git-lng-odp

git-lng-odp@lists.linaro.org

1 participants
1249 discussions

[COMMIT] branch master updated. v1.12.0.0-18-gb18dc64

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, master has been updated via b18dc645a68fd0ae50f1a361aa408b3e74e46c0d (commit) from 29dde30264da8106d64328b0d81eb19dad0065ac (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit b18dc645a68fd0ae50f1a361aa408b3e74e46c0d Author: Joe Savage <joe.savage(a)arm.com> Date: Wed Jan 4 14:52:35 2017 +0000 helper: fix odph_ipv4_csum functions for L3 offset 0 Fix a bug where odph_ipv4_csum_{valid, update} would interpret an L3 offset of zero as a sign that the packet has no L3 offset. Signed-off-by: Joe Savage <joe.savage(a)arm.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/helper/include/odp/helper/ip.h b/helper/include/odp/helper/ip.h index 4cfc00f..ba6e675 100644 --- a/helper/include/odp/helper/ip.h +++ b/helper/include/odp/helper/ip.h @@ -101,17 +101,18 @@ ODP_STATIC_ASSERT(sizeof(odph_ipv4hdr_t) == ODPH_IPV4HDR_LEN, */ static inline int odph_ipv4_csum_valid(odp_packet_t pkt) { + uint32_t offset; odp_u16be_t res = 0; uint16_t *w; int nleft = sizeof(odph_ipv4hdr_t); odph_ipv4hdr_t ip; odp_u16be_t chksum; - if (!odp_packet_l3_offset(pkt)) + offset = odp_packet_l3_offset(pkt); + if (offset == ODP_PACKET_OFFSET_INVALID) return 0; - odp_packet_copy_to_mem(pkt, odp_packet_l3_offset(pkt), - sizeof(odph_ipv4hdr_t), &ip); + odp_packet_copy_to_mem(pkt, offset, sizeof(odph_ipv4hdr_t), &ip); w = (uint16_t *)(void *)&ip; chksum = ip.chksum; @@ -137,10 +138,10 @@ static inline odp_u16sum_t odph_ipv4_csum_update(odp_packet_t pkt) odph_ipv4hdr_t *ip; int nleft = sizeof(odph_ipv4hdr_t); - if (!odp_packet_l3_offset(pkt)) + ip = (odph_ipv4hdr_t *)odp_packet_l3_ptr(pkt, NULL); + if (ip == NULL) return 0; - ip = (odph_ipv4hdr_t *)odp_packet_l3_ptr(pkt, NULL); w = (uint16_t *)(void *)ip; ip->chksum = odph_chksum(w, nleft); return ip->chksum; ----------------------------------------------------------------------- Summary of changes: helper/include/odp/helper/ip.h | 11 ++++++----- 1 file changed, 6 insertions(+), 5 deletions(-) hooks/post-receive --

9 years, 7 months

[COMMIT] branch api-next updated. v1.12.0.0-194-g29b05df

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, api-next has been updated via 29b05df013426838fe166084fed9f1d56c59ebed (commit) from 54c15b2da776d8c71366a2f952998be23933d6de (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 29b05df013426838fe166084fed9f1d56c59ebed Author: Christophe Milard <christophe.milard(a)linaro.org> Date: Wed Dec 21 13:55:56 2016 +0100 linux-gen: _ishm: fixing typos Fixing a set of iritating typos. just in comments. Signed-off-by: Christophe Milard <christophe.milard(a)linaro.org> Reviewed-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/_ishm.c b/platform/linux-generic/_ishm.c index 6d5b752..23f620d 100644 --- a/platform/linux-generic/_ishm.c +++ b/platform/linux-generic/_ishm.c @@ -76,7 +76,7 @@ /* * Maximum number of internal shared memory blocks. * - * This the the number of separate ISHM areas that can be reserved concurrently + * This is the number of separate ISHM areas that can be reserved concurrently * (Note that freeing such blocks may take time, or possibly never happen * if some of the block ownwers never procsync() after free). This number * should take that into account) @@ -240,7 +240,7 @@ static void procsync(void); * Take a piece of the preallocated virtual space to fit "size" bytes. * (best fit). Size must be rounded up to an integer number of pages size. * Possibly split the fragment to keep track of remaining space. - * Returns the allocated fragment (best_fragmnt) and the corresponding address. + * Returns the allocated fragment (best_fragment) and the corresponding address. * External caller must ensure mutex before the call! */ static void *alloc_fragment(uintptr_t size, int block_index, intptr_t align, @@ -286,11 +286,11 @@ static void *alloc_fragment(uintptr_t size, int block_index, intptr_t align, /* * if there is room between previous fragment and new one, (due to - * alignement requirement) then fragment (split) the space between + * alignment requirement) then fragment (split) the space between * the end of the previous fragment and the beginning of the new one: */ if (border - (uintptr_t)(*best_fragmnt)->start > 0) { - /* frangment space, i.e. take a new fragment descriptor... */ + /* fragment space, i.e. take a new fragment descriptor... */ rem_fragmnt = ishm_ftbl->unused_fragmnts; if (!rem_fragmnt) { ODP_ERR("unable to get shmem fragment descriptor!\n."); @@ -320,7 +320,7 @@ static void *alloc_fragment(uintptr_t size, int block_index, intptr_t align, if (remainder == 0) return (*best_fragmnt)->start; - /* otherwise, frangment space, i.e. take a new fragment descriptor... */ + /* otherwise, fragment space, i.e. take a new fragment descriptor... */ rem_fragmnt = ishm_ftbl->unused_fragmnts; if (!rem_fragmnt) { ODP_ERR("unable to get shmem fragment descriptor!\n."); @@ -515,7 +515,7 @@ static void delete_file(ishm_block_t *block) * performs the mapping, possibly allocating a fragment of the pre-reserved * VA space if the _ODP_ISHM_SINGLE_VA flag was given. * Sets fd, and returns the mapping address. - * This funstion will also set the _ODP_ISHM_SINGLE_VA flag if the alignment + * This function will also set the _ODP_ISHM_SINGLE_VA flag if the alignment * requires it * Mutex must be assured by the caller. */ @@ -736,7 +736,7 @@ static void procsync(void) last = ishm_proctable->nb_entries; while (i < last) { - /* if the procecess sequence number doesn't match the main + /* if the process sequence number doesn't match the main * table seq number, this entry is obsolete */ block = &ishm_tbl->block[ishm_proctable->entry[i].block_index]; @@ -1065,7 +1065,7 @@ static int block_free(int block_index) } /* - * Free and unmap internal shared memory, intentified by its block number: + * Free and unmap internal shared memory, identified by its block number: * return -1 on error. 0 if OK. */ int _odp_ishm_free_by_index(int block_index) @@ -1081,7 +1081,7 @@ int _odp_ishm_free_by_index(int block_index) } /* - * free and unmap internal shared memory, intentified by its block name: + * free and unmap internal shared memory, identified by its block name: * return -1 on error. 0 if OK. */ int _odp_ishm_free_by_name(const char *name) @@ -1492,8 +1492,8 @@ static int do_odp_ishm_term_local(void) * Go through the table of visible blocks for this process, * decreasing the refcnt of each visible blocks, and issuing * warning for those no longer referenced by any process. - * Note that non-referenced blocks are nor freeed: this is - * deliberate as this would imply that the sementic of the + * Note that non-referenced blocks are not freed: this is + * deliberate as this would imply that the semantic of the * freeing function would differ depending on whether we run * with odp_thread as processes or pthreads. With this approach, * the user should always free the blocks manually, which is @@ -1699,7 +1699,7 @@ int _odp_ishm_status(const char *title) fragmnt; fragmnt = fragmnt->next) nb_unused_frgments++; - ODP_DBG("ishm: %d fragment used. %d fragements unused. (total=%d)\n", + ODP_DBG("ishm: %d fragment used. %d fragments unused. (total=%d)\n", nb_used_frgments, nb_unused_frgments, nb_used_frgments + nb_unused_frgments); ----------------------------------------------------------------------- Summary of changes: platform/linux-generic/_ishm.c | 24 ++++++++++++------------ 1 file changed, 12 insertions(+), 12 deletions(-) hooks/post-receive --

