Media Transport Library V22.09 release

Signed-off-by: Du, Frank <[email protected]>

Author: frankdjx

CHANGELOG.md

@@ -1,5 +1,25 @@
 # Change log
 
+## Change log for 22.09:
+* License: update to BSD-3
+* dpdk: update DPDK to v22.07
+* ice: update driver to 1.9.11
+* CSC: Color format SIMD convert API from CPU little endian format to RFC4175 YUV422 10bit BE, see st_convert_api.h.
+* AF_XDP: introduce AF_XDP PMD experimental support, see af_xdp.md.
+* pipeline: introduce pipeline friendly API for both st20 tx and rx, see st_pipeline_api.h for detail.
+* tasklet: support unregister now.
+* iova: add map/unmap support for IO device, see st_dma_map and st_dma_unmap.
+* Header split: introduce hdr split offload experimental support, see header_split.md.
+* Ext frame: external frame mode to support user allocated memory, see st20_ext_frame and ST20_TX_FLAG_EXT_FRAME/ST20P_TX_FLAG_EXT_FRAME.
+* st30: support 250us, 333us, and 4ms packet time.
+* Windows: support DSA for rx offload.
+* Windows: add TAP driver support.
+* API: change st20_frame_meta to st20_rx_frame_meta, also add st20_tx_frame_meta for get_next_frame of st20 tx.
+* API: change st_frame_meta to st_frame, also change the callback arg of st20p.notify_frame_done from void* to struct st_frame*.
+* plugin: add kahawai as a plugin to OBS, only rx path now.
+* tasklet: add thread and sleep option for core usage, see ST_FLAG_TASKLET_THREAD and ST_FLAG_TASKLET_SLEEP.
+* fps: add 120(119.88) fps support, see ST_FPS_P119_88.
+
 ## Change log for 22.06:
 * User frame mode with uframe_pg_callback for rx video session.
 * Color format SIMD convert API between RFC4175 YUV422 10bit BE and other LE format, see st_convert_api.h.

README.md

@@ -11,11 +11,15 @@ The Media Transport Library(Kahawai) is a solution based on DPDK prepared for tr
 
 ## 2. Build:
 Please refer to [build guide](doc/build.md) for how to build DPDK, the library and the sample application.
-Please refer to [build guide](doc/build_WIN.md) for how to build DPDK, the library and the sample application in Windows.
+
+For Windows, please refer to [Win build guide](doc/build_WIN.md) for how to build.
 
 ## 3. Run:
 Please refer to [run guide](doc/run.md) for how to setup and run the demo pipeline application.
-Please refer to [run guide](doc/run_WIN.md) for how to setup and run the demo pipeline application in Windows.
+
+For Windows, please refer to [Win run guide](doc/run_WIN.md) for how to setup and run the demo.
+
+For VF and VM support under Linux, please refer to [vf guide](doc/vf.md) for how to setup VF based on SRIOV, [vm guide](doc/vm.md) for how to setup VM based on VF passthrough.
 
