dma: Scatter gather test

Adds a scatter gather test against memory to memory transfers. Initially
excludes all platforms as they are all failing.

Signed-off-by: Tom Burdick <thomas.burdick@intel.com>
This commit is contained in:
Tom Burdick 2022-03-21 12:34:36 -05:00 committed by Carles Cufí
commit 2ec3222426
6 changed files with 204 additions and 0 deletions

View file

@ -0,0 +1,8 @@
# SPDX-License-Identifier: Apache-2.0
cmake_minimum_required(VERSION 3.20.0)
find_package(Zephyr REQUIRED HINTS $ENV{ZEPHYR_BASE})
project(scatter_gather)
FILE(GLOB app_sources src/*.c)
target_sources(app PRIVATE ${app_sources})

View file

@ -0,0 +1,14 @@
# Copyright (c) 2022 Intel Corporation
# SPDX-License-Identifier: Apache-2.0
mainmenu "DMA Scatter Gather Test"
source "Kconfig.zephyr"
config DMA_SG_DRV_NAME
string "DMA device name to use for test"
default "DMA_0"
config DMA_SG_CHANNEL_NR
int "DMA channel to use"
default 0

View file

@ -0,0 +1,4 @@
CONFIG_ZTEST=y
CONFIG_DMA=y
CONFIG_LOG=y
CONFIG_DMA_LOG_LEVEL_INF=y

View file

@ -0,0 +1,18 @@
/*
* Copyright (c) 2022 Intel Corporation
*
* SPDX-License-Identifier: Apache-2.0
*/
#include <zephyr.h>
#include <ztest.h>
extern void test_dma_m2m_sg(void);
void test_main(void)
{
ztest_test_suite(dma_m2m_sg_test,
ztest_unit_test(test_dma_m2m_sg));
ztest_run_test_suite(dma_m2m_sg_test);
}