9 years, 7 months

[COMMIT] branch next updated. v1.12.0.0-121-g26e0820

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, next has been updated via 26e0820a7bc833239a8a66bc15d2eab5fd3edb87 (commit) via 22c01aac0d2ff868a02a20d394e4c763b1093ec1 (commit) via cbe195a283c703d84eceaa38e81bc305627074c1 (commit) via 6f7d32d9fa1ff48f3907f8ae9c9d61d02c1e5eee (commit) via 66e87bb0ae69813a7d69a0a357347ebe60f99aba (commit) from 4f87885fe0d7301cc8ff7da6d925c18a09e7e6be (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 26e0820a7bc833239a8a66bc15d2eab5fd3edb87 Author: Christophe Milard <christophe.milard(a)linaro.org> Date: Wed Dec 21 12:11:33 2016 +0100 test: shm: checking exported vs imported block length Checking that the block size returned by odp_shm_info() matches the exported block length. Signed-off-by: Christophe Milard <christophe.milard(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/linux-generic/validation/api/shmem/shmem_odp2.c b/test/linux-generic/validation/api/shmem/shmem_odp2.c index e39dc76..7d8c682 100644 --- a/test/linux-generic/validation/api/shmem/shmem_odp2.c +++ b/test/linux-generic/validation/api/shmem/shmem_odp2.c @@ -28,6 +28,7 @@ int main(int argc, char *argv[]) odp_instance_t odp1; odp_instance_t odp2; odp_shm_t shm; + odp_shm_info_t info; test_shared_data_t *test_shared_data; /* odp init: */ @@ -59,6 +60,13 @@ int main(int argc, char *argv[]) return 1; } + /* check that the read size matches the allocated size (in other ODP):*/ + if ((odp_shm_info(shm, &info)) || + (info.size != sizeof(*test_shared_data))) { + fprintf(stderr, "error: odp_shm_info failed.\n"); + return 1; + } + test_shared_data = odp_shm_addr(shm); if (test_shared_data == NULL) { fprintf(stderr, "error: odp_shm_addr failed.\n"); commit 22c01aac0d2ff868a02a20d394e4c763b1093ec1 Author: Christophe Milard <christophe.milard(a)linaro.org> Date: Wed Dec 21 12:11:32 2016 +0100 linux-gen: _ishm: exporting/importing user len and flags The size of the shared memory and its user flag set, as given at reserve time, are exported and imported so that odp_shm_info() return proper values on imported blocks Signed-off-by: Christophe Milard <christophe.milard(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/_ishm.c b/platform/linux-generic/_ishm.c index 33ef731..8b54be2 100644 --- a/platform/linux-generic/_ishm.c +++ b/platform/linux-generic/_ishm.c @@ -125,7 +125,9 @@ #define EXPORT_FILE_LINE3_FMT "file: %s" #define EXPORT_FILE_LINE4_FMT "length: %" PRIu64 #define EXPORT_FILE_LINE5_FMT "flags: %" PRIu32 -#define EXPORT_FILE_LINE6_FMT "align: %" PRIu32 +#define EXPORT_FILE_LINE6_FMT "user_length: %" PRIu64 +#define EXPORT_FILE_LINE7_FMT "user_flags: %" PRIu32 +#define EXPORT_FILE_LINE8_FMT "align: %" PRIu32 /* * A fragment describes a piece of the shared virtual address space, * and is allocated only when allocation is done with the _ODP_ISHM_SINGLE_VA @@ -481,7 +483,11 @@ static int create_file(int block_index, huge_flag_t huge, uint64_t len, new_block->filename); fprintf(export_file, EXPORT_FILE_LINE4_FMT "\n", len); fprintf(export_file, EXPORT_FILE_LINE5_FMT "\n", flags); - fprintf(export_file, EXPORT_FILE_LINE6_FMT "\n", align); + fprintf(export_file, EXPORT_FILE_LINE6_FMT "\n", + new_block->user_len); + fprintf(export_file, EXPORT_FILE_LINE7_FMT "\n", + new_block->user_flags); + fprintf(export_file, EXPORT_FILE_LINE8_FMT "\n", align); fclose(export_file); } @@ -806,6 +812,10 @@ int _odp_ishm_reserve(const char *name, uint64_t size, int fd, else new_block->name[0] = 0; + /* save user data: */ + new_block->user_flags = user_flags; + new_block->user_len = size; + /* If a file descriptor is provided, get the real size and map: */ if (fd >= 0) { fstat(fd, &statbuf); @@ -921,9 +931,11 @@ int _odp_ishm_find_exported(const char *remote_name, pid_t external_odp_pid, FILE *export_file; uint64_t len; uint32_t flags; + uint64_t user_len; + uint32_t user_flags; uint32_t align; int fd; - int ret; + int block_index; /* try to read the block description file: */ snprintf(export_filename, ISHM_FILENAME_MAXLEN, @@ -953,7 +965,13 @@ int _odp_ishm_find_exported(const char *remote_name, pid_t external_odp_pid, if (fscanf(export_file, EXPORT_FILE_LINE5_FMT " ", &flags) != 1) goto error_exp_file; - if (fscanf(export_file, EXPORT_FILE_LINE6_FMT " ", &align) != 1) + if (fscanf(export_file, EXPORT_FILE_LINE6_FMT " ", &user_len) != 1) + goto error_exp_file; + + if (fscanf(export_file, EXPORT_FILE_LINE7_FMT " ", &user_flags) != 1) + goto error_exp_file; + + if (fscanf(export_file, EXPORT_FILE_LINE8_FMT " ", &align) != 1) goto error_exp_file; fclose(export_file); @@ -970,13 +988,17 @@ int _odp_ishm_find_exported(const char *remote_name, pid_t external_odp_pid, flags &= ~(uint32_t)_ODP_ISHM_EXPORT; /* reserve the memory, providing the opened file descriptor: */ - ret = _odp_ishm_reserve(local_name, 0, fd, align, flags, 0); - if (ret < 0) { + block_index = _odp_ishm_reserve(local_name, 0, fd, align, flags, 0); + if (block_index < 0) { close(fd); - return ret; + return block_index; } - return ret; + /* set inherited info: */ + ishm_tbl->block[block_index].user_flags = user_flags; + ishm_tbl->block[block_index].user_len = user_len; + + return block_index; error_exp_file: fclose(export_file); diff --git a/test/linux-generic/validation/api/shmem/shmem_linux.c b/test/linux-generic/validation/api/shmem/shmem_linux.c index 39473f3..2f4c762 100644 --- a/test/linux-generic/validation/api/shmem/shmem_linux.c +++ b/test/linux-generic/validation/api/shmem/shmem_linux.c @@ -102,7 +102,8 @@ */ static int read_shmem_attribues(uint64_t ext_odp_pid, const char *blockname, char *filename, uint64_t *len, - uint32_t *flags, uint32_t *align) + uint32_t *flags, uint64_t *user_len, + uint32_t *user_flags, uint32_t *align) { char shm_attr_filename[PATH_MAX]; FILE *export_file; @@ -130,6 +131,12 @@ static int read_shmem_attribues(uint64_t ext_odp_pid, const char *blockname, if (fscanf(export_file, "flags: %" PRIu32 " ", flags) != 1) goto export_file_read_err; + if (fscanf(export_file, "user_length: %" PRIu64 " ", user_len) != 1) + goto export_file_read_err; + + if (fscanf(export_file, "user_flags: %" PRIu32 " ", user_flags) != 1) + goto export_file_read_err; + if (fscanf(export_file, "align: %" PRIu32 " ", align) != 1) goto export_file_read_err; @@ -192,6 +199,8 @@ int main(int argc __attribute__((unused)), char *argv[]) char shm_filename[PATH_MAX];/* shared mem device name, under /dev/shm */ uint64_t len; uint32_t flags; + uint64_t user_len; + uint32_t user_flags; uint32_t align; int shm_fd; test_shared_linux_data_t *addr; @@ -231,7 +240,8 @@ int main(int argc __attribute__((unused)), char *argv[]) /* read the shared memory attributes (includes the shm filename): */ if (read_shmem_attribues(odp_app1, ODP_SHM_NAME, - shm_filename, &len, &flags, &align) != 0) + shm_filename, &len, &flags, + &user_len, &user_flags, &align) != 0) test_failure(fifo_name, fifo_fd, odp_app1); /* open the shm filename (which is either on /tmp or on hugetlbfs) commit cbe195a283c703d84eceaa38e81bc305627074c1 Author: Matias Elo <matias.elo(a)nokia.com> Date: Wed Dec 21 13:27:22 2016 +0200 validation: test creating pool and timer pool with no name Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Bill Fischofer <bill.fischofer(a)linaro.org> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/common_plat/validation/api/pool/pool.c b/test/common_plat/validation/api/pool/pool.c index d48ac2a..8687941 100644 --- a/test/common_plat/validation/api/pool/pool.c +++ b/test/common_plat/validation/api/pool/pool.c @@ -8,19 +8,14 @@ #include "odp_cunit_common.h" #include "pool.h" -static int pool_name_number = 1; static const int default_buffer_size = 1500; static const int default_buffer_num = 1000; static void pool_create_destroy(odp_pool_param_t *params) { odp_pool_t pool; - char pool_name[ODP_POOL_NAME_LEN]; - snprintf(pool_name, sizeof(pool_name), - "test_pool-%d", pool_name_number++); - - pool = odp_pool_create(pool_name, params); + pool = odp_pool_create(NULL, params); CU_ASSERT_FATAL(pool != ODP_POOL_INVALID); CU_ASSERT(odp_pool_to_u64(pool) != odp_pool_to_u64(ODP_POOL_INVALID)); diff --git a/test/common_plat/validation/api/timer/timer.c b/test/common_plat/validation/api/timer/timer.c index 0007639..1945afa 100644 --- a/test/common_plat/validation/api/timer/timer.c +++ b/test/common_plat/validation/api/timer/timer.c @@ -156,7 +156,7 @@ void timer_test_odp_timer_cancel(void) tparam.num_timers = 1; tparam.priv = 0; tparam.clk_src = ODP_CLOCK_CPU; - tp = odp_timer_pool_create("timer_pool0", &tparam); + tp = odp_timer_pool_create(NULL, &tparam); if (tp == ODP_TIMER_POOL_INVALID) CU_FAIL_FATAL("Timer pool create failed"); commit 6f7d32d9fa1ff48f3907f8ae9c9d61d02c1e5eee Author: Matias Elo <matias.elo(a)nokia.com> Date: Wed Dec 21 13:27:23 2016 +0200 api: move ODP_*_NAME_LEN definitions from API to implementation Enables the implementations to choose the define values. Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/pool.h b/include/odp/api/spec/pool.h index af2829b..c0de195 100644 --- a/include/odp/api/spec/pool.h +++ b/include/odp/api/spec/pool.h @@ -36,8 +36,10 @@ extern "C" { * Invalid pool */ -/** Maximum pool name length in chars including null char */ -#define ODP_POOL_NAME_LEN 32 +/** + * @def ODP_POOL_NAME_LEN + * Maximum pool name length in chars including null char + */ /** * Pool capabilities diff --git a/include/odp/api/spec/shared_memory.h b/include/odp/api/spec/shared_memory.h index 074c883..1a9c129 100644 --- a/include/odp/api/spec/shared_memory.h +++ b/include/odp/api/spec/shared_memory.h @@ -40,8 +40,10 @@ extern "C" { * Synonym for buffer pool use */ -/** Maximum shared memory block name length in chars including null char */ -#define ODP_SHM_NAME_LEN 32 +/** + * @def ODP_SHM_NAME_LEN + * Maximum shared memory block name length in chars including null char + */ /* * Shared memory flags: diff --git a/include/odp/api/spec/timer.h b/include/odp/api/spec/timer.h index 46a4369..75f9db9 100644 --- a/include/odp/api/spec/timer.h +++ b/include/odp/api/spec/timer.h @@ -90,8 +90,10 @@ typedef enum { ODP_TIMER_NOEVENT = -3 } odp_timer_set_t; -/** Maximum timer pool name length in chars including null char */ -#define ODP_TIMER_POOL_NAME_LEN 32 +/** + * @def ODP_TIMER_POOL_NAME_LEN + * Maximum timer pool name length in chars including null char + */ /** Timer pool parameters * Timer pool parameters are used when creating and querying timer pools. diff --git a/platform/linux-generic/include/odp/api/plat/pool_types.h b/platform/linux-generic/include/odp/api/plat/pool_types.h index 4e39de5..6baff09 100644 --- a/platform/linux-generic/include/odp/api/plat/pool_types.h +++ b/platform/linux-generic/include/odp/api/plat/pool_types.h @@ -30,6 +30,8 @@ typedef ODP_HANDLE_T(odp_pool_t); #define ODP_POOL_INVALID _odp_cast_scalar(odp_pool_t, 0xffffffff) +#define ODP_POOL_NAME_LEN 32 + /** * Pool type */ diff --git a/platform/linux-generic/include/odp/api/plat/shared_memory_types.h b/platform/linux-generic/include/odp/api/plat/shared_memory_types.h index 4d8bbcc..afa0bf9 100644 --- a/platform/linux-generic/include/odp/api/plat/shared_memory_types.h +++ b/platform/linux-generic/include/odp/api/plat/shared_memory_types.h @@ -31,6 +31,8 @@ typedef ODP_HANDLE_T(odp_shm_t); #define ODP_SHM_INVALID _odp_cast_scalar(odp_shm_t, 0) #define ODP_SHM_NULL ODP_SHM_INVALID +#define ODP_SHM_NAME_LEN 32 + /** Get printable format of odp_shm_t */ static inline uint64_t odp_shm_to_u64(odp_shm_t hdl) { diff --git a/platform/linux-generic/include/odp/api/plat/timer_types.h b/platform/linux-generic/include/odp/api/plat/timer_types.h index 68d6f6f..8821bed 100644 --- a/platform/linux-generic/include/odp/api/plat/timer_types.h +++ b/platform/linux-generic/include/odp/api/plat/timer_types.h @@ -30,6 +30,8 @@ typedef struct odp_timer_pool_s *odp_timer_pool_t; #define ODP_TIMER_POOL_INVALID NULL +#define ODP_TIMER_POOL_NAME_LEN 32 + typedef ODP_HANDLE_T(odp_timer_t); #define ODP_TIMER_INVALID _odp_cast_scalar(odp_timer_t, 0xffffffff) commit 66e87bb0ae69813a7d69a0a357347ebe60f99aba Author: Matias Elo <matias.elo(a)nokia.com> Date: Wed Dec 21 13:27:21 2016 +0200 api: unify ODP_*_NAME_LEN specifications Unify name length definitions to always include terminating null character. Signed-off-by: Matias Elo <matias.elo(a)nokia.com> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/classification.h b/include/odp/api/spec/classification.h index 0e442c7..0e1addd 100644 --- a/include/odp/api/spec/classification.h +++ b/include/odp/api/spec/classification.h @@ -44,7 +44,7 @@ extern "C" { /** * @def ODP_COS_NAME_LEN - * Maximum ClassOfService name length in chars + * Maximum ClassOfService name length in chars including null char */ /** diff --git a/include/odp/api/spec/pool.h b/include/odp/api/spec/pool.h index 041f4af..af2829b 100644 --- a/include/odp/api/spec/pool.h +++ b/include/odp/api/spec/pool.h @@ -36,7 +36,7 @@ extern "C" { * Invalid pool */ -/** Maximum queue name length in chars */ +/** Maximum pool name length in chars including null char */ #define ODP_POOL_NAME_LEN 32 /** diff --git a/include/odp/api/spec/queue.h b/include/odp/api/spec/queue.h index b0c5e31..7972fea 100644 --- a/include/odp/api/spec/queue.h +++ b/include/odp/api/spec/queue.h @@ -44,7 +44,7 @@ extern "C" { /** * @def ODP_QUEUE_NAME_LEN - * Maximum queue name length in chars + * Maximum queue name length in chars including null char */ /** diff --git a/include/odp/api/spec/schedule.h b/include/odp/api/spec/schedule.h index f976a4c..8244746 100644 --- a/include/odp/api/spec/schedule.h +++ b/include/odp/api/spec/schedule.h @@ -42,7 +42,7 @@ extern "C" { /** * @def ODP_SCHED_GROUP_NAME_LEN - * Maximum schedule group name length in chars + * Maximum schedule group name length in chars including null char */ /** diff --git a/include/odp/api/spec/shared_memory.h b/include/odp/api/spec/shared_memory.h index 885751d..074c883 100644 --- a/include/odp/api/spec/shared_memory.h +++ b/include/odp/api/spec/shared_memory.h @@ -40,7 +40,7 @@ extern "C" { * Synonym for buffer pool use */ -/** Maximum shared memory block name length in chars */ +/** Maximum shared memory block name length in chars including null char */ #define ODP_SHM_NAME_LEN 32 /* diff --git a/include/odp/api/spec/timer.h b/include/odp/api/spec/timer.h index 49221c4..46a4369 100644 --- a/include/odp/api/spec/timer.h +++ b/include/odp/api/spec/timer.h @@ -90,7 +90,7 @@ typedef enum { ODP_TIMER_NOEVENT = -3 } odp_timer_set_t; -/** Maximum timer pool name length in chars (including null char) */ +/** Maximum timer pool name length in chars including null char */ #define ODP_TIMER_POOL_NAME_LEN 32 /** Timer pool parameters ----------------------------------------------------------------------- Summary of changes: include/odp/api/spec/classification.h | 2 +- include/odp/api/spec/pool.h | 6 ++-- include/odp/api/spec/queue.h | 2 +- include/odp/api/spec/schedule.h | 2 +- include/odp/api/spec/shared_memory.h | 6 ++-- include/odp/api/spec/timer.h | 6 ++-- platform/linux-generic/_ishm.c | 38 +++++++++++++++++----- .../include/odp/api/plat/pool_types.h | 2 ++ .../include/odp/api/plat/shared_memory_types.h | 2 ++ .../include/odp/api/plat/timer_types.h | 2 ++ test/common_plat/validation/api/pool/pool.c | 7 +--- test/common_plat/validation/api/timer/timer.c | 2 +- .../validation/api/shmem/shmem_linux.c | 14 ++++++-- .../validation/api/shmem/shmem_odp2.c | 8 +++++ 14 files changed, 73 insertions(+), 26 deletions(-) hooks/post-receive --

9 years, 7 months

[COMMIT] branch master updated. v1.12.0.0-17-g29dde30

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, master has been updated via 29dde30264da8106d64328b0d81eb19dad0065ac (commit) from 92ae191876c39a3f5d4fba63e58dd9e8c444239b (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 29dde30264da8106d64328b0d81eb19dad0065ac Author: Stanislaw Kardach <kda(a)semihalf.com> Date: Wed Dec 28 16:12:52 2016 +0100 helper: do not break odp_term_global protocol As per API documentation, odp_term_global() should be a final ODP call when terminating the ODP application in a controlled way. Therefore ODP threads or sub-processes created using ODP helper API should not call this function but leave it to the main application thread/process. Signed-off-by: Stanislaw Kardach <kda(a)semihalf.com> Reviewed-by: Christophe Milard <christophe.milard(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/helper/linux.c b/helper/linux.c index 7bd0b07..1f009cd 100644 --- a/helper/linux.c +++ b/helper/linux.c @@ -42,8 +42,6 @@ static void *odp_run_start_routine(void *arg) if (ret < 0) ODPH_ERR("Local term failed\n"); - else if (ret == 0 && odp_term_global(thr_params->instance)) - ODPH_ERR("Global term failed\n"); return ret_ptr; } @@ -277,8 +275,6 @@ static void *odpthread_run_start_routine(void *arg) if (ret < 0) ODPH_ERR("Local term failed\n"); - else if (ret == 0 && odp_term_global(thr_params->instance)) - ODPH_ERR("Global term failed\n"); /* for process implementation of odp threads, just return status... */ if (start_args->linuxtype == ODPTHREAD_PROCESS) ----------------------------------------------------------------------- Summary of changes: helper/linux.c | 4 ---- 1 file changed, 4 deletions(-) hooks/post-receive --

9 years, 8 months

[COMMIT] branch next updated. v1.12.0.0-116-g4f87885

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, next has been updated via 4f87885fe0d7301cc8ff7da6d925c18a09e7e6be (commit) from ec84962de6f2b392bbde9e8ebc802781979b95dc (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 4f87885fe0d7301cc8ff7da6d925c18a09e7e6be Author: Bill Fischofer <bill.fischofer(a)linaro.org> Date: Wed Dec 28 14:33:12 2016 -0600 api: pktio: pktio documentation typo correction Signed-off-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/packet_io.h b/include/odp/api/spec/packet_io.h index 2dabfcc..85cd6d1 100644 --- a/include/odp/api/spec/packet_io.h +++ b/include/odp/api/spec/packet_io.h @@ -189,7 +189,7 @@ typedef struct odp_pktin_queue_param_t { /** Number of input queues to be created * - * When classifier is enabled in odp_ipktin_queue_config() this + * When classifier is enabled in odp_pktin_queue_config() this * value is ignored, otherwise at least one queue is required. * More than one input queues require flow hashing configured. * The maximum value is defined by pktio capability 'max_input_queues'. ----------------------------------------------------------------------- Summary of changes: include/odp/api/spec/packet_io.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) hooks/post-receive --

9 years, 8 months

[COMMIT] branch api-next updated. v1.12.0.0-193-g54c15b2

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, api-next has been updated via 54c15b2da776d8c71366a2f952998be23933d6de (commit) from c237528646417ba4056118b9e95ca1f690489274 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 54c15b2da776d8c71366a2f952998be23933d6de Author: Bill Fischofer <bill.fischofer(a)linaro.org> Date: Wed Dec 28 14:33:12 2016 -0600 api: pktio: pktio documentation typo correction Signed-off-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/packet_io.h b/include/odp/api/spec/packet_io.h index 2dabfcc..85cd6d1 100644 --- a/include/odp/api/spec/packet_io.h +++ b/include/odp/api/spec/packet_io.h @@ -189,7 +189,7 @@ typedef struct odp_pktin_queue_param_t { /** Number of input queues to be created * - * When classifier is enabled in odp_ipktin_queue_config() this + * When classifier is enabled in odp_pktin_queue_config() this * value is ignored, otherwise at least one queue is required. * More than one input queues require flow hashing configured. * The maximum value is defined by pktio capability 'max_input_queues'. ----------------------------------------------------------------------- Summary of changes: include/odp/api/spec/packet_io.h | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) hooks/post-receive --

