_QUEUE_FIXED, this will be 0 or n only.\n+ */\n+static __rte_always_inline unsigned int\n+__rte_ring_do_rts_dequeue_elem(struct rte_ring *r, void *obj_table,\n+\tuint32_t esize, uint32_t n, enum rte_ring_queue_behavior behavior,\n+\tuint32_t *available)\n+{\n+\tuint32_t entries, head;\n+\n+\tn = __rte_ring_rts_move_cons_head(r, n, behavior, &head, &entries);\n+\n+\tif (n != 0) {\n+\t\t__rte_ring_dequeue_elems(r, head, obj_table, esize, n);\n+\t\t__rte_ring_rts_update_tail(&r->rts_cons);\n+\t}\n+\n+\tif (available != NULL)\n+\t\t*available = entries - n;\n+\treturn n;\n+}\n+\n+#endif /* _RTE_RING_RTS_ELEM_PVT_H_ */\n", "prefixes": [ "v2", "3/3" ] }