View file

@ -0,0 +1,155 @@
/*
* Copyright (c) 2022 Intel Corporation.
*
* SPDX-License-Identifier: Apache-2.0
*/
/**
* @file
* @brief Verify zephyr dma memory to memory transfer loops with scatter gather
* @details
* - Test Steps
* -# Set dma configuration for scatter gather enable
* -# Set direction memory-to-memory with two block transfers
* -# Start transfer tx -> rx
* - Expected Results
* -# Data is transferred correctly from src buffers to dest buffers without
* software intervention.
*/
#include <kernel.h>
#include <drivers/dma.h>
#include <ztest.h>
#define XFERS 2
#define XFER_SIZE 64
#if CONFIG_NOCACHE_MEMORY
static const char TX_DATA[] = "The quick brown fox jumps over the lazy dog";
static __aligned(32) char tx_data[XFER_SIZE] __used
__attribute__((__section__(".nocache")));
static __aligned(32) char rx_data[XFERS][XFER_SIZE] __used
__attribute__((__section__(".nocache.dma")));
#else
/* this src memory shall be in RAM to support usingas a DMA source pointer.*/
static const char tx_data[] = "The quick brown fox jumps over the lazy dog";
static __aligned(32) char rx_data[XFERS][XFER_SIZE] = { { 0 } };
#endif
#define DMA_DEVICE_NAME CONFIG_DMA_SG_DRV_NAME
K_SEM_DEFINE(xfer_sem, 0, 1);
static struct dma_config dma_cfg = {0};
static struct dma_block_config dma_block_cfgs[XFERS];
static void dma_sg_callback(const struct device *dma_dev, void *user_data,
uint32_t channel, int status)
{
if (status) {
TC_PRINT("callback status %d\n", status);
} else {
TC_PRINT("giving xfer_sem\n");
k_sem_give(&xfer_sem);
}
}
static int test_sg(void)
{
const struct device *dma;
static int chan_id;
TC_PRINT("DMA memory to memory transfer started on %s\n",
DMA_DEVICE_NAME);
TC_PRINT("Preparing DMA Controller\n");
#if CONFIG_NOCACHE_MEMORY
memset(tx_data, 0, sizeof(tx_data));
memcpy(tx_data, TX_DATA, sizeof(TX_DATA));
#endif
memset(rx_data, 0, sizeof(rx_data));
dma = device_get_binding(DMA_DEVICE_NAME);
if (!dma) {
TC_PRINT("Cannot get dma controller\n");
return TC_FAIL;
}
dma_cfg.channel_direction = MEMORY_TO_MEMORY;
dma_cfg.source_data_size = 4U;
dma_cfg.dest_data_size = 4U;
dma_cfg.source_burst_length = 4U;
dma_cfg.dest_burst_length = 4U;
#ifdef CONFIG_DMAMUX_STM32
dma_cfg.user_data = (struct device *)dma;
#else
dma_cfg.user_data = NULL;
#endif /* CONFIG_DMAMUX_STM32 */
dma_cfg.dma_callback = dma_sg_callback;
dma_cfg.block_count = XFERS;
dma_cfg.head_block = dma_block_cfgs;
dma_cfg.complete_callback_en = false; /* per block completion */
#ifdef CONFIG_DMA_MCUX_TEST_SLOT_START
dma_cfg.dma_slot = CONFIG_DMA_MCUX_TEST_SLOT_START;
#endif
chan_id = dma_request_channel(dma, NULL);
if (chan_id < 0) {
TC_PRINT("Platform does not support dma request channel,"
" using Kconfig DMA_SG_CHANNEL_NR\n");
chan_id = CONFIG_DMA_SG_CHANNEL_NR;
}
memset(dma_block_cfgs, 0, sizeof(dma_block_cfgs));
for (int i = 0; i < XFERS; i++) {
dma_block_cfgs[i].source_gather_en = 1U;
dma_block_cfgs[i].block_size = XFER_SIZE;
dma_block_cfgs[i].source_address = (uint32_t)(tx_data);
dma_block_cfgs[i].dest_address = (uint32_t)(rx_data[i]);
TC_PRINT("dma block %d block_size %d, source addr %x, dest addr %x\n",
i, XFER_SIZE, dma_block_cfgs[i].source_address,
dma_block_cfgs[i].dest_address);
if (i < XFERS - 1) {
dma_block_cfgs[i].next_block = &dma_block_cfgs[i+1];
TC_PRINT("set next block pointer to %p\n", dma_block_cfgs[i].next_block);
}
}
TC_PRINT("Configuring the scatter-gather transfer on channel %d\n", chan_id);
if (dma_config(dma, chan_id, &dma_cfg)) {
TC_PRINT("ERROR: transfer config (%d)\n", chan_id);
return TC_FAIL;
}
TC_PRINT("Starting the transfer on channel %d and waiting completion\n", chan_id);
if (dma_start(dma, chan_id)) {
TC_PRINT("ERROR: transfer start (%d)\n", chan_id);
return TC_FAIL;
}
if (k_sem_take(&xfer_sem, K_MSEC(1000)) != 0) {
TC_PRINT("timed out waiting for xfers\n");
return TC_FAIL;
}
TC_PRINT("Verify RX buffer should contain the full TX buffer string.\n");
for (int i = 0; i < XFERS; i++) {
TC_PRINT("rx_data[%d] %s\n", i, rx_data[i]);
if (strncmp(tx_data, rx_data[i], sizeof(rx_data[i])) != 0) {
return TC_FAIL;
}
}
TC_PRINT("Finished: DMA Scatter-Gather\n");
return TC_PASS;
}
/* export test cases */
void test_dma_m2m_sg(void)
{
zassert_true((test_sg() == TC_PASS), NULL);
}

View file

@ -0,0 +1,5 @@
tests:
drivers.dma.scatter_gather:
depends_on: dma
tags: drivers dma
skip: True