9 years, 8 months

[COMMIT] branch api-next updated. v1.12.0.0-192-gc237528

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, api-next has been updated via c237528646417ba4056118b9e95ca1f690489274 (commit) via 82fe6add4af42dcbc44ffc14a7b693004791d03f (commit) from 913d9b1a83ddd6a03daa181eeb6f62fd70f506e8 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit c237528646417ba4056118b9e95ca1f690489274 Author: Christophe Milard <christophe.milard(a)linaro.org> Date: Wed Dec 21 12:11:33 2016 +0100 test: shm: checking exported vs imported block length Checking that the block size returned by odp_shm_info() matches the exported block length. Signed-off-by: Christophe Milard <christophe.milard(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/linux-generic/validation/api/shmem/shmem_odp2.c b/test/linux-generic/validation/api/shmem/shmem_odp2.c index e39dc76..7d8c682 100644 --- a/test/linux-generic/validation/api/shmem/shmem_odp2.c +++ b/test/linux-generic/validation/api/shmem/shmem_odp2.c @@ -28,6 +28,7 @@ int main(int argc, char *argv[]) odp_instance_t odp1; odp_instance_t odp2; odp_shm_t shm; + odp_shm_info_t info; test_shared_data_t *test_shared_data; /* odp init: */ @@ -59,6 +60,13 @@ int main(int argc, char *argv[]) return 1; } + /* check that the read size matches the allocated size (in other ODP):*/ + if ((odp_shm_info(shm, &info)) || + (info.size != sizeof(*test_shared_data))) { + fprintf(stderr, "error: odp_shm_info failed.\n"); + return 1; + } + test_shared_data = odp_shm_addr(shm); if (test_shared_data == NULL) { fprintf(stderr, "error: odp_shm_addr failed.\n"); commit 82fe6add4af42dcbc44ffc14a7b693004791d03f Author: Christophe Milard <christophe.milard(a)linaro.org> Date: Wed Dec 21 12:11:32 2016 +0100 linux-gen: _ishm: exporting/importing user len and flags The size of the shared memory and its user flag set, as given at reserve time, are exported and imported so that odp_shm_info() return proper values on imported blocks Signed-off-by: Christophe Milard <christophe.milard(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/_ishm.c b/platform/linux-generic/_ishm.c index efdb7bc..6d5b752 100644 --- a/platform/linux-generic/_ishm.c +++ b/platform/linux-generic/_ishm.c @@ -125,7 +125,9 @@ #define EXPORT_FILE_LINE3_FMT "file: %s" #define EXPORT_FILE_LINE4_FMT "length: %" PRIu64 #define EXPORT_FILE_LINE5_FMT "flags: %" PRIu32 -#define EXPORT_FILE_LINE6_FMT "align: %" PRIu32 +#define EXPORT_FILE_LINE6_FMT "user_length: %" PRIu64 +#define EXPORT_FILE_LINE7_FMT "user_flags: %" PRIu32 +#define EXPORT_FILE_LINE8_FMT "align: %" PRIu32 /* * A fragment describes a piece of the shared virtual address space, * and is allocated only when allocation is done with the _ODP_ISHM_SINGLE_VA @@ -481,7 +483,11 @@ static int create_file(int block_index, huge_flag_t huge, uint64_t len, new_block->filename); fprintf(export_file, EXPORT_FILE_LINE4_FMT "\n", len); fprintf(export_file, EXPORT_FILE_LINE5_FMT "\n", flags); - fprintf(export_file, EXPORT_FILE_LINE6_FMT "\n", align); + fprintf(export_file, EXPORT_FILE_LINE6_FMT "\n", + new_block->user_len); + fprintf(export_file, EXPORT_FILE_LINE7_FMT "\n", + new_block->user_flags); + fprintf(export_file, EXPORT_FILE_LINE8_FMT "\n", align); fclose(export_file); } @@ -806,6 +812,10 @@ int _odp_ishm_reserve(const char *name, uint64_t size, int fd, else new_block->name[0] = 0; + /* save user data: */ + new_block->user_flags = user_flags; + new_block->user_len = size; + /* If a file descriptor is provided, get the real size and map: */ if (fd >= 0) { fstat(fd, &statbuf); @@ -921,9 +931,11 @@ int _odp_ishm_find_exported(const char *remote_name, pid_t external_odp_pid, FILE *export_file; uint64_t len; uint32_t flags; + uint64_t user_len; + uint32_t user_flags; uint32_t align; int fd; - int ret; + int block_index; /* try to read the block description file: */ snprintf(export_filename, ISHM_FILENAME_MAXLEN, @@ -953,7 +965,13 @@ int _odp_ishm_find_exported(const char *remote_name, pid_t external_odp_pid, if (fscanf(export_file, EXPORT_FILE_LINE5_FMT " ", &flags) != 1) goto error_exp_file; - if (fscanf(export_file, EXPORT_FILE_LINE6_FMT " ", &align) != 1) + if (fscanf(export_file, EXPORT_FILE_LINE6_FMT " ", &user_len) != 1) + goto error_exp_file; + + if (fscanf(export_file, EXPORT_FILE_LINE7_FMT " ", &user_flags) != 1) + goto error_exp_file; + + if (fscanf(export_file, EXPORT_FILE_LINE8_FMT " ", &align) != 1) goto error_exp_file; fclose(export_file); @@ -970,13 +988,17 @@ int _odp_ishm_find_exported(const char *remote_name, pid_t external_odp_pid, flags &= ~(uint32_t)_ODP_ISHM_EXPORT; /* reserve the memory, providing the opened file descriptor: */ - ret = _odp_ishm_reserve(local_name, 0, fd, align, flags, 0); - if (ret < 0) { + block_index = _odp_ishm_reserve(local_name, 0, fd, align, flags, 0); + if (block_index < 0) { close(fd); - return ret; + return block_index; } - return ret; + /* set inherited info: */ + ishm_tbl->block[block_index].user_flags = user_flags; + ishm_tbl->block[block_index].user_len = user_len; + + return block_index; error_exp_file: fclose(export_file); diff --git a/test/linux-generic/validation/api/shmem/shmem_linux.c b/test/linux-generic/validation/api/shmem/shmem_linux.c index 39473f3..2f4c762 100644 --- a/test/linux-generic/validation/api/shmem/shmem_linux.c +++ b/test/linux-generic/validation/api/shmem/shmem_linux.c @@ -102,7 +102,8 @@ */ static int read_shmem_attribues(uint64_t ext_odp_pid, const char *blockname, char *filename, uint64_t *len, - uint32_t *flags, uint32_t *align) + uint32_t *flags, uint64_t *user_len, + uint32_t *user_flags, uint32_t *align) { char shm_attr_filename[PATH_MAX]; FILE *export_file; @@ -130,6 +131,12 @@ static int read_shmem_attribues(uint64_t ext_odp_pid, const char *blockname, if (fscanf(export_file, "flags: %" PRIu32 " ", flags) != 1) goto export_file_read_err; + if (fscanf(export_file, "user_length: %" PRIu64 " ", user_len) != 1) + goto export_file_read_err; + + if (fscanf(export_file, "user_flags: %" PRIu32 " ", user_flags) != 1) + goto export_file_read_err; + if (fscanf(export_file, "align: %" PRIu32 " ", align) != 1) goto export_file_read_err; @@ -192,6 +199,8 @@ int main(int argc __attribute__((unused)), char *argv[]) char shm_filename[PATH_MAX];/* shared mem device name, under /dev/shm */ uint64_t len; uint32_t flags; + uint64_t user_len; + uint32_t user_flags; uint32_t align; int shm_fd; test_shared_linux_data_t *addr; @@ -231,7 +240,8 @@ int main(int argc __attribute__((unused)), char *argv[]) /* read the shared memory attributes (includes the shm filename): */ if (read_shmem_attribues(odp_app1, ODP_SHM_NAME, - shm_filename, &len, &flags, &align) != 0) + shm_filename, &len, &flags, + &user_len, &user_flags, &align) != 0) test_failure(fifo_name, fifo_fd, odp_app1); /* open the shm filename (which is either on /tmp or on hugetlbfs) ----------------------------------------------------------------------- Summary of changes: platform/linux-generic/_ishm.c | 38 +++++++++++++++++----- .../validation/api/shmem/shmem_linux.c | 14 ++++++-- .../validation/api/shmem/shmem_odp2.c | 8 +++++ 3 files changed, 50 insertions(+), 10 deletions(-) hooks/post-receive --

9 years, 8 months

[COMMIT] branch api-next updated. v1.12.0.0-190-g913d9b1

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, api-next has been updated via 913d9b1a83ddd6a03daa181eeb6f62fd70f506e8 (commit) from dcb3c06f4c894c7e6eca93a56233a4dc00493ea3 (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit 913d9b1a83ddd6a03daa181eeb6f62fd70f506e8 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Fri Dec 23 12:33:36 2016 +0200 api: ipsec: add algorithm capabilities Added the same algorithm capabilities API that was added to crypto API. IPSEC capabilities may be more limited than bulk crypto, since IPSEC RFCs limit the set of valid key lengths, etc options. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/ipsec.h b/include/odp/api/spec/ipsec.h index 86f66e6..255bb79 100644 --- a/include/odp/api/spec/ipsec.h +++ b/include/odp/api/spec/ipsec.h @@ -441,6 +441,46 @@ typedef struct odp_ipsec_sa_param_t { int odp_ipsec_capability(odp_ipsec_capability_t *capa); /** + * Query supported IPSEC cipher algorithm capabilities + * + * Outputs all supported configuration options for the algorithm. Output is + * sorted (from the smallest to the largest) first by key length, then by IV + * length. Use this information to select key lengths, etc cipher algorithm + * options for SA creation (odp_ipsec_crypto_param_t). + * + * @param cipher Cipher algorithm + * @param[out] capa Array of capability structures for output + * @param num Maximum number of capability structures to output + * + * @return Number of capability structures for the algorithm. If this is larger + * than 'num', only 'num' first structures were output and application + * may call the function again with a larger value of 'num'. + * @retval <0 on failure + */ +int odp_ipsec_cipher_capability(odp_cipher_alg_t cipher, + odp_crypto_cipher_capability_t capa[], int num); + +/** + * Query supported IPSEC authentication algorithm capabilities + * + * Outputs all supported configuration options for the algorithm. Output is + * sorted (from the smallest to the largest) first by digest length, then by key + * length. Use this information to select key lengths, etc authentication + * algorithm options for SA creation (odp_ipsec_crypto_param_t). + * + * @param auth Authentication algorithm + * @param[out] capa Array of capability structures for output + * @param num Maximum number of capability structures to output + * + * @return Number of capability structures for the algorithm. If this is larger + * than 'num', only 'num' first structures were output and application + * may call the function again with a larger value of 'num'. + * @retval <0 on failure + */ +int odp_ipsec_auth_capability(odp_auth_alg_t auth, + odp_crypto_auth_capability_t capa[], int num); + +/** * Initialize IPSEC configuration options * * Initialize an odp_ipsec_config_t to its default values. ----------------------------------------------------------------------- Summary of changes: include/odp/api/spec/ipsec.h | 40 ++++++++++++++++++++++++++++++++++++++++ 1 file changed, 40 insertions(+) hooks/post-receive --