 ## 4. Programmers guide:
 For how to develop application quickly based on Kahawai library, pls refer to [sample code](app/sample).
@@ -24,4 +28,4 @@ For how to develop application quickly based on Kahawai library, pls refer to [s
 Run below command before opening a PR.
 ```bash
 ./format-coding.sh
-```
+```

README_WIN_ENV.md

@@ -1,7 +1,7 @@
 # SPDX-License-Identifier: BSD-3-Clause
 # Copyright 2022 Intel Corporation
 
-project('Intel(R) ST DPDK sample app', 'c', default_options: ['c_std=gnu11', 'buildtype=release'], version: '22.06',)
+project('Intel(R) ST DPDK sample app', 'c', default_options: ['c_std=gnu11', 'buildtype=release'], version: '22.09',)
 
 # allow experimental api
 add_global_arguments('-DALLOW_EXPERIMENTAL_API', language : 'c')
@@ -55,7 +55,6 @@ if is_windows
   libjsonccflags = run_command('pkg-config', '--cflags', 'json-c')
   libsdl2cflags = run_command('pkg-config', '--cflags', 'sdl2')
   libsdl2_ttfcflags = run_command('pkg-config', '--cflags', 'sdl2-ttf')
-  libdpdklinkflags = run_command('pkg-config', '--static', '--libs', 'libdpdk')
   libopensslcflags = run_command('pkg-config', '--cflags', 'openssl')
 
   app_c_args += [libdpdkcflags.stdout().strip().split()]
@@ -72,9 +71,6 @@ app_ld_args = []
 
 if is_windows
   app_ld_args += ['-lmingw32','-lws2_32','-lDbghelp','-lSetupapi','-ldinput8', '-ldxguid', '-ldxerr8', '-lwinmm', '-limm32','-lversion','-lwpcap','-lPacket','-lSDL2','-lSDL2Main','-lmman','-ljson-c','-ldl']
-  app_ld_args += ['-Wl,--whole-archive']
-  app_ld_args += [libdpdklinkflags.stdout().strip().split()]
-  app_ld_args += ['-Wl,--no-whole-archive']
 endif
 
 # build executable
@@ -192,7 +188,13 @@ executable('TxSliceVideoSample', video_tx_slice_sample_sources,
   dependencies: [libst_dpdk, libpthread]
 )
 
-executable('RxSt20TxSt22Fwd', rx_st20_tx_st22_fwd_sources,
+executable('RxSt20pTxSt22pFwd', rx_st20p_tx_st22p_fwd_sources,
+  c_args : app_c_args,
+  link_args: app_ld_args,
+  dependencies: [libst_dpdk, libpthread]
+)
+
+executable('RxSt20pTxSt20pFwd', rx_st20p_tx_st20p_fwd_sources,
   c_args : app_c_args,
   link_args: app_ld_args,
   dependencies: [libst_dpdk, libpthread]
@@ -204,6 +206,18 @@ executable('RxSt20TxSt20Fwd', rx_st20_tx_st20_fwd_sources,
   dependencies: [libst_dpdk, libpthread]
 )
 
+executable('TxSt20PipelineSample', pipeline_tx_st20_sample_sources,
+  c_args : app_c_args,
+  link_args: app_ld_args,
+  dependencies: [libst_dpdk, libpthread]
+)
+
+executable('RxSt20PipelineSample', pipeline_rx_st20_sample_sources,
+  c_args : app_c_args,
+  link_args: app_ld_args,
+  dependencies: [libst_dpdk, libpthread]
+)
+
 executable('DmaSample', dma_sample_sources,
   c_args : app_c_args,
   link_args: app_ld_args,

app/meson.build

@@ -16,12 +16,12 @@
 #define DMA_PORT_BDF "0000:e7:01.0"
 #endif
 
-#if 0 /* icelake */
+#if 1 /* icelake */
 #define NIC_PORT_BDF "0000:4b:00.0"
 #define DMA_PORT_BDF "0000:00:01.0"
 #endif
 
-#if 1 /* clx */
+#if 0 /* clx */
 #define NIC_PORT_BDF "0000:af:00.0"
 #define DMA_PORT_BDF "0000:80:04.0"
 #endif
@@ -94,6 +94,20 @@ static int perf_cvt_422_10_pg2_be_to_le(st_handle st, int w, int h, int frames,
   printf("scalar, time: %f secs with %d frames(%dx%d,%fm@%d buffers)\n", duration, frames,
          w, h, planar_size_m, fb_cnt);
 
+  if (cpu_level >= ST_SIMD_LEVEL_AVX2) {
+    start = clock();
+    for (int i = 0; i < frames * 1; i++) {
+      pg_be_in = pg_be + (i % fb_cnt) * (fb_pg2_size / sizeof(*pg_be));
+      pg_le_out = pg_le + (i % fb_cnt) * (fb_pg2_size / sizeof(*pg_le));
+      st20_rfc4175_422be10_to_422le10_simd(pg_be_in, pg_le_out, w, h, ST_SIMD_LEVEL_AVX2);
+    }
+    end = clock();
+    float duration_simd = (float)(end - start) / CLOCKS_PER_SEC;
+    printf("avx2, time: %f secs with %d frames(%dx%d@%d buffers)\n", duration_simd,
+           frames, w, h, fb_cnt);
+    printf("avx2, %fx performance to scalar\n", duration / duration_simd);
+  }
+
   if (cpu_level >= ST_SIMD_LEVEL_AVX512) {
     start = clock();
     for (int i = 0; i < frames * 1; i++) {