9 years, 8 months

[COMMIT] branch next updated. v1.12.0.0-115-gec84962

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, next has been updated via ec84962de6f2b392bbde9e8ebc802781979b95dc (commit) via 6b07d871100f5eb92a14e8b5eec56a4bf014c60a (commit) via 5ea562d5d0c4f292ede6d366e205641b0db6c454 (commit) via 9fc7566c08ea131ea24c8f95e5ac92b1b45c1aa4 (commit) via 88339aef26f169fe01f06e158835b1d2f3617db7 (commit) via fffeaa22246f15a4b7c6c41d4bcfd7943c2b33cc (commit) via 02ab1ec8f2c3e9fd6934c296d68f22ba33620706 (commit) via 3f5429ed2383056b0a085b99db6d180c39afa85e (commit) from b3ef83fd050912b82dd2dcaa2063ef1e8c12ee4d (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit ec84962de6f2b392bbde9e8ebc802781979b95dc Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:08 2016 +0200 linux-gen: config: increase max num of segments Higher segment count enables optimizations (no data copy) on operations that modifying packet length. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/include/odp_config_internal.h b/platform/linux-generic/include/odp_config_internal.h index e89a6a3..e7d84c9 100644 --- a/platform/linux-generic/include/odp_config_internal.h +++ b/platform/linux-generic/include/odp_config_internal.h @@ -75,7 +75,7 @@ extern "C" { /* * Maximum number of segments per packet */ -#define CONFIG_PACKET_MAX_SEGS 2 +#define CONFIG_PACKET_MAX_SEGS 6 /* * Maximum packet segment size including head- and tailrooms commit 6b07d871100f5eb92a14e8b5eec56a4bf014c60a Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:07 2016 +0200 validation: packet: add new concat and extend tests Added new tests for better packet concat and extend test coverage. Both small and large data length are needed to create various segmentation scenarios. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/common_plat/validation/api/packet/packet.c b/test/common_plat/validation/api/packet/packet.c index 25252a6..cf11c01 100644 --- a/test/common_plat/validation/api/packet/packet.c +++ b/test/common_plat/validation/api/packet/packet.c @@ -54,6 +54,38 @@ static void _packet_compare_data(odp_packet_t pkt1, odp_packet_t pkt2) } } +static int fill_data_forward(odp_packet_t pkt, uint32_t offset, uint32_t len, + uint32_t *cur_data) +{ + uint8_t buf[len]; + uint32_t i, data; + + data = *cur_data; + + for (i = 0; i < len; i++) + buf[i] = data++; + + *cur_data = data; + + return odp_packet_copy_from_mem(pkt, offset, len, buf); +} + +static int fill_data_backward(odp_packet_t pkt, uint32_t offset, uint32_t len, + uint32_t *cur_data) +{ + uint8_t buf[len]; + uint32_t i, data; + + data = *cur_data; + + for (i = 0; i < len; i++) + buf[len - i - 1] = data++; + + *cur_data = data; + + return odp_packet_copy_from_mem(pkt, offset, len, buf); +} + int packet_suite_init(void) { odp_pool_param_t params; @@ -1289,6 +1321,459 @@ void packet_test_concatsplit(void) odp_packet_free(pkt); } +void packet_test_concat_small(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt, pkt2; + int ret; + uint8_t *data; + uint32_t i; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_concat", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + pkt = odp_packet_alloc(pool, 1); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + data = odp_packet_data(pkt); + *data = 0; + + for (i = 0; i < len - 1; i++) { + pkt2 = odp_packet_alloc(pool, 1); + CU_ASSERT_FATAL(pkt2 != ODP_PACKET_INVALID); + + data = odp_packet_data(pkt2); + *data = i + 1; + + ret = odp_packet_concat(&pkt, pkt2); + CU_ASSERT(ret >= 0); + + if (ret < 0) { + odp_packet_free(pkt2); + break; + } + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) + CU_ASSERT(buf[i] == (i % 256)); + + odp_packet_free(pkt); + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_concat_extend_trunc(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt, pkt2; + int i, ret; + uint32_t alloc_len, ext_len, trunc_len, cur_len; + uint32_t len = 1900; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + alloc_len = len / 8; + ext_len = len / 4; + trunc_len = len / 3; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_concat", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + pkt = odp_packet_alloc(pool, alloc_len); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + cur_len = odp_packet_len(pkt); + + for (i = 0; i < 2; i++) { + pkt2 = odp_packet_alloc(pool, alloc_len); + CU_ASSERT_FATAL(pkt2 != ODP_PACKET_INVALID); + + ret = odp_packet_concat(&pkt, pkt2); + CU_ASSERT(ret >= 0); + + if (ret < 0) + odp_packet_free(pkt2); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + alloc_len)); + cur_len = odp_packet_len(pkt); + } + + ret = odp_packet_extend_tail(&pkt, ext_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + ext_len)); + cur_len = odp_packet_len(pkt); + + ret = odp_packet_extend_head(&pkt, ext_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + ext_len)); + cur_len = odp_packet_len(pkt); + + pkt2 = odp_packet_alloc(pool, alloc_len); + CU_ASSERT_FATAL(pkt2 != ODP_PACKET_INVALID); + + ret = odp_packet_concat(&pkt, pkt2); + CU_ASSERT(ret >= 0); + + if (ret < 0) + odp_packet_free(pkt2); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + alloc_len)); + cur_len = odp_packet_len(pkt); + + ret = odp_packet_trunc_head(&pkt, trunc_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len - trunc_len)); + cur_len = odp_packet_len(pkt); + + ret = odp_packet_trunc_tail(&pkt, trunc_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len - trunc_len)); + cur_len = odp_packet_len(pkt); + + odp_packet_free(pkt); + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_extend_small(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt; + int ret, round; + uint8_t *data; + uint32_t i, seg_len; + int tail = 1; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_extend", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + for (round = 0; round < 2; round++) { + pkt = odp_packet_alloc(pool, 1); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + data = odp_packet_data(pkt); + *data = 0; + + for (i = 0; i < len - 1; i++) { + if (tail) { + ret = odp_packet_extend_tail(&pkt, 1, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } else { + ret = odp_packet_extend_head(&pkt, 1, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } + + if (ret < 0) + break; + + if (tail) { + /* assert needs brackets */ + CU_ASSERT(seg_len == 1); + } else { + CU_ASSERT(seg_len > 0); + } + + *data = i + 1; + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) { + if (tail) { + /* assert needs brackets */ + CU_ASSERT(buf[i] == (i % 256)); + } else { + CU_ASSERT(buf[len - 1 - i] == (i % 256)); + } + } + + odp_packet_free(pkt); + + tail = 0; + } + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_extend_large(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt; + int ret, round; + uint8_t *data; + uint32_t i, seg_len, ext_len, cur_len, cur_data; + int tail = 1; + int num_div = 16; + int div = 1; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_extend", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + for (round = 0; round < 2 * num_div; round++) { + ext_len = len / div; + cur_len = ext_len; + + div++; + if (div > num_div) { + /* test extend head */ + div = 1; + tail = 0; + } + + pkt = odp_packet_alloc(pool, ext_len); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + cur_data = 0; + + if (tail) { + ret = fill_data_forward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } else { + ret = fill_data_backward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } + + while (cur_len < len) { + if ((len - cur_len) < ext_len) + ext_len = len - cur_len; + + if (tail) { + ret = odp_packet_extend_tail(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } else { + ret = odp_packet_extend_head(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } + + if (ret < 0) + break; + + if (tail) { + /* assert needs brackets */ + CU_ASSERT((seg_len > 0) && + (seg_len <= ext_len)); + ret = fill_data_forward(pkt, cur_len, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } else { + CU_ASSERT(seg_len > 0); + CU_ASSERT(data == odp_packet_data(pkt)); + ret = fill_data_backward(pkt, 0, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } + + cur_len += ext_len; + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) { + if (tail) { + /* assert needs brackets */ + CU_ASSERT(buf[i] == (i % 256)); + } else { + CU_ASSERT(buf[len - 1 - i] == (i % 256)); + } + } + + odp_packet_free(pkt); + } + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_extend_mix(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt; + int ret, round; + uint8_t *data; + uint32_t i, seg_len, ext_len, cur_len, cur_data; + int small_count; + int tail = 1; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_extend", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + for (round = 0; round < 2; round++) { + small_count = 30; + ext_len = len / 10; + cur_len = ext_len; + + pkt = odp_packet_alloc(pool, ext_len); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + cur_data = 0; + + if (tail) { + ret = fill_data_forward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } else { + ret = fill_data_backward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } + + while (cur_len < len) { + if (small_count) { + small_count--; + ext_len = len / 100; + } else { + ext_len = len / 4; + } + + if ((len - cur_len) < ext_len) + ext_len = len - cur_len; + + if (tail) { + ret = odp_packet_extend_tail(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + CU_ASSERT((seg_len > 0) && + (seg_len <= ext_len)); + ret = fill_data_forward(pkt, cur_len, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } else { + ret = odp_packet_extend_head(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + CU_ASSERT(seg_len > 0); + CU_ASSERT(data == odp_packet_data(pkt)); + ret = fill_data_backward(pkt, 0, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } + + cur_len += ext_len; + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) { + if (tail) { + /* assert needs brackets */ + CU_ASSERT(buf[i] == (i % 256)); + } else { + CU_ASSERT(buf[len - 1 - i] == (i % 256)); + } + } + + odp_packet_free(pkt); + + tail = 0; + } + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + void packet_test_align(void) { odp_packet_t pkt; @@ -1414,6 +1899,11 @@ odp_testinfo_t packet_suite[] = { ODP_TEST_INFO(packet_test_copy), ODP_TEST_INFO(packet_test_copydata), ODP_TEST_INFO(packet_test_concatsplit), + ODP_TEST_INFO(packet_test_concat_small), + ODP_TEST_INFO(packet_test_concat_extend_trunc), + ODP_TEST_INFO(packet_test_extend_small), + ODP_TEST_INFO(packet_test_extend_large), + ODP_TEST_INFO(packet_test_extend_mix), ODP_TEST_INFO(packet_test_align), ODP_TEST_INFO(packet_test_offset), ODP_TEST_INFO_NULL, diff --git a/test/common_plat/validation/api/packet/packet.h b/test/common_plat/validation/api/packet/packet.h index 10a377c..9bc3d63 100644 --- a/test/common_plat/validation/api/packet/packet.h +++ b/test/common_plat/validation/api/packet/packet.h @@ -30,6 +30,11 @@ void packet_test_add_rem_data(void); void packet_test_copy(void); void packet_test_copydata(void); void packet_test_concatsplit(void); +void packet_test_concat_small(void); +void packet_test_concat_extend_trunc(void); +void packet_test_extend_small(void); +void packet_test_extend_large(void); +void packet_test_extend_mix(void); void packet_test_align(void); void packet_test_offset(void); commit 5ea562d5d0c4f292ede6d366e205641b0db6c454 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:06 2016 +0200 linux-gen: packet: optimize concat Optimized concat operation to avoid packet copy when destination packet has room to link source packet segments. Since concat uses extend tail it was also modified to handle variable segment sizes. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/include/odp_buffer_internal.h b/platform/linux-generic/include/odp_buffer_internal.h index 4cc51d3..6149290 100644 --- a/platform/linux-generic/include/odp_buffer_internal.h +++ b/platform/linux-generic/include/odp_buffer_internal.h @@ -75,7 +75,7 @@ struct odp_buffer_hdr_t { uint32_t size; /* Initial buffer data pointer and length */ - void *base_data; + uint8_t *base_data; uint32_t base_len; uint8_t *buf_end; diff --git a/platform/linux-generic/odp_packet.c b/platform/linux-generic/odp_packet.c index 2d9e3e6..58b6f32 100644 --- a/platform/linux-generic/odp_packet.c +++ b/platform/linux-generic/odp_packet.c @@ -74,6 +74,15 @@ static inline void *packet_tail(odp_packet_hdr_t *pkt_hdr) return pkt_hdr->buf_hdr.seg[last].data + seg_len; } +static inline uint32_t seg_headroom(odp_packet_hdr_t *pkt_hdr, int seg) +{ + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint8_t *base = hdr->base_data; + uint8_t *head = pkt_hdr->buf_hdr.seg[seg].data; + + return CONFIG_PACKET_HEADROOM + (head - base); +} + static inline uint32_t seg_tailroom(odp_packet_hdr_t *pkt_hdr, int seg) { uint32_t seg_len = pkt_hdr->buf_hdr.seg[seg].len; @@ -297,7 +306,7 @@ static inline int num_segments(uint32_t len) return num; } -static inline void copy_all_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from) +static inline void add_all_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from) { int i; int n = to->buf_hdr.segcount; @@ -313,52 +322,53 @@ static inline void copy_all_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from) } static inline void copy_num_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from, - int num) + int first, int num) { int i; for (i = 0; i < num; i++) { - to->buf_hdr.seg[i].hdr = from->buf_hdr.seg[num + i].hdr; - to->buf_hdr.seg[i].data = from->buf_hdr.seg[num + i].data; - to->buf_hdr.seg[i].len = from->buf_hdr.seg[num + i].len; + to->buf_hdr.seg[i].hdr = from->buf_hdr.seg[first + i].hdr; + to->buf_hdr.seg[i].data = from->buf_hdr.seg[first + i].data; + to->buf_hdr.seg[i].len = from->buf_hdr.seg[first + i].len; } to->buf_hdr.segcount = num; } -static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, - uint32_t len, int head) +static inline odp_packet_hdr_t *alloc_segments(pool_t *pool, int num) { - pool_t *pool = pool_entry_from_hdl(pkt_hdr->buf_hdr.pool_hdl); - odp_packet_hdr_t *new_hdr; - int num, ret; - uint32_t seg_len, offset; + odp_buffer_t buf[num]; + int ret; - num = num_segments(len); + ret = buffer_alloc_multi(pool, buf, NULL, num); + if (odp_unlikely(ret != num)) { + if (ret > 0) + buffer_free_multi(buf, ret); - if ((pkt_hdr->buf_hdr.segcount + num) > CONFIG_PACKET_MAX_SEGS) return NULL; + } - { - odp_buffer_t buf[num]; + return init_segments(buf, num); +} - ret = buffer_alloc_multi(pool, buf, NULL, num); - if (odp_unlikely(ret != num)) { - if (ret > 0) - buffer_free_multi(buf, ret); +static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, + pool_t *pool, uint32_t len, + int num, int head) +{ + odp_packet_hdr_t *new_hdr; + uint32_t seg_len, offset; - return NULL; - } + new_hdr = alloc_segments(pool, num); - new_hdr = init_segments(buf, num); - } + if (new_hdr == NULL) + return NULL; seg_len = len - ((num - 1) * pool->max_seg_len); offset = pool->max_seg_len - seg_len; if (head) { /* add into the head*/ - copy_all_segs(new_hdr, pkt_hdr); + add_all_segs(new_hdr, pkt_hdr); /* adjust first segment length */ new_hdr->buf_hdr.seg[0].data += offset; @@ -374,7 +384,7 @@ static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, int last; /* add into the tail */ - copy_all_segs(pkt_hdr, new_hdr); + add_all_segs(pkt_hdr, new_hdr); /* adjust last segment length */ last = packet_last_seg(pkt_hdr); @@ -387,49 +397,60 @@ static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, return pkt_hdr; } +static inline void free_bufs(odp_packet_hdr_t *pkt_hdr, int first, int num) +{ + int i; + odp_buffer_t buf[num]; + + for (i = 0; i < num; i++) + buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[first + i].hdr); + + buffer_free_multi(buf, num); +} + static inline odp_packet_hdr_t *free_segments(odp_packet_hdr_t *pkt_hdr, int num, uint32_t free_len, uint32_t pull_len, int head) { - int i; - odp_buffer_t buf[num]; - int n = pkt_hdr->buf_hdr.segcount - num; + int num_remain = pkt_hdr->buf_hdr.segcount - num; if (head) { odp_packet_hdr_t *new_hdr; + int i; + odp_buffer_t buf[num]; for (i = 0; i < num; i++) buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[i].hdr); /* First remaining segment is the new packet descriptor */ new_hdr = pkt_hdr->buf_hdr.seg[num].hdr; - copy_num_segs(new_hdr, pkt_hdr, n); + + copy_num_segs(new_hdr, pkt_hdr, num, num_remain); packet_seg_copy_md(new_hdr, pkt_hdr); /* Tailroom not changed */ new_hdr->tailroom = pkt_hdr->tailroom; - /* No headroom in non-first segments */ - new_hdr->headroom = 0; + new_hdr->headroom = seg_headroom(new_hdr, 0); new_hdr->frame_len = pkt_hdr->frame_len - free_len; pull_head(new_hdr, pull_len); pkt_hdr = new_hdr; + + buffer_free_multi(buf, num); } else { - for (i = 0; i < num; i++) - buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[n + i].hdr); + /* Free last 'num' bufs */ + free_bufs(pkt_hdr, num_remain, num); /* Head segment remains, no need to copy or update majority * of the metadata. */ - pkt_hdr->buf_hdr.segcount = n; + pkt_hdr->buf_hdr.segcount = num_remain; pkt_hdr->frame_len -= free_len; - pkt_hdr->tailroom = seg_tailroom(pkt_hdr, n - 1); + pkt_hdr->tailroom = seg_tailroom(pkt_hdr, num_remain - 1); pull_tail(pkt_hdr, pull_len); } - buffer_free_multi(buf, num); - return pkt_hdr; } @@ -530,19 +551,10 @@ void odp_packet_free(odp_packet_t pkt) odp_packet_hdr_t *pkt_hdr = odp_packet_hdr(pkt); int num_seg = pkt_hdr->buf_hdr.segcount; - if (odp_likely(CONFIG_PACKET_MAX_SEGS == 1 || num_seg == 1)) { + if (odp_likely(CONFIG_PACKET_MAX_SEGS == 1 || num_seg == 1)) buffer_free_multi((odp_buffer_t *)&pkt, 1); - } else { - odp_buffer_t buf[num_seg]; - int i; - - buf[0] = (odp_buffer_t)pkt; - - for (i = 1; i < num_seg; i++) - buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[i].hdr); - - buffer_free_multi(buf, num_seg); - } + else + free_bufs(pkt_hdr, 0, num_seg); } void odp_packet_free_multi(const odp_packet_t pkt[], int num) @@ -676,25 +688,277 @@ void *odp_packet_push_head(odp_packet_t pkt, uint32_t len) return packet_data(pkt_hdr); } +static inline uint32_t pack_seg_head(odp_packet_hdr_t *pkt_hdr, int seg) +{ + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint32_t len = pkt_hdr->buf_hdr.seg[seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[seg].data; + uint8_t *dst = hdr->base_data; + + if (dst != src) { + memmove(dst, src, len); + pkt_hdr->buf_hdr.seg[seg].data = dst; + } + + return len; +} + +static inline uint32_t pack_seg_tail(odp_packet_hdr_t *pkt_hdr, int seg) +{ + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint32_t len = pkt_hdr->buf_hdr.seg[seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[seg].data; + uint8_t *dst = hdr->base_data + hdr->base_len - len; + + if (dst != src) { + memmove(dst, src, len); + pkt_hdr->buf_hdr.seg[seg].data = dst; + } + + return len; +} + +static inline uint32_t fill_seg_head(odp_packet_hdr_t *pkt_hdr, int dst_seg, + int src_seg, uint32_t max_len) +{ + uint32_t len = pkt_hdr->buf_hdr.seg[src_seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[src_seg].data; + uint32_t offset = pkt_hdr->buf_hdr.seg[dst_seg].len; + uint8_t *dst = pkt_hdr->buf_hdr.seg[dst_seg].data + offset; + + if (len > max_len) + len = max_len; + + memmove(dst, src, len); + + pkt_hdr->buf_hdr.seg[dst_seg].len += len; + pkt_hdr->buf_hdr.seg[src_seg].len -= len; + pkt_hdr->buf_hdr.seg[src_seg].data += len; + + if (pkt_hdr->buf_hdr.seg[src_seg].len == 0) { + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[src_seg].hdr; + + pkt_hdr->buf_hdr.seg[src_seg].data = hdr->base_data; + } + + return len; +} + +static inline uint32_t fill_seg_tail(odp_packet_hdr_t *pkt_hdr, int dst_seg, + int src_seg, uint32_t max_len) +{ + uint32_t src_len = pkt_hdr->buf_hdr.seg[src_seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[src_seg].data; + uint8_t *dst = pkt_hdr->buf_hdr.seg[dst_seg].data; + uint32_t len = src_len; + + if (len > max_len) + len = max_len; + + src += src_len - len; + dst -= len; + + memmove(dst, src, len); + + pkt_hdr->buf_hdr.seg[dst_seg].data -= len; + pkt_hdr->buf_hdr.seg[dst_seg].len += len; + pkt_hdr->buf_hdr.seg[src_seg].len -= len; + + if (pkt_hdr->buf_hdr.seg[src_seg].len == 0) { + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[src_seg].hdr; + + pkt_hdr->buf_hdr.seg[src_seg].data = hdr->base_data; + } + + return len; +} + +static inline int move_data_to_head(odp_packet_hdr_t *pkt_hdr, int segs) +{ + int dst_seg, src_seg; + uint32_t base_len, len, free_len; + uint32_t moved = 0; + + base_len = pkt_hdr->buf_hdr.base_len; + + for (dst_seg = 0; dst_seg < segs; dst_seg++) { + len = pack_seg_head(pkt_hdr, dst_seg); + moved += len; + + if (len == base_len) + continue; + + free_len = base_len - len; + + for (src_seg = dst_seg + 1; src_seg < segs; src_seg++) { + len = fill_seg_head(pkt_hdr, dst_seg, src_seg, + free_len); + moved += len; + + if (len == free_len) { + /* dst seg is full */ + break; + } + + /* src seg is empty */ + free_len -= len; + } + + if (moved == pkt_hdr->frame_len) + break; + } + + /* last segment which have data */ + return dst_seg; +} + +static inline int move_data_to_tail(odp_packet_hdr_t *pkt_hdr, int segs) +{ + int dst_seg, src_seg; + uint32_t base_len, len, free_len; + uint32_t moved = 0; + + base_len = pkt_hdr->buf_hdr.base_len; + + for (dst_seg = segs - 1; dst_seg >= 0; dst_seg--) { + len = pack_seg_tail(pkt_hdr, dst_seg); + moved += len; + + if (len == base_len) + continue; + + free_len = base_len - len; + + for (src_seg = dst_seg - 1; src_seg >= 0; src_seg--) { + len = fill_seg_tail(pkt_hdr, dst_seg, src_seg, + free_len); + moved += len; + + if (len == free_len) { + /* dst seg is full */ + break; + } + + /* src seg is empty */ + free_len -= len; + } + + if (moved == pkt_hdr->frame_len) + break; + } + + /* first segment which have data */ + return dst_seg; +} + +static inline void reset_seg(odp_packet_hdr_t *pkt_hdr, int first, int num) +{ + odp_buffer_hdr_t *hdr; + void *base; + int i; + uint32_t base_len = pkt_hdr->buf_hdr.base_len; + + for (i = first; i < first + num; i++) { + hdr = pkt_hdr->buf_hdr.seg[i].hdr; + base = hdr->base_data; + pkt_hdr->buf_hdr.seg[i].len = base_len; + pkt_hdr->buf_hdr.seg[i].data = base; + } +} + int odp_packet_extend_head(odp_packet_t *pkt, uint32_t len, void **data_ptr, uint32_t *seg_len) { odp_packet_hdr_t *pkt_hdr = odp_packet_hdr(*pkt); - odp_packet_hdr_t *new_hdr; - uint32_t headroom = pkt_hdr->headroom; + uint32_t frame_len = pkt_hdr->frame_len; + uint32_t headroom = pkt_hdr->headroom; + int ret = 0; if (len > headroom) { - push_head(pkt_hdr, headroom); - new_hdr = add_segments(pkt_hdr, len - headroom, 1); + pool_t *pool = pool_entry_from_hdl(pkt_hdr->buf_hdr.pool_hdl); + int num; + int segs; - if (new_hdr == NULL) { - /* segment alloc failed, rollback changes */ - pull_head(pkt_hdr, headroom); + if (odp_unlikely((frame_len + len) > pool->max_len)) return -1; - } - *pkt = packet_handle(new_hdr); - pkt_hdr = new_hdr; + num = num_segments(len - headroom); + segs = pkt_hdr->buf_hdr.segcount; + + if (odp_unlikely((segs + num) > CONFIG_PACKET_MAX_SEGS)) { + /* Cannot directly add new segments */ + odp_packet_hdr_t *new_hdr; + int new_segs = 0; + int free_segs = 0; + uint32_t base_len = pkt_hdr->buf_hdr.base_len; + uint32_t offset; + + num = num_segments(frame_len + len); + + if (num > segs) { + /* Allocate additional segments */ + new_segs = num - segs; + new_hdr = alloc_segments(pool, new_segs); + + if (new_hdr == NULL) + return -1; + + } else if (num < segs) { + free_segs = segs - num; + } + + /* Pack all data to packet tail */ + move_data_to_tail(pkt_hdr, segs); + reset_seg(pkt_hdr, 0, segs); + + if (new_segs) { + add_all_segs(new_hdr, pkt_hdr); + packet_seg_copy_md(new_hdr, pkt_hdr); + segs += new_segs; + + pkt_hdr = new_hdr; + *pkt = packet_handle(pkt_hdr); + } else if (free_segs) { + new_hdr = pkt_hdr->buf_hdr.seg[free_segs].hdr; + packet_seg_copy_md(new_hdr, pkt_hdr); + + /* Free extra segs */ + free_bufs(pkt_hdr, 0, free_segs); + + segs -= free_segs; + pkt_hdr = new_hdr; + *pkt = packet_handle(pkt_hdr); + } + + frame_len += len; + offset = (segs * base_len) - frame_len; + + pkt_hdr->buf_hdr.seg[0].data += offset; + pkt_hdr->buf_hdr.seg[0].len -= offset; + + pkt_hdr->buf_hdr.segcount = segs; + pkt_hdr->frame_len = frame_len; + pkt_hdr->headroom = offset + pool->headroom; + pkt_hdr->tailroom = pool->tailroom; + + /* Data was moved */ + ret = 1; + } else { + void *ptr; + + push_head(pkt_hdr, headroom); + ptr = add_segments(pkt_hdr, pool, len - headroom, + num, 1); + + if (ptr == NULL) { + /* segment alloc failed, rollback changes */ + pull_head(pkt_hdr, headroom); + return -1; + } + + *pkt = packet_handle(ptr); + pkt_hdr = ptr; + } } else { push_head(pkt_hdr, len); } @@ -705,7 +969,7 @@ int odp_packet_extend_head(odp_packet_t *pkt, uint32_t len, if (seg_len) *seg_len = packet_first_seg_len(pkt_hdr); - return 0; + return ret; } void *odp_packet_pull_head(odp_packet_t pkt, uint32_t len) @@ -769,30 +1033,96 @@ void *odp_packet_push_tail(odp_packet_t pkt, uint32_t len) } int odp_packet_extend_tail(odp_packet_t *pkt, uint32_t len, - void **data_ptr, uint32_t *seg_len) + void **data_ptr, uint32_t *seg_len_out) { odp_packet_hdr_t *pkt_hdr = odp_packet_hdr(*pkt); - void *ret; - uint32_t tailroom = pkt_hdr->tailroom; - uint32_t tail_off = pkt_hdr->frame_len; + uint32_t frame_len = pkt_hdr->frame_len; + uint32_t tailroom = pkt_hdr->tailroom; + uint32_t tail_off = frame_len; + int ret = 0; if (len > tailroom) { - push_tail(pkt_hdr, tailroom); - ret = add_segments(pkt_hdr, len - tailroom, 0); + pool_t *pool = pool_entry_from_hdl(pkt_hdr->buf_hdr.pool_hdl); + int num; + int segs; - if (ret == NULL) { - /* segment alloc failed, rollback changes */ - pull_tail(pkt_hdr, tailroom); + if (odp_unlikely((frame_len + len) > pool->max_len)) return -1; + + num = num_segments(len - tailroom); + segs = pkt_hdr->buf_hdr.segcount; + + if (odp_unlikely((segs + num) > CONFIG_PACKET_MAX_SEGS)) { + /* Cannot directly add new segments */ + odp_packet_hdr_t *new_hdr; + int new_segs = 0; + int free_segs = 0; + uint32_t base_len = pkt_hdr->buf_hdr.base_len; + uint32_t offset; + + num = num_segments(frame_len + len); + + if (num > segs) { + /* Allocate additional segments */ + new_segs = num - segs; + new_hdr = alloc_segments(pool, new_segs); + + if (new_hdr == NULL) + return -1; + + } else if (num < segs) { + free_segs = segs - num; + } + + /* Pack all data to packet head */ + move_data_to_head(pkt_hdr, segs); + reset_seg(pkt_hdr, 0, segs); + + if (new_segs) { + /* Add new segs */ + add_all_segs(pkt_hdr, new_hdr); + segs += new_segs; + } else if (free_segs) { + /* Free extra segs */ + free_bufs(pkt_hdr, segs - free_segs, free_segs); + + segs -= free_segs; + } + + frame_len += len; + offset = (segs * base_len) - frame_len; + + pkt_hdr->buf_hdr.seg[segs - 1].len -= offset; + + pkt_hdr->buf_hdr.segcount = segs; + pkt_hdr->frame_len = frame_len; + pkt_hdr->headroom = pool->headroom; + pkt_hdr->tailroom = offset + pool->tailroom; + + /* Data was moved */ + ret = 1; + } else { + void *ptr; + + push_tail(pkt_hdr, tailroom); + + ptr = add_segments(pkt_hdr, pool, len - tailroom, + num, 0); + + if (ptr == NULL) { + /* segment alloc failed, rollback changes */ + pull_tail(pkt_hdr, tailroom); + return -1; + } } } else { push_tail(pkt_hdr, len); } if (data_ptr) - *data_ptr = packet_map(pkt_hdr, tail_off, seg_len, NULL); + *data_ptr = packet_map(pkt_hdr, tail_off, seg_len_out, NULL); - return 0; + return ret; } void *odp_packet_pull_tail(odp_packet_t pkt, uint32_t len) @@ -1199,19 +1529,38 @@ int odp_packet_align(odp_packet_t *pkt, uint32_t offset, uint32_t len, int odp_packet_concat(odp_packet_t *dst, odp_packet_t src) { - uint32_t dst_len = odp_packet_len(*dst); - uint32_t src_len = odp_packet_len(src); - - if (odp_packet_extend_tail(dst, src_len, NULL, NULL) >= 0) { - (void)odp_packet_copy_from_pkt(*dst, dst_len, - src, 0, src_len); - if (src != *dst) + odp_packet_hdr_t *dst_hdr = odp_packet_hdr(*dst); + odp_packet_hdr_t *src_hdr = odp_packet_hdr(src); + int dst_segs = dst_hdr->buf_hdr.segcount; + int src_segs = src_hdr->buf_hdr.segcount; + odp_pool_t dst_pool = dst_hdr->buf_hdr.pool_hdl; + odp_pool_t src_pool = src_hdr->buf_hdr.pool_hdl; + uint32_t dst_len = dst_hdr->frame_len; + uint32_t src_len = src_hdr->frame_len; + + /* Do a copy if resulting packet would be out of segments or packets + * are from different pools. */ + if (odp_unlikely((dst_segs + src_segs) > CONFIG_PACKET_MAX_SEGS) || + odp_unlikely(dst_pool != src_pool)) { + if (odp_packet_extend_tail(dst, src_len, NULL, NULL) >= 0) { + (void)odp_packet_copy_from_pkt(*dst, dst_len, + src, 0, src_len); odp_packet_free(src); - return 0; + /* Data was moved in memory */ + return 1; + } + + return -1; } - return -1; + add_all_segs(dst_hdr, src_hdr); + + dst_hdr->frame_len = dst_len + src_len; + dst_hdr->tailroom = src_hdr->tailroom; + + /* Data was not moved in memory */ + return 0; } int odp_packet_split(odp_packet_t *pkt, uint32_t len, odp_packet_t *tail) commit 9fc7566c08ea131ea24c8f95e5ac92b1b45c1aa4 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:05 2016 +0200 validation: packet: concat-split test bug fix Successful concat return values are 0 and >0. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/common_plat/validation/api/packet/packet.c b/test/common_plat/validation/api/packet/packet.c index 3ad00ed..25252a6 100644 --- a/test/common_plat/validation/api/packet/packet.c +++ b/test/common_plat/validation/api/packet/packet.c @@ -1232,7 +1232,7 @@ void packet_test_concatsplit(void) CU_ASSERT(pkt_len == odp_packet_len(pkt)); CU_ASSERT(pkt_len == odp_packet_len(pkt2)); - CU_ASSERT(odp_packet_concat(&pkt, pkt2) == 0); + CU_ASSERT(odp_packet_concat(&pkt, pkt2) >= 0); CU_ASSERT(odp_packet_len(pkt) == pkt_len * 2); _packet_compare_offset(pkt, 0, pkt, pkt_len, pkt_len); @@ -1262,7 +1262,7 @@ void packet_test_concatsplit(void) _packet_compare_offset(splits[0], 0, segmented_test_packet, pkt_len / 2, odp_packet_len(splits[0])); - CU_ASSERT(odp_packet_concat(&pkt, splits[0]) == 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[0]) >= 0); _packet_compare_offset(pkt, 0, segmented_test_packet, 0, pkt_len / 2); _packet_compare_offset(pkt, pkt_len / 2, segmented_test_packet, pkt_len / 2, pkt_len / 2); @@ -1279,9 +1279,9 @@ void packet_test_concatsplit(void) CU_ASSERT(odp_packet_len(splits[0]) + odp_packet_len(splits[1]) + odp_packet_len(splits[2]) + odp_packet_len(pkt) == pkt_len); - CU_ASSERT(odp_packet_concat(&pkt, splits[2]) == 0); - CU_ASSERT(odp_packet_concat(&pkt, splits[1]) == 0); - CU_ASSERT(odp_packet_concat(&pkt, splits[0]) == 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[2]) >= 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[1]) >= 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[0]) >= 0); CU_ASSERT(odp_packet_len(pkt) == odp_packet_len(segmented_test_packet)); _packet_compare_data(pkt, segmented_test_packet); commit 88339aef26f169fe01f06e158835b1d2f3617db7 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:04 2016 +0200 linux-gen: packet: improve packet print Added segmentation and head-/tailroom information to packet print out. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/odp_packet.c b/platform/linux-generic/odp_packet.c index 10fbded..2d9e3e6 100644 --- a/platform/linux-generic/odp_packet.c +++ b/platform/linux-generic/odp_packet.c @@ -1395,6 +1395,7 @@ int odp_packet_move_data(odp_packet_t pkt, uint32_t dst_offset, void odp_packet_print(odp_packet_t pkt) { + odp_packet_seg_t seg; int max_len = 512; char str[max_len]; int len = 0; @@ -1421,6 +1422,25 @@ void odp_packet_print(odp_packet_t pkt) len += snprintf(&str[len], n - len, " input %" PRIu64 "\n", odp_pktio_to_u64(hdr->input)); + len += snprintf(&str[len], n - len, + " headroom %" PRIu32 "\n", + odp_packet_headroom(pkt)); + len += snprintf(&str[len], n - len, + " tailroom %" PRIu32 "\n", + odp_packet_tailroom(pkt)); + len += snprintf(&str[len], n - len, + " num_segs %i\n", odp_packet_num_segs(pkt)); + + seg = odp_packet_first_seg(pkt); + + while (seg != ODP_PACKET_SEG_INVALID) { + len += snprintf(&str[len], n - len, + " seg_len %" PRIu32 "\n", + odp_packet_seg_data_len(pkt, seg)); + + seg = odp_packet_next_seg(pkt, seg); + } + str[len] = '\0'; ODP_PRINT("\n%s\n", str); commit fffeaa22246f15a4b7c6c41d4bcfd7943c2b33cc Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:03 2016 +0200 linux-gen: packet: fix bug in tailroom calculation Tailroom is calculated from the end of the last segment, not from the first. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/odp_packet.c b/platform/linux-generic/odp_packet.c index 0d3fd05..10fbded 100644 --- a/platform/linux-generic/odp_packet.c +++ b/platform/linux-generic/odp_packet.c @@ -74,6 +74,15 @@ static inline void *packet_tail(odp_packet_hdr_t *pkt_hdr) return pkt_hdr->buf_hdr.seg[last].data + seg_len; } +static inline uint32_t seg_tailroom(odp_packet_hdr_t *pkt_hdr, int seg) +{ + uint32_t seg_len = pkt_hdr->buf_hdr.seg[seg].len; + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint8_t *tail = pkt_hdr->buf_hdr.seg[seg].data + seg_len; + + return hdr->buf_end - tail; +} + static inline void push_head(odp_packet_hdr_t *pkt_hdr, uint32_t len) { pkt_hdr->headroom -= len; @@ -383,11 +392,12 @@ static inline odp_packet_hdr_t *free_segments(odp_packet_hdr_t *pkt_hdr, uint32_t pull_len, int head) { int i; - odp_packet_hdr_t *new_hdr; odp_buffer_t buf[num]; int n = pkt_hdr->buf_hdr.segcount - num; if (head) { + odp_packet_hdr_t *new_hdr; + for (i = 0; i < num; i++) buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[i].hdr); @@ -413,8 +423,7 @@ static inline odp_packet_hdr_t *free_segments(odp_packet_hdr_t *pkt_hdr, * of the metadata. */ pkt_hdr->buf_hdr.segcount = n; pkt_hdr->frame_len -= free_len; - pkt_hdr->tailroom = pkt_hdr->buf_hdr.buf_end - - (uint8_t *)packet_tail(pkt_hdr); + pkt_hdr->tailroom = seg_tailroom(pkt_hdr, n - 1); pull_tail(pkt_hdr, pull_len); } commit 02ab1ec8f2c3e9fd6934c296d68f22ba33620706 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:02 2016 +0200 api: packet: src and dst packet must not be the same Concat and copy_from_pkt operations must be called with src and dst packet handles which refer to the same packet. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-by: Balasubramanian Manoharan <bala.manoharan(a)linaro.org> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/packet.h b/include/odp/api/spec/packet.h index faf62e2..4a86eba 100644 --- a/include/odp/api/spec/packet.h +++ b/include/odp/api/spec/packet.h @@ -781,7 +781,8 @@ uint32_t odp_packet_seg_data_len(odp_packet_t pkt, odp_packet_seg_t seg); * Concatenate all packet data from 'src' packet into tail of 'dst' packet. * Operation preserves 'dst' packet metadata in the resulting packet, * while 'src' packet handle, metadata and old segment handles for both packets - * become invalid. + * become invalid. Source and destination packet handles must not refer to + * the same packet. * * A successful operation overwrites 'dst' packet handle with a new handle, * which application must use as the reference to the resulting packet @@ -928,6 +929,9 @@ int odp_packet_copy_from_mem(odp_packet_t pkt, uint32_t offset, * Copy 'len' bytes of data from 'src' packet to 'dst' packet. Copy starts from * the specified source and destination packet offsets. Copied areas * (offset ... offset + len) must not exceed their packet data lengths. + * Source and destination packet handles must not refer to the same packet (use + * odp_packet_copy_data() or odp_packet_move_data() for a single packet). + * * Packet is not modified on an error. * * @param dst Destination packet handle commit 3f5429ed2383056b0a085b99db6d180c39afa85e Author: Balasubramanian Manoharan <bala.manoharan(a)linaro.org> Date: Fri Nov 4 23:04:30 2016 +0530 api: pktio: adds further definition for classification configuration Updates classification configuration documentation. Signed-off-by: Balasubramanian Manoharan <bala.manoharan(a)linaro.org> Reviewed-by: Petri Savolainen <petri.savolainen(a)nokia.com> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/packet_io.h b/include/odp/api/spec/packet_io.h index d46e405..2dabfcc 100644 --- a/include/odp/api/spec/packet_io.h +++ b/include/odp/api/spec/packet_io.h @@ -189,12 +189,11 @@ typedef struct odp_pktin_queue_param_t { /** Number of input queues to be created * - * When classifier is enabled the number of queues may be zero - * (in odp_pktin_queue_config() step), otherwise at least one - * queue is required. More than one input queues require either flow - * hashing or classifier enabled. The maximum value is defined by - * pktio capability 'max_input_queues'. Queue type is defined by the - * input mode. The default value is 1. */ + * When classifier is enabled in odp_ipktin_queue_config() this + * value is ignored, otherwise at least one queue is required. + * More than one input queues require flow hashing configured. + * The maximum value is defined by pktio capability 'max_input_queues'. + * Queue type is defined by the input mode. The default value is 1. */ unsigned num_queues; /** Queue parameters @@ -202,7 +201,9 @@ typedef struct odp_pktin_queue_param_t { * These are used for input queue creation in ODP_PKTIN_MODE_QUEUE * or ODP_PKTIN_MODE_SCHED modes. Scheduler parameters are considered * only in ODP_PKTIN_MODE_SCHED mode. Default values are defined in - * odp_queue_param_t documentation. */ + * odp_queue_param_t documentation. + * When classifier is enabled in odp_pktin_queue_config() this + * value is ignored. */ odp_queue_param_t queue_param; } odp_pktin_queue_param_t; @@ -887,6 +888,8 @@ int odp_pktio_mac_addr(odp_pktio_t pktio, void *mac_addr, int size); * * @retval 0 on success * @retval <0 on failure + * + * @note The default_cos has to be unique per odp_pktio_t instance. */ int odp_pktio_default_cos_set(odp_pktio_t pktio, odp_cos_t default_cos); ----------------------------------------------------------------------- Summary of changes: include/odp/api/spec/packet.h | 6 +- include/odp/api/spec/packet_io.h | 17 +- .../linux-generic/include/odp_buffer_internal.h | 2 +- .../linux-generic/include/odp_config_internal.h | 2 +- platform/linux-generic/odp_packet.c | 542 +++++++++++++++++---- test/common_plat/validation/api/packet/packet.c | 500 ++++++++++++++++++- test/common_plat/validation/api/packet/packet.h | 5 + 7 files changed, 977 insertions(+), 97 deletions(-) hooks/post-receive --

9 years, 8 months

[COMMIT] branch api-next updated. v1.12.0.0-189-gdcb3c06

by git＠git-us.linaro.org

This is an automated email from the git hooks/post-receive script. It was generated because a ref change was pushed to the repository containing the project "". The branch, api-next has been updated via dcb3c06f4c894c7e6eca93a56233a4dc00493ea3 (commit) via 1f5f12f6c7d3862f1c5c29b0773a1473c3f24b06 (commit) via 0f329751b46ce9b0378c10bf8b0040cd010736c8 (commit) via db0c6cd9dc0dd3821aaa1320faa24c12ec578c41 (commit) via 894bc720a8a849a6bd9b41cf20d483f9da19793c (commit) via 0b9981a71186cd065a7429cb8fd116e70b20fc23 (commit) via 910f5cda1a83fcfcc8f14d491b2956dceea4e827 (commit) from 3b20a21d0a45af4f78db993449052462590db22d (commit) Those revisions listed above that are new to this repository have not appeared on any other notification email; so we list those revisions in full, below. - Log ----------------------------------------------------------------- commit dcb3c06f4c894c7e6eca93a56233a4dc00493ea3 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:08 2016 +0200 linux-gen: config: increase max num of segments Higher segment count enables optimizations (no data copy) on operations that modifying packet length. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/include/odp_config_internal.h b/platform/linux-generic/include/odp_config_internal.h index c494660..dadd59e 100644 --- a/platform/linux-generic/include/odp_config_internal.h +++ b/platform/linux-generic/include/odp_config_internal.h @@ -75,7 +75,7 @@ extern "C" { /* * Maximum number of segments per packet */ -#define CONFIG_PACKET_MAX_SEGS 2 +#define CONFIG_PACKET_MAX_SEGS 6 /* * Maximum packet segment size including head- and tailrooms commit 1f5f12f6c7d3862f1c5c29b0773a1473c3f24b06 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:07 2016 +0200 validation: packet: add new concat and extend tests Added new tests for better packet concat and extend test coverage. Both small and large data length are needed to create various segmentation scenarios. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/common_plat/validation/api/packet/packet.c b/test/common_plat/validation/api/packet/packet.c index 25252a6..cf11c01 100644 --- a/test/common_plat/validation/api/packet/packet.c +++ b/test/common_plat/validation/api/packet/packet.c @@ -54,6 +54,38 @@ static void _packet_compare_data(odp_packet_t pkt1, odp_packet_t pkt2) } } +static int fill_data_forward(odp_packet_t pkt, uint32_t offset, uint32_t len, + uint32_t *cur_data) +{ + uint8_t buf[len]; + uint32_t i, data; + + data = *cur_data; + + for (i = 0; i < len; i++) + buf[i] = data++; + + *cur_data = data; + + return odp_packet_copy_from_mem(pkt, offset, len, buf); +} + +static int fill_data_backward(odp_packet_t pkt, uint32_t offset, uint32_t len, + uint32_t *cur_data) +{ + uint8_t buf[len]; + uint32_t i, data; + + data = *cur_data; + + for (i = 0; i < len; i++) + buf[len - i - 1] = data++; + + *cur_data = data; + + return odp_packet_copy_from_mem(pkt, offset, len, buf); +} + int packet_suite_init(void) { odp_pool_param_t params; @@ -1289,6 +1321,459 @@ void packet_test_concatsplit(void) odp_packet_free(pkt); } +void packet_test_concat_small(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt, pkt2; + int ret; + uint8_t *data; + uint32_t i; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_concat", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + pkt = odp_packet_alloc(pool, 1); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + data = odp_packet_data(pkt); + *data = 0; + + for (i = 0; i < len - 1; i++) { + pkt2 = odp_packet_alloc(pool, 1); + CU_ASSERT_FATAL(pkt2 != ODP_PACKET_INVALID); + + data = odp_packet_data(pkt2); + *data = i + 1; + + ret = odp_packet_concat(&pkt, pkt2); + CU_ASSERT(ret >= 0); + + if (ret < 0) { + odp_packet_free(pkt2); + break; + } + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) + CU_ASSERT(buf[i] == (i % 256)); + + odp_packet_free(pkt); + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_concat_extend_trunc(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt, pkt2; + int i, ret; + uint32_t alloc_len, ext_len, trunc_len, cur_len; + uint32_t len = 1900; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + alloc_len = len / 8; + ext_len = len / 4; + trunc_len = len / 3; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_concat", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + pkt = odp_packet_alloc(pool, alloc_len); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + cur_len = odp_packet_len(pkt); + + for (i = 0; i < 2; i++) { + pkt2 = odp_packet_alloc(pool, alloc_len); + CU_ASSERT_FATAL(pkt2 != ODP_PACKET_INVALID); + + ret = odp_packet_concat(&pkt, pkt2); + CU_ASSERT(ret >= 0); + + if (ret < 0) + odp_packet_free(pkt2); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + alloc_len)); + cur_len = odp_packet_len(pkt); + } + + ret = odp_packet_extend_tail(&pkt, ext_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + ext_len)); + cur_len = odp_packet_len(pkt); + + ret = odp_packet_extend_head(&pkt, ext_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + ext_len)); + cur_len = odp_packet_len(pkt); + + pkt2 = odp_packet_alloc(pool, alloc_len); + CU_ASSERT_FATAL(pkt2 != ODP_PACKET_INVALID); + + ret = odp_packet_concat(&pkt, pkt2); + CU_ASSERT(ret >= 0); + + if (ret < 0) + odp_packet_free(pkt2); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len + alloc_len)); + cur_len = odp_packet_len(pkt); + + ret = odp_packet_trunc_head(&pkt, trunc_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len - trunc_len)); + cur_len = odp_packet_len(pkt); + + ret = odp_packet_trunc_tail(&pkt, trunc_len, NULL, NULL); + CU_ASSERT(ret >= 0); + + CU_ASSERT(odp_packet_len(pkt) == (cur_len - trunc_len)); + cur_len = odp_packet_len(pkt); + + odp_packet_free(pkt); + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_extend_small(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt; + int ret, round; + uint8_t *data; + uint32_t i, seg_len; + int tail = 1; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_extend", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + for (round = 0; round < 2; round++) { + pkt = odp_packet_alloc(pool, 1); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + data = odp_packet_data(pkt); + *data = 0; + + for (i = 0; i < len - 1; i++) { + if (tail) { + ret = odp_packet_extend_tail(&pkt, 1, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } else { + ret = odp_packet_extend_head(&pkt, 1, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } + + if (ret < 0) + break; + + if (tail) { + /* assert needs brackets */ + CU_ASSERT(seg_len == 1); + } else { + CU_ASSERT(seg_len > 0); + } + + *data = i + 1; + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) { + if (tail) { + /* assert needs brackets */ + CU_ASSERT(buf[i] == (i % 256)); + } else { + CU_ASSERT(buf[len - 1 - i] == (i % 256)); + } + } + + odp_packet_free(pkt); + + tail = 0; + } + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_extend_large(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt; + int ret, round; + uint8_t *data; + uint32_t i, seg_len, ext_len, cur_len, cur_data; + int tail = 1; + int num_div = 16; + int div = 1; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_extend", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + for (round = 0; round < 2 * num_div; round++) { + ext_len = len / div; + cur_len = ext_len; + + div++; + if (div > num_div) { + /* test extend head */ + div = 1; + tail = 0; + } + + pkt = odp_packet_alloc(pool, ext_len); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + cur_data = 0; + + if (tail) { + ret = fill_data_forward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } else { + ret = fill_data_backward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } + + while (cur_len < len) { + if ((len - cur_len) < ext_len) + ext_len = len - cur_len; + + if (tail) { + ret = odp_packet_extend_tail(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } else { + ret = odp_packet_extend_head(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + } + + if (ret < 0) + break; + + if (tail) { + /* assert needs brackets */ + CU_ASSERT((seg_len > 0) && + (seg_len <= ext_len)); + ret = fill_data_forward(pkt, cur_len, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } else { + CU_ASSERT(seg_len > 0); + CU_ASSERT(data == odp_packet_data(pkt)); + ret = fill_data_backward(pkt, 0, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } + + cur_len += ext_len; + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) { + if (tail) { + /* assert needs brackets */ + CU_ASSERT(buf[i] == (i % 256)); + } else { + CU_ASSERT(buf[len - 1 - i] == (i % 256)); + } + } + + odp_packet_free(pkt); + } + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + +void packet_test_extend_mix(void) +{ + odp_pool_capability_t capa; + odp_pool_t pool; + odp_pool_param_t param; + odp_packet_t pkt; + int ret, round; + uint8_t *data; + uint32_t i, seg_len, ext_len, cur_len, cur_data; + int small_count; + int tail = 1; + uint32_t len = 32000; + uint8_t buf[len]; + + CU_ASSERT_FATAL(odp_pool_capability(&capa) == 0); + + if (capa.pkt.max_len && capa.pkt.max_len < len) + len = capa.pkt.max_len; + + odp_pool_param_init(&param); + + param.type = ODP_POOL_PACKET; + param.pkt.len = len; + param.pkt.num = 100; + + pool = odp_pool_create("packet_pool_extend", &param); + CU_ASSERT(packet_pool != ODP_POOL_INVALID); + + for (round = 0; round < 2; round++) { + small_count = 30; + ext_len = len / 10; + cur_len = ext_len; + + pkt = odp_packet_alloc(pool, ext_len); + CU_ASSERT_FATAL(pkt != ODP_PACKET_INVALID); + + cur_data = 0; + + if (tail) { + ret = fill_data_forward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } else { + ret = fill_data_backward(pkt, 0, ext_len, &cur_data); + CU_ASSERT(ret == 0); + } + + while (cur_len < len) { + if (small_count) { + small_count--; + ext_len = len / 100; + } else { + ext_len = len / 4; + } + + if ((len - cur_len) < ext_len) + ext_len = len - cur_len; + + if (tail) { + ret = odp_packet_extend_tail(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + CU_ASSERT((seg_len > 0) && + (seg_len <= ext_len)); + ret = fill_data_forward(pkt, cur_len, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } else { + ret = odp_packet_extend_head(&pkt, ext_len, + (void **)&data, + &seg_len); + CU_ASSERT(ret >= 0); + CU_ASSERT(seg_len > 0); + CU_ASSERT(data == odp_packet_data(pkt)); + ret = fill_data_backward(pkt, 0, ext_len, + &cur_data); + CU_ASSERT(ret == 0); + } + + cur_len += ext_len; + } + + CU_ASSERT(odp_packet_len(pkt) == len); + + len = odp_packet_len(pkt); + + memset(buf, 0, len); + CU_ASSERT(odp_packet_copy_to_mem(pkt, 0, len, buf) == 0); + + for (i = 0; i < len; i++) { + if (tail) { + /* assert needs brackets */ + CU_ASSERT(buf[i] == (i % 256)); + } else { + CU_ASSERT(buf[len - 1 - i] == (i % 256)); + } + } + + odp_packet_free(pkt); + + tail = 0; + } + + CU_ASSERT(odp_pool_destroy(pool) == 0); +} + void packet_test_align(void) { odp_packet_t pkt; @@ -1414,6 +1899,11 @@ odp_testinfo_t packet_suite[] = { ODP_TEST_INFO(packet_test_copy), ODP_TEST_INFO(packet_test_copydata), ODP_TEST_INFO(packet_test_concatsplit), + ODP_TEST_INFO(packet_test_concat_small), + ODP_TEST_INFO(packet_test_concat_extend_trunc), + ODP_TEST_INFO(packet_test_extend_small), + ODP_TEST_INFO(packet_test_extend_large), + ODP_TEST_INFO(packet_test_extend_mix), ODP_TEST_INFO(packet_test_align), ODP_TEST_INFO(packet_test_offset), ODP_TEST_INFO_NULL, diff --git a/test/common_plat/validation/api/packet/packet.h b/test/common_plat/validation/api/packet/packet.h index 10a377c..9bc3d63 100644 --- a/test/common_plat/validation/api/packet/packet.h +++ b/test/common_plat/validation/api/packet/packet.h @@ -30,6 +30,11 @@ void packet_test_add_rem_data(void); void packet_test_copy(void); void packet_test_copydata(void); void packet_test_concatsplit(void); +void packet_test_concat_small(void); +void packet_test_concat_extend_trunc(void); +void packet_test_extend_small(void); +void packet_test_extend_large(void); +void packet_test_extend_mix(void); void packet_test_align(void); void packet_test_offset(void); commit 0f329751b46ce9b0378c10bf8b0040cd010736c8 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:06 2016 +0200 linux-gen: packet: optimize concat Optimized concat operation to avoid packet copy when destination packet has room to link source packet segments. Since concat uses extend tail it was also modified to handle variable segment sizes. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/include/odp_buffer_internal.h b/platform/linux-generic/include/odp_buffer_internal.h index 4cc51d3..6149290 100644 --- a/platform/linux-generic/include/odp_buffer_internal.h +++ b/platform/linux-generic/include/odp_buffer_internal.h @@ -75,7 +75,7 @@ struct odp_buffer_hdr_t { uint32_t size; /* Initial buffer data pointer and length */ - void *base_data; + uint8_t *base_data; uint32_t base_len; uint8_t *buf_end; diff --git a/platform/linux-generic/odp_packet.c b/platform/linux-generic/odp_packet.c index 2d9e3e6..58b6f32 100644 --- a/platform/linux-generic/odp_packet.c +++ b/platform/linux-generic/odp_packet.c @@ -74,6 +74,15 @@ static inline void *packet_tail(odp_packet_hdr_t *pkt_hdr) return pkt_hdr->buf_hdr.seg[last].data + seg_len; } +static inline uint32_t seg_headroom(odp_packet_hdr_t *pkt_hdr, int seg) +{ + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint8_t *base = hdr->base_data; + uint8_t *head = pkt_hdr->buf_hdr.seg[seg].data; + + return CONFIG_PACKET_HEADROOM + (head - base); +} + static inline uint32_t seg_tailroom(odp_packet_hdr_t *pkt_hdr, int seg) { uint32_t seg_len = pkt_hdr->buf_hdr.seg[seg].len; @@ -297,7 +306,7 @@ static inline int num_segments(uint32_t len) return num; } -static inline void copy_all_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from) +static inline void add_all_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from) { int i; int n = to->buf_hdr.segcount; @@ -313,52 +322,53 @@ static inline void copy_all_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from) } static inline void copy_num_segs(odp_packet_hdr_t *to, odp_packet_hdr_t *from, - int num) + int first, int num) { int i; for (i = 0; i < num; i++) { - to->buf_hdr.seg[i].hdr = from->buf_hdr.seg[num + i].hdr; - to->buf_hdr.seg[i].data = from->buf_hdr.seg[num + i].data; - to->buf_hdr.seg[i].len = from->buf_hdr.seg[num + i].len; + to->buf_hdr.seg[i].hdr = from->buf_hdr.seg[first + i].hdr; + to->buf_hdr.seg[i].data = from->buf_hdr.seg[first + i].data; + to->buf_hdr.seg[i].len = from->buf_hdr.seg[first + i].len; } to->buf_hdr.segcount = num; } -static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, - uint32_t len, int head) +static inline odp_packet_hdr_t *alloc_segments(pool_t *pool, int num) { - pool_t *pool = pool_entry_from_hdl(pkt_hdr->buf_hdr.pool_hdl); - odp_packet_hdr_t *new_hdr; - int num, ret; - uint32_t seg_len, offset; + odp_buffer_t buf[num]; + int ret; - num = num_segments(len); + ret = buffer_alloc_multi(pool, buf, NULL, num); + if (odp_unlikely(ret != num)) { + if (ret > 0) + buffer_free_multi(buf, ret); - if ((pkt_hdr->buf_hdr.segcount + num) > CONFIG_PACKET_MAX_SEGS) return NULL; + } - { - odp_buffer_t buf[num]; + return init_segments(buf, num); +} - ret = buffer_alloc_multi(pool, buf, NULL, num); - if (odp_unlikely(ret != num)) { - if (ret > 0) - buffer_free_multi(buf, ret); +static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, + pool_t *pool, uint32_t len, + int num, int head) +{ + odp_packet_hdr_t *new_hdr; + uint32_t seg_len, offset; - return NULL; - } + new_hdr = alloc_segments(pool, num); - new_hdr = init_segments(buf, num); - } + if (new_hdr == NULL) + return NULL; seg_len = len - ((num - 1) * pool->max_seg_len); offset = pool->max_seg_len - seg_len; if (head) { /* add into the head*/ - copy_all_segs(new_hdr, pkt_hdr); + add_all_segs(new_hdr, pkt_hdr); /* adjust first segment length */ new_hdr->buf_hdr.seg[0].data += offset; @@ -374,7 +384,7 @@ static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, int last; /* add into the tail */ - copy_all_segs(pkt_hdr, new_hdr); + add_all_segs(pkt_hdr, new_hdr); /* adjust last segment length */ last = packet_last_seg(pkt_hdr); @@ -387,49 +397,60 @@ static inline odp_packet_hdr_t *add_segments(odp_packet_hdr_t *pkt_hdr, return pkt_hdr; } +static inline void free_bufs(odp_packet_hdr_t *pkt_hdr, int first, int num) +{ + int i; + odp_buffer_t buf[num]; + + for (i = 0; i < num; i++) + buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[first + i].hdr); + + buffer_free_multi(buf, num); +} + static inline odp_packet_hdr_t *free_segments(odp_packet_hdr_t *pkt_hdr, int num, uint32_t free_len, uint32_t pull_len, int head) { - int i; - odp_buffer_t buf[num]; - int n = pkt_hdr->buf_hdr.segcount - num; + int num_remain = pkt_hdr->buf_hdr.segcount - num; if (head) { odp_packet_hdr_t *new_hdr; + int i; + odp_buffer_t buf[num]; for (i = 0; i < num; i++) buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[i].hdr); /* First remaining segment is the new packet descriptor */ new_hdr = pkt_hdr->buf_hdr.seg[num].hdr; - copy_num_segs(new_hdr, pkt_hdr, n); + + copy_num_segs(new_hdr, pkt_hdr, num, num_remain); packet_seg_copy_md(new_hdr, pkt_hdr); /* Tailroom not changed */ new_hdr->tailroom = pkt_hdr->tailroom; - /* No headroom in non-first segments */ - new_hdr->headroom = 0; + new_hdr->headroom = seg_headroom(new_hdr, 0); new_hdr->frame_len = pkt_hdr->frame_len - free_len; pull_head(new_hdr, pull_len); pkt_hdr = new_hdr; + + buffer_free_multi(buf, num); } else { - for (i = 0; i < num; i++) - buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[n + i].hdr); + /* Free last 'num' bufs */ + free_bufs(pkt_hdr, num_remain, num); /* Head segment remains, no need to copy or update majority * of the metadata. */ - pkt_hdr->buf_hdr.segcount = n; + pkt_hdr->buf_hdr.segcount = num_remain; pkt_hdr->frame_len -= free_len; - pkt_hdr->tailroom = seg_tailroom(pkt_hdr, n - 1); + pkt_hdr->tailroom = seg_tailroom(pkt_hdr, num_remain - 1); pull_tail(pkt_hdr, pull_len); } - buffer_free_multi(buf, num); - return pkt_hdr; } @@ -530,19 +551,10 @@ void odp_packet_free(odp_packet_t pkt) odp_packet_hdr_t *pkt_hdr = odp_packet_hdr(pkt); int num_seg = pkt_hdr->buf_hdr.segcount; - if (odp_likely(CONFIG_PACKET_MAX_SEGS == 1 || num_seg == 1)) { + if (odp_likely(CONFIG_PACKET_MAX_SEGS == 1 || num_seg == 1)) buffer_free_multi((odp_buffer_t *)&pkt, 1); - } else { - odp_buffer_t buf[num_seg]; - int i; - - buf[0] = (odp_buffer_t)pkt; - - for (i = 1; i < num_seg; i++) - buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[i].hdr); - - buffer_free_multi(buf, num_seg); - } + else + free_bufs(pkt_hdr, 0, num_seg); } void odp_packet_free_multi(const odp_packet_t pkt[], int num) @@ -676,25 +688,277 @@ void *odp_packet_push_head(odp_packet_t pkt, uint32_t len) return packet_data(pkt_hdr); } +static inline uint32_t pack_seg_head(odp_packet_hdr_t *pkt_hdr, int seg) +{ + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint32_t len = pkt_hdr->buf_hdr.seg[seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[seg].data; + uint8_t *dst = hdr->base_data; + + if (dst != src) { + memmove(dst, src, len); + pkt_hdr->buf_hdr.seg[seg].data = dst; + } + + return len; +} + +static inline uint32_t pack_seg_tail(odp_packet_hdr_t *pkt_hdr, int seg) +{ + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint32_t len = pkt_hdr->buf_hdr.seg[seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[seg].data; + uint8_t *dst = hdr->base_data + hdr->base_len - len; + + if (dst != src) { + memmove(dst, src, len); + pkt_hdr->buf_hdr.seg[seg].data = dst; + } + + return len; +} + +static inline uint32_t fill_seg_head(odp_packet_hdr_t *pkt_hdr, int dst_seg, + int src_seg, uint32_t max_len) +{ + uint32_t len = pkt_hdr->buf_hdr.seg[src_seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[src_seg].data; + uint32_t offset = pkt_hdr->buf_hdr.seg[dst_seg].len; + uint8_t *dst = pkt_hdr->buf_hdr.seg[dst_seg].data + offset; + + if (len > max_len) + len = max_len; + + memmove(dst, src, len); + + pkt_hdr->buf_hdr.seg[dst_seg].len += len; + pkt_hdr->buf_hdr.seg[src_seg].len -= len; + pkt_hdr->buf_hdr.seg[src_seg].data += len; + + if (pkt_hdr->buf_hdr.seg[src_seg].len == 0) { + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[src_seg].hdr; + + pkt_hdr->buf_hdr.seg[src_seg].data = hdr->base_data; + } + + return len; +} + +static inline uint32_t fill_seg_tail(odp_packet_hdr_t *pkt_hdr, int dst_seg, + int src_seg, uint32_t max_len) +{ + uint32_t src_len = pkt_hdr->buf_hdr.seg[src_seg].len; + uint8_t *src = pkt_hdr->buf_hdr.seg[src_seg].data; + uint8_t *dst = pkt_hdr->buf_hdr.seg[dst_seg].data; + uint32_t len = src_len; + + if (len > max_len) + len = max_len; + + src += src_len - len; + dst -= len; + + memmove(dst, src, len); + + pkt_hdr->buf_hdr.seg[dst_seg].data -= len; + pkt_hdr->buf_hdr.seg[dst_seg].len += len; + pkt_hdr->buf_hdr.seg[src_seg].len -= len; + + if (pkt_hdr->buf_hdr.seg[src_seg].len == 0) { + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[src_seg].hdr; + + pkt_hdr->buf_hdr.seg[src_seg].data = hdr->base_data; + } + + return len; +} + +static inline int move_data_to_head(odp_packet_hdr_t *pkt_hdr, int segs) +{ + int dst_seg, src_seg; + uint32_t base_len, len, free_len; + uint32_t moved = 0; + + base_len = pkt_hdr->buf_hdr.base_len; + + for (dst_seg = 0; dst_seg < segs; dst_seg++) { + len = pack_seg_head(pkt_hdr, dst_seg); + moved += len; + + if (len == base_len) + continue; + + free_len = base_len - len; + + for (src_seg = dst_seg + 1; src_seg < segs; src_seg++) { + len = fill_seg_head(pkt_hdr, dst_seg, src_seg, + free_len); + moved += len; + + if (len == free_len) { + /* dst seg is full */ + break; + } + + /* src seg is empty */ + free_len -= len; + } + + if (moved == pkt_hdr->frame_len) + break; + } + + /* last segment which have data */ + return dst_seg; +} + +static inline int move_data_to_tail(odp_packet_hdr_t *pkt_hdr, int segs) +{ + int dst_seg, src_seg; + uint32_t base_len, len, free_len; + uint32_t moved = 0; + + base_len = pkt_hdr->buf_hdr.base_len; + + for (dst_seg = segs - 1; dst_seg >= 0; dst_seg--) { + len = pack_seg_tail(pkt_hdr, dst_seg); + moved += len; + + if (len == base_len) + continue; + + free_len = base_len - len; + + for (src_seg = dst_seg - 1; src_seg >= 0; src_seg--) { + len = fill_seg_tail(pkt_hdr, dst_seg, src_seg, + free_len); + moved += len; + + if (len == free_len) { + /* dst seg is full */ + break; + } + + /* src seg is empty */ + free_len -= len; + } + + if (moved == pkt_hdr->frame_len) + break; + } + + /* first segment which have data */ + return dst_seg; +} + +static inline void reset_seg(odp_packet_hdr_t *pkt_hdr, int first, int num) +{ + odp_buffer_hdr_t *hdr; + void *base; + int i; + uint32_t base_len = pkt_hdr->buf_hdr.base_len; + + for (i = first; i < first + num; i++) { + hdr = pkt_hdr->buf_hdr.seg[i].hdr; + base = hdr->base_data; + pkt_hdr->buf_hdr.seg[i].len = base_len; + pkt_hdr->buf_hdr.seg[i].data = base; + } +} + int odp_packet_extend_head(odp_packet_t *pkt, uint32_t len, void **data_ptr, uint32_t *seg_len) { odp_packet_hdr_t *pkt_hdr = odp_packet_hdr(*pkt); - odp_packet_hdr_t *new_hdr; - uint32_t headroom = pkt_hdr->headroom; + uint32_t frame_len = pkt_hdr->frame_len; + uint32_t headroom = pkt_hdr->headroom; + int ret = 0; if (len > headroom) { - push_head(pkt_hdr, headroom); - new_hdr = add_segments(pkt_hdr, len - headroom, 1); + pool_t *pool = pool_entry_from_hdl(pkt_hdr->buf_hdr.pool_hdl); + int num; + int segs; - if (new_hdr == NULL) { - /* segment alloc failed, rollback changes */ - pull_head(pkt_hdr, headroom); + if (odp_unlikely((frame_len + len) > pool->max_len)) return -1; - } - *pkt = packet_handle(new_hdr); - pkt_hdr = new_hdr; + num = num_segments(len - headroom); + segs = pkt_hdr->buf_hdr.segcount; + + if (odp_unlikely((segs + num) > CONFIG_PACKET_MAX_SEGS)) { + /* Cannot directly add new segments */ + odp_packet_hdr_t *new_hdr; + int new_segs = 0; + int free_segs = 0; + uint32_t base_len = pkt_hdr->buf_hdr.base_len; + uint32_t offset; + + num = num_segments(frame_len + len); + + if (num > segs) { + /* Allocate additional segments */ + new_segs = num - segs; + new_hdr = alloc_segments(pool, new_segs); + + if (new_hdr == NULL) + return -1; + + } else if (num < segs) { + free_segs = segs - num; + } + + /* Pack all data to packet tail */ + move_data_to_tail(pkt_hdr, segs); + reset_seg(pkt_hdr, 0, segs); + + if (new_segs) { + add_all_segs(new_hdr, pkt_hdr); + packet_seg_copy_md(new_hdr, pkt_hdr); + segs += new_segs; + + pkt_hdr = new_hdr; + *pkt = packet_handle(pkt_hdr); + } else if (free_segs) { + new_hdr = pkt_hdr->buf_hdr.seg[free_segs].hdr; + packet_seg_copy_md(new_hdr, pkt_hdr); + + /* Free extra segs */ + free_bufs(pkt_hdr, 0, free_segs); + + segs -= free_segs; + pkt_hdr = new_hdr; + *pkt = packet_handle(pkt_hdr); + } + + frame_len += len; + offset = (segs * base_len) - frame_len; + + pkt_hdr->buf_hdr.seg[0].data += offset; + pkt_hdr->buf_hdr.seg[0].len -= offset; + + pkt_hdr->buf_hdr.segcount = segs; + pkt_hdr->frame_len = frame_len; + pkt_hdr->headroom = offset + pool->headroom; + pkt_hdr->tailroom = pool->tailroom; + + /* Data was moved */ + ret = 1; + } else { + void *ptr; + + push_head(pkt_hdr, headroom); + ptr = add_segments(pkt_hdr, pool, len - headroom, + num, 1); + + if (ptr == NULL) { + /* segment alloc failed, rollback changes */ + pull_head(pkt_hdr, headroom); + return -1; + } + + *pkt = packet_handle(ptr); + pkt_hdr = ptr; + } } else { push_head(pkt_hdr, len); } @@ -705,7 +969,7 @@ int odp_packet_extend_head(odp_packet_t *pkt, uint32_t len, if (seg_len) *seg_len = packet_first_seg_len(pkt_hdr); - return 0; + return ret; } void *odp_packet_pull_head(odp_packet_t pkt, uint32_t len) @@ -769,30 +1033,96 @@ void *odp_packet_push_tail(odp_packet_t pkt, uint32_t len) } int odp_packet_extend_tail(odp_packet_t *pkt, uint32_t len, - void **data_ptr, uint32_t *seg_len) + void **data_ptr, uint32_t *seg_len_out) { odp_packet_hdr_t *pkt_hdr = odp_packet_hdr(*pkt); - void *ret; - uint32_t tailroom = pkt_hdr->tailroom; - uint32_t tail_off = pkt_hdr->frame_len; + uint32_t frame_len = pkt_hdr->frame_len; + uint32_t tailroom = pkt_hdr->tailroom; + uint32_t tail_off = frame_len; + int ret = 0; if (len > tailroom) { - push_tail(pkt_hdr, tailroom); - ret = add_segments(pkt_hdr, len - tailroom, 0); + pool_t *pool = pool_entry_from_hdl(pkt_hdr->buf_hdr.pool_hdl); + int num; + int segs; - if (ret == NULL) { - /* segment alloc failed, rollback changes */ - pull_tail(pkt_hdr, tailroom); + if (odp_unlikely((frame_len + len) > pool->max_len)) return -1; + + num = num_segments(len - tailroom); + segs = pkt_hdr->buf_hdr.segcount; + + if (odp_unlikely((segs + num) > CONFIG_PACKET_MAX_SEGS)) { + /* Cannot directly add new segments */ + odp_packet_hdr_t *new_hdr; + int new_segs = 0; + int free_segs = 0; + uint32_t base_len = pkt_hdr->buf_hdr.base_len; + uint32_t offset; + + num = num_segments(frame_len + len); + + if (num > segs) { + /* Allocate additional segments */ + new_segs = num - segs; + new_hdr = alloc_segments(pool, new_segs); + + if (new_hdr == NULL) + return -1; + + } else if (num < segs) { + free_segs = segs - num; + } + + /* Pack all data to packet head */ + move_data_to_head(pkt_hdr, segs); + reset_seg(pkt_hdr, 0, segs); + + if (new_segs) { + /* Add new segs */ + add_all_segs(pkt_hdr, new_hdr); + segs += new_segs; + } else if (free_segs) { + /* Free extra segs */ + free_bufs(pkt_hdr, segs - free_segs, free_segs); + + segs -= free_segs; + } + + frame_len += len; + offset = (segs * base_len) - frame_len; + + pkt_hdr->buf_hdr.seg[segs - 1].len -= offset; + + pkt_hdr->buf_hdr.segcount = segs; + pkt_hdr->frame_len = frame_len; + pkt_hdr->headroom = pool->headroom; + pkt_hdr->tailroom = offset + pool->tailroom; + + /* Data was moved */ + ret = 1; + } else { + void *ptr; + + push_tail(pkt_hdr, tailroom); + + ptr = add_segments(pkt_hdr, pool, len - tailroom, + num, 0); + + if (ptr == NULL) { + /* segment alloc failed, rollback changes */ + pull_tail(pkt_hdr, tailroom); + return -1; + } } } else { push_tail(pkt_hdr, len); } if (data_ptr) - *data_ptr = packet_map(pkt_hdr, tail_off, seg_len, NULL); + *data_ptr = packet_map(pkt_hdr, tail_off, seg_len_out, NULL); - return 0; + return ret; } void *odp_packet_pull_tail(odp_packet_t pkt, uint32_t len) @@ -1199,19 +1529,38 @@ int odp_packet_align(odp_packet_t *pkt, uint32_t offset, uint32_t len, int odp_packet_concat(odp_packet_t *dst, odp_packet_t src) { - uint32_t dst_len = odp_packet_len(*dst); - uint32_t src_len = odp_packet_len(src); - - if (odp_packet_extend_tail(dst, src_len, NULL, NULL) >= 0) { - (void)odp_packet_copy_from_pkt(*dst, dst_len, - src, 0, src_len); - if (src != *dst) + odp_packet_hdr_t *dst_hdr = odp_packet_hdr(*dst); + odp_packet_hdr_t *src_hdr = odp_packet_hdr(src); + int dst_segs = dst_hdr->buf_hdr.segcount; + int src_segs = src_hdr->buf_hdr.segcount; + odp_pool_t dst_pool = dst_hdr->buf_hdr.pool_hdl; + odp_pool_t src_pool = src_hdr->buf_hdr.pool_hdl; + uint32_t dst_len = dst_hdr->frame_len; + uint32_t src_len = src_hdr->frame_len; + + /* Do a copy if resulting packet would be out of segments or packets + * are from different pools. */ + if (odp_unlikely((dst_segs + src_segs) > CONFIG_PACKET_MAX_SEGS) || + odp_unlikely(dst_pool != src_pool)) { + if (odp_packet_extend_tail(dst, src_len, NULL, NULL) >= 0) { + (void)odp_packet_copy_from_pkt(*dst, dst_len, + src, 0, src_len); odp_packet_free(src); - return 0; + /* Data was moved in memory */ + return 1; + } + + return -1; } - return -1; + add_all_segs(dst_hdr, src_hdr); + + dst_hdr->frame_len = dst_len + src_len; + dst_hdr->tailroom = src_hdr->tailroom; + + /* Data was not moved in memory */ + return 0; } int odp_packet_split(odp_packet_t *pkt, uint32_t len, odp_packet_t *tail) commit db0c6cd9dc0dd3821aaa1320faa24c12ec578c41 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:05 2016 +0200 validation: packet: concat-split test bug fix Successful concat return values are 0 and >0. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/test/common_plat/validation/api/packet/packet.c b/test/common_plat/validation/api/packet/packet.c index 3ad00ed..25252a6 100644 --- a/test/common_plat/validation/api/packet/packet.c +++ b/test/common_plat/validation/api/packet/packet.c @@ -1232,7 +1232,7 @@ void packet_test_concatsplit(void) CU_ASSERT(pkt_len == odp_packet_len(pkt)); CU_ASSERT(pkt_len == odp_packet_len(pkt2)); - CU_ASSERT(odp_packet_concat(&pkt, pkt2) == 0); + CU_ASSERT(odp_packet_concat(&pkt, pkt2) >= 0); CU_ASSERT(odp_packet_len(pkt) == pkt_len * 2); _packet_compare_offset(pkt, 0, pkt, pkt_len, pkt_len); @@ -1262,7 +1262,7 @@ void packet_test_concatsplit(void) _packet_compare_offset(splits[0], 0, segmented_test_packet, pkt_len / 2, odp_packet_len(splits[0])); - CU_ASSERT(odp_packet_concat(&pkt, splits[0]) == 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[0]) >= 0); _packet_compare_offset(pkt, 0, segmented_test_packet, 0, pkt_len / 2); _packet_compare_offset(pkt, pkt_len / 2, segmented_test_packet, pkt_len / 2, pkt_len / 2); @@ -1279,9 +1279,9 @@ void packet_test_concatsplit(void) CU_ASSERT(odp_packet_len(splits[0]) + odp_packet_len(splits[1]) + odp_packet_len(splits[2]) + odp_packet_len(pkt) == pkt_len); - CU_ASSERT(odp_packet_concat(&pkt, splits[2]) == 0); - CU_ASSERT(odp_packet_concat(&pkt, splits[1]) == 0); - CU_ASSERT(odp_packet_concat(&pkt, splits[0]) == 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[2]) >= 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[1]) >= 0); + CU_ASSERT(odp_packet_concat(&pkt, splits[0]) >= 0); CU_ASSERT(odp_packet_len(pkt) == odp_packet_len(segmented_test_packet)); _packet_compare_data(pkt, segmented_test_packet); commit 894bc720a8a849a6bd9b41cf20d483f9da19793c Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:04 2016 +0200 linux-gen: packet: improve packet print Added segmentation and head-/tailroom information to packet print out. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/odp_packet.c b/platform/linux-generic/odp_packet.c index 10fbded..2d9e3e6 100644 --- a/platform/linux-generic/odp_packet.c +++ b/platform/linux-generic/odp_packet.c @@ -1395,6 +1395,7 @@ int odp_packet_move_data(odp_packet_t pkt, uint32_t dst_offset, void odp_packet_print(odp_packet_t pkt) { + odp_packet_seg_t seg; int max_len = 512; char str[max_len]; int len = 0; @@ -1421,6 +1422,25 @@ void odp_packet_print(odp_packet_t pkt) len += snprintf(&str[len], n - len, " input %" PRIu64 "\n", odp_pktio_to_u64(hdr->input)); + len += snprintf(&str[len], n - len, + " headroom %" PRIu32 "\n", + odp_packet_headroom(pkt)); + len += snprintf(&str[len], n - len, + " tailroom %" PRIu32 "\n", + odp_packet_tailroom(pkt)); + len += snprintf(&str[len], n - len, + " num_segs %i\n", odp_packet_num_segs(pkt)); + + seg = odp_packet_first_seg(pkt); + + while (seg != ODP_PACKET_SEG_INVALID) { + len += snprintf(&str[len], n - len, + " seg_len %" PRIu32 "\n", + odp_packet_seg_data_len(pkt, seg)); + + seg = odp_packet_next_seg(pkt, seg); + } + str[len] = '\0'; ODP_PRINT("\n%s\n", str); commit 0b9981a71186cd065a7429cb8fd116e70b20fc23 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:03 2016 +0200 linux-gen: packet: fix bug in tailroom calculation Tailroom is calculated from the end of the last segment, not from the first. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/platform/linux-generic/odp_packet.c b/platform/linux-generic/odp_packet.c index 0d3fd05..10fbded 100644 --- a/platform/linux-generic/odp_packet.c +++ b/platform/linux-generic/odp_packet.c @@ -74,6 +74,15 @@ static inline void *packet_tail(odp_packet_hdr_t *pkt_hdr) return pkt_hdr->buf_hdr.seg[last].data + seg_len; } +static inline uint32_t seg_tailroom(odp_packet_hdr_t *pkt_hdr, int seg) +{ + uint32_t seg_len = pkt_hdr->buf_hdr.seg[seg].len; + odp_buffer_hdr_t *hdr = pkt_hdr->buf_hdr.seg[seg].hdr; + uint8_t *tail = pkt_hdr->buf_hdr.seg[seg].data + seg_len; + + return hdr->buf_end - tail; +} + static inline void push_head(odp_packet_hdr_t *pkt_hdr, uint32_t len) { pkt_hdr->headroom -= len; @@ -383,11 +392,12 @@ static inline odp_packet_hdr_t *free_segments(odp_packet_hdr_t *pkt_hdr, uint32_t pull_len, int head) { int i; - odp_packet_hdr_t *new_hdr; odp_buffer_t buf[num]; int n = pkt_hdr->buf_hdr.segcount - num; if (head) { + odp_packet_hdr_t *new_hdr; + for (i = 0; i < num; i++) buf[i] = buffer_handle(pkt_hdr->buf_hdr.seg[i].hdr); @@ -413,8 +423,7 @@ static inline odp_packet_hdr_t *free_segments(odp_packet_hdr_t *pkt_hdr, * of the metadata. */ pkt_hdr->buf_hdr.segcount = n; pkt_hdr->frame_len -= free_len; - pkt_hdr->tailroom = pkt_hdr->buf_hdr.buf_end - - (uint8_t *)packet_tail(pkt_hdr); + pkt_hdr->tailroom = seg_tailroom(pkt_hdr, n - 1); pull_tail(pkt_hdr, pull_len); } commit 910f5cda1a83fcfcc8f14d491b2956dceea4e827 Author: Petri Savolainen <petri.savolainen(a)nokia.com> Date: Thu Dec 22 16:33:02 2016 +0200 api: packet: src and dst packet must not be the same Concat and copy_from_pkt operations must be called with src and dst packet handles which refer to the same packet. Signed-off-by: Petri Savolainen <petri.savolainen(a)nokia.com> Reviewed-by: Balasubramanian Manoharan <bala.manoharan(a)linaro.org> Reviewed-and-tested-by: Bill Fischofer <bill.fischofer(a)linaro.org> Signed-off-by: Maxim Uvarov <maxim.uvarov(a)linaro.org> diff --git a/include/odp/api/spec/packet.h b/include/odp/api/spec/packet.h index faf62e2..4a86eba 100644 --- a/include/odp/api/spec/packet.h +++ b/include/odp/api/spec/packet.h @@ -781,7 +781,8 @@ uint32_t odp_packet_seg_data_len(odp_packet_t pkt, odp_packet_seg_t seg); * Concatenate all packet data from 'src' packet into tail of 'dst' packet. * Operation preserves 'dst' packet metadata in the resulting packet, * while 'src' packet handle, metadata and old segment handles for both packets - * become invalid. + * become invalid. Source and destination packet handles must not refer to + * the same packet. * * A successful operation overwrites 'dst' packet handle with a new handle, * which application must use as the reference to the resulting packet @@ -928,6 +929,9 @@ int odp_packet_copy_from_mem(odp_packet_t pkt, uint32_t offset, * Copy 'len' bytes of data from 'src' packet to 'dst' packet. Copy starts from * the specified source and destination packet offsets. Copied areas * (offset ... offset + len) must not exceed their packet data lengths. + * Source and destination packet handles must not refer to the same packet (use + * odp_packet_copy_data() or odp_packet_move_data() for a single packet). + * * Packet is not modified on an error. * * @param dst Destination packet handle ----------------------------------------------------------------------- Summary of changes: include/odp/api/spec/packet.h | 6 +- .../linux-generic/include/odp_buffer_internal.h | 2 +- .../linux-generic/include/odp_config_internal.h | 2 +- platform/linux-generic/odp_packet.c | 542 +++++++++++++++++---- test/common_plat/validation/api/packet/packet.c | 500 ++++++++++++++++++- test/common_plat/validation/api/packet/packet.h | 5 + 6 files changed, 967 insertions(+), 90 deletions(-) hooks/post-receive --

9 years, 8 months

Jump to page:

2026

2025

2024

2023

2022

2021

2020

2019

2018

2017

2016

git-lng-odp